I-Appliance BBS The Official Source for Internet Appliance Upgrades and Mods Click Here! BBS Main List | Sign In | Sign Up | Search | Help | Linux-Hacker.net Reply to Thread | Printer |

Home / I-Opener Areas / I-Opener Technical Stuff K6 III+ 450mhz $59 How about something new

K6 III+ 450mhz $59 (modified 0 times) vladgur Profile http://www.pcliquidator.com/amdk63450.htm I wonder if this one will work in an Iopener Here are some specs The operating voltage spec is 1.9V - 2.1V, with absolute max of 2.2V. The maximum power dissipation is 16W @ 2V. 09-30-2000 19:15:03

RE:K6 III+ 450mhz $59 (modified 0 times) iogenie Profile I blew up that picture and the chip says 2.4V Core/3.3 V I/O. I don't understand what they're talking about on the page. Also, the designation on the chip is AHX or AMX. Can't tell which but I could find no AMX on AMD's site, and AHX is rated to 65 degrees case temp, where an AFK is rated to 80, which is why people on this site are such big fans. I'm going to call pcliquidator tomorrow and see what the part number is... 10-01-2000 09:21:03

RE:K6 III+ 450mhz $59 (modified 0 times) WanneBe Profile Has anyone bought one of these? Even if it's not a mobile, this is about 1/2 the market rate for a normal K6-3 and a good price. I e-mailed them about the discrepancies in the core voltages and got no reply yet. 10-04-2000 14:23:23

RE:K6 III+ 450mhz $59 (modified 0 times) iogenie Profile Talked to PCliquidator..."pay no attention to the Picture", i was told. Here's the specs: K63+450ACZ Case Temp. : 0-85 deg. C Core: 1.9-2.1V I/O: 3.135-3.6V Maximum design power for thermal packaging: 16W Tech Doc: http://www.amd.com/K6/k6docs/pdf/23535.pdf Disclaimer: I am in no way affiliated with AMD or pcliquidator.com. Seems pretty interesting...low power consumption and excellent temperature rating. The question of the day: What do we know about pcliquidator.com? Are they legit? 10-04-2000 16:57:29

RE:K6 III+ 450mhz $59 (modified 0 times) UhClem Profile I found it a little off-putting that they said the CPUs were "AS IS". That means they don't even guarantee it against DOA. Because "we expect you are probably going to fry it". 10-05-2000 10:10:27

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email OK, I just got my 3 K6-III+'s. Put one in my new I-2000 (AKA V5) and set the SW4 off/off (2V), put on my Lazagna cooler and a patched V5 chip and booted into Win98. Ran for 5 minutes, restarted and check the temps in the BIOS- 95 degrees F. No glitches so far. Now all I have to do is wade through all the threads on how to set the multipliers and bus speed and I'll be set. This little sucker might well go all the way up! Specs say it only draws 14 watts at 450MHz. If someone wants to save me some time and post the hack for the V5 to get it to 450MHz, it would be nice. I wasn't real interested until this thread started, so I sort of lost track. Did some preliminary searches and found lots of stuff on the V1-V4's. It is probably burried in there, just was too eager to get this going to really read. 10-10-2000 17:46:54

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Found the settings I need. Duval had a good thread a while back on SW1 and SW2 settings and I confirmed it with a later post. I'll keep the group posted. 10-11-2000 05:28:28

RE:K6 III+ 450mhz $59 (modified 0 times) SomeHuman Profile | Email BadFlash: Did you buy from PcLiquidator ? If so, how was the service? -= SomeHuman =- 10-11-2000 12:20:02

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Service wasn't bad. Automated response, sent me a tracking number, and the package arrived about a week later via 2nd day air. 10-11-2000 19:08:37

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email The agony of defeat. Looks like I order the dip switches to experiment. I went for broke, set to 100 MHZ FSB, X4.5 and it wouldn't do didly. Set to 75 MHZ FSB X 4.5 and it started to post, but tripped out. Set to 66 MHZ X 4.5 and it booted to windows, ran for about 5 minutes and then tripped out. At 66 X 3 I can run quite a while, probably forever. Looks like I need to do some power supply mods. 10-11-2000 19:13:20

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile BadFlash: Just short out R302 with a piece of soldier wick or large wire to get the max current. The stock value will give you only about 6 amps where the max is over 7 amps. I have just ordered four K6-III+ modules to do more testing but it sounds like I won't get them till next week. It does sound promising and dropping the vcore to 1.9 or even 1.8 may also be necessary to get the max speed. I will know more when I drop them into my modified iopeners and try different vcore values. 10-11-2000 20:33:39

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email I guess I better read up on the power supply thread, Turbo3. My V5 trips out immediately with R302 shorted. Must be something I missed. 10-12-2000 16:35:21

RE:K6 III+ 450mhz $59 (modified 0 times) AMDclocker Profile Badflash, I would try to set the multiplier to 2X and FSB @ 66 this may set the chip to 6X internally. It does on the AMD K3, I've only tested this in a desktop PC. When set to 2X sandra reports 399 MHz. I haven't done the SW1 and SW2 mod's yet, if your works I'll do them sooner than later. Good luck 10-12-2000 16:58:11

RE:K6 III+ 450mhz $59 (modified 0 times) AMDclocker Profile Sorry make that a K6-III not K3 10-12-2000 17:03:45

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Well, I'm running with the 4.5X jumper and the 66MHZ FSB now. What I did was rummage through my closet and found a spool of 24 Gauge speaker wire. Got out my wire resitiance table and cut 2.2' of wire to make a .06 ohm resitor. Using my finger I put it in a non-inductive loop wind and soldered it in parallel to R302, (makes a .012 ohm resistor) and fired it up. Now I'll take your suggestion and pull the spec sheet on the K6-III+ and see how it interprets the multipliers. The BIOS reports CPU temp at 113 F and system temp at 150F. Anyone know if this is accurate? 10-12-2000 17:47:04

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Ran all night at 300 MHz. Final temps by the bios indicated 120F for the processor and 158F for the system, what ever that means. I'll be boosting the FSB from 66Mhz to 75Mhz (337.5), then to 100Mhz (450) this weekend. 10-13-2000 05:16:20

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email K6-III+ uses the 2X multiplier as 2X according to the data sheet. Could be wrong, but that is waht it says. They don't list a 6X, but do show a 2X. I'll stick with the 4.5X and boost the FSB up. I'll get more performance out of the memory that way. 10-13-2000 06:15:20

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Someone posted earler that the IO only likes 66 or 100 Mhz, so it might be unstable or unbootable at 75 Mhz. (Waiting on parts to mod a V2). 10-13-2000 09:05:17

RE:K6 III+ 450mhz $59 (modified 0 times) Tackhead Profile Badflash: You got 120F CPU and 150F system after a night at 300 MHz? (== 48C CPU, 65C system) Questions: 1) What was VCore? 2) Did you have a fan on the CPU? 3) Was the case (and RF shield) in place when you did your test? 4) Were you running any software-based CPU cooling solution? 5) Were you running something moderately CPU-intensive (e.g. MP3 playback with visualization?) or was the machine idle? Your temperatures sound close to what I got running a K6-III-200 at 1.9V with the case closed, or a K6-III-300 at 1.9V with the RF shield and back of the case removed, with stock heatsink and no "software CPU cooler" in effect. The single largest jump in temperature that I saw was when I sealed up the unit by putting the RF shield and plastic backshell in place. It upped my temperatures by ~20C or ~30F. 10-13-2000 12:55:14

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Vcore is 2.0, stock switch settings on V5 Have Lasagna cooler on cpu. No RF shield, but had the back on and it was sitting flat on the workbench, just propped up by the power plug and USB mouse. No software cooling. Software testing was not intense. I played solitare and lost, I ran the "mystify your mind" screen saver. Now I have a question. Did you get your FSB to 100 MHz? I just did and the system appears to boot, but I get a dark screen. I suspect the PCI bus is not being clocked down properly for the display. Probably has a /2 from 66Mhz and needs a /3 for 100 Mhz. Am I on the right track, and if so anyone know how to do it? 10-13-2000 16:31:34

RE:K6 III+ 450mhz $59 (modified 0 times) Tackhead Profile Badflash: Wow, if that's a K6-III+-450 underclocked to 300, it's a *heck* of a lot cooler than the K6-III-333AFK! Either that, or (this is quite possible!) my high K6-III-333 (at 300) motherboard temperatures are being seriously boosted by my use of passive cooling of the CPU (the stock heatsink). As for 100 MHz FSB, I think you're on the right track, and I *think* (when I tried 2.5x100 = 250) the PCI bus is being divided down correctly. That said, I was unable to get my IO to run *stably* at 95 or 100 MHz FSB when I did the big pile of benchmarks in the "really technical" section. It may be a case of YMMV. (I was using the stock PC100 32M DIMM, but who knows if the rest of the motherboard on a V1 is happy with it... or if my undervoltaging down to 1.9V was at fault.) I did get some results at 285 (95*3) and 250 (100*2.5), however, so I'm pretty sure it's not the PCI divider. Bus speeds of 75 and 83.3 might be problematic, though. 10-13-2000 17:37:55

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email I found an old post by Duvell the shows I am barking up the wrong tree. My setting would have given me a /3. Looks like I am just out of zots. I dropped the FSB down to 75Mhz and it seems to be running OK. It beat me at solitare again and I am running the screen saver now to see what happens. The system temp is up to 166F after about 30 minutes. We'll see. 337 may be as high as I can get without adding some juice, but with the K6-III+ that is actually pretty quick. I don't have any good software on the drive to really test it. What would you recommend? 10-13-2000 18:17:58

RE:K6 III+ 450mhz $59 (modified 0 times) WOODi Profile How does $47.99 + $6-8 for shipping (ground) sound? http://www.pcliquidator.com/amdk63450private.htm 10-14-2000 00:19:08

RE:K6 III+ 450mhz $59 (modified 0 times) ckbone Profile Badflash.....If you want a free program that is CPU intensive (excessively so), download Windows Media Player 7.0. Play mp3s for several hours....see what happens. For a good temperature measuring program, try the new version Motherboard Monitor.....there's a link in Tips #6 (9-12). This program makes tittle dashboard guages......works good. 10-14-2000 04:43:01

RE:K6 III+ 450mhz $59 (modified 0 times) ckbone Profile Hummmm....There's a name for that type of mis-spelling.....wishful thinking? 10-14-2000 04:59:50

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Looks like 75X4.5 is the limit without more power. 75X5 won't boot- black screen, but drive spins up. Looks like I've run out of 2V current. I'll download & run some better benchmark and torture software to test. What does anyone suggest for the method to directly verify FSB speed and CPU speed? Right now I'm just using soldered jumpers to verify as the post information in the bios won't talk to a K6-III+. The boot up alternately displays 60 & 66 MHz, and only that it has found an MMX processor. The 75 MHz FSB setting runs hotter & faster than 66MHz by the right amount, so I'm sure it is working. 130F CPU and 172F system are the numbers I get after several hours of running. 10-14-2000 07:10:19

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile WCPUID by H. Oda! http://hp.vector.co.jp/authors/VA002374/src/download.html 10-14-2000 08:02:02

RE:K6 III+ 450mhz $59 (modified 0 times) AMDclocker Profile Badflash, have you tried any tests without the hard drive attached? I‘ve had significantly different results when using older and newer hard drives. I’m sure some drives use much less current than others. Just curious. 10-14-2000 08:03:05

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email No, I am using an old clunky 840 meg HD thick as a brick. I don't think it is the HD though. The 5V's hangs in there and the drive spins up fine, but the 2V can't hack it so it doesn't boot up. I'm shooting for a general solution here, not the best possible. I suppose I could set up the sandisk for some special tests, but that will be later. Right now I'm pretty happy where I'm at. 300Mhz, rock stable, 100Mhz FSB. Shield's up, back is on, sitting flat on the bench just propped up by the power jack and USB mouse. The CPU reads 130F and the system temp is 170F. This is after several hours of run time. Should be cooler after I mount it correctly. Not bad for 2 tiny snips of wire on SW2 and one long one on R302. Talk about a cheap hack. Just the cost of the new CPU. With a K6-III+ this is close to 600Mhz on a K6-II without the cashe. I can live with that. It will be a few days before I have real benchmarks. I've located a cheap external power supply. It puts out +3.3V @19.5 amps and +12V at 3A for $9.95. Should be able to rig something. Recap: K6-III+ & Lasagna Cooler. soldered 2 jumpers (snippets of wire) on SW2 for R229 and R230 Made a wire 2.2' long from 24 guage speaker wire, looped up back & forth so it would not be an electro-magnet and soldered it in parallel to R302. That is all it takes to run at 100X3. 10-14-2000 16:39:57

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Turbo3, Tackhead, et al: Have you guys doped out the whole power supply to the I-O? I found a $10 power supply that looks like it could replace most of the I-O power guts. It has +3.3V@8A, +5V @9A, +12V@1A and -12V@.2A. Does anything use the output from Q27 other than Q16 and Q17? I would think Q17 and Q16 should work at 3.3V vs. 3.5V and Q16 would be a lot cooler with the lower voltage. I'd like to bypass the 19V feed altogether and just bring in the power from the external power supply. Does anything use the 19V directly? 10-15-2000 11:15:10

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile No, I have not tried using an external supply for the 3.5v and 5v ones in the iopener. But 3.3 at 8 amps is not much more than what the standard iopener can supply which is 7+ amps. I would try dropping the vcore to 1.9v before going to the trouble of adding an external supply. 10-15-2000 20:53:08

RE:K6 III+ 450mhz $59 (modified 0 times) anty Profile How about making our own 2V supply? I don't know exactly what the circuit for the IO supply looks like, but it sounds like it just doesn't have enough juice. It does not have to be an outside source. Actually in my IO, I just tapped the 19V from the pwr connector and regulated 12V for my HD. The same thing can be done for this. Although 8A+ sounds like a lot. But, I'm sure a mini switching regulator could do it as long as the MOSFET and coil is of the proper rating. What do you guys think? 10-16-2000 09:00:35

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email I don't think the 19V supply can hack much over 30 watts total. We are looking for 14 watts just for the K6-III+. That just leaves just 16W for the rest of the beast. It is possible to build a beefier external supply, which is the route I was going down, but I don't want multiple power boxes and I don't want to spend a bunch of money. If the main Vcore regulator will run on 3.3V rather than 3.5V it would heat up a bit less. Right now for 2V out we are dropping 1.5V at 7Amps, or 10.5 watts. At 3.3 volts it would be just 9.1 watts. Either way, that puts us at around 25 watts just to run the K6-III+ and the regulator.I don't think the HD, screen, and motherboard can live on 5 watts. This makes me think that getting juice from an external feed is the way to go. A separate 2V regulator circuit is likely the way to go rather than smoke the existing Vcore regulator. Lets see what Turbo3 comes up with with dropping the voltage down to 1.9 volts. I don't see how that reduces the current all that much, but that would be great if it did. 1.9V heats up the Vcore regulator more, but if the cpu draws less current it will help. 10-16-2000 09:31:23

RE:K6 III+ 450mhz $59 (modified 0 times) Tackhead Profile Badflash: I haven't tried rigging up external power supplies to the IO either. The main risk I'm seeing at this point is thermal on Q16 - 170F = 77C. Given that the capacitor between Q16 and the thermistor is probably only rated at 80C, you may be seriously shortening the life of that cap, and by extension, the IO. I'm still trying to heatsink Q16 properly. I haven't had much luck in significantly reducing temperatures of the motherboard sensor by sticking heatsinks onto Q16. I may end up biting the proverbial bullet and moving Q16 off the motherboard. (with suitably heavy-gauge wire!) Any suggestions other than the "main" heatsink (my hard drive's mounted there) and the RF shield (which I have to add/remove fairly frequently whenever swapping hard drives). Maybe screwing a 1x1" metal plate between the modem and one of those support posts? 10-16-2000 11:10:57

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email This is why I am curious about the 3.5V feed to Q16 and Q17. If we want 2.0 V out of Q16, it would make a lot more sense to drop the 3.5V supply down to say 2.7 volts. If nothing else is using that I think the mosfets will hack it. This does a couple of things, like cutting the power dissapation of Q16 in half, and giving us back some power we are currently throwing away. Could be we need a piggy-back supply for the lower voltage feed, but if you've done any work on the 3.5V side, I'd like to see it. I am also on the trail of a 19V 3 amp supply. Another option is to build a small low voltage supply, and separately feed Q16 from it. With the extra current from the bigger supply we could get away with it. I don't think we have the power budget with the existing supply. The problem with Q16 is that the motherboard IS the heat sink. Unless you de-solder it and re-sink it elsewhere, that is where the heat will go, not out through the plastic. 10-16-2000 13:27:44

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Ok, got my K6-III+ 450MHz order in. I have dropped vcore to 1.80v and have run Sandra Pro. Here are the numbers: CPU..........1059 FPU...........557 I-MMX........1098 FP/3D Now....1648 The iopener boots at 66MHz and is then switched to 100MHz with SoftFSB. This is a test iopener (V3) setup for max speed testing. CPU cooling is a stock socket 7 12v heatsink/fan driven off a bench 12v supply. The iopener is laying LCD down with an external monitor attached to CN1. It has 32MB RAM, a 6GB Hitachi drive, and is running Win98SE. This unit has NO heatsink on Q16 so I did not run it for long. Just long enough to see that running Sandra did not cause the power supply to auto shut down. I did have an external fan blowing air over it. I will add a small heat sink to Q16 before doing more testing. I just tried a vcore of 1.75v, which is the lower limit of this unit, and at 66MHz it boots to Windows ok. I was able to switch it to 95MHz (477MHz) with SoftFSB but Q16 seemed real hot so I shut it down. But at least the power supply seems capable of running it if I can keep Q16's temp down with better cooling. 10-16-2000 21:12:10

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Heat Sink added to Q16. Here are the 5.0x100 Sandra Pro numbers at 1.75v vcore. Speed.........450......500 ========================== CPU..........1059.....1172 FPU...........557......620 I-MMX........1098.....1224 FP/3D Now....1648.....1837 For the next test I will plug in my USB Ethernet adapter and see if I can get on the net. 10-16-2000 21:42:26

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile This note is being entered from the iopener running at 500MHz. I was able to drop the voltage a little more to 1.73v vcore and I had to switch from 66MHz to 100MHz before I plugged in the USB Hub with the Ethernet and Philips USB speakers. There is still a fan blowing air on the back of the iopener and the temp is 45c (small heatsink on Q16). 10-16-2000 21:58:36

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Were you able to set a 4.5x100 Mhz. multiplier= 450 Mhz.? possible without overheating? Someone in another thread had tried replacing Q16 with another part having a higher heat tolerance. Moved it off the mainboard & attached to the big stock heatsink. 10-16-2000 21:58:53

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile I have had no luck booting the iopener at 100MHz. I always boot at 66 then use SoftFSB to do the Warp jump to 100MHz. Windows Media Player is running as I type this note at 5.0x100MHz. Moving Q16 off the board would help keep from cooking the capacitor which is the real danger at the high temps. Q16 is good to 175c so there is no danger of damaging it no matter what you do. The system seems stable at this speed. Temp is now 46c. I will now turn off the external fan and see how hot it gets... 10-16-2000 22:09:16

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile 5 minites = 56c 10 minites = 56.5c Will now try a disk scan and see what happens. 10-16-2000 22:21:21

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile After 30 minites temp hit around 60c but I lost the network connection and shortly after it powered down. So I must be close to the power limit, but that is to be expected at these speeds. I have turned on the external fan and will let it run overnight and see if it is still working in the morning. I am using an R302 of 0.012 ohms and that can be reduced a little more to see if I can get a few 100 milliamps more of 3.5v. But it seems like 450MHz is no problem. 10-16-2000 23:02:07

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Is your HD off an external supply or from the I-O? What is the current draw off the HD if it is on the I-O supply? Could you boot at 300 and switch to 450 above 1.8V? Might be an easy fix to add a little 5V supply to run the fan & HD & tap off the 19V feed. I sharpend my pencil on the external supply. It is a 35 watt rated supply. I found a 60W 19V power supply, but can't buy them unless I want to get 10K units (hey guys group buy oportunity hehe).If the problem is at the 5V supply off loading those feeds might help. Sure wish I had time to chase down the entire power supply scheme... 10-17-2000 04:37:18

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Seven and a half hours at 500MHz and still going strong (Windows Media Player going all night). A bit windy with the external fan. Temp is only 42c. HDA is a Hitachi DK239A-65 rated at a max 5v of 0.5A mounted I-MOD2 like (where are my other five kits Codeman?? coming soon I hope) and power comes from the iopener. Only the CPU fan is externally powered and at 1.73v core I don't think the CPU needs much cooling. Some things to do next are: 1. Install Sunon iopener powered fan on K6-III+. 2. Put another resistor on top of the 0.012 ohm R302. 3. Add wires to R302 so I can monitor the current. 4. Increase resistance of R342 so I can drop vcore down to at least 1.5v. 5. Move Q16 off the motherboard. I may get a new Q16 just to have the full mounting tab. Anyone checked Radio Shack's MOSFETs for a good match? 10-17-2000 07:02:07

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Not yet on the mosfet from rat shack. How hot is the 19V power adapter getting? I'll try the 1.8V core. Looks like a 180K ohm resistor on top of R343 will do it. I don't plan on going over 450, so that should give me some margin. Checked with tenmax and their fan only draws .08 amps, so that probably wasn't my problem. Likely was the washing machine I was using for a hard drive sucking away the power. I like your 66 on boot up with warping to 100 using SOFTFSB. That allows the fan and drive to spin up without drawing power. How does it do if the drive is spun down (idle) and spins back up? Specs some of the drives say they draw about 5 watts when that happens. You really think that puppy will kick at 1.5V? 10-17-2000 08:07:24

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email http://rocky.digikey.com/USCatalog/V2/261.PDF Pick your poison. Looks like quite a few will work. If you are going to re-locate, probably 2SK2312-ND or 2SK2614-ND would be my guess. 10-17-2000 09:27:49

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Here is a brain storming idea. Suppose you break the connection to Q16 from the 3.5V feed (I haven't looked at the board, so I don't know how hard this is) and insert 5.2' of 24 guage wire (my favorite). This inserts .143 ohms of resistance in series with Q16 and would lower the voltage applied to Q16 by about 1 volt. The wire, not Q16 would then absorb most of the power. With over 5' of wire this could be spread out and use the RF shield as a heat sink. The break-down at 7 amps would put 7 watts in the wire and 3.5 watts in Q16. That would make it run really cool. 10-17-2000 10:24:20

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile As I reported in other threads I have paralleled the original 0.015 R302 with the 0.012 one to give an effective resistance of 0.006666. I fired up the K6-III+ iopener and switched to 500MHz. Measuring the voltage across the pair gives a reading of 42mv at a vcore 1.73v for a total 3.5v current draw of 6.3 amps. This is well within the capacity of the iopener. Next step is to lower vcore some more and see where it dies. Yes, using a large wattage resistor to drop some of the voltage might be a good solution for the hot Q16. Radio Shack has some 5W 0.47 ones ($0.59 each) and if you parallel three of them you get 0.156 ohms at 15W. The resistors would be easier to mount than hot wire. I decided to measure the 5v supply as well. Only 10.6mv across R306 for a total current of 530ma. This will go up a little when I connect the fan to the +5 and the disk is accessed. This gives us a power of 22 watts of 3.5v and 2.7 watts of 5v for a total of about 25 watts. The power adapter can put out 19v at 1.84 amps or 35 watts so we are 10 watts below max when running at 500MHz. One unknown to me is where the LCD backlight power comes from. If it taps off of the 19v supply then it is not included in these power numbers. Ok, since I have an external monitor attached I turned off the backlight and remeasured both voltages. No change, so the backlight comes off the 19v (or at least not the 3.5 and 5 that I am measuring). The 5 volt current did go up when the disk drive did some accesses. I may fire up one of the unmodifed stock iopener V5s and check the current for a reference. I'll do that now... 10-17-2000 20:47:29

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Stock V5 iopener with Rise chip as shipped from Netpliance uses about 3.2 amps of 3.5v and 350ma of 5v. 10-17-2000 20:57:12

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Changed R342 from 47K to 102K, same as R343. Adjusted vcore to 1.66v. Current dropped from 6.3A@1.73v to 5.7A@1.66v. Network is up and posting from it at 500MHz. Media Player running too. This K6-III+ loves to run at low voltage. With the R342 change I could try as low as 1.3v. 10-17-2000 23:40:06

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email I'll see if I can find the power resitors at rat shack. Sounds like you have this knocked. I haven't had time to get back & open up my V5 and look at the traces. Can the feed to Q16 be easily cut & re-routed, or will this require moving Q16? 10-18-2000 05:37:00

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile What can I say. 1.5v vcore and still going.... amazing. If I can get down to 0v vcore the Q16 temp problem is solved Music is playing from the internet and explorer is working just fine. Ran Sandra pro and still ok.. However, your must boot at 66 MHz with the multiplier set at 5.0x then use SoftFSB to go to 100MHz. I also just swapped in a 64MB PC100 CAS2 memory for the Netpliance 32MB SODIMM. Current is down to 5.25 amps from the 3.5v supply. Max temp is 48c but I still have the external fan going. (site seems to have gone down, can't post...back again). Ran all night while waiting for site to come back. 10-18-2000 07:11:49

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile RS number for 0.47 5W resistors is 271-130 (page 275 in catalog). 10-18-2000 07:18:07

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile Hey, any chance either Turbo3 or Badflash would be willing to start a clean thread in the moderated technical reference area/update existing mod pages to document this hack? I wouldn't ask such a hefty task except that you guys have deviated from your own hack instructions, and I'm *just now* getting a replacement bios for my machine. I'm still having black screen probs, and I need a known good bios (e-mail's in your inbox badflash). I just ordered a K6-3 450+ today, expect it in on friday. 500 mHz out of this machine is simply mind-blowing. Good work! :) Just figure everything else on this board is well documented. I figured that this hack deserves a clean write up. There's over 50 replies on this one, digging out the needed info is quite the chore, true? Numbski 10-18-2000 10:30:56

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Once we get this thing worked out I'm sure it will happen. Last piece of the puzzle is getting the cooling resistors in series with Q16 to cool it off. Turbo3 has proven everything else works. For a V4 or V5, looks to me like all that is needed is 2- .47ohm power resitors for Q16, a resistor between 70K and 80K across R343 to drop the core voltage to 1.6V, and the reduction of R302 down to .012 or even less. Use jumpers on the multiplier to get X4.5 and use SOFTFSB to jump from 66MHZ FSB to 100MHZ FSB after you are booted up. Just got to see where to splice in the resistors for Q16. To get to 450 you don't need to de-solder anything and all parts are available at Rat-Shack. 10-18-2000 13:22:00

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Finally reached the limit at 1.45v. At 1.45v I can boot and get to 500MHz but plugging in the USB causes program checks and machine lockups. For now I am going to use 1.6v as a safe vcore value. 10-18-2000 21:10:02

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Installed the Sunon fan powered from the iopener and turned off the external fan. Still face down with back/shield off. At 30 minutes temp is 62c. This is not really that bad BUT as others have pointed out, when you install the shield, back, and mount it back on the stand alot of the hot air stays inside and the temp goes up up up. My plan now is to install dropping resistors between Q16 and the 3.5v bus. But removing Q16 is probably more difficult than getting R302 out and we need Q16 in working shape after it is out. So, I plan on leaving it on the board and just cutting the copper foil between it and the capacitor with an exacto knife. Then just add thick wires between Q16 and the copper foil on the other side of the capacitor going to the resistors. 10-18-2000 22:02:23

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile New pictures (hi-res) of 500MHz iopener setup. http://photos.yahoo.com/bc/turbo3pictures 10-18-2000 23:00:37

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile Is that right...? On V1/V2 (they're pretty much identical hardware-wise) you don't actually have to remove any parts, and the parts you need have radio shack part numbers? Wow guys, I'm almost afraid to say this, but I'm glad I waited. :) I wonder if the next-gen IO's are going to be 400mHz now. LOL Numbski 10-19-2000 06:01:04

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Can you see if using the Actiontec modem has any negative effect on operation at 500 Mhz? Thanks. 10-19-2000 06:16:48

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email With the PCI bus running at 1/3 vs 1/2 and the connection being via what appears to be a serieal port, I kinda doubt it. From Turbo3's picture it loks like he was hooked using the modem. V1/V2 owner need to do the split voltage mod to support the K6-III+. They also need to unhook resistors to allow the 2V or lower operation. 10-19-2000 10:12:41

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile BadFlash: You lost me on that one. If I look at my pictures it is clear I am not using the modem. Both phone plugs are empty. There is a USB cable plugged in that goes to an Ethernet adapter. I am connected through my home network to a DSL line. At 100MHz the PCI bus is running at the standard speed of 33MHz. I don't see why the modem would not work at that speed. In fact the modem is attached to the serial port and does not even see the CPU clock or any other system clock. I will look at running the phone line to the unit and verifying that the modem works. 10-19-2000 11:37:40

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email What I think you understood is not what I meant, I think... The Modem is plugged into something on the motherboard that looks like a serial port. I agree and I beleive I said that the PCI bus should work fine because of the 100/3 setting so the modem should work fine even if it is plugged into the PCI bus as should the serial port. I have not researched how the modem talks to the motherboard. There are two pictures, one front and one back. From the front it is clear that you are on-line. I did not assume the picture of the back was taken with you on line. I mistakenly assumed the picture at the back was taken at some other time so the plugs or not plugs had no effect on my assumption. I stand corrected. Bottom line, I agree the modem should work, but can't prove it. Have you tried the .47ohm resistors yet? I have them, but haven't had time to implement. 10-19-2000 14:34:42

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Actually the front picture is from the attached CRT monitor (the external monitor cable is at the bottom of the picture next to the HDA. This way I can run the iopener face down and still see the screen. Yes the back shot is of the actual running iopener (you can see the fan spinning) just seconds after the screen shot was taken. I went to Fry's today and got a pair of 5W 0.27 ohms resistors for $0.89. But I have not installed them yet. When I do I will switch back to the Rise chip then Fry's K6-III 333MHz before I risk the more expensive K6-III+ 450MHz. 10-19-2000 20:01:29

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile vailr: I am posting this message from the K6-III+ running at 500MHz using the built-in ActionTec modem. No problems, I switched to 100MHz clock speed and then let it auto detect the modem. The only problem is it takes so long to download this thread. Much faster when running over DSL. I have nothing plugged into the USB port at this time. 10-19-2000 21:01:59

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile Any progress? 10-23-2000 11:12:02

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile started further hacking on my V5 first thing to do is lower reference below 2.0V set up divider for U16 vIN R342/R343 R342 = 10K R343 = 10K U16 pin 3 (vIN/vREF) = 1.25v *this should have given me 1.25v it gave me 1.12v (resisters used were scavenged from a defective HD and were likely out of tolerence) turned it on to test if my soldering was suitable, not expecting it to actually turn on. it powers up and works fine at 200Mhz 1.12v Jumped R327/R328 for 4.5x multiplier boots and works just fine at 300Mhz 1.12v tried to softFSB to 100Mhz - machine locks tried to softFSB to 75 - machine locks tried to softFSB to 66.8 - no problem to date I have not done anything to boost current. my v3b will do the jump to 450 with 1.48v, without the current boost. ???????????????????????????? does this sound like a current problem, or a voltage problem, or perhaps I should try setting jumpers for the 100Mhz bus? ???????????????????????????? See info about V3b at http://www.linux-hacker.net/cgi-bin/UltraBoard/UltraBoard.pl?Action=ShowPost&Board=technical&Post=1736 10-23-2000 12:05:17

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Sounds like a current problem to me. Same thing happened to me until I reduced the value of R302. I can't believe this thing runs with voltage that low. I had to work all weekend and then had to ship the codeman brackets (they finally came in!) I am exhausted. I need some sleep before I finish my hack. 10-23-2000 13:22:17

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Check my picture site for my first version of the Q16 cooling fix. I have now gone to placing the dropping resistors totally outside of the iopener case. The first fix had four steps. 1. Mount Q16 vertically so you can place dropping resistor in series with it and to keep the heat out of the board and away from the capacitor. 2. Fan fold a 4 mil copper sheet into a replacement heat sink for Q16 since it is now up in the air. Placed so it gets the air coming off the CPU cooling fan. 3. Place resistor between old Q16 board pad and Q16. Originally a 0.1 ohm 5W mounted internally as shown in the picture. But it got to over 110c and caused the machine to lock up after a while. Switched to two 0.1 ohms in series (0.2 total) currently hanging out the memory door. 4. Sheet of mylar hot glued to fan so Q16 copper cooler (which has 2.5-3.5 volts on it) does not touch the RF shield (ground) when closed. Also keeps one of the stand mounting screws from touching the copper as well. All closed up (memory door still off) and it would only run for about 20 minutes until the mother board heat sensor reach about 47c and then it would lock up. Solve this problem by dropping vcore to 1.5v. The two board sensors stayed at 37.5c and 38.5c for over an hour. So dropping the vcore really helps. As you can see in the pictures my vcore adjust pot can be adjusted with the case closed. Almost forgot, after the lockups I dropped the speed to 450MHz from 500MHz. Will do more testing tonight at the higher speed to see if the lower vcore helps that case as well. You definitely need to move the dropping resistors outside to keep the heat from building up in the closed case and the lower the vcore the better. All tests are done with a USB Ethernet adapter installed and Explorer running along with the Media Player playing music from the internet connection. If the music stops I know I have a problem. 10-23-2000 14:21:54

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Turbo3: Re: V1, V2 IOpeners using the K6-3+ 450 Mhz. @ ~1.6volts vCore. Could you make a schematic and/or short description of exactly what the best-option setup for COOLEST RUNNING of the Q16, as far as pots, resistors. Tackhead's solution (originally designed for K63-333) lifts pin 3 of U16 & adds a 100k resistor & a 50k pot. Your solution uses a 500k pot in place of R202. Would prefer to NOT require lifting the U16 #3 pin. Then there's the question of R342, R343: replace with what values? Anyway, thanks for your efforts! 10-24-2000 22:55:42

RE:K6 III+ 450mhz $59 (modified 0 times) DMasier Profile If you are looking for another source of 450 CPUs AMD K6-III+ 450-$64.95 http://www.softwareandstuff.com/h_comp_amdk63plus.html I've been in the store. They exist. Never bought anything but they are probably reliable. Glen 10-24-2000 23:44:21

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile The best way to cool Q16 is with an external dropping resistor and mounting it vertically. I have a schematic of a V5 vcore supply on my picture page if that helps. The changes for my current V3 setup are: 1. R342 is a 102K resistor which is the same as R343. This gives me a reference voltage on Pin 3 of U16 of about 1.25 and sets the lower limit for vcore. 2. R202 is the 500K pot 3. There is no R347 on a V3 so it does not exist. Do you have an R342 in a V1/V2? Does it go between Pin 3 and 5 of U16? If it does then you probably only need to add an R343 which is just a resistor from Pin 3 to ground (Pin 4). But I recall there being some problem with the 2.5v reference in a V1/V2 being very week and adding the resistor to ground pulled the voltage down. That is why Tackhead got his 2.5v from a different place on the board. If you have an R342 you could remove it instead of lifting Pin 3. Then you need to add a resistor divider (R342/R343) using two 100K resistors with one end going to Tackhead's 2.5v, the junction of the resistors to Pin 3 and the last resistor end going to Pin 4 (ground). Then you can add a pot for R202. Do you understand the function of the lower opamp that controls Q16? What makes this a little confusing is that you can adjust vcore by varying the voltage on either Pin 3 or Pin 2 of U16 or both. U16 just tries to make the two voltages it sees on those two pins equal by varying the output. Be sure to do all your testing WITHOUT the CPU installed. The CPU is put in only after you have verified with a volt meter the vcore is within range for your CPU At little status update: Ran at 500MHz for about 3 hours at 1.6v with temps in the 47c range. Iopener was closed up and on the stand but the memory door was off due to the dropping resistors being outside. This still seems to hot so today I switched back to 450MHz and 1.5v vcore. I sealed up the memory door opening (wires still come out) and have it running Explorer and Media Player. Temps are 44c and 46c. The problem now appears to be getting better air flow so the hot air gets out. I will modify the RF shield so it better matches the plastic cover and probably make a round whole in the RF shield and plastic for the fan air intake. I would like to see less than 40c after 8 hours. 10-24-2000 23:54:57

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email OK, I've got my cooling resistors installed. I mounted mine (two .47 ohm 10 watt resistors from ratshack) in the space between the Lasagna cooler and the hard drive. I used 24 guage wire to run over to the group of little holes by Q17 and soldered in one end. Using a 40 watt iron I heated the tab on Q16 and once the solder melted I bent it to stand up. I soldered the other end to the back of Q16. Things seem to be working just fine, but I've run out of time for tonight. Once I'm done it will be all parts from ratshack, I'm using jumpers, not dip switches. An aside here. It looks like the K6-III+ will accept multiplier changes on the fly, not just at post. Maybe SOFTFSB can be enhanced to support that, and no jumpers would be needed at all? 10-25-2000 17:57:01

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Not sure if mine is a V2 or a V3. Came with a gooped bios & "male voice". R342 & R343 spaces both exist; only R342 is occupied: resistor is marked "103" in very small numerals. My understanding was that putting a 500k pot on R202 was only for adjusting vCore upwards from 2.0 volts. Is this incorrect? 10-25-2000 18:56:52

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email That arrangement can only get down to 2.5V. Add a pot to where R343 goes. This will make a voltage divider that would allow you to adjust the voltage down. If I read your number correctly, R343 is 10K. What is the value of R202 on your board? 10-26-2000 06:00:12

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email If you use a 50K pot for R343 it should allow you to go from 2.3V down to zip. A 25K pot should take you from 2V down to zip. 10-26-2000 06:14:13

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile OK, how does this look: replace R342 & 343 positions with 50k pots (or would 2x 100k pots be better?). Replace R202 with 500k trim pot. Now: is lifting of U16 pin#3 still also required, or because R342 & 343 are present=not required? 10-26-2000 16:50:20

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email You are doing too much. Have you looked at Turbo3's pictures? I could be full of it, but I think your unit is close enough to mine, so... U16 pin 3 is one input to the op amp that drives Q16. Don't mess with it. R342 connects to the 2.5V reference. You want to add your pot to the junction where R343 belongs. 50K pot is fine. By reducing the resistance of your pot you should be able to set the voltage any where you want it. The way this thing works is that the reference voltage at pin 3 of U16 must match the voltage at pin 2 of U16. The output of the op amp will force Q16 to conduct more or less to make that true. If you lower what pin 3 sees, pin 2 will be driven down to follow it. On the V1-V3's the voltage was set close to 2.8V on Vcore. R202 and R203 formed a voltage divider so that Vcore had to be a little higher than 2.5V to match up. As long as R202 is there Vore will always be higher than what you set Vref to. 10-26-2000 17:48:21

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Per Turbo3's mod, had already removed R202. Should this be replaced with zero ohm, 500k trim pot, or left empty? Target is ~1.6 volts vCore, with adjustability from 1.5 to 2.1 volts. Later, will probably also try the Q16 cooling. Lift Q16 from mboard, add 2x 0.10 ohm, 5 watt resistors for better heat dissipation; but need clearer idea of which pins to connect the resistors to? Wish an IO schematic was somewhere to be found... 10-26-2000 20:45:04

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile If you use a zero ohm for R202 you will get a vcore of 3.5v. Not what you want. There is a schematic on my picture page with all you need to know. Here is what I have on my V3 which gets me down to a vcore of 1.4v. R342 & R343= 102K R203= 10K R202= 500K pot No pins need to be lifted. Once you put in R342/R343 power it up without a CPU and measure U16 Pin3. That will be your lowest vcore value and should be around 1.25v. I have now mounted the two 0.1 ohm dropping resistors inside the case. I placed them above the CPU socket so they would get the cool air coming off the CPU fan as it exists the top of the iopener. Heat rises so this gets the heat out of the case in the most direct way with a little boost from the fan. It is away from any other components since the resistors can reach 80-100c. After 50 minutes of stress testing the temp is only 110F and 112F (43/44c) for motherboard/"CPU" sensors. I cut a round hole in the RF shield and case for the fan. It just did not make sense to put a fan in and then block more than 50% of the air from getting to it. It makes less noise. Except for the l.5" round hole in the back it looks just like a stock iopener but runs at 450MHz. I think the temp is equal to or less than stock too. 10-26-2000 23:38:51

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile What about scavenging a fan off of another heatsink, or possibly even a lasagna cooler, and mounting a second fan to the RF sheild? Cut a second hole, have it blowing in. That way we have a full circle of circulation, and get an arrangement that we know we won't have to worry about high temps? Just a thought. Badflash: I keep saying that I have a v1/v2 IO. I need the v1 bios. Am I confusing you somehow? And I sent you a new e-mail...send it c/o the friends I'm staying with. The post office must be looking at the name on the address. Grrr.... Numbski 10-27-2000 08:03:51

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile numbski: After 8 hours of running my latest mods with a totally closed up iopener the temp is only 101.1/105.4/81.9F (38.4/40.8/27.7C) at a vcore of 1.5v. This is so cool there is no value in another fan. Speed is 450MHz, USB Ethernet connect to DSL line, Windows Media Player and Explorer running. At a vcore of 1.6v the temps are 107.5/110.6/82.6F (41.9/43.7/28.1C). I will bump the speed up to 500MHz tonight. (BadFlash, If I change the multiplier value on the fly it does not change the speed on my machine. Are you sure this worked for you?) 10-27-2000 08:29:54

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Maybe cut a 2" fan opening in the plastic case & add on a thin chrome metal finger guard? 10-27-2000 09:17:18

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Actually, a 1.5" hole is all you need. It is far enough from the edges that a finger gaurd is not absolutely needed. Without the Media Player going and only Explorer running the max temp I get now is 101F (450MHz). 10-27-2000 09:50:27

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email numbski: I'll ship you a V1 bios, I tried to as you know. You probably don't want one, but if that is what you want, OK. I keep trying toget people to understand that the V5 patched bios works on ALL the I-O's no matter what version, and does a better job of booting off hard drives without requiring any odd formatting. Yes it supports sound on the V1's and V3's. That is what I'm running on mine. Turbo3: I'm not sure of anything at this point. Changing dip switches on the fly doesn't do a thing for sure, they are only read at POST. What I referred to was something in the data sheets that said the K6-III+ could be clocked down on the fly via the multiplier. I was hoping that softfsb could implement, are you saying they do but it doesn't work? I'm re-locating the cooling resistors away from the hard drive. I didn't like how much heat I was getting there. I'll probably move the resistors and Q16 to a separate perf board and move it over behind the fan. I'm with you though, I think 1.5 volts is the right place and the heat load is down low enough for the lasagna to move it out. 10-27-2000 09:52:33

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile Oops...if that's what works best then send me that. :) I dropped out of reading the board for a long while...I only knew what worked for my box in april. :) Numbski 10-27-2000 10:54:38

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Revision #2: Replace R202 & R 342 with fixed-value 100k resistors. Put 50k pot where R343 (empty position) is now. Will this allow 1.5 to 2.1 volt vCore adjust? (Trying to save another parts order). 10-27-2000 11:10:05

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email R202 is not needed unless you want to have a core voltage above 2.5 volts. You can leave it in or take it out, it doesn't matter. Life is simpler if it is removed as that means Vcore will equal the voltage at the junction of R342 and R343 (your pot). If your R342 is 10K, that is fine, you don't need to change it. If it is 25K, that is fine too. Put your 50K pot where R343 is. With R202 removed you can adjust the voltage: 2.5V*R343/(R343+R342)=Vcore. If R342 is 10K your max voltage is 2.08V, you can adjust to zero. With R342=25K max votage is 10-27-2000 13:36:21

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email R202 is not needed unless you want to have a core voltage above 2.5 volts. You can leave it in or take it out, it doesn't matter. Life is simpler if it is removed as that means Vcore will equal the voltage at the junction of R342 and R343 (your pot). If your R342 is 10K, that is fine, you don't need to change it. If it is 25K, that is fine too. Put your 50K pot where R343 is. With R202 removed you can adjust the voltage: 2.5V*R343/(R343+R342)=Vcore. If R342 is 10K your max voltage is 2.08V, you can adjust to zero. With R342=25K max votage is 1.65 volts. WITH R202 in place you can go from about 2.3volts with R342=10K or 1.83Volts with R342=25K. 10-27-2000 13:39:28

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile New pictures of latest mod: http://photos.yahoo.com/bc/turbo3pictures 10-27-2000 15:03:56

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile OK: Going to leave R202 empty, put a 10k resistor at (currently empty)R342 position & a 50k pot at (from factory empty)R343 position. When the Q16 mod (lift from board & add resistors) is done, where on Turbo3's schematic are 2x in series: 0.10 ohm 5 watt resistors inserted? Between 3.5 volt supply & Q16? Thanks for the help. 10-27-2000 15:07:32

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile The resistors are used to make the connection that was broken by lifting up Q16. So one end of the resistors goes to the large pad on the board and the other to Q16. You should also provide some type of cooling fin for Q16 since it still does dissipate some heat. Now testing at 500MHz (1.64v) with a temp of 114/115F after 1 hour of stress. 10-27-2000 15:28:14

RE:K6 III+ 450mhz $59 (modified 0 times) vailr Profile Would it be possible to utilize the (wasted) current going to the 5 watt resistors to be instead: utilized to increase the voltage going to the CPU fan? My Tennmax Lasagna fan doesn't spin very fast with only 5 volts. 10-27-2000 16:06:09

RE:K6 III+ 450mhz $59 (modified 0 times) 02U2 Profile Got my order for the K6-III-450 from PC Liquidator yesterday. Little slow response but the parts were well wrapped in esd protection and I'm happy! 10-28-2000 05:05:53

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Vailr, check your connection on the lasagna cooler. Sometimes the connectors don't fit tight. No you can't do anything with the wasted current unless you were to build a small switching power supply that would prut out the voltage you need. My latest trick was to take a chunk of copper clad board and cut it to size and re-locate Q16 to it. I mounted it using the heat sink mounts outboard of the lasagna cooler. This gives plenty of heat sink area and uses the air movement from the lasagna cooler. I also have the power resistors there and they push up against the RF shield to absorb heat too. I'm going to try a torture run at 2V 500MHz and see what happens. Keep you posted. 10-28-2000 08:37:44

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile BadFlash: Any chance of getting a picture of your current Q16/resistor setup? I too have just started a 2v 500MHz run for comparison. Tried 105 and 110 MHz (at 1.74v) and both ran Sandra CPU tests without the USB ethernet installed but caused blue screen checks shortly after installing USB ethernet and starting Media player or Explorer. Sandra numbers are: 525MHz (5x105) 1232,651,1283,1926 550MHz (5x110) 1254,683,1346,2020 I believe it is very important to widen the top holes in the RF shield into slots to match the plastic grill. There is an impressive amount of heat pouring out of the top of the iopener at these speeds. You need to run for at least an hour before the heat reaches equlibrium. The RF shield might look like a good heat sink but it is surrounded by a thick layer of plastic. So for a while it will absorb the heat but at some point it will become heat soaked. You want to be able to feel the heat coming out the top and not have it traped inside. So any cooling method must include the CPU Fan to transfer the heat to the air and push it out the top. This said your mounting of Q16 is probably better than mine since mine directs the Q16 cooling air downward and it sounds like yours is towards the top. After 22 minutes of 2v 500MHz temps are 124/114/82F (51.3/45.4/27.7C) 10-28-2000 09:32:37

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile WARNING!!!! My iopener power brick just died while testing at 2v core and 500MHz. It had been running a stress test for a couple of hours with a temp reading of 134/125/84F (56.5/51.8/29C). I had to go out and pick up my daughter and when I got back the iopener was off. No green light on power brick. Replaced with another and the iopener is up and running. I will stick with 1.8v or lower vcore from now on and open up the power brick to see what part died. 10-28-2000 13:31:24

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Added picture of fried power brick components to my picture page. You need a Torx 10TP driver to open the power brick (iopener uses 8TP but I bought both sizes at Fry's... good thing I did). Looking at a good brick R8A-D are all the same part, a 5R6 which is perhaps a 0.6 ohm resistor. Will take one off and measure it. There is a 2A fuse inside but it is still good. 10-28-2000 14:09:52

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Replacing the four 6.0 ohm resistors with two 3.0 ohm 1 watt resistors did not bring the power brick back to life. Instead the 2 amp fuse blow. So it may be one of the MOSFETs that is cooked. 10-28-2000 22:36:34

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email My 2V run didn't work, so I dropped the voltage down to 1.5 and the I-O got the blue screen of death when windows started to load. I jumped the voltage back to 1.6V, and got no response. Went to 2V and got no response. Something else gave up the ghost and died. Power supply is fine, except my 3.5 volt power supply seems to be a 3.3V power supply. I didn't check that until it died. Anyway, the mod is pretty simple, but I don't have a good way to post a picture of it. Pretty simple and it did work unltil I tried to outfox the power supply. I used the heat sink as a template for length and holes, then cut out the proper width to fit between the RF shield and the CPU on the heat sink posts. I moved Q16 to the middle of it and ran wires back to where it was. This seems to be a good cooling solution. No cooling resitors. Be carefull removing Q16. Use solder wick to remove excess solder. Heat pin 1 first and bend it up with a jewelers' screw driver. Do the same with the outboard lead. You can then heat the tab and lift it off the board. 10-29-2000 09:54:36

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile The dropping resistors are nice because they provide an easy way to dump the heat near the top of the iopener. They have a large surface area and can take the heat. I want to keep Q16 in its original position so I do not introduce any garbage into the vcore going to the CPU. There are usually very strict wiring rules that must be following when placing these parts. I have been able to run for many many hours at 450 to 500MHz with a 1.6-1.7 vcore. It is only when I went to 2.0v core that I had problems and it was not with the iopener itself but with the power brick. 10-29-2000 15:32:27

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Me thinks that all K6-III's and all I-Ops are not created equal. We are exploiting engineering margin which is not guarenteed. You have done your work well and have also been luckk, Turbo3 (goes with doing your work well). I think the best bet is to mount the cooling resitors outboard rather than moving Q16. I had no choice as I got carried away with my soldering iron and lost some traces. .25hms or there abouts is the correct resistance to move the most heat. Outboard of the CPU socket is the best place for it too. Sometime during the week I'll try another shot at it. BTW- my wire is just as good of a resistor as anything you can buy. Just cheaper and easier to solder. I've used wire in industrial settings for resitors for years. 10-29-2000 19:36:32

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile BadFlash...I'm willing to test your idea of moving the cooling resistors. I wish you could a take a picture. I have a digital camera, what I'll probably do is take pics of what I'm doing and post them. If I'm understanding right, you would have me desolder part of q16, solder wire leads to it, then take a piece of breadboard, put the cooling resistors on that breadboard and mount it someplace north of the cpu? Or directly above the processor fan so it's blowing directly across them? Numbski 11-03-2000 08:48:02

RE:K6 III+ 450mhz $59 (modified 0 times) Tackhead Profile Turbo3/BadFlash/Numbski: What would you think of mounting the resistors in the "side channel" - that nice empty space of plastic between the RF shield and the outside world where the speakers live? I'm thinking a small hole or two drilled in the top and bottom of the case would allow enough airflow through convection to keep the heat away from the rest of the board, especially with the physical barrier of the RF shield in place. (Going by my IO pics, that would put the resistors roughly where my "audio hack" is mounted. Bad news for me. Good news for anyone who hasn't drilled holes already...) 11-03-2000 12:29:37

RE:K6 III+ 450mhz $59 (modified 0 times) Tackhead Profile Turbo3/BadFlash/Numbski: What would you think of mounting the resistors in the "side channel" - that nice empty space of plastic between the RF shield and the outside world where the speakers live? I'm thinking a small hole or two drilled in the top and bottom of the case would allow enough airflow through convection to keep the heat away from the rest of the board, especially with the physical barrier of the RF shield in place. (Going by my IO pics, that would put the resistors roughly where my "audio hack" is mounted. Bad news for me. Good news for anyone who hasn't drilled holes already...) 11-03-2000 12:29:39

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile I perfer the forced air approach I am using where the exhaust air from the CPU Fan blows the resistor heat out the top. Opening up the RF Shield holes to match the grill gives a large wide area for the air to get out. If you place it on the side you will need large holes top and bottom and even then I don't think you will get the same amount of heat out that using the fan gives you. Have a look at my picture page to see what it looks like. I was able to run it at very high power levels (2v 500MHz) for several hours before I burned up the power brick and still have a low internal temp. I normally run at around 1.64 volts which uses much less power and has temps of 40c and below so I am very pleased with this resistor location. 11-03-2000 13:29:26

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile My question is where on earth did you get copper foil, and how is it attached to q16??? I also have to wonder if double or triple thick aluminum foil from the grocery store could yield the same effect. I imagine it would, but I still have no idea how you have it attached. I'm still wondering about relocating the cooling resistors onto some breadboard. Dunno. I need to go parts shopping for my V1's in the morning. Dips, resistors, and such. A shopping list would be nice. :) Anyway, It's going to be freaky to see this thing running at 500 mHz. I'm happy with that K6-3+ running in my main machine. I'm going to wind up buying 2 more of those chips! Actually, I still need to find a 9mm hard drive for this thing. I really don't want to pay for a new one. :\ Ah well...I bought into this hack, I guess I'll see it through. Numbski 11-03-2000 14:27:03

RE:K6 III+ 450mhz $59 (modified 0 times) numbski Profile Oh, and another question...how do you feel safe screwing with that vCore pot with the case closed and without being able to use a multimeter to know what the voltage is as you turn it? That never did make sense to me. :) Numbski 11-03-2000 14:41:14

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile The copper sheet is from the hardware store (OSH) and is used to keep snails from climbing up fruit tree trunks. It comes as one long strip (6'?) folded up with notches every inch. The copper is soldered to Q16 so it is hot with 3.5 volts. That is why I have the mylar sheet above it so it won't short out on the RF sheild. The trouble with aluminum is how to attach it to Q16 with good thermal contact. Copper is also a better conductor of heat. 11-03-2000 14:46:45

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email You can use the heat sink posts a a piece of perf board to mount the cooling resistors over there tackhead. This puts them in contact with the RF shield and gets a little draft from the fan. I had that working with my V5 before it dies from causes unknown. I'm going to try it again this weekend with a V3. Numbski, you measure the voltage before you close it up & put the CPU back. Once you have it working you don't need to adjust it. 11-03-2000 14:54:27

RE:K6 III+ 450mhz $59 (modified 0 times) WanneBe Profile I've been playing with the K6-III outside the i-opener. For those of you who have pushed the CPU speed to 550 MHZ, setting the multiplier to 2X will give you a 6X. Maximum speed of 600 MHZ. To get to that speed, the core voltage needs to be increased( K6-III+ -- 2.2V someone mentioned that's the safe maximum voltage for the K6-III+ ). Throughput test showed it was running at 600 MHZ. 11-19-2000 16:19:45

RE:K6 III+ 450mhz $59 (modified 0 times) BadFlash Profile | Email Absolute max safe rating of 2.2V is listed at 450Mhz. 600Mhz you better have some damn good cooling. 11-19-2000 20:41:13

RE:K6 III+ 450mhz $59 (modified 0 times) WanneBe Profile I was curious when I read about it and tried it in a desktop. The cooling and power problems in an I-Opener is beyond my capability, time, and budget. I've followed some of the threads where going above 300 MHZ requires quite a bit of creativity and work. At this point, 300 MHZ is fine for me. For cooling, I didn't see anybody open the back and mount a PC case fan just above Q16 ( with a heat sink ) and use a normal size CPU heat sink/fan to see if the cooling was sufficient. To be safe, I would probably drive the fans externally. To prevent burning up the power supply, it would probably be best to use a bench supply. At 300 MHZ the I-opener power supply runs hot. If this worked and the power supply problem was solved, it all could be mounted inside the deeper custom back shell that someone was selling for $20. The end result would be like a hotrod. Big and noisy...not very useful for day-day activities. I repeat that this is beyond me and I'm just throwing it out there as an open suggestion. 11-20-2000 01:44:21

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile k6 III+ 450 @450 with no power system modifications!!!! note that I have not benchmarked or thermally tested this. on a v3b that has not previously ably to make the jump to 450 using softFSB, I have found another answer. First, I set the FSB to 100Mhz, and the multiplier to 3x(default) then I ran the k6clk program during autoexec to go to 4.5x (5.0 would sometimes pass and sometimes not pass the change of the clock. when I went to 5.0 in windows I got an immediate blue screen; this is probably vlotage related, but 450's OK with me) it appears that there is some extra draw when trying to change the FSB, but not when changing the multiplier internal to the CPU. MB Rev I mods: split pane voltage mult to 4.5 configured R342/3 for 1.46Volts tried softfsb (unable to pass 70Mhz) mult to 3.0 using k6clk 11-20-2000 07:05:07

RE:K6 III+ 450mhz $59 (modified 0 times) WanneBe Profile Programmer: I've found that with the AMD devices and overclocking, it's sometimes necessary to jump the core voltage up. I've tried dropping .1-.2 V and that didn't help. Boosting .1-.2 V ( as long as it borders on safe ) seems to do the trick. I'll try the reverse and drop it to 1.5 V and see if it can still run at 600 MHZ on my desktop. What's k6clk and where did you get it? I did a search on altavista and this board and could not find it. Does this mean that with SoftFSB and a V5, a user would not have to do any dipswitch hacks? This would eliminate a lot of the work associated with going to a faster speed. 11-20-2000 09:02:18

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile With SoftFSB, there is apparantly a Current problem as the reconfiguration of all of the system clocks is required; and this draws too much and trips the system. With k6clk from here http://www.geocities.com/liamsay/index.html the only thing that changes is the multiplier internal to the CPU. Increasing the voltage may be useful for going to a higher speed, but at present going to 450 without altering the overload protection seems like a good deal to me. my v3B has a 100Mhz bus jumpered in with a 3x multiplier, I am using k6clk in autoexec to bump it to 4.5x This does mean that you can play with your clock settings without opening the case. presumably you could downgrade the multiplier then upclock the bus frequency, then upclock the multiplier, but IMHO it's much easier to set a base bus speed that you want with a smaller multiplier and then push the multiplier higher when needed. I will probably be doing this same mod to a v5 tonight 11-20-2000 11:33:57

RE:K6 III+ 450mhz $59 (modified 0 times) ASPguy Profile | Email programmer did you ever figure out the problem to the iopener with the black screen i sold to you? as for running an iopener with an amd mod at 450 without the extra cooling, was it my misunderstanding that you've figured out a way to do this? 11-23-2000 14:07:38

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile nope.. never figured it out.. went over it with a fine tooth comb and never found it.. got bored removed all chips, transisters, etc, am sending it to Turbo3 to trace connections to the chipset.. and Yes I have a v3b and a v5 running at 450 without having lowered the resistance on the current limiter, This means ONLY that I am running on currents within the original tolerance, not that Q16 isn't heating up. v3b CPU k6III+ 450@450 vcore 1.46V requires heatsink and fan unknown temps because it works for now.. no problems - boots at 3.0x100Mhz using k6clk in autoexec to go to 450 v5 vcore 1.12V requires heatsink and fan unknown temps because it works for now.. k6clk locks machine boots fine at 4.5x100 can not go to standby mode intermittant locks 11-24-2000 10:07:26

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Programmer: Any estimate as to when you would have time to mail the stripped board to me? Thanks. 11-24-2000 10:38:00

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile it's packaged and ready to go.. I'll call UPS in the morning myself to schedule a pickup.. my wife has been sitting on 3 packages for well over a week now, and I'm finally getting a chance to look around and see what's up... 11-24-2000 20:11:12

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile I have now completed all the modifications on my V3 and have it running rock solid at 500MHz. One key to the puzzle was Programmer finding that K6clk works a lot better than SoftFSB in getting to a 100MHz clock frequency. With the help of K6clk in autoexec.bat the iopener boots right to 500MHz without any user intervention. One big problem I had with SoftFSB was the need to have the USB Ethernet adapter disconnnected before using SoftFSB. Now I not only have the USB Ethernet adapter but also a CD Rom drive, Compact Flash/SmartMedia adapter, and Philips USB Speakers connected through a 4 port hub. The CPU is the K6-III+ 450MHz running at 1.8 volts. Dropping it to 1.7 volts gave me an unstable system under stress testing. A higher voltage puts too much stress on the power supply. I have replaced the original power brick with a 19v 2.5 amp one from Jameco(#169076 $20). This is a must since there are peaks above 2 amps and I have already burned out one of the stock bricks with my testing. Even though the ratings between the two are close (1.84 amps vs 2.5 amps) the Jameco one is twice the size. It uses real heatsinks not bent sheet metal. If you don't mind lockups and strange behavior then stick with the stock brick. But if you are looking for a stable system you can let others use then spend the $20 and get the best. There are two parts to cooling the Vcore supply while running it at 1.8 volts. You need to add a dropping resistor in series with Q16 to help drop the 3.5 v bulk voltage down to 1.8v. Not only does this help keep Q16 running cool but the resistors that are dropping the bulk of the power can be placed close to the top of the iopener case and get cooled by the heatsink fan. At the end of the day you need to get the heat out of the case and into the surrounding air. I have made stress runs for over 10 hours with MBprobe reporting 123/117/84F. The second part of cooling is moving the air through the case and out again. For this I use a Sunon 7mm CPU cooler which is thin enough that it does not touch the RF shield so it produces less noise. I cut a matching round hole in both the RF shield and plastic case. Not much sense in adding a fan if you block over 2/3 of the intake area. It also helps to reduce the noise. Now we need a way to get the hot air out. I cut wide slots in the top edge of the RF shield to match the plastic grill openings. This also helps get the heat from the HDA out as well. I have already posted pictures of the copper heat sink I added to Q16 since it is not connected to the board but still gets hot. I am now very pleased with the results and will start modifying my other iopeners to match these changes. 11-26-2000 23:32:02

RE:K6 III+ 450mhz $59 (modified 0 times) Turbo3 Profile Just to make it clear, with K6clk you can fix the clock at 100MHz with SW2 so your machine boots at 100MHz with SW1 set at the stock value of 3x (for an initial speed of 300MHz). You then place K6clk in autoexec.bat to do the jump to 5.0x. Autoexec.bat gets executed early in the Windows boot process even before you get to the "Safe Boot" option step so most of your Windows boot goes at 500MHz. And don't forget to put "mode co80" at the start of your autoexec.bat file. 11-26-2000 23:43:38

RE:K6 III+ 450mhz $59 (modified 0 times) PhotonWarrior Profile Excellent work Turbo! I have been looking for a while now, but all the sources mentioned in this for this chip have all dried up. I'm probably being dense about this, but from what I have read in this thread, it seems that the processors needed of this are special versions of the K63+450? Is a special rating required ACZ, AHX, AMX, AFK? Or will any one of the following processors work? http://www.tigerdirect.com/applications/Category/category_slc.asp?id=12080102 http://www.pricewatch.com/1/3/1738-1.htm TIA! -Photon 11-27-2000 12:47:20

RE:K6 III+ 450mhz $59 (modified 0 times) Programmer Profile No special rating required, get them from the cheapest source you can find... My 2 were bought on ebay and are in fact grey market. The original serial was burned off and rewritten, and the aluminum plate was not etched in a high quality factory. Mine are very likely chips that were out of tolerance and were disposed of; recovered by scavangers; labelled and sold as real. The chips available from tiger are likely to be genuine; the ones elsewhere really depend on where they got them from. I have dealt with many wholesalers that do not even realise that they are dealing with counterfeit merchandise.. (give me a deal and I'll not look TOO close) A REAL amd chip will have no scarring around the numbers at the top left and right of the chip; and the writing on the aluminum will be clear and very legible particularly the designed for windows in the lower right hand corner. there should be no burrs from the etching and NOTHING less than perfect. But for our purposes, even rejects will work very well.. 1.12 volts at 450Mhz is tough to complain about (even though it can't wake up from sleep (Halt)) Do make sure you're getting a + though.. 11-27-2000 13:31:32

RE:K6 III+ 450mhz $59 (modified 0 times) nesraLmiK Profile | Email To squeze a bit more power out of the 2.0V (or whatever you need for the CPU) put a capacitor 1000 micro Farad or more on the supply. Best result is between ground and the point between the resistors and Q16. This will reduce the power peeks and keep the Core Voltage more stable. 12-20-2000 17:09:12

RE:K6 III+ 450mhz $59 (modified 0 times) Technosat Profile Turbo3 : Any chance you could post some pictures of your mod somewhere? I would love to see the cool(ing) techniques you used to keep you IO to an acceptable temperature. 12-20-2000 18:21:48

Reply to Thread | Printer | All times are PST Powered by UltraBoard v1.62

I-Appliance BBS General - NEWS, Breaking NEWS - linux-hacker.net BBS on CD Questions - User Groups in Your Town - E-Bay, Yahoo Auction LINKS ONLY (MODERATED) - Jokes & Humor - usdtv box I-Opener Areas - I-Opener General Posts - I-Opener Technical Stuff - I-Opener Technical Info Reference (MODERATED) WebSurfer Areas - WebSurfer General Posts - WebSurfer Technical Stuff - WebSurfer Technical Info Reference (MODERATED) Other I-Appliances - Virgin Webplayer - WebPal - MailStation - 3Com Audrey - ThinkNIC / Oracle NIC - Gateway/AOL Connected Touch Pad - Philips/AOL-TV - ACER NT-150 STB - MSN Companion - Compaq IA - Emachines IA - IBM NetVista IA - Intel dot.station - MSNTV2 Network Computers - Fujitsu-Siemens Scovery Linux Net Computers - GallantWeb Server Consoles - Dreamcast - xbox hacking Web Pads - Epods One General Post - Panasonic CF-01 - Fujitsu Stylistic 1200 - Cybiko Thin Clients - Thin Clients DigitalVideoRecorders - DTV Recorders, Replay, TiVo, Etc... - Tivo Hacks Linux - Midori Linux General Board - ITX MotherBoards Single Board Computers - Earth Computers mARMalade Elite Member Area MISC Areas - The Swap Meet WE TAKE NO RESPONSIBILITY FOR ANY TRANSACTION HERE AND NO COMMERCIAL POST PERIOD !!! - Wireless Net - Cameras - Robotics - Routers/Gateways - Mattel JuiceBox