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

Home / I-Opener Areas / I-Opener Technical Stuff Turbo3 Pictures Vcore Mod Pictures from Turbo3

Turbo3 Pictures (modified 0 times) Turbo3 Profile I have set up a yahoo photos page to post pictures of my Vcore modifications. If you would like specific pics post the requests here. http://photos.yahoo.com/bc/turbo3pictures You can see in the Mod Vcore picture that R342 has been changed to a 47K resistor (of the wrong size, 805 instead of 603, since that is all I had on hand) that allows the vcore to go down to 1.71 volts. The two yellow wires go off to the pot. 08-25-2000 21:20:54

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Request: Could I see what's on or near R342 or R343? I'm currently trying to take a V1 all the way up to K6-III and have found all the parts except for these two. On my unit, the R60/61 and R184/185 seems to be identical (enabling split-plane voltage), and the feedback divider around U16 (R202 and R203) seems right. I can't find R342 or R343 to save my life, though. My hunch is they're numbered differently on a V1/V2. The multiplier stuff (R130, R327, R328) appears to be in place, as does the current-limiting R302 1W resistor. That leaves R342 and R343 as the obstacles to a successful V1 hack... 08-26-2000 00:31:13

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Scratch that request: http://www.dominateweb.com/iopener/MVC-427Xb.JPG Serves me right for not reading the night's threads all the way through before posting. R342 and R343 are *definitely* not present on my V1, and are probably not present on a V2 either. CB68 (the SMT cap below the red wire) is present, however, and I'm pretty sure the change for R342 and R343 is doable with some cutting/jumpering on my V1. I'll take a closer look over the weekend when I'm awake and my hands are steady... 08-26-2000 00:46:47

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile Tackhead: Most likely pins 3 and 5 of U16 are connected together and come from Q28 which is the 2.5 reference supply. You need to isolate pin3 from 2.5 ref but leave pin 5 connected to 2.5 ref. Then you just need a simple resistor divider to set the lowest Vcore value you want at pin 3. You could use regular 1/4 watt resistors and wire wrap wire to connect it back in. For a K6-III I recommend a 47K tied to 2.5 ref (say form U16 pin 5) and a 100K tied to ground. The other ends of each resistor are connected to U16 pin 3. Use a 500K pot for R202 and you can have any Vcore you want. Remove the CPU then power up and adjust for the Vcore you want. I am having the best luck at a 1.90 Vcore for 5x66 (333MHz) operation. Probably best to just heat and lift the pad for pin 3 of U16. Then you don't need to worry about the routing of 2.5 being changed. Put a piece of tape between the pin and motherboard and you are all set. 08-26-2000 05:19:58

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Turbo3: I looked more closely at your pics and compared them with my V1 - the vias on the parts side of your V3/4/5 look like there are traces on the back that make up the divider you describe. (Yeah, you probably checked with a multimeter, but seeing the V1 right beside the V3/4/5 makes it pretty clear what they're doing.) My V1 has the big wide trace (that goes to one end of R342, the other end of R342 connecting with R343, presumably to GND on your unit) going straight to pin 3 of U16. Pins 3 and 5 are connected as you suspected. I'll have proof tomorrow, but it looks like everything will work. Just as a sanity check, on your unit, is the point where R342 and R343 are joined - is that connected to pin 3 of U16 on your unit (minimum Vcore), with pin 5 of U16 being the 2.5Vref on the input side of R342. 08-26-2000 22:53:10

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile Tackhead: You got it. Yes, junction of R342 and R343 go only to U16 pin 3 and 2.5 ref goes to U16 pin 5. 08-26-2000 23:24:35

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile OK, what am I missing here? Starting with my V1: 1) Move R60/61 to R184/5 2) Isolate Pin 3 U16 3) Pin 5 of U16 -> 47K -> 100K -> GND 4) Joint of 47K and 100K resistor -> Pin 3 U16 5) Remove R202 for Vcore == Vcore-min I get 1.5V on my pin 5 (2.5V-ref) and 1.02V on my pin 3 (and VCore=1.02V on the non-3.3V pad of removed R60/61) Swapping 47K for 23K, I get 1.14V as VCore and 1.41ishV on pin 5. Going nuts, I swap 47K for 10K, and get 1.2V for VCore and 1.3V on pin 5. Possible explanations: 1) Pilot error: Did I somehow confused R342 and R343? 2) Does my V1 have something weird going on with R203? (R203 and R202 measured 5K in-circuit. Removing R202, confirmed as a 10K resistor, resulted in a measurement across R203 changing to 7.5K.) What obvious thing am I missing here? (My hunch is that it's something obvious... I'm just not thinking straight enough to figure it out.) 08-27-2000 14:10:44

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile The basic problem seems to be with the reference 2.5v going to U16 pin 5. Check Q28 and make sure you see 2.5v there then investigate why it is not going to U16 Pin 5. Everything else seemed ok given the ref was so low. You should be doing all this without the CPU installed, right? If you are trying to run a K6-III then the min vcore with a 47K R342 is probably too low. You will need some value of resistor for R202 or better yet a 500K pot at least until you determine what value of R202 you need. You can use a big old pot if you have one. It just sits out next to the iopener. 08-27-2000 15:48:09

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile U16 Pin 5 does not generate the 2.5v reference. It uses it to regulate Q17 which is the bulk 2.5v MOSFET. The source of the 2.5v ref is from Q28 in the power supply area. Perhaps you have disconnected U16 pin 5 from the 2.5 ref supply. 08-27-2000 15:55:40

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Nope, I have continuity between pin5-U16 and the 2.5V pin of Q28. When I power up with both my R342/R343 substitutes removed (and pin 3 floating and no R202), I get 2.5V on pin 5. As soon as I put R342/343 in place (with or without a 470K resistor in R202), I get the 1.4V / 1.1V split. I vaguely recall some discussion of load on the supply pulling things down, but am unsure if it's relevant to this problem. What are your values for R342/343, and where's your pot for R202 set. (Um, and do you have an R203?) I see empty pads for R320/321 near the CPU, and R253/254 in the power supply region. Are these unpopulated on your unit? 08-27-2000 19:01:42

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile Something strange here. First, yes I have the same empty pads on my V3. The values I use for R342 and R343 are 48,000 ohms and 102,000 ohms. If you apply 2.5 volts across 150,000 ohms you are only going to draw 166 microamps. That is nothing. No way can that pull down the 2.5 ref. Please confirm that you do not have the CPU installed when doing these tests. 08-27-2000 19:44:41

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Confirmed: CPU is not installed while doing the tests. (Sorry for not mentioning that in my last post.) Silly question - on my end, it looks like the pads (R60/61, R184/185) for split voltage are common - that is - both zero-ohm resistors go from the same place to the same place. That's counterintuitive (why have two jumpers where one would do, unless there's a serious current limitation on the 0-ohm resistors, which doesn't make much sense either). Could I have shorted something out way back when I was moving R60/61 to R184/5? Next step will be tomorrow evening. First, I'll verify that I can get this V1 back to a working (Winchip-200) state. When I've confirmed that today's messing around was harmless to *this* V1, I'll attempt to replicate my results on my other V1. FWIW, I'm as confused as you are. The only thing I've got going for me is the fact that my two weird voltages always add up to 2.5V. 08-27-2000 20:30:50

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile As for the two resistors it is because of the high current (5-7 Amps). There are two, not because the resistor is the limit but because of the limits of the thin copper foil leading up to the resistors. The thickness is fixed and the width of the pad going to the resistor is also fixed and too small for the current. The only option is to use two resistors and have twice as much copper to handle the current. As for the fact that the voltages always add up to 2.5v, I don't understand why that would be. You can't really add them up since they are both relative to ground. The clue is the different ref voltage levels you get for different resistances. 147K gets you 1.5 and 110K gets you 1.3. What you get for vcore is just the divider effect so it does not help explain what is going on. So a 37K difference changes the voltage 0.2 volts. One thing I am not sure I told you is that on my V3s which have pads for R342 and R343 only R342 was populated. It had a 10K resistor whose function I don't understand without there being an R343. Since the opamp inputs draw very very little current pins 3 and 5 are both at 2.5 volts. One yours I take it the two pins were actually tied together. Could you try leaving pins 3 and 5 tied together to the 2.5v ref and add in the 47k/102K resistors and see what happens? The junction of the 47K and 102K would not be connect to anything. By the way, U16 pin 1 does go to R204 (10ohms) and then to Q16, right? 08-27-2000 21:29:12

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Yes, R204 = 10 ohms, and it sits in series between U16-1 and Q16. I'll try the pins 3/5 tied together thing, just in case. (Though I agree it sounds weird.) I thought the deal for a resistor divider was: Vin ---/\/\/\---/\/\/\--- GND ...........R1....|....R2 ..................Vout (Printed in non-proportional font) Where Vout = Vin * (R2 / (R1+R2)) With R2=100K and R1=47K (50K), that's Vout = 2.5 * (100/150) = 1.66 Somehow you get 2.0V and I get 1.02V. Weird... I'll doublecheck R184/5 to be sure both resistors are in place. Though without a CPU, there's no load on it so it shouldn't matter. I'll also make another check for shorts. 08-28-2000 07:50:31

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile You are correct about the divider when using 47K you will get about 1.66 volts on Pin3 of U16. I then use the 500K pot on R202 to adjust it back up to what ever I want. You can not get a Vcore of 2.0 using a 47K resistor unless you also have R202 installed. To get 2.0 without an R202 you need to replace the 47K with a 25K one. I use the 47K because I need even lower Vcores for the K6-2+ and to run the K6-III 333 at full speed (1.91vcore). 08-28-2000 12:10:00

RE:Turbo3 Pictures (modified 0 times) Zogg Profile Turbo3, I accidentally pulled up the pad of R202 closest to the SODIMM connector, and I'm trying to find another spot to solder to. It looks like the trace goes to R203 right beside it and then to pin 2 of U16. Can you confirm this connection for me? Thanks. 08-28-2000 13:39:58

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Re: R202. I now grok your comment about "to get 2.0V use 25K instead of 47K" - fair enough, 2.0V = 2.5V * (100/(100+25)) You've got 1.66Vcoremin and are boosting to 1.91 with R202. Suppose I tried a very small (5-10K) R202. Could I use this to boost my VCore significantly above my goofy 1.0Vcore-min? If I did so, would I run a risk of frying anything (I have no idea where most of the current flows through this circuit) Your 1.66 to 1.91 isn't much of a stretch, but my gut feel is that trying to turn 1.2 to 2.0 (without any knowledge of how R202/3/4 are supposed to work) could be pushing it. This is more a kludge than a fix, but if all else fails... Before I do that, I'll try it with my R342 (my 47K resistor) = 10K, and R343 (my 100K resistor) being open-circuit, with pin 3 still isolated. That should duplicate your original V3 setup, right? Failing that, I'll try leaving pin 3 sitting on the 2.5V connected to pin 5. Then with VCore=2.5, I'll try plunking resistors from pin 5 and/or Pin 3 to GND. (Obviously a resistor betwen pin 5 and pin 3 won't do anything with a solid trace connecting the two pins together beneath the chip Do you have a link to the datasheet for U16? It'd be nice if I knew what pin 5 and pin 3 were for, so I could try to ohm out the schematic and get a better idea of how all these resistors fit together. (I have the datasheets for Q16 and Q17) I'll also ohm out all the U16 pins. Silly thought - what if pin 3 and pin 5 weren't the only pins tied together. (Or if there were pins *not* tied together on my V1 that *are* tied together on your V3/4/5?) 08-28-2000 16:57:07

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile Zogg: Yes, R202 (the side not connected to ground) goes to R203, U16 Pin 2, and C173 (and on a V4b,V5 it also goes to R347). 08-28-2000 18:24:14

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile E-yoooooooo-reka! V1 with VCore modified for a K6-III! Your first hunch - beefing up the 100K part of the divider - paid off. For those new to the story, V1 IOs don't have R342 or R343. Turbo3 revealed that R342 = resistor from pin 5 to R343, and R343 = resistor from pin 3 to GND (pin 4). These form a voltage divider that should bring the 2.5V reference down low enough that it can be used with a 2.2V (or lower!) K6-III chip. When I tried to duplicate his results on my V1, I got _much_ lower VCore numbers than theory would have predicted for the divider network. Most of this thread has been Turbo3 and I looking at my results and saying "WTF?!". After much WTFing, I managed to get a safe core voltage and successfully boot a K6-III in my V1. On my V1 with R202 removed: R342=100K, R343=100K, VCore=0.8V R342= 10K, R343=100K, VCore=1.2V R342= any, R343=removed, VCore=2.5V So I beefed up R343: R342=100K, R343=750K, VCore=1.97V (woo-hoo!) R342= 25K, R343=750K, VCore=2.15V R342= 0K, R343=750K, VCore=2.20V I had a 1M SMT resistor that fit nicely between pins 3 and 4: R342=100K, R343=1M, VCore=2.08V R342=350K, R343=1M, VCore=1.73V With VCore=1.97V, I was able to boot my K6-III-333 at 200 MHz and verify my VCore as 1.97V and the other voltages in BIOS as within 10% of their normal values. I was running without a heatsink on the CPU, let alone Q16, so I didn't do any serious testing, just enough to make sure it booted. I suspect I'll be lazy and swap my R342 with a pot and leave R202 empty. (Do you see any thermal issues with Q16 with this route, as compared to setting my VCore_min for 1.7V and using R202 to bring it back up?) Finally, Turbo3, thanks for all your patience with this. Debugging remotely is bad enough, it's well nigh impossible without the actual hardware to play with. (And I'm still as confused as you are that a ~3:1 divider was fine when the resistances got big enough. Heck, for all I know, when I do my second IO, it'll "work right the first time" with the smaller resistors Now that I've got a K6-III running in a V1, I'll finalize the K6 hack and start playing with multiplier settings, little heatsinks for Q16, and what-not. 08-28-2000 20:22:12

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile I have posted a copy of the schematic for the 2.5 and Vcore Power Supplies on the photos page. 08-28-2000 20:34:30

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile The values you pick for R342,R343,R202, and R203 only set the voltage that goes into the opamp and the inputs draw very little current. The pin 3 leg is fixed at some value controlled by R342/R343 and the pin 2 leg is proportional to the Vcore value based on R203/R202. The opamp will drive its output higher and higher until these two inputs are equal or it reaches it limit. The stress on Q16 is based only on the Vcore value and the amount of current the CPU needs at that Vcore setting. The resistors you are changing change the Vcore value but they do not directly put any load on Q16. The resistors just make up a divider network and the ratio is what ends up setting the Vcore value. If you have no R202 then pin 2 sees Vcore and the opamps will drive Q16 until it equals the value on pin 3. So yes, you can use a pot in the R342/R343 to control Vcore. It seems like your 2.5 ref is very limited in the current it can supply since you have to use a 750K-1M resistor. Any chance of getting a picture of the components around the 2.5v ref supply? Where abouts in the country are you? 08-28-2000 21:03:47

RE:Turbo3 Pictures (modified 0 times) Linuxguru Profile | Email I had similar problems with low-value resistors on a V3 and deduced that the voltage reference was getting loaded excessively (this can happen, for instance, if the Vref is obtained from a Zener which is biased with a very low current, i.e high resistance from Vcc). I did not have time to investigate it further, but I solved the problem by using larger values for R342 and R343. The data posted by Tackhead seems to confirm that this is an issue for V1/V2 as well. If the biasing resistor is located, it can be swapped out with a lower value to fix this problem. 08-29-2000 00:48:10

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email I ran into exactly the same thing as tackhead with my V1 iopener. Following Turbo3's instructions just pulled the 2.5V line way to low! I used what I had available in my 1/4w resistors and came up with the following: R342=100K,  R343=500K,  R202=Empty BIOS CPU Feature Now Reports with my 2.2v KIII/333 processor: VCore : 1.75 V   2.5V : 2.12 V   3.3V : 3.27 V      5V : 5.05 V                          12V : 11.88 V I tried a few different settings for the MB frequency and multiplier and settled on switch 2: 66MHz, and Switch 1: 4.5X (all 3 on). I'm now running (apparently reliably) at 300MHz! Sandra reports the processor at 301MHz, CPU Temp: 199.6° (with the original heatsink and I shut down when it got that high!) and the core voltage at 2.77 (I guess because the 2.5 reference was so low?). I guess the next step is to put in a fan! Las_Vegas http://iopener.how.to/ 08-29-2000 02:25:30

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile I'll get some pics up in the next day or two. As for others who've reported loading problems with 2.5Vref, I'm probably in that group too - ISTR seeing my 2.5V reported as ~2.3V when I booted to BIOS and checked the voltages. This is a unit where most of the voltages (before my mods) reported slightly lower than normal, so I thought nothing of it last night. Turbo3: Thanks for the reassuring notes and schematic work. FYI, I'm in CA. I'm gonna take a few close-up pics of the P/S area around Q28, as well as ohm some stuff out to see where/if my setup differs from yours. Las_Vegas: 200 degrees? (F or C?) If C, that's gotta be getting close to melting the solder :) Also, with VCore=1.75, why did Sandra report 2.77? (Umm, and that 2.5V in BIOS = 2.12V looks suspiciously familiar, see my comments earlier in this post about how I overlooked my lower 2.5V result.) 08-29-2000 10:28:27

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email 200° F. Sorry. I don't usually pay any attention to Celcius measurements. My actual measurements came to about 1.75 VCore and 2.05 for the 2.5 reference. With the parts I had this was the closest I could get to what I wanted. Doesn't seem like close to 2/3 Megohm should pull Q28 down that much, but it does! I initially considered simply using a pot there and center tap my core voltage there, but I don't have a pot that large! Maybe I'll hit RS today and get a Megohm pot and see what I get... Las_Vegas http://iopener.how.to/ 08-29-2000 11:57:00

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email I've come up with a solution for the V1 situation. The 2.5V Reference at pin 5 of U16 just doesn't have the curent to allow Turbo's S4 substitution, but the 2.5V Reference U27 (MAX1631 at the power supply) does! I connected a 50K to the top of C257 and a 100K to ground side of C248 and tied the ends of the resistors together. This gave me a reference of about 1.7V. Running that to the lifted pin 3 of U16 and replacing R202 with a 25K gave me a vCore voltage of a solid 2.14V. Not quite the 2.2V I was going for, but close enough for the KIII/333 I'm running with it! A picture of the mod can be found here: http://www.lvcm.com/lna/images/vcoreadj.jpg I will be adding this hack to my page soon. Las_Vegas http://iopener.how.to/ 08-29-2000 16:45:32

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Las_Vegas: Great work! Out of curiosity, in BIOS, what do you get for your 2.5V level with your new mod in place? (i.e. closer to 2.5V than the 2.12V you reported last time?) 08-29-2000 17:14:48

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email Well, my iopener is back together and running with the K6 III (I dropped the multiplier down to 4X until I get a fan. Sandra reports the proccessor speed at 267MHz (obviously) and core voltage at 3.26 (I guess it's not actually looking at the vCore!). After about 20 minutes the temperature reached 93°C/199.4°F At which point I shut down. Is this too high, or am I reading the temp of the MOSFET? Las_Vegas http://iopener.how.to/ 08-29-2000 17:18:02

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email Tackhead: The BIOS reports: Vcore : 2.15 V       2.5V : 2.48 V  3.3V : 3.27 V          5V : 5.02 V   12V : 11.88 V Las_Vegas http://iopener.how.to/ 08-29-2000 17:21:48

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile Las_Vegas: Similar numbers for me on my other V1. Damn Q16 gets hot, and the K6-III runs cool. On my V1, the stock (flat) heatsink is a perfect fit for the K6 with the little foampad. Even without ANY heatsink, the K6-III at 200 was only "warm" to the touch after 5-10 minutes. But Q16... ye gods. I almost burned my finger on it. Since I'm using the stock heatsink as a hard drive mount, I'm thinking of using a big screw or nut and drilling (tapping/threading) through the heatsink. I'll then sand off the end of the screw, put a dab of thermal goop on it, and hand-tighten it for decent thermal contact between Q16 and the heatsink. The other option would be to solder a piece of copper to the joint of Q16 to the PCB board, pass the copper over the surface of Q16, and extend it for some length past Q16. I've got two units, so I may try one of each. Cooling Q16 is definitely an interesting problem. RT2 is in the same place on a V1 as on the later models. My Q16 appears to be a 6030L. 08-30-2000 07:52:34

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email I've got a small heatsink attached to Q16 with a piece of thermal tape. The fins were too high with the IO's plate above it, so I bent the fins 90°. I have another unbent if I ever throw a fan in here. (COMDEX is in November. Should be able to find some great fan coolers there!) I like the idea of a screw though... Perhaps a brass screw to improve heat transfer (better than a zinc plated screw anyway.). Las_Vegas http://iopener.how.to/ 08-30-2000 12:44:54

RE:Turbo3 Pictures (modified 0 times) Zogg Profile Turbo3, First of all thanks for all the info you have given us. It has been a tremendous help to me and others I'm sure. I'm still fighting my V3 after doing the K6-III mods, but it seems to be fairly stable now at 200MHz. However, I had to bump the core voltage up to 2.3V to get that. I have tried as high as 2.4V trying to get it to run at 233MHz but no joy. I did the original mods you suggested (25K & 100K resistor divider) to get a reference voltage of 2.06V, and a 500K pot on R202 to adjust with. I did also replace R302 with a 0.012. I tried reloading Win98SE but I got the same lockups, which occur right after the boot process is complete. Usually just moving the mouse will make it lock up, but sometimes I can start CPUcool or something like that and it will then lock up. It never takes long after a boot of Win98SE, but I can go to DOS or Safe Mode and it has never locked up there. So, is there anything else I can try other than a different CPU? I am running with a modified stock heatsink and I'm not unhappy with the 200MHz performance, it just seems odd that it is taking such a high voltage to get there. Thanks! 08-30-2000 14:26:26

RE:Turbo3 Pictures (modified 0 times) pmarcus Profile Zogg, That is the same problem I have been having. Look at my question post: http://www.linux-hacker.net/cgi-bin/UltraBoard/UltraBoard.pl?Action=ShowPost&Board=technical&Post=1627&Idle=0&Sort=0&Order=Descend&Page=0&Session= 08-30-2000 20:27:08

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile zogg: This is a long shot, but are you sure it's not the CPU Cooler that's locking you up? My experiences today at 200 MHz K6-III, with W98SE installed on a WinChip 200, had the thing locking hard from both CPUIdle and Rain. CPUIdle was version 6.0 (and I tried 5.3 with the same results). I forget the Rain version I was using, but it locked hard even after I used RAININST.EXE to specify a K6. (Though it could be an old enough version of Rain that it couldn't correctly identify the K6-III) Just like your experiences, I was fine booting into safe mode. (Attempting to run CPUIdle or Rain from safe mode resulted in a hard lock again.) Just like your experiences, I was able to boot W98SE and even spawn a DOS window in the few seconds before the CPU cooler started up. Just like your experiences, the machine then locked hard. Possibly *unlike* your experiences, I was plenty stable when I removed the software coolers. (Did you do a fresh reinstall of W98SE?) My guess regarding my experience is that the software coolers are smart enough to configure themselves to whatever CPU the original Windoze install, but the two I was using weren't designed with on-the-fly CPU changes in mind. If I can't find a software cooling solution in the next day or so, I'll reinstall 'doze and see if my old CPU coolers still blow up. (With luck, everything will be fine and I'll fail to reproduce your results... but if my reinstall turns out to be unstable, well, at least you'll know you're not going crazy :) 08-30-2000 21:49:00

RE:Turbo3 Pictures (modified 0 times) Turbo3 Profile K6-III Status Update: I now have my daughter's V3 iopener running at 300MHz using 1.91 Vcore and my V4 is running at 333MHz using the 2.0 vcore setting for SW4. I have not had any trouble running in the 300-333MHz range using the lower Vcores. The advantage of the lower Vcores is a cooler running Q16. The voltage drop is more but the current is much less resulting in the lower temps for Q16. On the V4 I tried the 2.2 vcore setting and the temp went up 3c. The V3 is running at 5x66 MHz and the V4 at 3.5x95MHz. Both machines boot and start windows at these settings. I do not use SoftFSB. The big change was the 0.012 ohm R302. If there is someone in the San Jose area with a V1 I would be interested in takeing a look at it to see why the 2.5 ref is being pulled down so much. 08-30-2000 23:53:34

RE:Turbo3 Pictures (modified 0 times) Las_Vegas Profile | Email You're welcome to come visit Las Vegas and look at mine anytime Turbo3! lol Las_Vegas http://iopener.how.to/ 08-31-2000 01:37:00

RE:Turbo3 Pictures (modified 0 times) Linuxguru Profile | Email Zogg, 2.3 or 2.4 V is a little high for Vcore on the K6-III. I think your stability will improve if you lower Vcore to between 1.9 to 2.1 V, and it will run cooler, too. My V3 mods are: R342/R343 are 54.9K/200K, R202 removed, R302 is a .025 in parallel with .015 that was already there (it was easier just to solder another resistor on top, rather than try to desolder R302, which is tightly positioned between L32 and C256 and soldered on very broad copper islands as well). The voltage reference gives 2.49 V (measured with DMM) and Vcore is 1.96 V (which matches the expected value). However, the voltage at pin 3 of U16 was 1.89 V as measured with the DMM. This is almost certainly a measurement artifact, I think the input impedance of my el cheapo DMM is not high enough in the 2 V range and is measurably loading the junction of R342/R343. When the probe is placed at the source of Q16 (28N03L), I get 1.96 V as expected. This mod is very stable at 200 MHz (as expected), and should allow an easy 266 MHz, which I don't really need (I'd prefer a cool-running Q16, although it's rated to 175 c). 08-31-2000 02:16:04

RE:Turbo3 Pictures (modified 0 times) Zogg Profile Tackhead: I'm positive it isn't the cooler program. Many times it will lock up right after I start moving the mouse, without the cooler running. I also tried a fresh install and that, too, didn't work. I did originally think it was Rain doing it to me but quickly found out otherwise. Linuxguru: I've seen Turbo3's luck running at a low Vcore, and now yours. I'm certainly willing to try it, but since it took 2.1V to get 167MHz stable and 2.3V to get 200MHz stable it didn't seem logical to take it lower. Perhaps, as Turbo3 pointed out, the lower current draw would help stability. I'll give it a go when I get back to the office (where everything is) next Tuesday. 08-31-2000 10:29:05

RE:Turbo3 Pictures (modified 0 times) Zogg Profile pmarcus: I just went back and re-read your thread. This is exactly the same problem I'm having. One time I got the cpu stability test up and running with CPUCool at the same time, at 200MHz, and it ran for an hour at 2.25V I think. Stopped the test and it immediately locked up. Reinstalling Win98SE had no effect. I also can't get the second USB port to work. Windows detects a device but doesn't recognize it. I checked the circuit paths yesterday and they are good, so I think maybe I have a capacitor in the wrong range. I wouldn't think that would cause the lockups as most of my testing is with nothing plugged in. It might be spurious interrupts though. I'm beginning to think that maybe some of the K6-III's are marginable. Wouldn't be the first time. 08-31-2000 10:55:17

RE:Turbo3 Pictures (modified 0 times) ASPguy Profile | Email Turbo3 or anyone else: How do you bring up the mhz to 300 mhz or more? 1.) I have a V4 with sw4 switches and was able to insert the amd 333 mhz AFK chip in and it works. but someone told me that it runs in the 200 ish range. 2.) Where do i purchase the resistors and switches needed to make the v2 and v3 able to use the amd 333 mhz AFK chip? I've been reading the posts since march and have tried to find the answers, but.. i haven't been able to figure it out on my own. though someone might help me. Thanks. -AspGuy 08-31-2000 11:49:10

RE:Turbo3 Pictures (modified 0 times) Tackhead Profile ASPguy: on parts availability The resistor for R302 is to allow more current for the CPU. It probably needs to be sourced from Digi-Key or a parts house, as it's a weird part - high-precision, much-less-than-one-ohm resistors aren't commonly available. The DIP switches are for convenience. If you ohm out the DIP switch pads, you'll see there are empty resistor pads, one end going to GND (like one end of the DIP switch) and the other end going to the other end of the DIP switch. So you can either install the DIP switches, or you can install jumpers over the resistor pads that correspond to the individual switches on the DIP switches. Zogg: Yes, Turbo3 has also mentioned that he's seen inconsistencies between K6-IIIs. The rumor is that AMD dumped an overstock to Fry's. Turbo3: San Jose area, huh? I'll be in SJ for the Vintage Computer Festival and CA Extreme towards the end of September (Sept. 30th and Oct 1st IIRC). Might be a good excuse to have a Bay Area IO get-together and parts-swap. Fry's in Fremont and Sunnyvale will also be relatively close by. Suggest we take that discussion over to the "User Groups" board. 09-01-2000 17:13:47

RE:Turbo3 Pictures (modified 0 times) Zogg Profile I decided to update to this thread as the 'SW3 SW4...' thread is pretty long. I had previously modified my V3 for dual plane voltage, and Turbo3's lower core voltage mods, but my IO keeps locking up. Today I replaced R342 with a 47K resistor (already have a 500K pot on R202) so I can totally control the core voltage. I tried running the IO at 1.91V, and it locks up regularly right after the cursor changes from the hourglass to the normal arrow (Win98SE) at both 167MHz and 200MHz. Same with 2.0V. In previous testing, I had to go to the 2.2-2.28V range to get 167MHz and 200MHz stable, and never could get anything higher to not lock up. This is all with no cooling program running. Turbo3 previously mentioned that some of the K6-III's he picked up work better than others. So, I'm off to Fry's again tomorrow to pick up another CPU, assuming that the one I have is at the bad end of the compatibility spectrum. I guess I could have messed up the IO, but everything I test checks out okay so I don't think that's it. I'll let you know tomorrow if it's the K6-III or my work. 09-05-2000 18:35:16

RE:Turbo3 Pictures (modified 0 times) Zogg Profile Okay, I surrender. Picked up a new K6-III at Fry's, got it back, and it was dead as a doornail. Nothing. Wasted another hour returning that chip and getting another one. This one worked, but just as before with the other chip. It takes 2.25V to get 200MHz stable, and I gave up at 2.35V trying to get it stable at 233MHz. There must be something with my IO (or my work) that is causing this, but it was rock solid with the WinChip and the voltages are all okay as best I can tell. So, I give up and will just run it at 200MHz. It is definitely zippier than the WinChip, so I guess all this effort has been worth it. 09-06-2000 15:09:08

RE:Turbo3 Pictures (modified 0 times) mp3boombox Profile | Email Hi, ok I'm still waiting on orderd parts but this is kind of off topic. I'v got 2 io's a v3 and a v1. But I need a new screen for the v1 due to ether my cat or my roomate tipping the io over and nether are talking :/ (ok bad joke) but in any case I'd like to try to do the mod on the v1 also just need a dang screen. There is a guy local in town who is getting an io also but the version is unkown to me. If any one can give me a hand on my prob drop me a line berinc@acadia.net icq 25133744 I'm also working on a VERY simple Ni-Cd battery mod for the io's wich last's for about 41 minets on a stock IO clocked at 210mhz 3x70 with a 12gig 9mm HD and mini fan with full interactive use on the computer. cost's aprox $25 per battery pack. cheep and efective :) The duration should incress with a better cpu like a winchip2 or amd and no fan by about an extra 20 minets. wich is not bad for such an easy mod. the battery packs are half premade already :) I'll post my results when I'm finished 09-30-2000 09:40:30

RE:Turbo3 Pictures (modified 0 times) ASPguy Profile | Email My iopener would get a blank screen or restart on its own. When i restart it, i have to wait several seconds eitherwise the harddrive won't start and i'll just get a blank screen. I've tried installing windows98 instead of win95 but the same problem exist. it even happens in the booting up process before the os startup. I had read turbo3's comments of the jumper settings he set his v3 at: "The V3 is running at 5x66 MHz and the V4 at 3.5x95MHz. " I have a v3, and currently it's set at 3.5x95mhz. Wonder if i should set it to 5x66mhz, but then the io wouldn't be running at 100 or 95 mhz bus. Wondering why my problems are happening, and if i should make the jumper settings. I have a sunon fan 5v powered through the serial pins of 9 and 10. And i have the amd 333 mhz mod. Thanks. 12-09-2000 04:51:34

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