3rd Update: 310ED & WinChip at 240 MHz (and beyond?)

[toddyjoe]

Continuing my updates with the Thinkpad 310ED processor upgrade, here is my latest report:

After playing around with the AMD K6-2 380 (AFK mobile version) some more without consistent success above 210 MHz, I decided to try the WinChip C6 240 provided by AlphaKilo again. Previously, I had no luck getting CPU speed above 200 MHz with the few voltage DIP settings I found. Since then, I found more voltage settings above 3.1V through rough translations from a few Japanese sites. Unfortunately, I believe none of the new voltage settings reach the 3.52V that the WinChip uses as standard Vcore. [When I post a final update and report, I will list the various voltage DIP settings I have found.]

After further experimentation, I am now able to get the WinChip C6 240 up to a CPU speed of at least 240 MHz. The issue I needed to overcome was heating. The flat top profile of the WinChip was not making any contact with the aluminum heat sink on the thermal plate of the 310ED. When I recognized the gap, I used a thin shim piece of square copper sheet between the WinChip top and heat sink bottom so that there was direct contact. After that, I was able to get the computer to turn on, post, boot and operate crash-free at a CPU speed of 240 MHz using the 1.5-multiplier setting to get a 4.0 multiplier and the 60 MHz bus setting. The voltage was set to the 3.2V or 3.3V DIP setting described in my last update (on the S1 DIP box, switch 1 is set to On and switches 2, 3 and 4 to Off).

Feeling adventurous, I tried the maximum setting the WinChip would be capable of in the 310ED: 266 MHz using a 66 MHz bus and the 1.5 multiplier to get a 4.0 multiplier. Using various voltage settings above 3.1V, I was able to get the 310ED to turn on, post and start to boot Windows 95. However, the computer consistently locks up when it reaches the cloud-image Windows 95 boot screen. Given that I have been operating my make-shift copper heat sink dry without thermal compound, getting this last bit of performance could still be a heat issue. I am going to pick up some thermal compound this week and see if I can overclock at 266 MHz with the WinChip 240. I will post a follow-up post here if it works.

I have not yet abandoned the hope of getting 400 MHz with the AMD K6-2. After deciphering some Japanese sites, there is some suggestion that you need either a full battery or a higher-current AC adapter to get an AMD K6-2 to boot above 200 MHz in the 310 and 315 series laptops. This makes some sense when you consider that the maximum core current of the AMD is 8.0A and the maximum of the stock Pentium 133 MMX is about 3.3A. My original battery is presently dead and incapable of holding one second of charge, which may be why my attempts at getting the AMD above 210 MHz have failed thus far. Taking the cheaper option, I have ordered an inexpensive, upgrade 72W AC adapter capable of providing the same 16V but at 4.5A instead of the 2.2A of the original 310ED AC adapter. Further updates will follow when the adapter arrives. Another experiment on the table will be trying to go beyond the claimed-maximum 64MB of memory modules.

Stay tuned for further updates.

[AlphaKilo470]

Sounds great. I personally wouldn't try pushing the WinChip past 240 though if you can get it to work at 266, that'd be awesome. My personal recomendation though would be 233mhz on a 66mhz bus as opposed to the factory setting of 240mhz at 60mhz. 6 extra mhz on the bus ususally does better than 7 mhz added on the clock.

[toddyjoe]

Unfortunately, the 310ED does not have a 3.5 multiplier setting to get 233 MHz with a 66 MHz bus. The standard multiplier settings on the computer are 1.5, 2.0, 2.5 and 3.0. I am only able to get a 4.0 multiplier for 240 MHz because the WinChip interprets the 1.5 multiplier as 4.0.

During the lunch hour today, I bought some run-of-the-mill Radio Shack heat sink thermal compound to try with the WinChip. I put a thin layer between the processor top and the underside of the copper-sheet wafer, and another thin layer between the top of the copper sheet and the bottom of the heat sink on the Thinkpad's thermal plate. Unfortunately, it still will not fully boot at the 266 MHz setting. It locks up at the same cloud-image Windows 95 boot screen. It look likes I will not be able to overclock the WinChip 240 because it will always be under-volted in a 310ED.

As a summary for the WinChip C6 240, the socketed processor in the 310ED (and likely any 310 or 315 series) can be upgraded with a WinChip C6 240 with no required software or hardware modifications. The only recommended hardware option would be to add a piece of copper sheet metal between the WinChip and the heat sink on the Thinkpad's thermal plate to allow proper heat transfer. The two best WinChip options are 200 MHz (66 MHz bus with 3.0 multiplier) and 240 (60 MHz bus with 1.5/4.0 multiplier), both set for a core voltage of about 3.2V to 3.3V (a voltage setting not found in the hardware manual but noted in a post above). The 310ED will work fine at 200 MHz with no copper plate, but the copper shim appears to be necessary to consistently get 240 MHz. Hopefully my next update with note even better final results with the AMD K6-2.

Thanks again, AlphaKilo, for the WinChip transaction. None of these experiments would have been possible without your help.

[rRobert]

... first of all sorry for my poor english.

I first tried to upgrade my 310ED with a K6-2 450MHz (desktop version 2.2/3.3V ) ... and it worked. Once with 66MHz x1,5(3,5) = 233 MHz , then 50MHz x 2(6) = 300 MHz. I don't remmember if I tried 6 X 66MHz.

Today I got a K6-2 500MHz ( 2.2/3.3V ) and I see that it works on 66MHz x2(6) = 400MHz ( the proc. is seen as 399 or 399,4 by sisoft sandra or cpuZ ). So ... the japanesse are right ... it realy works. Too bad that I can't use it because it's very hot.

About the voltages ... there are 4 switches 2^4 > 16 combinations ... but when I measured voltages for each combination I saw that SW1 have not any effect. So you can obtain 8 diferent voltages :
2 ON 3 ON 4 ON > 2,25 V
2 ON 3 ON 4 OFF > 2,13 >with this voltage I have 400MHz
2 ON 3 OFF 4 ON > 2,63
2 ON 3 OFF 4 OFF > 2,47
2 OFF 3 ON 4 ON > 2,46
2 OFF 3 ON 4 OFF > 2,32
2 OFF 3 OFF 4 ON > 3,06
2 OFF 3 OFF 4 OFF > 2,85

...these voltages are measured by me ... I think they can vary +/- 0,03 V.


So ... I wish that I was useful ... a non-japanesse is telling you that you can have 400MHz on 310ED . I'm courious if with a mobile K6-2 I can use the laptop with his own cooling.

[toddyjoe]

About the voltages ... there are 4 switches 2^4 > 16 combinations ... but when I measured voltages for each combination I saw that SW1 have not any effect.

This is interesting. I had assumed switch 1 of the S1 DIP switch box had some effect on the overall core voltage. When I was experimenting, I had been getting the computer to boot in some cases with switch 1 in one position but not being able to get a full post and boot with switch 1 in a different position. Perhaps I was making an incorrect assumption and my 310ED was simply unstable at the time because I was switching the various DIP switches back and forth. One Japanese site I ran across in my search showed different voltage settings based on how switch 1 of the S1 voltage box was set, so that reinforced my assumption. I will see if I can find that website again. I will also dig out a multimeter and try to do some accurate core voltage measurements. Thank you for the information!

EDIT: Here is one of the Japanese sites I had found that listed different core voltages depending on how switch 1 of the S1 DIP switch box was set: http://www.geocities.co.jp/SiliconValle ... 310cpu.htm Look about half of the way down the page for the list. I know I saw another similar page as well, but I cannot find that one at present. I think that one had information about soldering different resistors to switch 1 to get even lower core voltages under 2.0V to operate AMD K6-2+ and K6-3 processors. Anybody else seen that site?

[toddyjoe]

I have one more addendum to the use of the WinChip C6 240 at 240 MHz. After trying to get extended periods of use out of the 310ED at 240 MHz, the computer started shutting down after about a half an hour of steady use with MIDI plug-ins. The main culprit is likely heat, because the WinChip and even the thermal plate above it were quite hot to the touch afterwards. I believe the use of a copper (or maybe aluminum) wafer and thermal compound may not be optional but may be required. In addition, additional cooling measures likely need to be taken (perhaps a better fan or even more creative measures?). I have dialed back the CPU speed to 200 MHz (66 MHz bus at 3.0 multiplier) for pretty similar speed and performance but better heat control. I will see if my mobile version of the AMD K6-2 380 has similar heat issues when my new AC adapter arrives.

[serverbook]

i was running a open tec warrior laptop with skt7 pentium mmx tilliamook 266 @240 stable only up to 60fsb because of its pathedic isa bus only chipset(makes me wonder how opentec rorted the military $$$$ of such pathedicly slow laptops which supposed to deliver max priority performance and heat retension(ok leave it in the freezer for 2 days and it boots fine,but what about hot temp efficiency +operation?
(and that staggering max speed of 2.4mps from any given hdd.that almost makes my usb 1.1 memory card look like a scuzzi drive in comparison)

regardless i fitted an amd 350 cpu and somehow manipulated the jumpers
to force amds special 2x =6x setting ,and it worked a treat @360/6x60fsb.everything went fine up until the 28th minute of sever stress test benchtest program which it crashed from overheating because open tech mob were too scabby to fit a decent heatsink with a fan attached,and regardless of compounds of clampdown techniques ,could not go beyong 28 minutes so i switched back to the pentium and left it at that.