Where the hell can I find a ~200 ohm pot?

[SuperFarStucker]

I want to vmod my 7800 gtx and radioshack doesn't have any at all, only resistors in that range, which wouldn't do me much good. Apparently they like 7800gt vmods better at radioshack because they have a metric ****load per fortnight of 10k trimmers which I mistakenly picked up

Also, the circuit schematic on the back of the 100k trimmer I have indicates the 'third' leg is for wiper? I don't know that much about circuits, I've only taken a physics course that covers elementary circuit design and they never said anything about wipers! I've seen some pics with a leg clipped off but which leg do I clip (they didn't have any blue pots at the shack, just these funky ones with a black knob on the front).

 

[four4875]

the wiper is the thinger that physically moves inside the pot, the wiper moves between the other 2 legs. when close to one leg, the resistance betwen it and that leg is low, and the other is high. move the wiper the other way and it reverses.

as for the 800 ohm... i dont see why one of a higher resistance wouldnt work, just be touchier on the setting i would think, if it is the same number of turns that is. this is because the pot covers a larger range, from 10 k to 0 ohms, as opposed to 800 to 0.

at my local radioshack, the only multi turn pots they have are 10k, so im kinda screwed there as well.

you might try mouser electronics at www.mouser.com for something more specific, or other large vendors, whos addys are slippinmy mind right about now. also, the more turns the finer you'll be able to adjust the voltage, so somethin you might want to keep in mind.

 

[crimedog]

www.hosfelt.com cheaper than mouser

 

[SuperFarStucker]

I finally got the parts. I've got all the stuff soldered together and calibrated to the correct resistances. Only thing I have to do is solder the wires to the contacts and tack it down with conformal coating. What are the recommended safe limits on the vgpu/vmem? I've read 1.6gpu/2(1?).25mem but I think they were talking about air?

 

[four4875]

i would just make sure to stay below a certain temp and you should be fine for any cooling method.

 

[SuperFarStucker]

I got it all together, just looking for my as5. My conformal coating job is pretty bad and the soldering is even worse but everything appears to be in order when looked at under 5x magnification. Here's to hoping the magic smoke doesn't come out.

 

[SuperFarStucker]

Success. I nearly ruined the whole thing. I checked the schematic one last time before I powered it on and I realized I had put the trimmers on the wrong circuits(the 200 ohm on the 100k circuit and the 100k on the 200 ohm circuit) !!! my memory would have been blown off the the BGA pads. I have a question about the readings though. I get a 2.17 volt measurement for the memory and 1.5 volt for the core. Are these correct default measurements? I get about .08 - .09 v per turn on the core, is that normal?

pics & results to follow.

Also, something happened to my radioshack dual wattage iron. right before I finished the job the plating turned purple and it started to act like it was running at at least 30w (solder was melting instantaneously). Did it die?

 

[SuperFarStucker]

What is usually taken as the clock speed? The geometric domain speeds or the ROP/shader domain speeds? What does the clock delta mean? Is there any way you can independantly adjust the ROP/Shader/Geometric domains or are they in a fixed relation?

 

[Revivalist]

Success. I nearly ruined the whole thing. I checked the schematic one last time before I powered it on and I realized I had put the trimmers on the wrong circuits(the 200 ohm on the 100k circuit and the 100k on the 200 ohm circuit) !!! my memory would have been blown off the the BGA pads.

Wow! . . . Good thing you double checked before you powered it up!

I have a question about the readings though. I get a 2.17 volt measurement for the memory and 1.5 volt for the core. Are these correct default measurements? I get about .08 - .09 v per turn on the core, is that normal?

That sounds perfect!

pics & results to follow.

Looking forward to see it. I'm going to be doing the same thing to a 7800GTX in the next week or so. By the way, did you take a look at this thread for assistance?

Also, something happened to my radioshack dual wattage iron. right before I finished the job the plating turned purple and it started to act like it was running at at least 30w (solder was melting instantaneously). Did it die?

Perhaps you just left it sitting for a while. It gets a lot hotter when it just sits there since there's no solder on it to absorb the heat.

 

[four4875]

my RS irons all turn a blueish color when i first use them, never had a problem from them, even left one plugged in for like 3 days straight before i realised it (it was in the basement)

 

[SuperFarStucker]

I just looked at the pics on malvescorner for diagrams. The only really hard point to solder is the gpu point. There is a whole bunch of small points near it and it wouldn't be hard to bridge them all together by using too much solder. My hands have always been pretty bad and i still got it after a couple tries. I imagine with one of those micro irons it would be effortless.

I hit 9500 bogomarks05 on my winchester clocked to 2.5 GHz / 240 2-4-4-8-2t mem on default driver settings. My old score on the winchester at 505/1360 was about 8400. I didn't have my opty in because I was trouble shooting some compatibility problems that I thought were due to the DC (turns out the new 81.98's (feb 7) + bf2 + vsync = bsod). I imagine I will gain a few hundred points when i put it back in too.

I can't tweak it anymore until next week because I sold my TCCD modules on eBay since you really don't get any extra performance out of them running them in 4x512. Have to wait until the 2x1024 modules get here I suppose. I played a bit of battlefield 2 and it seems the extra core/memory clocks gave it a nice bump. It felt quite a bit smoother as I think my minimum fps jumped at least 10 running all high & 2x AA (default driver settings). Between 55-70 fps running averages depending on scene complexity.

My clock speed was 600 MHz and the memory was at 1420 when I ran the bench (artifact free). I've seen a lot better memory speeds and I'm wondering if they run them at looser timings to get that because their is absolutely no way I could even come close to 1600 on the memory. I got like 10 extra MHz going from 2.25 to 2.35. I'll get some pics up tonight I suppose but it looks pretty damn messy. No way i'll ever be able to get all that conformal coating off. Max Load temps were 46c (1.71 v) and the memory sinks were just shy of burning hot (2.25v). I've got a amd reference cooler fan zip tied to the voltage regulator although I'm not sure it is helping any, silverprop fusion HL for cooler. Overall not too shabby for 35.00 worth of mats (i had to get one of those helping hands and a smaller iron) and next to no experience soldering

 

[Revivalist]

. . . I've got a amd reference cooler fan zip tied to the voltage regulator although I'm not sure it is helping any, silverprop fusion HL for cooler. Overall not too shabby for 35.00 worth of mats (i had to get one of those helping hands and a smaller iron) and next to no experience soldering

Well, it sounds like you're doing pretty well for just starting out . . .

So you zip tied a fan to the voltage regulator heatsink. . . . That's an interesting idea. I was trying to figure out how to get a fan attached there. I really like the way Viper John gets them on there. I would like to do something similar and also let the fan be powered off the card so that fan speed will be automatically controlled by the card. Also, if the fan could actively cool the ramsinks as well, that would be nice.

 

[Beginner]

if its been mentioned already i apologize. are you near a fry's? they sell 1k-200k variable/adjustable pots cheaper than radio shack.

radio shack - blue
fry's - cream colored.

 

[SuperFarStucker]

pictures

Beginner: I ordered them from hosfelt.com as crimedog suggested.

Now for the pics

VMOD1
comments: This gives an overview of the layout at a pretty high resolution. As you can see, I went a 'bit' overboard on conformal coating but I figured it was better to be safe than sorry in this case as my wires do have quite a bit of exposed copper on each end. I plan on cutting some shrink wrap in half and sealing off the points coming off the potentiometers sometime. The reason one is upside down is because when I realized I had them on in the wrong configuration I desoldered them and then put one on 'backwards', by this time I was fairly frustrated and didn't want to deal with it anymore. I also failed to cut my wires to the shortest possible length but the messiness doesn't bother me so I don't think I'll be redoing it.

VMOD2
comments: This is a close up of the vgpu mod area. As you can see I had some exposed wire where the iron briefly touched it. This actually happened when I was soldering the point, not the capacitor. I just put a good coat of conformal coating below the wire in the affected area and let it dry for a bit and then covered the bare wire with conformal coating, so it is sufficiently insulated. The point also has a considerable amount of exposed wire which was due to the insulation shrinking away while I was trying to solder the point. Again I just made liberal use of conformal coating before tacking the wire down. This was the hardest point to solder of all of them because there is a lot of connections which could easily be bridged within a few millimetres

VMOD3
comments: This was the last point I soldered and I consider it the best. My criterion for an 'acceptable' joint was that I should be able to lift the card up from the joint without it breaking. I think I used to much solder but I wasn't too keen on messing with it anymore.

AAR (After Action Review): Well, it works, but I think I learned a lot of lessons which would enable me to do a much better job next time around. I really should have practiced on some spare dead disk drives before I started to do the mod at least but I'm a rather impatient person and I've had some experience soldering together much larger objects (although I've never been good at that either). I don't think I would run the trimmers to the edge of the board if I did it again. It would be much easier to just tack them down in place and I think I could make it look cleaner. Of course, having a screwdriver near exposed conductors could be bad news if you slipped off the trimmer adjustment knob. In any event I'm extremely satisfied with the result and can't wait until I get my 'new' memory modules to start tweaking it some more.

dark days have fallen over the memory socket.

edit: Revivalist: you can see how i've done the fan, I've tried both ways and I believe it works best when it is pulling air off the vreg rather than pushing air onto it. It is actually pushing air onto it in this configuration and I plan on flipping it back over. I briefly considered pushing air across the board like the reference cooler does but I doubt it would yield any better results on the memory and my case has quite a bit of airflow in it. I don't have it on their very tight because I have a feeling it would place a lot of stress on the pcb which would be very not good. The fan is plugged into one of my motherboards headers and I believe I could adjust it if needed but I just leave it at full speed since the 120x38mm 115 cfm panaflos drown out all the other noise anyway.

 

[Revivalist]

Wow, thanks for the detailed review of the proccess! . . .

Do you think you could just post the full size pictures? I don't know why, but clicking on thumbnail pictures in this forum causes the picture to come up still the size of a thumbnail. . . .

As for running the trimmers to the edge of the board, I think that was a good idea. In fact, I'm planning to solder a molex to the voltage measuring points as well. Last time I did a volt-mod, I ended up killing the card later when I tried to measure the voltage and the probe accidentally shorted two points on the card just as you mentioned. . . . So be careful when you're measuring the voltages later! . . .

As for the fan on the voltage regular heatsink, did you consider using the holes that the heatsink is mounted on the PCB with in order to also mount the fan? . . . I was checking out my GTX today and I realized that this would work perfectly. The holes are separated by the exact distance for a 40mm fan to be mounted. I'm just going to remove the plastic pins that the heatsink is using, run some longer screws through there instead to mount the fan and the heastink to the card. Plastic washers should help to elevate the fan and to prevent the screws or nuts from shorting anything. . . .I'm looking forward to settingthat up.

 

[SuperFarStucker]

I guess it isn't very clear from my layout but the underlined text are links to the larger images. You can't hotlink images on this forum, the moderators have it disabled for whatever eason or they need to update the system, or perhaps I just don't know the proper syntax. In any event, the text that is directly below the thumbs links to the image. Counter intuitive but I can't think of a better solution short of directly linking which is hideous.

 

[Revivalist]

I guess it isn't very clear from my layout but the underlined text are links to the larger images. You can't hotlink images on this forum, the moderators have it disabled for whatever eason or they need to update the system, or perhaps I just don't know the proper syntax. In any event, the text that is directly below the thumbs links to the image. Counter intuitive but I can't think of a better solution short of directly linking which is hideous.

Oh, I see. That's actually a good idea to put links below the thumbnails. I think I'll adpot that myself, if you don't mind. We could just put a little note to the reader that full images can be seen at the link below the thumbnail . . .

Looking at the larger pics I see now that you did indeed solder a molex for measuring the voltage points. I'm glad you did in order to avoid the foolish mistake that I made. . .

By the way, I was just wondering why you used such long wires since shorter wires would be cleaner, need less conformal coating and be less likely to snag. . . . Or is the extra slack due to swapping the resistors later on?

All in all, you did a great job! . . . I'm looking forward to seeing what kind of clock speeds you get out of that thing. . . . What kind of cooling will you be using on it?

 

[SuperFarStucker]

I've finished playing around with it and I think I've reached as much as I'm going to get at reasonable voltage levels. 600/1470 1.695vgpu/2.35vdd which works out to 9.6k bogomark5 or so @ default driver settings. I did some runs at 630/1500 (the next shader level) and got 10.3k but there was artifacting and it took over 1.75 volts to make the core stable at that speed and 2.4 volts to make the memory stable. From my understanding a slight gain is possible by tweaking the memory settings, which I will probably do when I get around to flashing the card to these speeds.

As for games, you wouldn't think a 100 MHz on the core and memory would make much of a difference but my framerate in battlefield 2 has jumped at least 20 frames for the running average and my minimum has jumped about as much as well. Running average ranges between 80-100 fps on high settings, 2x aa, 1680x1050, absolute minimum I saw was 54 when an artillery shell landed on top of me. Games like psychonauts seem to run a lot smoother as well with high levels of aa. I'm using watercooling and my load temps are about 42c.

I imagine if I pelted it 650+ may be possible at same voltage but there is just too much **** you need to buy to do a pelted setup.

 

[Revivalist]

Congradulations! Very nice speeds and scores you're getting there.

And thank you for reporting your findings. That gives others a reference to go by, like myself when I get around to v-modding my 7800GTXs. . . . The only difference is that I'm planning to pelt them right away. . . .

Could you also test your FPS in BF2 with 1680x1050 resolution and all of the settings turned up in the video card propertiers (filters at 16x, trilinear mapping, etc.)? . . .

The reason I ask is because I found that with this 7800GT running 500MHz/1150MHz I'm also able to run 1680x1050 with the highest setting in BF2 and get around 50-70FPS. . . . But when I turn up all the filters and setting in the video card properties (like 16x filters, trilinear mapping, etc.) then the FPS drop to 30-50. . . .

I'm wondering if we can start to see the extra power of your v-modded GTX over my 7800GT when the settings are maxed all the way in the display properties. . . .

 

[SuperFarStucker]

Sorry I didn't get back sooner. I tried everything on highest settings and apparently I was lowballing the fps because the average over an entire round was 61 fps. This is on 'high quality' with 16xaf, 4xaa forced trilinear mip maps and all settings in game on 'high' @ 1680x1050. Minimum is ~40, and the range is quite large. I don't think my plot really makes things very clear but I'll include it just so you can see for yourself.

Fraps only records the framerate as integers when you dump it to a csv, which is the reason you can see 'discrete' levels of framerate. It was over 50 minutes of gameplay and the exact numbers fraps compiled are: 0 fps min, 60.69 average, 120fps max.