Flawed Configuration: Update

immudium · 03-10-2002, 03:09 PM

I have been actively working on my peltier rig for a few months now and have recently completed some modifications and thought I would share the updated results. For all the sordid details of my past woes, you can check out the thread here: http://forums.procooling.com/vbb/sho...&threadid=2007

Since I had read that Danger Den radiators favored lower flow rates while the water block favored higher flow rates, I decided to add another Danger Den super cube in parallel to the one I already had. My thinking was that I would be effectively cutting the flow rate in half through the radiators while returning to a higher flow rate for the water block after the radiators.

Anyway, my results after doing this have turned out to be alot better than I thought. Where before my post-radiator temps were anywhere between 10-15 degrees C above ambient, my water temps are only 2-3 degrees C above ambient at most. As a result, I'm seeing much better CPU temps.

To be consistent, my cpu temps as reported by the bios are 0 deg idle, 5 deg under max load. Previous temps were roughly 15 deg idle, 20 deg load where ambient was around 28 deg. I have always suspected that my bios temps were somewhat inaccurate due to my condensation prevention. I have since added an external temp sensor as close to the core as I could get it and it reports temps between -12 idle and -9 load which I think is much more accurate.

I am currently running an Athlon XP 1600+ (1400 MHz) at 1762 MHz @ 1.90V. My previous record was 1725 @ 1.85. My most recent vcore mod failed to allow voltages above 1.90V. I have been able to successfully boot into windows at 1825 MHz so I suspect if I can refine the mod to allow > 2V it should become stable at 1800+ MHz. If anyone has any experience modding a Soyo Dragon Plus, please let me know since I am really not much of an electronics whiz at all and the only sites I can find are in German with rough English translations.

-Immudium

EMC2 · 03-10-2002, 04:28 PM

Haven't seen any Vmod writeups for the Dragon, but if you get me the part number off your Vcore controller IC, I would be able to help

BTW- congrats on lowering your temps by 10C

Brad · 03-10-2002, 11:58 PM

the dden results aren't too surprising, but still really cool.

for the board, is the vrm hip6301?

immudium · 03-11-2002, 10:06 PM

Yes, as a matter of fact it is the HIP6301CB. The mod instructions I am using are here: Dragon Mod

Basically I am using a really simple 50 kOhm pot. As far as I could tell the resistance didn't even change when adjusting the knob so maybe it's not very good. And instead of directly soldering it to the board I am using two circuit probe clips, that I just hook to pins 7 and 9 on the chip, but like I say, I am certainly not seeing seeing 2.40 V like they are reporting.

There is another mod for a different board using the same chip here: Shuttle Mod using 30 kOhm trimmers and connected to pin 10 and ground. Any idea what the differences are between the two different methods? Obviously the first didn't work, but would I have more luck with the second? I haven't tried that one yet, since it seems a little more complex since I may not be able to just clip on my little probes. Also are potentiometers and trimmers basically the same thing? Should I be expecting a slightly different resistance if I adjust my little potentiometer?

Thanks for any suggestions.

Immudium

EMC2 · 03-12-2002, 04:55 PM

Okay, here is how the circuit works and what the mods actually do

Pin 7 is the feedback resistor to the error amplifier. The ratio of it's resistance to the resistance of Rcomp that runs between pins 7 and 6 set the gain of the error amplifier (basic op amp operation).

The voltage level seen at pin 7 is amplified by the error amp and the controller IC adjusts the output voltage accordingly. The lower the voltage seen at pin 7, the higher the IC will drive the Vcore voltage level.

When you put a resistor between pin 7 and ground, you are decreasing the voltage level seen by the error amplifier, thus increasing the VCore voltage.

If that were the only part of the control loop, everything would be fine and dandy. But it isn't. The HIP6301 also has a VCore sense input for detecting under/over voltage conditions. This is pin 10. If the voltage level here gets either too high or too low (greater than 15% difference from the voltage programmed by the VID pins), the controller will turn off the VCore circuit, shutting down the system. To stay within this 15%, the max you could program the Vcore for is a 2.1V output when VCore is set to 1.85V in the BIOS.

Going above 2.1V for Vcore is where the 2nd resistor in the Shuttle mod you reference comes into play. By putting a resistor between pin 10 and ground, you also decrease slightly the voltage seen at the Vsense node. This prevents the controller from shutting down when you crank up the Vcore signal level.

Summary, depending upon how high you intend to crank the Vcore voltage, you could get away with the single resistor on the Vfeedback pin (7). However, if you go too high with Vcore (above ~2.1V) and start getting shutdowns, you will have to add the 2nd resistor that connects to the Vsense pin (10).

In regards to your question about the potentiometer your using - are you using the center tap and one of the two end taps?

A potentiometer has 3 connections, there is a connection to each end of the resistor and then the 3rd connection is hooked to the wiper arm that moves between the two ends of the resistor. If you accidently hook up to the two end connections, you will always read the value of the resistor, no matter how much you turn the dial (in your case, 50Kohms).

You should hook your jumpers up and measure BEFORE you connect to the circuit and make SURE of two things. 1) one connection is to the center tap (verified by watching the resistance change between your leads when you turn the dial. 2) that the initial resistance you have it set to is at least 45Kohms between your leads. Again, both of these checks should be made before you hook your clips up to your motherboard.

<edit> One other thing of importance - you need to keep the length of the leads between the clip going to the IC legs and the potentiometer as short as possible. Ideally you will have only enough wire coming out of the clip to solder the potentiometer leg to. This is because of the sensitivity of the feedback circuit. Remember, this is going into a high gain error amplifier. If you leave the leads long and start picking up a lot of noise, your Vcore will fluctuate

EMC2 · 03-12-2002, 04:58 PM

Ooops, forgot to answer one of your questions :O

Trimmers versus potentiometers - basically the same thing. This is kind of like the difference between a tube and a hose

Potentiometers is a general term applying to variable resistors. Trimmers tend to be small, low power, 10 to 20 turn potentiometers for making fine scale adjustments in low power applications (power relative to the trimmer). Rheostats are high power potentiometers.

(then you get into linear taper, log taper, .....)

Brad · 03-12-2002, 05:54 PM

emc2, great cut and paste there, where did you get it from? :P

EMC2 · 03-12-2002, 06:57 PM

ROFLMAO @ Brad... no cut and paste, sorry to disappoint. FYI - I'm a senior R&D engineer in RL with more years experience than we'll get into

immudium · 03-12-2002, 08:19 PM

EMC2, that is some really great info. I was pretty much doing everything wrong. Big long wires coming off the IC, wires soldered to the wrong leads of the potentiometer...I could keep going, but I'm too embarrased. If not for you, I would still be a pitiful, slobbering loozer to this day!

Thanks for setting me straight.

Anyway, I'm printing out a copy of what you wrote and giving it a go tonight. If you hear of a state-wide power outage in California tomorrow, then it is either another energy crisis or me doing something horribly wrong once again. Anyway, I'll let you know how I fare.

Immudium

decodeddiesel · 03-13-2002, 11:56 AM

nice work immudium, it is always encouraging to see a sucessful use of TEC(s) with an AMD rig.

Nice info EMC2, if I had a Dragon I would be even more appreciative, but still very helpfull.

03-10-2002, 03:09 PM	#1
immudium Cooling Neophyte Join Date: Jan 2002 Location: Salt Lake City, Utah Posts: 13	Flawed Configuration: Update I have been actively working on my peltier rig for a few months now and have recently completed some modifications and thought I would share the updated results. For all the sordid details of my past woes, you can check out the thread here: http://forums.procooling.com/vbb/sho...&threadid=2007 Since I had read that Danger Den radiators favored lower flow rates while the water block favored higher flow rates, I decided to add another Danger Den super cube in parallel to the one I already had. My thinking was that I would be effectively cutting the flow rate in half through the radiators while returning to a higher flow rate for the water block after the radiators. Anyway, my results after doing this have turned out to be alot better than I thought. Where before my post-radiator temps were anywhere between 10-15 degrees C above ambient, my water temps are only 2-3 degrees C above ambient at most. As a result, I'm seeing much better CPU temps. To be consistent, my cpu temps as reported by the bios are 0 deg idle, 5 deg under max load. Previous temps were roughly 15 deg idle, 20 deg load where ambient was around 28 deg. I have always suspected that my bios temps were somewhat inaccurate due to my condensation prevention. I have since added an external temp sensor as close to the core as I could get it and it reports temps between -12 idle and -9 load which I think is much more accurate. I am currently running an Athlon XP 1600+ (1400 MHz) at 1762 MHz @ 1.90V. My previous record was 1725 @ 1.85. My most recent vcore mod failed to allow voltages above 1.90V. I have been able to successfully boot into windows at 1825 MHz so I suspect if I can refine the mod to allow > 2V it should become stable at 1800+ MHz. If anyone has any experience modding a Soyo Dragon Plus, please let me know since I am really not much of an electronics whiz at all and the only sites I can find are in German with rough English translations. -Immudium

03-10-2002, 11:58 PM	#3
Brad Thermophile Join Date: Oct 2001 Location: Nuu Zeeelin Posts: 3,175	the dden results aren't too surprising, but still really cool. for the board, is the vrm hip6301? __________________ 2x P3 1100's at 1400, Abit VP6, 2x Corsair 256mb PC150 sticks, 20gb 'cuda ATA-III, 2x 40gb 'cuda ATA-IV in raid 0. 20" Trinitron. No fans 2x 2400+ at 2288mhz (16.0 x 143), Iwill MPX2, 2x Kingmax PC-3200 256mb sticks, 4x 20gb 60gxp in Raid 5 on a Promise SX6000. Asus Ti4200 320/630. Cooled by Water

03-12-2002, 04:55 PM	#5
EMC2 Cooling Savant Join Date: Feb 2002 Location: Dark Side of the Moon Posts: 365	Okay, here is how the circuit works and what the mods actually do Pin 7 is the feedback resistor to the error amplifier. The ratio of it's resistance to the resistance of Rcomp that runs between pins 7 and 6 set the gain of the error amplifier (basic op amp operation). The voltage level seen at pin 7 is amplified by the error amp and the controller IC adjusts the output voltage accordingly. The lower the voltage seen at pin 7, the higher the IC will drive the Vcore voltage level. When you put a resistor between pin 7 and ground, you are decreasing the voltage level seen by the error amplifier, thus increasing the VCore voltage. If that were the only part of the control loop, everything would be fine and dandy. But it isn't. The HIP6301 also has a VCore sense input for detecting under/over voltage conditions. This is pin 10. If the voltage level here gets either too high or too low (greater than 15% difference from the voltage programmed by the VID pins), the controller will turn off the VCore circuit, shutting down the system. To stay within this 15%, the max you could program the Vcore for is a 2.1V output when VCore is set to 1.85V in the BIOS. Going above 2.1V for Vcore is where the 2nd resistor in the Shuttle mod you reference comes into play. By putting a resistor between pin 10 and ground, you also decrease slightly the voltage seen at the Vsense node. This prevents the controller from shutting down when you crank up the Vcore signal level. Summary, depending upon how high you intend to crank the Vcore voltage, you could get away with the single resistor on the Vfeedback pin (7). However, if you go too high with Vcore (above ~2.1V) and start getting shutdowns, you will have to add the 2nd resistor that connects to the Vsense pin (10). In regards to your question about the potentiometer your using - are you using the center tap and one of the two end taps? A potentiometer has 3 connections, there is a connection to each end of the resistor and then the 3rd connection is hooked to the wiper arm that moves between the two ends of the resistor. If you accidently hook up to the two end connections, you will always read the value of the resistor, no matter how much you turn the dial (in your case, 50Kohms). You should hook your jumpers up and measure BEFORE you connect to the circuit and make SURE of two things. 1) one connection is to the center tap (verified by watching the resistance change between your leads when you turn the dial. 2) that the initial resistance you have it set to is at least 45Kohms between your leads. Again, both of these checks should be made before you hook your clips up to your motherboard. <edit> One other thing of importance - you need to keep the length of the leads between the clip going to the IC legs and the potentiometer as short as possible. Ideally you will have only enough wire coming out of the clip to solder the potentiometer leg to. This is because of the sensitivity of the feedback circuit. Remember, this is going into a high gain error amplifier. If you leave the leads long and start picking up a lot of noise, your Vcore will fluctuate Last edited by EMC2; 03-12-2002 at 05:03 PM.

03-12-2002, 05:54 PM	#7
Brad Thermophile Join Date: Oct 2001 Location: Nuu Zeeelin Posts: 3,175	emc2, great cut and paste there, where did you get it from? :P __________________ 2x P3 1100's at 1400, Abit VP6, 2x Corsair 256mb PC150 sticks, 20gb 'cuda ATA-III, 2x 40gb 'cuda ATA-IV in raid 0. 20" Trinitron. No fans 2x 2400+ at 2288mhz (16.0 x 143), Iwill MPX2, 2x Kingmax PC-3200 256mb sticks, 4x 20gb 60gxp in Raid 5 on a Promise SX6000. Asus Ti4200 320/630. Cooled by Water

03-13-2002, 11:56 AM	#10
decodeddiesel Cooling Savant Join Date: Oct 2001 Location: classified Posts: 534	nice work immudium, it is always encouraging to see a sucessful use of TEC(s) with an AMD rig. Nice info EMC2, if I had a Dragon I would be even more appreciative, but still very helpfull. __________________ ...i hurt... do me a favor, disconect me... they can re-work me but i'll never be top of the line again ...i'd rather be nothing...

03-10-2002, 04:28 PM	#2
EMC2 Cooling Savant Join Date: Feb 2002 Location: Dark Side of the Moon Posts: 365	Haven't seen any Vmod writeups for the Dragon, but if you get me the part number off your Vcore controller IC, I would be able to help BTW- congrats on lowering your temps by 10C

03-11-2002, 10:06 PM	#4
immudium Cooling Neophyte Join Date: Jan 2002 Location: Salt Lake City, Utah Posts: 13	Yes, as a matter of fact it is the HIP6301CB. The mod instructions I am using are here: Dragon Mod Basically I am using a really simple 50 kOhm pot. As far as I could tell the resistance didn't even change when adjusting the knob so maybe it's not very good. And instead of directly soldering it to the board I am using two circuit probe clips, that I just hook to pins 7 and 9 on the chip, but like I say, I am certainly not seeing seeing 2.40 V like they are reporting. There is another mod for a different board using the same chip here: Shuttle Mod using 30 kOhm trimmers and connected to pin 10 and ground. Any idea what the differences are between the two different methods? Obviously the first didn't work, but would I have more luck with the second? I haven't tried that one yet, since it seems a little more complex since I may not be able to just clip on my little probes. Also are potentiometers and trimmers basically the same thing? Should I be expecting a slightly different resistance if I adjust my little potentiometer? Thanks for any suggestions. Immudium

03-12-2002, 04:58 PM	#6
EMC2 Cooling Savant Join Date: Feb 2002 Location: Dark Side of the Moon Posts: 365	Ooops, forgot to answer one of your questions :O Trimmers versus potentiometers - basically the same thing. This is kind of like the difference between a tube and a hose Potentiometers is a general term applying to variable resistors. Trimmers tend to be small, low power, 10 to 20 turn potentiometers for making fine scale adjustments in low power applications (power relative to the trimmer). Rheostats are high power potentiometers. (then you get into linear taper, log taper, .....)

03-12-2002, 06:57 PM	#8
EMC2 Cooling Savant Join Date: Feb 2002 Location: Dark Side of the Moon Posts: 365	ROFLMAO @ Brad... no cut and paste, sorry to disappoint. FYI - I'm a senior R&D engineer in RL with more years experience than we'll get into

03-12-2002, 08:19 PM	#9
immudium Cooling Neophyte Join Date: Jan 2002 Location: Salt Lake City, Utah Posts: 13	EMC2, that is some really great info. I was pretty much doing everything wrong. Big long wires coming off the IC, wires soldered to the wrong leads of the potentiometer...I could keep going, but I'm too embarrased. If not for you, I would still be a pitiful, slobbering loozer to this day! Thanks for setting me straight. Anyway, I'm printing out a copy of what you wrote and giving it a go tonight. If you hear of a state-wide power outage in California tomorrow, then it is either another energy crisis or me doing something horribly wrong once again. Anyway, I'll let you know how I fare. Immudium

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)