no still waiting for parts. hopefully by the end of the week
no still waiting for parts. hopefully by the end of the week
After talking to pby5cat i have had a closer look at that mosfets. i was planning to use 2 of them to give me a max of 40 amps . i only need 12 ish. while 40 is far greater than 12 the nax continues wattage is a pitiful 28watts. also you can't simple wire 2 mostfets in parallel.
These mosfets are a much better option ranging from 300 to 340watts. i obviously want the the 340watt one as it has a much lower resistance but is sooo much more expensive especially since it's not in stock in my country
http://newzealand.rs-online.com/web/p/mosfet/7165558/
http://newzealand.rs-online.com/web/p/mosfet/7259212/
http://newzealand.rs-online.com/web/p/mosfet/7259363/
i would like to try these in a current limiting mode for interest sake it may cuase a fire though.
Right you guys will definitely want to use the last one as it will pass more current when powered by the Arduinos low 5 volts. I’ll be fine with the cheapest one as I use 24+ volts (to gain the maximum efficiency ) and therefor my current is lower.
Im Seriously considering doing a dual voltage controller to massively improve efficiency.
I would like to pwm 24 ish volts to about 50% then at that point apply 12volts at 100% then below that pwm it to 0.
Why you may ask ? Well that’s because pwm 24volts at 50% is not that same as applying 50% of 24volts. If the resistance was fixed then they would be the same. However in a TEC’s case the wattage used is greatly affected by the % of it’s max voltage you apply and this relationship is NOT a straight line.
A tec running at 100% of Umax(15.2v) will consume 395watts now if we PWM that to 50% or have it turned on ½ the time and then have it off ½ the time the wattage used is 50% of 395 or 197watts. HOWEVER if we apply 50% of umax the wattage used is 85watts !!
This a loan is a big enough reason for me to take the dual voltage route but there are more benefits!
The first thing you need to realise is that TEC’s conduct heat when they are NOT powered. If you have a multi TEC chiller then they can conduct heat rather well, as I wright this my Tec’s are all off and my PC is perfectly fine even through the height of summer. Now hopefully from my previous section you will have realized that PWM is simply turning the TEC’s on and off. This kind of control has significant issues with tec when the duty cycle is very low, which it will be when the pc is unloaded (idle). What happens (in Practice) is that when the tec’s are on it move say 50watts of heat from the cold side to the hot side while consuming maybe 20 watts so there is now 70watts on the hot side. But now the Tec’s turn off and because the TEC’s can conduct heat when they’re off, say 20watts will travel back across the tec’s to the cold side. Now when the TEC’s come back on again the tec’s will have to move 50 from the CPU and the 20 that travelled back. Now that the TEC’s are moving 70 watts they will consume more to do so maybe 30watts. So on the hot side there is now 100watts and when the TEC’s are off now 40watts will travel back. Now what you get is thermal run away. However as you increase the duty cycle the amount of time allowing the heat to transfer back is reduced until you reach the point of equilibrium. Obviously your system will be consuming significantly more power than it should be in theory.
SO if I reduce the voltage it helps this situation in 2 ways. The tec’s are more efficient so when the TEC’s are off there is less heat to be transferred back and the second is the duty cycle will be twice what it would be (when running at 12volts) than if I was PWM from 24 volts.
I probably will release all the code for only PWM one voltage to keep things simple. But im sure I’ll take the dual voltage route.
Could you use a low pass filter to convert the pwm to an analog voltage btw 0 and 24v instead of using two voltages?
Originally Posted by plzdontbetaken
Well im not an electronics engineer.
My understanding of a pass filter is it has a resister inline to the load and I think that would be a disaster to have a resister that was powerful enough to pass the 30 od amps were talking about through it ?
I can easily put a cap on the TEC load side but that wont reduce the peak of 24volts each cycle
maybe something like this ?
http://www.broadcastconcepts.com/ind...&products_id=7
going back to my highschool years,
using a resistor will just drop some voltage across teh resistor but will end result have the same voltage cycle, just less voltage going to the TEC,
I think you would have to use a combination of a coil and capacitor... but i think i'm grasping and details way to long ago gone foggy.
edit:
darn didn't see Ultrasonic2's post... but ya like that
"Great spirits have always encountered violent opposition from mediocre minds" - (Einstein)
right i could get 2 of these for 164,000uf and get my own wire and wind myself 0.02 ohms and that should work perfectly ( will result in a 1.8v ripple ). what i dont know is why the cap has 4 wires ?
http://www.aliexpress.com/product-fm...olesalers.html
i assume i need a very low resistance so it doesn't catch fire and therefore i need large capacitance.
opinions from the more educated plaese ?
i 've had a re think and i could use 2 of these in parallel to giving me 0.135 ohms which give me a voltage drop of 1.88v and they will have to dissipated 26.18 watts total
http://newzealand.rs-online.com/web/...power/7332694/
if i was to then apply 47,000uf that would give me a ripple of 0.945V
or if i apply 82,000uf a ripple of 0.52 Which is quite perfect
Oh also i have received all my parts and i've got the raw parts working
low pass.png
so this should be the final numbers
Across the 2 resisters there is a 1.6721v drop while dissipating 20.735watts of heat total.
Across the TEC there is a 24.768v drop while dissipating 306watts of heat total.
Across the Mosfets there is a 0.0285v drop while dissipating 0.353watts of heat total.
IMG_1280.JPG
the project is coming along nicely. i've got all the hard ware working except for the mosfets and i'd accurately coveted the thermistors to the correct temp and i've done the dew point calculations. the screen looks mint as, but the flash makes it look not so good.
Sorry I moved to a new house and had trouble getting internet installed so I couldn't respond. Looks like you are making good progress.
Um yes . Things have changed again with the low pass filter. I can now run the arduino at 62.5k instead of 1k meaning I only need a puny capacitor.
I’ve almost finished with the programming of the arduino and have started working on the communication between it and the PC. The pc wont be required but if you want to change any of the settings you’ll have to do so on the PC and send it via USB. Also I want to be able to read all of the Arduino’s parameters so I can log them and graph them. I will also be making this VB.net source code available.
I’ve also come up with a real cool way to simplify the wiring which I’ll implement later too
Pretty interesting reading all your stuff Ultrasonic, you and mindchill seem to be at the forefront of TEC's in PC's these days...
Just curious, i have a build planned with an i7 3930k and 3 GTX680's, I want to watercool this with 2 XSPC RX480 rads... if i was to add in your TEC Controller and have the 2 RX480's cool a TEC water chiller with say 4 TEC's, would it be able to keep the water on cold side sub ambient? Obviously 2 RX480s is a lot so just wandering if that is the best option, or should i have 1 RX480 cool the hotside and leave 1 RX480 in the cold side loop to help mitigate the load from GPU's? Just speculation at the moment but curious to hear your thoughts on Running TEC's with a massive amount of radiators, is it worth it? Also would it be more worthwhile to isolate the CPU on the chiller loop, and leave GPU's on a normal WC loop.
Really only asking because i want to OC the 3930k as high as possible, possibly 5.5ghz on one core (I want a fast single core for Solidworks) You reckon a TEC Chiller could give me the extra cooling I need to push it that high or do you think it won't net me much gain over the 2 RX480's?
Fascinating work, Ultrasonic!
The TEC’s should beable to keep your load sub ambient. I have a single GTS 240 cooling my Cpu and it’s able too. However I will be adding more rad soon. More rad is always better so having 2 480 cooling the hot side will be awesome . I’d never suggest anyone put a ran inline on the cold side as that largely defeats the purpose of having a TEC. You want or need more cooling you should just add more TEC’s not more RAD.
Definitely leave the GPUS out of the chiller loop. My experience has been virtually no oc gain on GPUS when chilled. I would suggest you do the same as I have where you have 2 loops the TEC cold side which only chills the CPU and the hot side has the GPUS and the rads in it
My htpc gets pretty hot I will tec the gpu just to keep case temp down.Thinking I should replace vfd w/ ours since drivers for it only works w/mce2k4 and partially for 2k5,I'll cycle through temps to playback/Rec info.
i didn't get that ?
You were saying tecing gpu is useless but w/small case temps rise so fast and cpu will be clocked and loaded almost 24 so max temps should be a little better.also I was saying I could use vfd for more than temp monitoring.
what is vfd
Vacuum fluorescent display
http://en.wikipedia.org/wiki/Vacuum_fluorescent_display
unlike LCDs, VFDs are not limited by the response time of rearranging liquid crystals and are thus able to function normally in cold, even sub-zero, temperatures,
Last edited by Hell Hound; 05-15-2012 at 06:20 PM.
Yeah, I realized that gpu's could go on hot side right after I wrote it, too lazy to edit... So that should be sweet, 2 RX480's Cooling 3 GPU's plus Motherboard Block and 4-6 Tecs, might even go closer to 0-5* on CPU if i can find a not so ugly way to insulate a motherboard. What sort of tubing for 0-5*? Case is inverted ATX so that might make it easier, Only have to insulate socket and VRM area. I wonder what the overclocking range is on the 3930k at that temp. I would only be justified in going this way if i can decently increase OC range.
Just found this:
http://www.overclock.net/t/1219738/h...motherboard/10
Looks like he is using a direct die TEC, not a high OC, only 4.6ghz and Temps are at 30-35 under full load. If I use a TEC Chiller setup and use Sin0822 LN2 insulation method around the socket, with an acrylic cover to keep it looking good then i could run this setup really COLD!!! that would be fun, see how much this chip could do 24/7 with high OC on single core and 5ghz for all cores.... Anyone know what the max multi is with Turbo enabled? might have to run all cores at same oc because of 1.25 bclock strap not allowing voltage offset and turbo ratios not working when OCing so high.
By the way im talking about water temps not core temps. To achieve core temps of around ambient AT LOAD you’d need 0 ish water and to get to 0 ish core temps you’d need -20 ish water temps This is just like normal water cooling where your water will be over ambient and the core temps can be 20c or more over that.
It would be very difficult to hold your water temps at 0c with 4 TEC’s but I can give you a better answer to that once I get my rig going again
My controller will beable to keep your water temps just over dew point so you wont have to worry about insulating.
Like most things gains will be small verses the cost but it’s still fun and unique
yeah, i'm with you there, i'm talking water temps too, well, alcohol or whatever at those temps, but once i saw the insulation setup i started thinking colder again. that guy had 0-5* cold side temps and he was hitting 30-35 load temps on cpu, so that shows the massive load SB-e puts out, so maybe 6 tecs, -10* cold side 10-20* CPU temps might be doable??
i would go 6 Definitely would be that low at real world loads 24/7 and with just one core at 100%.
Posting Permissions
Reply With Quote
Bookmarks