Thanks cp, I can't thank you enough for the detailed help. Wow, I thought I had my bases covered with these heatsinks. Let me post some pics just to be sure, because these babies have some mass to em'. The bottom (cold side), is going to have these solid copper (heavy little punks) blocks: The top will have these larger heatsinks with the copper heatpipe, and better fin arrangement for airflow: (Sorry if these pics are too big) Are those not big enough? I might have to attach an individual fan to each heatsink, but I'm not sure I can go much bigger.
I saw the little heatsink b4 You should test it! Just sit everything on top of each other, use a dab of grease on both sides of the peltier, and then rest one of your fans on the top of the heatsink. Slightly better temps will be had (~5-10c) when everything is clamped tight but the weight of that heatsink & fan should work okish and give us a good idea of what to expect. I'm guessing you might hit around 30-40c on the hot side, and maybe like 10c on the cold side like this, so around 0c-5c better with everything clamped tight - but maybe that's all you want? in which case you won't need better cooling. So what's this for? There must be some sort of load, and loads will bring the temps up...
I am definitely looking forward to running some tests. I wasn't able to find out the ΔT rating on the TECs from the seller, so I need to know what performance to expect. I'm into doing rather than talking when it comes to DIY projects. I need to get my stuff together and know for sure this is doable before I start promising pics. If I get caught up in a snag, I know where to come for help. You guys and gals have all been great, thabks for the assistance. -TTL Locke
Are the hot side heatsinks from a Dell Dimension by any chance? They look very familiar, I think I have one in my bits box The fans those come with are just awesome. Edit: here's how you do an adjustable current source (R1 = 1 ohm resistor followed by 100/120 ohm pot, with the pot wiper jumping 'over' the 1R resistor... basically so that the minimum resistance is 1R)
Good eye Cin, they are DELL cpu heatsinks, but I didn't get the fans. Current limitor looks spiffy, right now I don't understand everything that's going on, but I can study up on things before I pester you guys with n00b questions. Thanks again.
The basic LM317/LM117 current regulator is simple and works fine, but it's very wasteful and makes lots of heat at high currents. The chip drops about 1.5V minimum and the regulating resistor 1.25V, so from a 12V line your maximum output is about 9.25V. It's a good way of running something like Luxeon high-current LEDs from a very variable input voltage (like a car or boat alternator/battery, with an 11-15V range possible) but if you're starting with a stable input voltage, voltage or PWM control is more efficient.
The current limiter is quite simple. The regulator attempts to keep Vref (in this case 1.25V) between it's out and adj terminals. In this case, as you know the resistance (R1) and the voltage across it (1.25V) you can work out the current, I=V/R. Actually that current limiter might not work so well as it will require a potentiometer which can handle the current. There are certainly better regulators and indeed better methods yes, PWM might be one of them. Do you think the sudden on/off load of ~7.5 amps times however many TECs there are might be a bit stressful on the power supply?
I'm not sure to be completely honest with you. I'm trying to get a PSU from a reputable company that has enough juice on the 12V rails to handle what I can throw at it. The Zalman ZM750-HP has 80A on the 12V rails alone. I don't really know much about how stress on electric components affects them short/long term, but I am willing to listen to y'all's (yeah, I just typed that) input, and make changes if necessary. Would stress cause the PSU to heat up? This shouldn't be too much of a problem. If the device can be physically damaged, then I might have a bit of a problem on my hands. However I would assume that it would be rated to handle the on/off of a powerful PC, but this area is far from my specialty. What do you guys think?
Never mind, actually, it shouldn't be a problem... I had a theory about sagging due to the switch mode nature of the power supply, but it wasn't right High current draw will cause the PSU to heat up, yes. It may be prudent to add a short delay to allow the PSU to get going before dumping such a big load on it - at start up it will be stressed quite hard due to inrush, but if the PSU is chunky enough it should be OK without.
For a high power PWM controler, cars are a good place to look.. any that have climate control should have PWM or similar (rather than switched resistors on my car). The one I got from the breakers uses two dual opamp chips, though one amp is unused, with a 30A darlington. Its contolled by a 0-5v signal, usually from the car climate control system. Only issue with this is a darlington will take 1.2?v from your maximum output. Cost me £5 with the huge heatsink. I was planning to swap to the climate control system but looks a bit too complicated...
Yep you'll loose something on the order of a volt whatever transistor you use (~0.7V for a single one, 1.4V upwards for a Darlington (2x0.7, init))
No, a normal bi-polar loss is around 0.2V (VCEsat in its spec) when wired as a switch. Darlingtons are over double that, but will go over 1V at high current and can reach 4V drop. (But this application is high current). With MOSFETs you've only got the resistance Rds so they're usually more efficient.
Is it? Hmm. I may have been thinking of Vbe/Vbc which I thought (can remember reading, in fact) was essentially a diode (PN junction) so 0.7V
lineir regulation is a very bad idea here you need a switching power supply. Personaly I would buy a dedicated petier supply or build one myslef. You need to adjust the power supplied to the pelts using PWM or simlar. A dead simple solution would be variable voltage transformer > bridge rectifier > V. large capacitance bank (several farad) > pelts you would need a quite beafy transformer though. About 400VA and you will probobly need surge protection to stop such a large transformer and capacitor bank triping your mains when you turn it on. Very simple effeciant solution though.
The problem with that is variable transformers (Variacs) are expensive, and invariably large. Also, they are autoformers which means NO isolation from the mains at all - one power spike/dip and either you, or your equipment, could be finished.
This thread has taken a turn into some highly technical waters. (for my level of knowledge anyway, don't hate the n00b) So, trying to run a PC and the TECs off the PSU is a bad idea? Using CP's idea of: IC1 puts out a 12V "high" or 0V "low" signal and will turn any number of parallel n-MOSFETs on & off, each with a TEC load. IRF530 Cheesecake. I thought this woudn't be so much as linear regulation as it was linear control. I have abandoned the idea of voltage regulation in the way I originally thought, and will simpy be controlling whether the TECs are on or off. That way I can regulate power in roughly 70-75W incriments, up to 375W max. I probably have no idea what I'm talking about, am I even on the same page book library as you guys? F T W = Cheesecake huh? Interesting...
PWM is only good if you have a suitable powersupply to run them off. Otherwise it may be cheaper to have an adjustable SMPS - this maybe better than pwm anyway.. its more a case of finding out how easy it can be done. This is beond my knowledge, but I assume most SMPS can be modified to output any voltage within a resonable range, probably by adjusting a resistor or in the case of ATX supplys there is a feedback wire on the motherboard connector. I have used this before, however I am not sure if this is the correct way to achive a large adjustment. If you were to go down the PWM route, look at an automotive one, just swap the darlington for a single large mosfet.. although it would probably be ok just to keep the darlington.. as on full you might only lose a few watts of cooling. You would not be able to buy a better unit or just the parts for one at the same price. If you want it really simple, all you need is a 5 postion switch, 5 mosfets, couple of resistors... Since your TECs add upto a large amount of heating.. you should put some kind of thermal fuse on the heatsink, just incase the fan or pump stops.
The TEC's won't be actively cooling the CPU, so I don't have to worry about that last bit. Also, no pump.
