Author Topic: Peltier trick, tips and intrigue!  (Read 2954 times)

0 Members and 1 Guest are viewing this topic.

TrickEMethod

  • Guest
Peltier trick, tips and intrigue!
« on: October 31, 2002, 05:32:00 PM »
The usefullness of a compact, cheep, clean, controlable and non-consumable source of heat extraction is well understood.  I have read an untold number of posts speculating on the many uses of Peltier cooling devices for clandestined chemistry.  No doubt many bee's have tried using Peltiers, but I doubt many have actually been able to use one in production.  Let me know if you found different, but I found the little bastards to be somewhat of a tease the first half dozen times I tried to get them to do any real work. 

They would initially produce an impressive burst of temperature drop, shooting to 50-60 deg C below ambient, but quickly drift back up and in some cases come to rest ABOVE ambient.  Since a cooling device that works for less than a minuite is near useless, I kept trying and scraping idea's that involved peltiers until I learned 'the secret' to using peltiers.

'The secret' is simply the realization that these devices have inherent tendancy towards currrent stability.  They will draw far more current, and thereby deliver more heat to the 'hot' side, than can be removed by the heat exchanger attached to the device.  If offered mucho amps, they will pass mucho amps and happily fry themselves like a vietnamese monk.

Circuits are available to limit current at the supply, but without inteligent current controll based on actual sensor feedback, pessamistic assumptions must be made which will seriously limit the practicality of the device.  Since it is doubtfull that your average bee can look at a heat exchanger and guess its Watt transfer capabilities(which actually requires calculating the exchange metrics for two coupled heat exchange processes that start at ambient conditions and then cycle towards an equilibrium), limits can save the device from self destruction but are unlikely to produce an optimal result.

And since no feedback it provided to the operator as to the result being achieved or the final resulting temp, the operator has no idea what is being achieved, how much more could be achieved or what to do to make that happen. 

Inteligent control based on sensor input can change all this.  By sampling temperature data from the hot and cold sides of the device, or even better, the coolant entry and exit temp from the hot and cold sides of the device, the actual results can be calculated using the maximum ejection capabilities of the liquid/air heat exchanger as the limiting factor.

In addition to providing a maximized result based on the limitations of the current system, the control software can provide feedback to the operator about how much of the pumping capabilities of the chip are being utilized and what can be done to improve the situation.

I have already done about half of what I describe using a cheep and easy to use microcontroller called an OOPIC.  The OOPICE has numerous built it A/D converters and offeres easy current control through built in PWM(pulse width modulation) functionality.  PWM is the technique of toggeling the current on and off at a high frequency and varying the percentage of on-time vs. off-time to control the current provided to the device.  This is the recomended method for controlling a Peltier.

The OOPIC dev kit with everything necessary to program it, except for a windows PC, can be purchased for less than $100.  It connects to the PC via a serial cable that is supplied.  You program the device using your choice of; a simplified VB like syntax, a simplified C like syntax or a simplified Java like syntax.  once simplified, all three look almost alike anyway.  The main difference being whether you put a semicolon at the end of the lines and how you punctuate a function call.

If you are a first time programmer, you will have to make all the first time programmer mistakes, but you are not likely to find a less demanding environment to learn in than this.  Getting a temp reading takes a couple lines of code, setting the PWM percentage takes a single line of code.  The entire program with all the bells and whistles will likely be less than fifty to seventy lines of code unless someone gets inspired enough to add a complicated LCD or HTML realtime display screen.

Now is the time to make suggestions, if people have other controllers they like.  I also have the Rabbit 2000, which is much faster and has a more professional development environment.  But the dev kit is $200-500 with a limited use compiler that will likely need to be upgraded, and does not have built in A/D conversion.  So an A/D chip, mux and voltage reference circuite would need to be added, which is a pain in the ass compared to the OOPIC.  It also requires classic C programming skills, further reducing its usefullness.

Let me know what you think...
TrickE



And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

MnkyBoy78

  • Guest
Idea? Or an Answer!
« Reply #1 on: October 31, 2002, 07:44:00 PM »
Wavelength Electronics makes a temperature controller for TE coolers.  One model in particular is the LFI-3526.  With 2.5 A output, adjustable set-point (in K), current limiting setting...Every thing one would need.  It also has a I/O port for linking to your PC.  The TEC's are connected via a DB9 connector which suplies juice for the TEC and feedback back to the controiler. 

Price new? Beets the shit out of me...SWIM picked his up at a liquidation auction, along with the rest of the pallet of shit, for a meer 15 USD. 

I imagine its not a cheep unit (> 500 USD?). 

More info to come soon...If so desired!

Why 78...I just dont know? But the voices said it sounded good

Rhodium

  • Guest
Controlling Peltiers
« Reply #2 on: October 31, 2002, 08:07:00 PM »
It seems like voltage/current limiting is the way to go for Peltiers, rather than pulse width modulation. The following section is taken from

http://www.marlow.com/faq.htm



What is the best way to power and control a TEC?

Thermoelectric coolers require smooth dc current for optimum operation. A ripple factor of less than 10% will result in less than 1% degradation in delta T. Voltage or current limiting should be used in order to ensure that Imax for the TEC is not exceeded. A bipolar power supply is required for those applications requiring both heating and cooling. Pulse width modulation may be used at frequencies above 1 kHz. Linear proportional, PI or PID control can also be used. Marlow does not recommend ON/OFF control. While this is the simplest control technique, temperature cycling within the TEC as power is cycled from full ON to full OFF may result in premature failure.

On another note, where do you find CPU heat exchangers cheaply - I have only been able to find artistic ones for designer case-modding costing $50-100 for only the CPU cooling block.

ClearLight

  • Guest
Here it is...
« Reply #3 on: October 31, 2002, 08:10:00 PM »
The laser controller is not cheap...

 here is the application note on how to build your controller with a $3.75 IC plus caps/res.  You can request free samples for up to 10 parts...

http://www.maxim-ic.com/appnotes.cfm/appnote_number/1757



and here is a All in One chip for $16.00

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/3527




and here is the 20 page app note that tells you how to build/use it...you want the 1979 as that handles 6 vs 3 amps..

http://pdfserv.maxim-ic.com/arpdf/MAX1978-MAX1979.pdf



( I love process control stuff! )



Infinite Radiant Light - THKRA

hCiLdOdUeDn

  • Guest
You can get a peltier type cooling refrigerator ...
« Reply #4 on: October 31, 2002, 10:38:00 PM »
You can get a peltier type cooling refrigerator at a local store. Get the kind that has a metal reservoir and can hold water with a lid that opens up. Then just fill in water and let the peltier cool the water (down to atleast 40F) and run it through the condensor and recirculate. The problem with this type is it takes 4-5hours to cool the h2o temp to 40F. But it will keep the water cool for many many hours when using a medium sized condensor setup distilling high boiling liquids (150C)!!

Sink or SWIM

TrickEMethod

  • Guest
That is a cool product, which one did your get?
« Reply #5 on: November 01, 2002, 02:24:00 AM »
That is a cool product, which one did your get?  I would like to hear about it.  The standalone unit is very similar to what I would like to build, except I would feel more comfortable with a higher output (12-18v/0-15amp or around 200watt continuous). 

The biggest problem with products like the one you found is cost.  The 5.0 amp unit listed at ~$2400, without sensors, cables or other options.  And I am pretty sure that this is just a controller, not a power supply.  So you will need an external power supply, which can easily cost $500 for a similar style off the shelf model.  So you are up to $3k before buying the 2 solid surface to liquid heat exchangers, 2 coolant tank, 2 pumps and of course, a high capacity Peltier. 

And you will also need to connect the controller to a PC if you wish to automate staged transitions that are sensitive to various reaction conditions.  A PC would also be necessary to get any diagnostic feedback.  Feedback will be the key to finding and fixing performance limitations, like whether the primary or secondary coolant flow is inadequate or warning when secondary coolant runs out of ice and begins to spike.

I'll layout some additional reasons why I think a OOPIC driven controller is superior even without cost considerations after I go get some food...

TrickE

Ps. In response to Rhodiums comments, I'll also explain why I think PWM through a high current FET H-bridge is the way to go for current control.



And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

TrickEMethod

  • Guest
Why would a OOPIC temp controller be superior to ...
« Reply #6 on: November 01, 2002, 03:20:00 AM »
Why would a OOPIC temp controller be superior to off-the-shelf controllers...

A temp controller investment is not going to be made by many new bee's looking to do a Curbshot Nano(no disrespect intended).  Larger and more diverse reactions are to be expected, most of which would not be well served by 50 watts of cooling potential.  Overrunning a peltier based cooling system is something to be avoided, and the only way to avoid that is have a significant cooling headroom over and above the anticipated requirements.

If the cooling capabilities of a Peltier device are exceeded, it will not respond in a gracefully fashon.  The system needs to have the inteligence to REDUCE current in the event of an excessive enthalpy load.  Reducing current will reduce the cooling to some degree, which is not exactly what you would like to happen during a heat overload but is still better than overshoot which actually turns the peltier into a heater for a brief period before it self destructs and simply becomes a thermal barrier.  So the options are a partial reduction in the peltier component of the total cooling factor, or a near complete loss of cooling.  So the three options are bad, really bad and have a big enough reserve to avoid the issue.

I had hoped to deliver 500 watts of constant cooling capability between the peltier and the secondary coolant combination.  I found CPU heat exchangers for $40, which were a good size match for 4-50 watt peltiers in a 2 x 2 configuration.  Having four devices lowers the overall cost, and issolates fracture failures to some degree. 

The wattage would count as additive on top of the cooling offered by the secondary liquid cooling, which I have speculatively pegged at 300 watts.  (I will refine and post support for these numbers as things go along.)

By additive I mean:
Cooling is a function of deltaT, conductance and duration.  The higher the temperature differnence(delta T) between the primary coolant and the compound being cooled, the greater the volume of enthalpy or heat will be extracted.  The thermal coupling between the coolant and the compound, also plays a factor and can increase or reduce the cooling affect of a given delta T.  Since the primary coolant will be cooled by a peltier, which is in turn cooled by a secondary liquid(MeOH/ice for example), the deltaT at work on the compound is the summary result of the temperature drop of the coolant and peltier in combination.

Here is how I would expect to maintain headroom and affective temp control:
Lets say a reaction calls for maintaining a jacketed flask at 0 deg Celcius through a reaction that transitions from a short endothermic stage to a vigerous exothermic stage.  To have the maximum cooling potential available, MeOH/ice will be the secondary coolant which might float in the vacinity if -20 deg.  This means that without additional effort from the peltier, the cooling potential of the system exceeds the needs of the reaction.  If a less substancial secondary coolant is used such as H20/ice, then the inital stage is better served, but the subsequent exothermic stage will place a high demand on the peltier and risk overshoot.

This is where some clever control of the peltier can really pay off.  By reversing the current direction using the H-bridge the peltier can be used to pump heat from the secondary coolant into the primary coolant.  This means that the secondary cooling supply would not be diminished, to some degree it would actually be increased, and the primary coolant would deliver enthaply to the reaction to instagate the reaction.  Once the reaction had transitioned to the exothermic stage, as determined by sampling a thermocouple inserted into the flask, the current would be reversed.  Since rapid thermal transitions are the number one cause of peltier failures, rate limits need to be observed. 

One option to allow for quicker reversal of enthalpy flow is to drive each of the four peltiers on a seperate PWM channel, allowing a bank of two peltiers to be used for moderate cooling in one direction and two to be floated neutral and then activated for moderate cooling in the opposite direction affectively doubling the rate of transition.  A four stage transition with cancelation between reversed devices could transition more than four times quicker while obeying the same localized rate limits.  An additional benifit of four PWM channels is use of more commodity h-bridges.

I havn't found where any of these features are available from commercial controllers, and without something similar I don't know if I would trust a peltier powered device to control a reaction that could land be in jail or the cemitary.

And in response to the issues of PWM current ripple vs. linear supplies... we will likely perform pulse modulation at 50-200k cycles and smooth that with a moderately sized capacitor.  We have the advantage of not needing, or wanting, overly rapid voltage following so the capacitor delivers a low ripple supply without adversing affecting any performance characteristic that matters to us.

Thats enough for now, I think I will go take some pictures of what I already have...  believe it or not, its actually simpler than it all sounds.

TrickE

And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

goiterjoe

  • Guest
a PC is only necessary for loading software
« Reply #7 on: November 01, 2002, 03:26:00 AM »
You can write condition statements and sauter on switches to your controller board to tell the board how to react to temperature spikes and different fractions coming over.  You can also get controller boards with LCD displays to output temperature readings and other data you might need.

As far as the 4 transistor H-circuit goes, it should work fine as long as sampling rate is fairly low.

If at first you don't succeed, try, try again. Then give up. There's no use being a fool about it.

Rhodium

  • Guest
PIC IC
« Reply #8 on: November 01, 2002, 03:39:00 AM »
Do re really need much more than the peltiers, a few temperature sensors, transistor driver circuits and a Programmable PIC IC? The circuit can be simulated in the parallell port of a PC first, then burned to the PIC.

This would cost no more than $100...

TrickEMethod

  • Guest
I'm not sure exactly what you mean...
« Reply #9 on: November 01, 2002, 03:56:00 AM »
I'm not sure exactly what you mean...

Do you mean the controller devices from Wavelength can do what you describe without a PC, or an OOPIC can? 

I looked for that type of capability from Wavelength and didn't find it, but I by no means did an exhaustive search.  I just reviewed the PDF manual of the 'LFI3751 Digital Temperature Control Instrument'.  Which seemed to be their most advanced offering.

If you mean the OOPIC does not need a PC, I would have to agree 90% with that statement.  The OOPIC could control a reaction quite well by itself, and could report detailed findings via a LCD or dump them to a PC after the fact.  But providing an interface to input and edit reaction parameters via menus, buttons and the LCD screen would require significant work and would consume a significant amount of the FLASH and SRAM resources of the processor.  Not to mention the programming effort.

I was thinking about a simple Java editior to create reaction profiles as XML file defining states, transition points, control and alarm parameters.  Then anyone could play around with whatever interface they wanted to enter and store profiles on a PC, Mac or Palm pilot.  Reaction results could be recorded in a matching XML file and returned to the PC either realtime or as post game review.

By allowing the operator to insert additional data items, such as reagents, mole/weights/measurments in and out of the reaction basic record keeping can be encouraged.  This would be particularly usefull for accuratly comparing results online and establishing manditory sentencing.

TrickE

And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

Rhodium

  • Guest
PIC 16C
« Reply #10 on: November 01, 2002, 04:02:00 AM »

TrickEMethod

  • Guest
Almost everything I have discussed is just ...
« Reply #11 on: November 01, 2002, 04:14:00 AM »
Almost everything I have discussed is just software.  With the possible exception of seperating the peltiers h-bridges, which is cheeper than purchasing a single 200 watt h-bridge.

If you mean do you need a primary and secondary liquid coolant arrangement, I can say with confidence that you do.  I built several elaborate forced air heat exchangers that didn't even come close.  In fact the best didn't supply 5% of the heat ejection that would be needed to tackle NH3 condensation.

All in all, a liquid secondary cooler costs about $55.  I have one working great and will forward pictures tonight or tomorrow.  The back-2-back CPU coolers cost about $40 each, and the primary pump, various hoses and fittings costs maybe $20.  Four peltiers will likely run $10 each, plus 5$(or less) for the h-bridges and caps to drive them.  The OOPIC was something around $60, I think.  The powersupply will be whatever high capacity open frame unit is taking up too much shelf space at the surplus warehouse, call it $30.  Add to this $40 of unknown shit including some thermocouples, resisters.

So for a total of: $345

But you can built it all incrementally, start with the liquid coolant arrangement which is highly usefull all by itself, for $55.  Then pick up a OOPIC, CPU coolers and a single peltier for experimenting using whatever free wall wart powersupply you have lying around, for another $130.  Then when you are confident that the system can do what you want shell out the remaining $150 or so to jump the capacity to something that can do some real work.

And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

ClearLight

  • Guest
what???
« Reply #12 on: November 01, 2002, 09:06:00 AM »
No one looked at my posts on maxim...there's a 20 page pdf explaining the theory and the schematic, and you can do it for about $25.00, since the smarts are built into the single ic... unless you want to get really clever...  the problem with the OOPIC is that you have to program a PID loop to get the thing to converge to equilibrium, otherwise it will oscillate... that is not a trivial task.. If you have a fuzzy logic module, like some of the PIC Micro ones, it is a lot simpler to write the control logic..


Infinite Radiant Light - THKRA

hCiLdOdUeDn

  • Guest
Thermoelectric cooling/warming device
« Reply #13 on: November 01, 2002, 11:50:00 PM »


You can add approx. 1 gallon of water into the reservoir, and let it do its thing. I bought mine for $30 at local store :)  Its only 32 watts but it works great for condensor cooler especially when your distilling for long periods of time.

All you have to do is add water and have a water pump attach with the tubing and run it through the condensor!! If you do a lot of distilling this is cheaper than ice :)

Sink or SWIM

MnkyBoy78

  • Guest
TEC used on SWIM's CPU
« Reply #14 on: November 02, 2002, 02:30:00 AM »
When SWIM wanted to aid his ol' processor in cooling down, he stumbled across these [ommited for not wanting to post sources].  A few sizes to choose from (1.17 inch2 or 1.56 inch2.  There also is the choice of new or used.  Price: 17 - 26 USD

Note on ommited: PM me

Why 78...I just dont know? But the voices said it sounded good

TrickEMethod

  • Guest
WOW, I have to admit that I skipped over the ...
« Reply #15 on: November 03, 2002, 02:45:00 AM »
WOW, I have to admit that I skipped over the links wanting to reply to a couple of the other posts before I got distracted.  Also the 'laser' tie-in left the initial impression that the info would fall outside out intended purpose.

I have only just glanced at the links, but I was very interested.  They seem to be looking at the very problems that I ran into getting the little bastards to work.  From one perspective it's gratifying to see that I am not the only person who has bent their pick on the problem. 

I printed them all and will take them with me tonight.  It does look like a significant portion of what I wished to do has been implemented by Maxim in a simple and cheep descrete package.  Given the availablity of a chip component makes is pretty likely that it can be integrated with whatever microcontroller to make profile following possible.

I use a number of Maxim's chips regularly, and am on their site several times a year, but I had never seen anything on TEC/Peltier controll devices.  Live and learn...

Thanks for bringing this to my attention, and I will take a good look at the docs I printed and throw out some design options.  I still think that you will need some sort of controller steering the ship, but with the subtle majic needed by the peltier out of the way, things could get quite simple indeed.

Again, my appologies for letting the links you posted slip by.  I meant to come back and check them out, but I got distracted.  My bad, I hate it when people do that to me!

TrickE

And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

TrickEMethod

  • Guest
Well, shit. The Maxim products are interesting, ...
« Reply #16 on: November 03, 2002, 06:12:00 AM »
Well, shit.  The Maxim products are interesting, but not really designed for what we are trying to do.  At least not what I am trying to do. 

What the Maxim chips can do is take a voltage from a temp sensor and compare it to a target voltage and adjust the current fed to a pletier (hopefully coupled to the temp sensor) until the two voltages match.  Once they match it will do its damdest to keep them matched and trigger an alarm if it deviates beyond some amount.  This is all an entirely analog effort, it would have no idea whether it was keeping the device at 150 or -150 degrees.  It simply works to match two voltages in a wheatstone bridge.

If you need to stablize a peltier to a predetermined temperature setting that never changed, and did the design work upfront to know that your heat ejection capabilities were well matched to the load(pretty well describes the dedicated communications laser market) then this is a simple and affective solution.

If you wish to achieve digital temp control, then you would have to drive the controll voltage via a DAC, and sample a seperate thermal sensor to judge the results as feedback.  The downside of DAC circuites is the need for precision references and preamplification to match the voltage window's of the output device to the wheatstone bridge.  If the windows don't match then you loose precision and predictability, if the circuit worked at all.

Bottom line, digital control of the the Maxim requires:
   DAC
   Voltage reference
   Output preamp
   Temp sensor
   Software feedback loop

Digital control withtout the Maxim requires:
   PWM
   FET Hbridge
   Smoothing capacitor
   Temp sensor
   Software feedback loop

The Maxim device is undoubtably going to achieve a greater temperature stability real-time, but the non-Maxim solution will require less analog circuit complexity and fewer components and cost.

The fact that Maxim has a device out for Peltier control does make me think its worthwhile to do some more searching to see if someone else might have a similar solution with digital control.

TrickE

And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

TrickEMethod

  • Guest
Here is a cheesy text diagram of the proposed two ...
« Reply #17 on: November 03, 2002, 12:37:00 PM »
Here is a cheesy text diagram of the proposed two stage peltier driven coolant system.

          
               Secondary
               Coolant return
              /
        +----------------------------------------+     
        |+--------------------------------------+|
========||=========                             ||
||      \/       ||                             ||
||               ||                             ||
||               ||                             ||
||               ||            CPU Cooler #1    ||
||               ||                      \      ||
||               ||      Secondary        \     ||
||               ||      Coolant out       \    ||
||               ||     /                 _____ ||     Peltier
||               ||   +------------------|     |+|    /
||               ||   |+-----------------|     |-+   /         Primary
||                ||\ ||                ######### <--         / coolant
||               #|| -||-\            +--|     |-------------/    out 
||          +++++#+|      |           |+-|_____|-------------
||          +++++#+|      |           ||    \                    Primary
||               #|| ----/           /  \    CPU Cooler #2      / coolant
||                ||/   \            |  |--------------------- /   return
||               ||      \           |  |---------------------
 =================        Bilge      \__/
5 gal Gattoraid Cooler    pump         \
  w/spiggot at bottom                   \
                                         Windshield Washer pump



And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!

Disciple

  • Guest
So ...
« Reply #18 on: December 20, 2002, 12:47:00 AM »
I've heard you hinting in other threads about having completed the plans for this setup. but haven't received  enough interest. So, just registering my interest. Hurry up!!  :P

TrickEMethod

  • Guest
I have been devoting 30min - hour of time ...
« Reply #19 on: December 22, 2002, 05:42:00 AM »
I have been devoting 30min - hour of time everyday to hive related projects, but am having to complete with work committments.

The one getting the vast majority of my time is documenting the NH3 reaction/dryer/prechiller/condenser, and I have getting quite near complete on this.

I have already posted a number of pieces in different threads, and recieved some good advice on a alternative chemisty for the NH3 production which suffers less from H20 absorbtion loss than the Fert + NaOH + H20 method that I had been using.

I will put the instructions that I have completed together with the instructions that I have already posted together into a single document as an almost complete draft tomorrow night, at the latest.  I'm sorry about the delay, getting this information into play is very important to me, but I have hd my back to the wall on committments that I could not afford to ignore.

TrickE


And on the eight day, God created Meth...
... and hasn't done much of anything usefull since!