To do that you have to use a 2 stage control

Thanks devigt. Could you give me more details?

Using a 2 stage control , you have 2 output on it , you can set it to be OUTPUT 1 to heat and OUTPUT 2 to cool.

Then you set the setpoint (SP) , and can set a dead band to any value so if the actual temp is over the set point it will work on OUTPUT 2 , and if it is below the SP it will work on OUTPUT 1 ,if the dead band if great than 0 , it will not ON any OUTPUT between the setpoint and the setpoint less the dead band.

Each otput will "on" it's relay , one to send +- , the other will send -+ , to the peltier cell.

By the way , do you speak Spanish?

Where do you live?

I'm in Argentina.

Gabriel

I forgot to LOGIN , it is my answer to you .

Gabriel .

Yes, I speak spanish. I am Chilean.

I suggest that we continue developing this discussion in English, but if you want, please feel free to contact me in esteban@biolabware.com

thanks!

The only way I can think of to make this work is to use a relay instead of a current output. You reverse the current going into the Peltier device and cause it to alternately heat and cool. The ratio of heating to cooling will determine your temperature.

I haven't used Peltiers so I don't know if this is advisable so experiment at your own risk. I shall do my own experimentation when I can afford to buy one .

regards,

Vulcan

Simplest way, as Vulcan suggests, is to use a relay. A double-pole changeover relay can be wired to reverse the Peltier module connections - just drive it from the relay drive output of the controller, and use DIRECT mode.

This way, when its not heating, it must be cooling. You should set the dead-band wide enough so that the relay doesn't chatter or switch too often.

Alternatively, as you suggest, you can use an H bridge - but find one with a built-in driver, or use an external driving circuit, to make sure you don't turn on both transistors in one side of the H at the same time. Using an H-bridge means you can switch the voltage polarity as fast as you want. Again, you need to use a relay drive (or other ON-OFF type) output from the controller - an analog type output isn't suitable.

Just google H bridge and you'll get plenty of information. Note that most of the circuits you come across will be for driving motors. If you're building a bridge from discrete components, you can save a few pence by omitting the diodes in parallel with the transistors. The Peltier device isn't inductive.

Why do you have to have a digital controller? I suggest looking at the Linear Technology LTC1923 data sheet which provides all the information to produce what you're asking for. A circuit board is available from them which can be configured to meet individual needs.

I use these circuits in very precise TEC controls down to millidegrees. It isn't difficult to set a setpoint to control a temperature.

I will provide a link to the data sheet a little later, busy right now, sorry.

Dear Electronic Wiz,

Thansks for your post!. Really I was looking for that kind of information. Alternatives to the current solution.

I am using a controller because I am not an expert in electronics, and I suppose that is the easiest way, but now I know that is a messy way to control a Peltier under ambient temperature variations.

I have revised the next link (http://www.datasheetcatalog.org/datasheet/lineartechnology/1923f.pdf), that contains information about LTC1923. Is first time that I hear about this company and this product, so I have some questions.

1. is a PID controller?

2. have a digital led or lcd screen?

I will continue reading that, but I am not an electrical engineer, I am mechanical engineer.

Have you any picture or diagram of your aplication controlling a Peltier?

dear users,

Please review the picture below, that is the scheme of the current configuration. as you can seem, to change the polarity (to cool or to heat) I am using now a manuel switch, that should be used coordinated with the configuration of the controller.

Could Vulcan and JohnDG explain me again their solutions?

This is the arrangement using a relay:

It does just about the same as your circuit, but here the controller is operating the polarity inverter switch.

JohnDG beat me to it. Yes, that's the way I'd do it.

regards,

Vulcan

I would suggest a second relay to disconnect the power to the Peltier device when the temperature is within the dead band. Otherwise you are using power all the time, with the Peltier device oscillating between heating and cooling.

One simple way would use an op-amp to amplify the sensor voltage, with a bi-directional optocoupler that would connect the supply whenever the op-amp output exceeds the trigger voltage of either of the LEDs in the Optocoupler.

Esteban: Si Ud. necesita ayuda en este proyecto, estoy disponible en inglés o español Chileno. On the other hand, it appears that your command of English is pretty good!

Dick

Dear JohnDG,

Thanks for your answer. Could I deduce that with your proposal my problem is solved?, I am decided to implement it!

Could you tell me more details of the relay that I should purchase?.

From the scheme I read that the 14 and 22 are connected to -Vp, and 12 and 24 are connected to +Vp. 21 and 11 are connected to the Peltier. I need to check if the relay could deal the 6A that the Peltier draws.

A1 gets the outputs from the solid state relay. That is the part that I can not understand. Could you help me completing the attached diagram?. I have leaved an space for your work.

Thanks for your interest and support, really.

Hello estebanoliveros,

Try out this site. They seem to have a good choice of temp' controllers.

http://www.controlstation.com/?gclid=CKrcyvnQi5sCFaAA4wodu0-FoQ

Good luck and keep in touch please...............

bb

A very simple tip to avoid the reversal of the performance of the Peltier cooler consists of placing a heater on the heat sink of the Peltier, able to maintain the temperature of the hot side of the Peltier above 20°C. It's apparently silly (why increase the heat dissipation?), but it works!

Hi all , as I can see Esteban is a end USER , so he want , shall , and need, to use of the shelf devices.

The CAL 3300 series can be suited to fill it's need.

It will PID both stages.

It can be a 2 stage controller .

devitg: I've been looking at the CAL3300 user manual, and I agree - using two outputs (heating/cooling) is easy to implement with this controller, and a good way of going about it.

I'll pm you.

Hello, sorry I took so long to get back, here is the direct link to the part's general page, there are further links on the page to the data sheet and application data.

http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1142,C1095,P1893

Now that I know you are a mechanical engineer, that changes things a little. This part is not difficult to use and I think you could likely get it up and running with a just a little help from an electronic friend or help from us. Linear Technology is one of the best linear chip sources on the planet. I use them frequently.

To answer your question, no this is not a PID controller, while PIDs have very good application, their parameters must be carefully chosen or they will give you trouble, they are usually analog but there are digital hybrids of them as well (analog/digital).

The LTC1923 has a built in driver for an H-bridge output, however, since you are using at least a 12 volt supply, a higher voltage interface will have to be used (the LT1970 runs at 5 volts), listed in the data sheet and the application note AN89A. Two LTC1693-1 chips will provide the 12 volt drive interface. The system is capable of driving your 80W TEC with selection of the proper MOSFETs. An evaluation board is available from Linear which provides the basic system, data on this is also on this main page.

It is very stable and can control temperature very tightly if necessary. Your 20'C temperature will be no problem, a 10K thermistor will work nicely in a bridge configuration, I can give you the nominal resistor value needed for 20'C if you decide to try this. The eval module will save you a lot of wiring fuss.

I'm sure Linear has representatives in your region and a distributor, they are very helpful. Any questions, just ask, either here or by the private e-mail link here. I'm sure Linear will sample you out a couple of LTC1693-1 chips as well.

Dear JohnDG,

Thanks for your answer. Could I deduce that with your proposal my problem is solved?, I am decided to implement it!

Could you tell me more details of the relay that I should purchase?.

From the scheme I read that the 14 and 22 are connected to -Vp, and 12 and 24 are connected to +Vp. 21 and 11 are connected to the Peltier. I need to check if the relay could deal the 6A that the Peltier draws.

A1 gets the outputs from the solid state relay. That is the part that I can not understand. Could you help me completing the attached diagram?. I have leaved an space for your work.

Thanks for your interest and support, really.

First a couple of points.

Do you already have the controller? If so, which version?

If your version has a relay output, you don't need the solid state relay. If you only have SSR drive outputs, you'll need a solid state relay rated to switch DC - these are not too common. The SSR drive output (5V @ 15mA) is enough to drive a small mechanical relay could be used to drive the power relay - this would be a cheaper option.

The relay I had in mind is the Finder 40.52.7.012.0000, which will handle up to 8A. The connection numbering I've used relates to relay base Finder 95.05SPA - the base costs more than the relay, and isn't needed if you're assembling the bits on prototype board (e.g. Veroboard).

If you let me know the controller model, I'll finish the circuit & suggest a part no. for the SSR or 5V/15mA mechanical relay if you need one.

As he wants to have PID , i think a releay is not an option.

Surely it depends on the thermal time constant of the system being controlled. Switching between heating and cooling with a relay is, after all, just a form of pulse width modulation (albeit a slow one).

... and the thermal mass of a Peltier device capable of handling 80 W, even without considering the mass of the plate being controlled, removes any advantage there might otherwise be to using true PWM.