The big TEC driver thread!

[dnar]

I don’t know, I may be missing something here but the relay is either switching thru the modulation signal (for example pin 5 of the DB25 Ilda conector 0-5V + [red]) or open circuiting ilda side and putting 0V(gnd) on the modulation in of the LD Driver which should blank lasing? Right?

This was something I raised a few days ago, but went un-attended. Many people will have differential modulation signals (ie. balanced two wire). And from memory, while ILDA specifies a differential signal, it is not mandatory, rather preferred - correct me if i am wrong.

What FourDee is suggesting is use a DPDT relay, one pole per signal, and in the inactive state, short +ve to -ve on the output so the signal input does not just float.

[Solarfire]

This was something I raised a few days ago, but went un-attended.

Shit.. I missed that one OK.. no biggy, understood be back in little bit..


Cheers!

[dnar]

Shit.. I missed that one OK.. no biggy, understood be back in little bit..
Cheers!

Cool.

A question for anyone interested in these TEC controllers, what should the interlock window be? +/-2c? +/-5c?

Also, I feel the interlock should only be active when the controller is within this window for a given time, ie. a few seconds?

Please chime in.

[Solarfire]

Cool.

A question for anyone interested in these TEC controllers, what should the interlock window be? +/-2c? +/-5c?

Also, I feel the interlock should only be active when the controller is within this window for a given time, ie. a few seconds?

Please chime in.

Just a thought... we've got one IO left on the tiny24, make the window adjustable form like +/- 0.5°C to +/- 2.5°C. Depending on the application a tighter window may be a necesary.

[dnar]

Just a thought... we've got one IO left on the tiny24, make the window adjustable form like +/- 0.5°C to +/- 2.5°C. Depending on the application a tighter window may be a necesary.

Unless there is a need to use for holography applications I do not see a need for an adjustable inhibit window.

Having said that, let me say this; it makes sense to add another trim pot on the spare ADC input.

I do have concerns providing hard-coded PID tuning parameters, as they may not be optimal for all applications. The implications could range from oscillations to poor response and offset error depending on the thermal mass of the application.

I have contemplated a master PID gain adjustment to permit tuning to suit the application. To my thinking this would be more useful than adjustable TEC current limiting. Ideally, we would provide 3 individual P,I,D gain adjustments, however I see that proposition as problematic as PID tuning is an art in itself, where as a suitable PID ratio could be hard coded and a single gain provided that could easily be user tuned with the aid of the 2 over/under temp diagnostic LED's.

BTW, are you implementing multi-turn trim-pots in your PCB layout?

[Solarfire]

Unless there is a need to use for holography applications I do not see a need for an adjustable inhibit window.

Having said that, let me say this; it makes sense to add another trim pot on the spare ADC input.

I do have concerns providing hard-coded PID tuning parameters, as they may not be optimal for all applications. The implications could range from oscillations to poor response and offset error depending on the thermal mass of the application.

I have contemplated a master PID gain adjustment to permit tuning to suit the application. To my thinking this would be more useful than adjustable TEC current limiting. Ideally, we would provide 3 individual P,I,D gain adjustments, however I see that proposition as problematic as PID tuning is an art in itself, where as a suitable PID ratio could be hard coded and a single gain provided that could easily be user tuned with the aid of the 2 over/under temp diagnostic LED's.

BTW, are you implementing multi-turn trim-pots in your PCB layout?

The problem I see with no current limiting for the TEC is that you would be bound to using a 12V TEC of which most are in the higher wattage ranges. The only other solution with an H-Bridge would be a current limiting resistor.


As far as PID tuning for everybody, I’ve seen people biting the PCB’s and perform exorcisms!


Yes, all trimmers are multi turn.


So what do you think, let me know what you need..


Schematic update.. Implemented the DPDT Relay

[andythemechanic]

Hi,

shouldn't there be some filter caps at the TEC? As far as I know TECs dont't like switching at high frequencies.

Andreas

[FourDee]

Also, the under setpoint and over setpoint led could be replaced by a single led. Over setpoint could be blinking fast and under setpoint blinking slow for example.

[Solarfire]

Hi,

shouldn't there be some filter caps at the TEC? As far as I know TECs dont't like switching at high frequencies.

Andreas

I checked out some TEC Manufactures statements to ripple current on TEC’s and found on average the following statement.


“Ideally, thermoelectric coolers should operate on purely direct current for the best performance. However, a ripple factor of 10% will only result in 1% degradation in temperature difference.”


I’ve done some more simulations in view of different TEC loads, PWM frequencies, and found that smaller loads (higher Ri) will ironically produce a higher ripple current.


The ideal setup for this H-bridge was at 100kHz, 33uH/0R084/3.5A, 3u3/100V bipolar C (covers small load situations). The resulting min. ripple was <1% and max. @ <3%. In most standard applications <1% ripple will be achieved with C in place. So in order to stay broadband as to what TEC’s can be used with this controller a bipolar C of 3.3uF/100V will be implemented.

[dnar]

I checked out some TEC Manufactures statements to ripple current on TEC’s and found on average the following statement.


“Ideally, thermoelectric coolers should operate on purely direct current for the best performance. However, a ripple factor of 10% will only result in 1% degradation in temperature difference.”


I’ve done some more simulations in view of different TEC loads, PWM frequencies, and found that smaller loads (higher Ri) will ironically produce a higher ripple current.


The ideal setup for this H-bridge was at 100kHz, 33uH/0R084/3.5A, 3u3/100V bipolar C (covers small load situations). The resulting min. ripple was <1% and max. @ <3%. In most standard applications <1% ripple will be achieved with C in place. So in order to stay broadband as to what TEC’s can be used with this controller a bipolar C of 3.3uF/100V will be implemented.

A good solution. Have you factored peak current in the inductor chosen? Peak current will be higher than average current.