Hi All,

Just curious, but what are the main differences between using mosFETs and an H-bridge design, versus power op-amps for motor control or for peltier control?

I've only seen H-bridge designs before but I stumbled across the idea of using power op-amps which can be used to output a linear voltage.

With an H-bridge design, for a peltier cooler, I would pwm the FETs and then filter out the signal to get a nice, smooth voltage. It seems that using a DAC and power op-amps would also be another way of doing this.

What are the advantages/disadvantages of each?


Thanks!

 

I predict that cost is a big factor. An H-bridge to handle 10A is cheap, but I'm betting a 10A op-amp is not.

 

Also PWM is much more efficient than a linear output op amp. A Peltier cooler typically requires a lot of current and that could mean large power dissipation in the op amp, depending upon the relative difference between the power voltage and the cooler operating voltage.

You only need an H-bridge if you want to reverse the voltage across the load.

 

Thanks for the info guys.

I've also been researching on whether you need to filter the PWM output for the Peltier. I initially thought that you would have to because I thought joule heating increases by the square, but I did some calculations and I can't seem to understand why some people suggest filtering the signal and why others say that you don't have to filter it.

For example, if I'm controlling the voltage via PWM at 10V with a 50% duty cycle for a two second period, I expect the overall joule heating to be the same as if I applied 5V DC over 2 seconds (assuming the peltier is a constant resistance, which I know it's not, as it increases with increasing delta T).

So I can't seem to understand why filtering the PWM signal to get a 5V DC would work better than just PWM pulsing. Unless using the typical LC filter adds some extra efficiency because when the pulse is at zero, the inductor will release some of its' stored energy?

 

Low frequency PWM introduces the problem of heat-shock, as it causes the TEC to go through zillions of small heat-cool cycles. But that shouldn't be much of an issue for normal PWM frequencies.

A 10A TEC is more efficient at moving heat when operating at 5A. So 50% pulsing of a 10A current should be less efficient than a steady 5A current to that same device. Filters are not 100% efficient but I think there is an overall gain by filtering.

 

The reason you need to filter the PWM signal with an inductor for best efficiency is because the cooling of the Peltier is directly proportional to the current but the losses are proportional to the square of the current. Thus, if you don't filter the current, the high peak current of the PWM signal will cause more I^2 R losses than a steady DC current equal to the average of the PWM signal.

For example, suppose you used a 50% duty-cycle PWM. The peak current is then twice the average current and the peak power loss is 4 times the average. Dividing that by the 50% duty-cycle gives a power loss twice that of the DC average.

 

I realized I made a calculation error when I was looking at the Joule heating. Yes, they do generate more heat when PWMing at higher voltages/currents. I also took a closer look at the TEC datasheet and indeed, they are less efficient at higher currents (ie. doubling current does not double heat moving capabilities).

I think I get it now, thanks guys!

Now I've found a TEC controller that outputs a 1kHz PWM output which I'd like to filter. I simulated a 100uH and 1000uF LC filter (~500Hz cutoff), but it was still showing oscillation. I'm guessing I need to just beef these values up? However, if I do so, I'm going to end up with some pretty large inductors and caps. Is there a better way to filter this signal? This controller is a 3rd party OEM controller and I'm going to assume that changing the frequency is not an option.

 

I did some more simulations and using a 10mH inductor will smooth out the signal nicely. I couldn't find any fixed inductors that have high current capacity and have 10mH values, so I simulated a common mode choke (744825510) and only connected one side of the inductor.

I am not familiar with common mode chokes, but will using them this way cause any harm?

 

I did some more simulations and using a 10mH inductor will smooth out the signal nicely. I couldn't find any fixed inductors that have high current capacity and have 10mH values, so I simulated a common mode choke (744825510) and only connected one side of the inductor.

I am not familiar with common mode chokes, but will using them this way cause any harm?

It won't cause any harm but it won't work either. The current-rating for a common-mode choke is for two equal currents flowing in opposite directions through the two windings (so the net magnetic flux in the core is zero). The high common-mode inductance is only valid for small common-mode AC currents. If you run its rated current through only one winding the choke will saturate and the inductive impedance will drop to a very low value. You need to use a standard inductor rated for the current you need.

 

It won't cause any harm but it won't work either. The current-rating for a common-mode choke is for two equal currents flowing in opposite directions through the two windings (so the net magnetic flux in the core is zero). The high common-mode inductance is only valid for small common-mode AC currents. If you run its rated current through only one winding the choke will saturate and the inductive impedance will drop to a very low value. You need to use a standard inductor rated for the current you need.

Darn, that's too bad. I'm trying to look for a 5-10mH inductor that can withstand up to 5A current. Haven't been able to really find one. I might end up just not filtering the output, but getting a lower voltage power supply.

 

You will have hard times trying to find that, better use smaller inductor and larger capacitor or more caps in parallel.

 

If you used a higher PWM frequency the inductor value can be lowered proportionally.

 

If you used a higher PWM frequency the inductor value can be lowered proportionally.

Unfortunately, the controller I have only outputs at 1kHz. I've asked the manufacturer if that can be changed, but if it can't I guess I will have to live with some inefficiency from the non-filtered output. I know it will still work, but it just won't be ideal. I'll simply pick a power supply near the max voltage i will be operating at so as to minimize the inefficiencies.

Thanks for the help!

 

If you are going for an inductive filtering scheme, go for higher operating frequencies, then you can find a reasonable inductor that will handle the current without being too gigantic and expensive.

I would investigate some of the modern switch mode LED drive circuits, what you want is an adjustable current source that has high efficiency- same as what you need to drive high power LED's.