Q factor of LC filter for TEC driver

[Gibson486]

I am making a filter for a TEC (only needs to go one direction). The values I intend to use are an inductance of 33uH and a capacitance of around 270uF. I did this to minimize ripple at my frequency (~40khz PWM), so these values attenuate close to 60db at my frequency. I plotted the frequency analysis, but it showed that I have a high Q value and it peaks at around 1.5kHz. Since I am not really interested in the band of the peaked value, could it be ignored? I could dampen it and put a resistor in front of the filter, but the TEC is over 5 amps, so that would be a huge resistor.

Thanks!

[Benta]

https://en.wikipedia.org/wiki/TEC

[Gibson486]

Thermal Electric Cooler (aka peltier)

[duak]

I worked on a few projects that used PWM TEC drivers.  I don't remember the details as I didn't design that section but I don't recall any particular problems with using a simple LC filter.  I do remember that the designer used powdered metal inductors instead of ferrite inductors.

We'd have to know more about the circuit, especially if there's some sort of feedback from the filter output.  If the circuit is open loop, I wouldn't think there would be a problem because there should be anything to excite the LC resonance.

The load impedance has to be considered - have you determined what the TEC's dynamic resistance is over the range of drive currents?

If you need more damping, you might be able to add a parallel RC network across the filter cap, but it depends on the dynamic resistance of the TEC.

A small consideration is that TECs generate a bit of power when there is a temperature difference across them.  In one case, this sneak power source kept one of the power bus indicators on for a few minutes and caused a few minutes of head scratching.

[Gibson486]

It's just open loop. There is feedback, but for temperature, not the circuit itself.

I am not sure what a powered metal inductor is. I had to use this big one from coilcraft though because it needed to be able to handle the 5 amps.

[Kleinstein]

Usually there is no classical LC filter to drive a TEC. It is more like a buck converter, with a diode to take over the current. On the output side the load from the TEC provides quite some damping to possible ringing. To a good approximation the TEC is an ohmic load. The TEC is not that sensitive to a little residual ripple of a few percent. So there is not that much need for an extra filter stage - if at all this would be to stop EMI and this more for the higher frequency ringing from the switches with a much smaller inductor. This way the load can still dampen the initial "filter".

Because the inductor needs to store quite some energy powder cores can be a good choice, as they allow a higher fields before saturation. This is especially true for higher power, where the ohmic loss is less of a problem than saturation. So powder cores can be smaller (my estimate would be a factor of 3 if there is no limit by loss) than a ferrite core.

Powder cores are made from a kind of iron powder with insulation in between. They are limited in frequency, moderate µ_r (which is ok for a choke),  but higher in saturation than ferrites. It is nothing exotic, but more like the normal choice for lower frequency power inductors.

[ogden]

The values I intend to use are an inductance of 33uH and a capacitance of around 270uF. I did this to minimize ripple at my frequency (~40khz PWM), so these values attenuate close to 60db at my frequency.

33uH have 8.3Ohms reactance at 40KHz - kinda small. I would pick at least 100uH (25 Ohms @ 40KHz), ensure that current is far from saturation.

[Gibson486]

The values I intend to use are an inductance of 33uH and a capacitance of around 270uF. I did this to minimize ripple at my frequency (~40khz PWM), so these values attenuate close to 60db at my frequency.

33uH have 8.3Ohms reactance at 40KHz - kinda small. I would pick at least 100uH (25 Ohms @ 40KHz), ensure that current is far from saturation.

Can not seem to find any that will not heat up a lot at 5 Amps.

[Kleinstein]

A TEC that is made for 5 A should have a resistance of some 2-4 Ohms. This is a significant damping for the filter. Just to avoid confusion about the circuit show a simple circuit diagram so we are talking about the same circuit.

At 5 A 33 µH is already a lot.   If the voltage to start with is not too high, I would expect a lower inductance, more like 10 µH for the main choke and if present some 1 µH for an additional RF filter. The capacitor may have to be larger, possible in the 1000 µF range, if really low ripple is aimed for.

[T3sl4co1l]

Two things you're missing --

1. TEC is basically a resistor, so at frequencies where we can ignore junk like lead inductance, yeah, it's a resistor.  So, at 1.5kHz, that filter capacitor is probably going to have a low Q.  Even lower still if you match sqrt(L/C) = resistance.

2. Even if you aren't varying it over any meaningful kind of time scale (i.e., 1.5kHz+), you still have startup transients.  The inrush surge current is given by -- guess what, the same ratio, sqrt(L/C).   That is, if you have a step change from 0 to 10V (or 20V at 50% PWM, it's equivalent when F_PWM >> cutoff frequency), you'll get a peak current on the order of (10 - 0) / sqrt(L/C).  So, for 33uH and 270uF, 28.6A (assuming the transistors deliver it at that point?).

I strongly recommend using a current mode control.  Even just for running a dumb TEC, it's a good idea to use some kind of current limited switch.  Preferably a fast enough one to do PWM + filter, and preferably the PWM control inside too, and cheap.  Fortunately, these exist: just grab a switching regulator.  Yes indeed, put in a full and proper regulated supply just to run a TEC.  Why not?   (If you're scratching your head about control, just run a resistor from the control voltage input, to the feedback node.  The CV looks like extra output voltage, so more CV = less Vout -- it's an inverting amplifier, with gain set by the ratio of resistors.  If your source is ultimately PWM, an RCR filter will do.)

Tim

[ogden]

At 5 A 33 µH is already a lot.

Good point. When you have enough money to buy inductor - you better build current-regulated buck converter to power TEC. Obviously at higher than 40KHz, frequency.

[Gibson486]

I thought about using a switcher to control this, but all the circuits I saw for this look huge. Granted, I did a search with Texas instruments power supply configurator.

[ogden]

I thought about using a switcher to control this, but all the circuits I saw for this look huge.

In what sense they look huge? Some regulators are not much bigger than mosfet transistor+driver needed for PWM. Just an example:

[Gibson486]

I just looked at the amount of components and the area spec it gives. Perhaps they do a bad at conveying this?

Edit.....my tec is 24V and my supply is 24V....

[T3sl4co1l]

If you forgot to include the area of your existing filter, it may look overly significant...

Tim