Linear MOSFET TEC circuit

[rcheshire (Rowland)]

I'm putting this up here out of interest, in my pursuit of a linear MOSFET driver circuit. This is a copy of a power LED dimmer with a few additional components on the TEC side.

I've been reading about Trench MOSFETs and the capability to run linear. All I have done here is smooth things out - naively, possibly. This works over a range of PWM frequencies. Some component values need adjusting otherwise.

3R resistor is the TEC.

The 5R resistor prevents reverse current in the TEC at start up - 1.67uA is the current through the 2.2nf cap and resistor according to my Android app, Every Circuit.

Not sure of the heatsink requirements yet.

Here it is running at 70% duty cycle, which is typical. The Teensy board PWM output at ~3.75Mhz 0 - 5V.

http://www.candlepowerforums.com/vb/...rent-Regulator

[Garbz (Chris)]

This circuit has me confused. The bottom transistor has its base and emitter shorted. In effect I don't think it would ever forward bias and thus doesn't actually have any function at all.

Am I missing something or would the circuit run the same if it were removed? It looks like it should almost be a classic linear constant current sink, but in the normal ones you see in the textbook there's a resistor between the base and emitter and the current ends up being ~0.7V / resistor.

I'm confused.

[bojan]

Yep. lower transistor is redundant here (unless it is so called parasitic transistor, part of the chip)

[rcheshire (Rowland)]

Ah! It should look like this. My mistake. I will have to calculate an exact R value. I've also dropped the 1M resistor between the PNP and NPN transistors, which was experimental.

[rcheshire (Rowland)]

This is the final iteration. Some minor changes from the previous. Mainly, the 100pf cap on the gate side is not required - the resistor value is also reduced as per Garbz IRF2804 circuit. The cap on the Drain side is increased to 3N3 to further restrict current reversal in the TEC.

The prototype helped me sort out the errors in previous versions as well as fine tuning. I've used a lot of resistors in parallel to provide a bit more safety given the current requirements of the TEC.

This circuit achieves a full 3C advantage over the hard switched version with no gate switching interference. Coupled with careful choice of active heatsink my DSLR cooling system achieves an additional 7C differential. 12V 6A TEC 27C differential, which means imaging at 0 on a warm night.

However, there is technically no low PWM state and the PWM Pin must be hard set LOW to prevent cooling during setpoint temperature and target PWM calculations - small change in the Arduino code.

Very happy with this.

[Garbz (Chris)]

Just wondering what you're trying to do here because I think you've fundamentally changed the operation of the circuit.

The first one (well second one) is a classic text book constant current sink which is chopped by your input wave.
The most recent circuit has you again using the mosfet as a high frequency switch.

Both will likely work, just thought I'd point out they are fundamentally different ways of driving the load.

[rcheshire (Rowland)]

Thanks Chris. I see what you mean. Interestingly it works and there is no interference. However, I notice some anomolies/inconsistencies with the design tool I'm using. Just last night while playing around with component values and input types it reverted back to a square wave with the settings in the image below. Not to blame the tools, because I am a rank amateur with electronics. As you have pointed out this is just a switching circuit. So this has caught me out.

I have another program which is spice based but a pain to use. I will try that and see if I can replicate the behaviour.

I have tried to avoid use of the second transistor in this case but I think the program is displaying the previous behaviour. Got quite excited but it looks as though I need to go back to the drawing board.

Still on L plates.

[rcheshire (Rowland)]

This is bare minimum logic driven PWM and as close as I can get to no discernable interference camera side. The MOSFET temperature runs below touch detection with a small heatsink - almost imperceptable unless you pinch the heatsink. A small capacitor (100pf - 1nf) could be added gate to source for comfort, but it doesn't really need it. Temp modulates +/-0.5C of setpoint.

A 22uf capacitor gate to source will remove noise to undetectable levels, but the mosfet will overheat, with a change in properties and cooling accuracy.

MOSFET is an NXP PSMN1R1-30PL . 1.4milliohms Rds on at 4.5Vgs. 1.3mohm at 10Vgs. The IRF2804 gets hotter driven the same way.

[nebulosity. (Jo)]

Good you've got it working Rowland

It's a nice simple looking design, something I could probably make

Does it have a potentiometer or something so you can adjust the output? I built a PWM the other day from a Jaycar kit and it was working ok, but now when I turn the nob nothing changes must have killed it.

Electronics is not something I've taught myself much about, so if I took your diagram into Jaycar or someplace, would they be able to get the parts I need?

Cheers
Jo

[rcheshire (Rowland)]

Hi Jo. If you want hands off regulated cooling I can show you how to setup a microprocessor - code is complete and working. If you want to build a 555 timer PWM circuit there are a number of tutorials online and all the parts are available at jaycar. Just add my circuit by connecting your pwm output to the MOSFET gate. Importantly, make sure you earth any metal such as cold finger, casing and camera framework first.

I will draw up a schematic to show my complete set up.

EDIT: SPICE based circuit. I've changed the snubber design to the gate adding a small capacitor. OSC output Red channel, A Blue channel B

EDIT: With PNP transistor switching higher gate voltage as per Chris's spice circuit - not tested.

[Garbz (Chris)]

Originally Posted by nebulosity.

I built a PWM the other day from a Jaycar kit and it was working ok, but now when I turn the nob nothing changes must have killed it.

Is it stuck fully on?
Was it working before?
Can you post the schematic?

I'm sure we can talk you through fixing it. Heck I'm sure we can talk you through supercharging it and adding racing stripes too

[nebulosity. (Jo)]

Thanks for the schematic Rowland I'll see how it goes.

[nebulosity. (Jo)]

Originally Posted by Garbz

Is it stuck fully on?

Yes, it's stuck on full voltage.

Was it working before?

Yes I had it working well controlling my dew heater.

Can you post the schematic?

Yep, here's what was in the instructions.

Cheers
Jo

[mikerr (Michael)]

The "Silicon Chip" article is partially shown here.....

http://archive.siliconchip.com.au/cm...6/article.html

Michael.

[rcheshire (Rowland)]

Hi Jo.

It will be less of a headache to use a linear supply and map temperature differentials against voltage. That way you can make adjustments through the imaging session to account for night time temperature reduction.

Have you checked your soldering. Is there a chance you have damaged the speed control pot.

I am still working on refining the design specifically for DSLR TEC temperature control. I have followed all the advice given me along with lots of reading about soft and delayed switching. All helpful, except that DSLRs are very sensitive to any switching in the mosfet.

The circuit posted earlier is very quiet, but still needs improvement.

Short of strapping a 10 - 22uf capacitor across the pins to absorb the switching spikes I have not yet found a quiet low energy loss solution.