The big TEC driver thread!

[j4cbo]

Possibly do via the PWM output using PB5 (Reset) as an analog in for reading a set point via trim pot setting 0-5V. Hey Wayne, is this power up Reset circuit necessary? Some reading I’ve done says no? If we could use this input for current limiting.

The external power-on reset circuit is not needed for the AVR. However, be very cautious with reassigning the reset pin - it's also used to enable programming mode. There's a fuse in the AVR that can be set to allow use of RESET as a general-purpose I/O pin, but once that's done, the only way to reprogram the part is with high-voltage programming (which far fewer people have programmers for).

If you want to move to a bigger part, I'd suggest an ATtiny24. It only costs a few cents more than the '13, but has 14 pins rather than 8.

[dave]

Actually the thougt was a little more simple. A ref. voltage of 0-5V on an input (PB5) determins/limits the max % PWM after the PI(D) loop on the output. For example, since the H-Bridge outputs 3A at 100% PWM the 50% should = 1.5A. I don't think anything more would be neccesary.

Does the H-Bridge current limit? From the data sheet it doesn't really look like it does.

[dnar]

Actually the thougt was a little more simple. A ref. voltage of 0-5V on an input (PB5) determins/limits the max % PWM after the PI(D) loop on the output. For example, since the H-Bridge outputs 3A at 100% PWM the 50% should = 1.5A. I don't think anything more would be neccesary.

You need to think about the ramifications of doing this. Limiting the max PWM % is only throwing away resolution.

Why would you want to limit the full PI(D) control range?

If there are to be any adjustments, they should be:

1. Setpoint
2. Proportional gain
3. Integral gain

BTW, your theory of H-bridge output is flawed. At 50% PWM the output current is not determined by the bridge, rather a function of V and load. TEC loads vary with temperature differential also. You can't simplify and say X at 100%, X/2 at 50%.

The external power-on reset circuit is not needed for the AVR. However, be very cautious with reassigning the reset pin - it's also used to enable programming mode. There's a fuse in the AVR that can be set to allow use of RESET as a general-purpose I/O pin, but once that's done, the only way to reprogram the part is with high-voltage programming (which far fewer people have programmers for).

If you want to move to a bigger part, I'd suggest an ATtiny24. It only costs a few cents more than the '13, but has 14 pins rather than 8.

Agreed. It is pretty safe to reuse the RESET/SCK/MISO/MOSI pins as outputs, but not inputs. In fact the original KEG controller used a pair for debugging the timer period and reporting firmware version at boot time.

EDIT: I have recieved NO feedback regarding the new PI firmware. Once I get some feedback, I'll release whatever changes are necessary including the now implemented EEPROM setpoint adjustment. I have also been experimenting with adding derivative control for full PID, it does make the response time faster.

[Solarfire]

Ok.. Totally revised the H-Bridge version to implement the Tiny24.

Current sense (I_SENS) of the H-Bridge fed to tiny24 for current monitoring (4863mV = 3A of output).

Trim pot implemented fed to the tiny24 for realizing current limiting via software influence on the PWM out

Unless there’s a better idea to current limit the bipolar current of the H-Bridge?

EDIT: I have recieved NO feedback regarding the new PI firmware. Once I get some feedback, I'll release whatever changes are necessary including the now implemented EEPROM setpoint adjustment. I have also been experimenting with adding derivative control for full PID, it does make the response time faster.

@Wayne
I haven’t been able to do any testing on the Original board because I broke my F**in shoulder the end of December. That’s why I’m sitting here doing layouts and simulations (things I can do with one hand) and keeping the dust from settling on all the laser goodies lyin around here.. shit. I’ll probably be out of action for about another 2 weeks.


So take a look at the revised schematic, especially the pin assignments of the tiny24. If it looks good I’ll get a couple of PCBs prototyped and if you’re up to it I’ll send you one for testing.

What is frequency of the PWM out, need it to dimension the inductor?


Cheers!

[LaNeK779]

you could all be speaking in an ancient dialect from the big red dry plains of mars, i couldn't even tell. but all this tec-tech talk, i really like!!!

i have the growing suspicion that a bad-arse-keg-controller has emerged from the darkness???? mmm???

[dnar]

@Wayne
I haven’t been able to do any testing on the Original board because I broke my F**in shoulder the end of December. That’s why I’m sitting here doing layouts and simulations (things I can do with one hand) and keeping the dust from settling on all the laser goodies lyin around here.. shit. I’ll probably be out of action for about another 2 weeks.


So take a look at the revised schematic, especially the pin assignments of the tiny24. If it looks good I’ll get a couple of PCBs prototyped and if you’re up to it I’ll send you one for testing.

What is frequency of the PWM out, need it to dimension the inductor?

Cheers!

Hi mate, I had a quick look at the schematic, a few quick comments (I don't have much time at the moment to do a full review):

1. Don't use R10/R11, this set point can be scaled and range limited in software.
2. LD-Inhibit should not connect directly to the MCU! Use an NPN in open-collector configuration. - Can someone confirm OC output is ok for the inhibit loop?
3. LM2940 is overkill. I would prefer a smaller regulator with precision output.
4. Why not use an LT1014 quad op-amp, and use 2 op-amps as voltage followers to buffer the trim-pot setpoints?
5. Connect "-ve in" to "op" of any unused op-amps otherwise they will oscillate.
6. As for PWM frequency, this can be adjusted to suit. Typicall I have used 100uH inductors (must be rated for more than 3A) and also some C across the TEC. In the past I have used 220uF and this must be rated to handle the ripple current - but this won't fly with a bipolar output!!!!
7. Add 100nF of C from pot wipers to GND.
8. When you do the PCB layout, make sure the TEC ground (IC4 pin 7) return has it's own heavy track to the 0V input terminal, and a separate ground track to the regulator/CPU/op-amps etc. This is to reduce ground noise. Due to the currents involved, ground noise issues could be severe otherwise.

I'll take a closer look when I get some time.

Good work!

EDIT:

• Use either pin 5 (OC0A) or pin 6 (OC0B) for the PWM drive signal. This allows the use of the TIMER0 fastpwm mode.
• R13 is going o cause issues as you have drawn it's GND connection. The ground current from IC4 pin 7 are going to be substantial. I am not convinced current regulation is necessary, but I guess the hardware can be implemented regardless.

[Solarfire]

Update:

1.Don't use R10/R11, this set point can be scaled and range limited in software.
R10/R11 are there for error detection (possible short to Gnd/VCC or open circuit) in the software voltages <100mV or >4400mV should be acknowledged as an error. The current set VR is also feeding off these now.

2.LD-Inhibit should not connect directly to the MCU! Use an NPN in open-collector configuration. - Can someone confirm OC output is ok for the inhibit loop?
Decided on an open collector here, it’s probably a little more versatile to what’s connected here. What do you think?

3.LM2940 is overkill. I would prefer a smaller regulator with precision output.
Went back to the LP2950 this should be precise enough and I don’t see anything pulling more than 100mA on VCC.

4.Why not use an LT1014 quad op-amp, and use 2 op-amps as voltage followers to buffer the trim-pot setpoints?
Mainly trying to save PCB-space

5.Connect "-ve in" to "op" of any unused op-amps otherwise they will oscillate.
Is done, didn’t know yet if I was going to use the 2nd amp or not.

6.As for PWM frequency, this can be adjusted to suit. Typicall I have used 100uH inductors (must be rated for more than 3A) and also some C across the TEC. In the past I have used 220uF and this must be rated to handle the ripple current - but this won't fly with a bipolar output!!!!
100uH is quite a lot, I did some simulations and depending on frequency I was getting good results with 15-47uH. What can the tiny24 output on frequency? Again here looking at saving PCB space.

7.Add 100nF of C from pot wipers to GND.
Do you think it’s necessary? Again here looking at saving PCB space.

8.When you do the PCB layout, make sure the TEC ground (IC4 pin 7) return has it's own heavy track to the 0V input terminal, and a separate ground track to the regulator/CPU/op-amps etc. This is to reduce ground noise. Due to the currents involved, ground noise issues could be severe otherwise.

Agreed..


EDIT:
9.Use either pin 5 (OC0A) or pin 6 (OC0B) for the PWM drive signal. This allows the use of the TIMER0 fastpwm mode.
Changed it.

10.R13 is going o cause issues as you have drawn it's GND connection. The ground current from IC4 pin 7 are going to be substantial. I am not convinced current regulation is necessary, but I guess the hardware can be implemented regardless.

As far as current limiting goes I think this may be absolutly neccesary, I'm looking a little closer at this issue at the moment?

Yeah! The drawing also go cleaned up some..

Currently the PCB will be about 50mm x 40mm!

Ahh.. nice workin with ja dude!

Cheers!

[dave]

Use an NPN in open-collector configuration. - Can someone confirm OC output is ok for the

I would have thought a fet would be a better choice.

[dnar]

Update:

1.Don't use R10/R11, this set point can be scaled and range limited in software.
R10/R11 are there for error detection (possible short to Gnd/VCC or open circuit) in the software voltages <100mV or >4400mV should be acknowledged as an error. The current set VR is also feeding off these now.

2.LD-Inhibit should not connect directly to the MCU! Use an NPN in open-collector configuration. - Can someone confirm OC output is ok for the inhibit loop?
Decided on an open collector here, it’s probably a little more versatile to what’s connected here. What do you think?

3.LM2940 is overkill. I would prefer a smaller regulator with precision output.
Went back to the LP2950 this should be precise enough and I don’t see anything pulling more than 100mA on VCC.

4.Why not use an LT1014 quad op-amp, and use 2 op-amps as voltage followers to buffer the trim-pot setpoints?
Mainly trying to save PCB-space

5.Connect "-ve in" to "op" of any unused op-amps otherwise they will oscillate.
Is done, didn’t know yet if I was going to use the 2nd amp or not.

6.As for PWM frequency, this can be adjusted to suit. Typicall I have used 100uH inductors (must be rated for more than 3A) and also some C across the TEC. In the past I have used 220uF and this must be rated to handle the ripple current - but this won't fly with a bipolar output!!!!
100uH is quite a lot, I did some simulations and depending on frequency I was getting good results with 15-47uH. What can the tiny24 output on frequency? Again here looking at saving PCB space.

7.Add 100nF of C from pot wipers to GND.
Do you think it’s necessary? Again here looking at saving PCB space.

8.When you do the PCB layout, make sure the TEC ground (IC4 pin 7) return has it's own heavy track to the 0V input terminal, and a separate ground track to the regulator/CPU/op-amps etc. This is to reduce ground noise. Due to the currents involved, ground noise issues could be severe otherwise.

Agreed..


EDIT:
9.Use either pin 5 (OC0A) or pin 6 (OC0B) for the PWM drive signal. This allows the use of the TIMER0 fastpwm mode.
Changed it.

10.R13 is going o cause issues as you have drawn it's GND connection. The ground current from IC4 pin 7 are going to be substantial. I am not convinced current regulation is necessary, but I guess the hardware can be implemented regardless.

As far as current limiting goes I think this may be absolutly neccesary, I'm looking a little closer at this issue at the moment?

Yeah! The drawing also go cleaned up some..

Currently the PCB will be about 50mm x 40mm!

Ahh.. nice workin with ja dude!

Cheers!

Cool. A few more thoughts.

PWM frequency, I'll have to check out the tiny14 and see what is achievable, but what I know from the tiny13 it should not be an issue to set the frequency to suit 47uH.

Re: the Interlock/Inhibit. If this was to be wired in series with say the Interlock of a Stanwax ILDA board and Flexmod P3 drivers, it would be best if we provided a pair of dry contact relay contacts for wiring in series with the Interlock loop? Dave, Dr Lava please chime in here!

Inhibit range, we should decide what the Inhibit window will be before finalizing the setpoint pot/ADC range. I would suggest +/- 3c would be a good starting point. This being the case, the NTC range should be able to swing +/-3c of the setpoint as a minimum. Not always true with your current NTC gain block and setpoint range limit resistors.

I don't see the pot resistors are necessary, I have not done such before and never had a problem.

Pot capacitors. Yes, absolutely. Add them please. They won't take up much more space.

Diagnostics. Why not add LED's for:

1. Under setpoint
2. Above setpoint
3. Current limit active
4. Inhibit state (directly off the MCU output).

The first 3 can all be run off the SCK, MISO and MOSI lines. Use high efficiency 3mm LED's @ 5mA for example with the MCU sinking (not sourcing).

[Solarfire]

Re: the Interlock/Inhibit. If this was to be wired in series with say the Interlock of a Stanwax ILDA board and Flexmod P3 drivers, it would be best if we provided a pair of dry contact relay contacts for wiring in series with the Interlock loop? Dave, Dr Lava please chime in here!
.

From what I understand the LD-Inhibit and the ilda interlock are 2 totaly different features. LD-Inhibit preventing the diode from lasing outside of a safe temp. level. Mostly implemented on boards with both LD-Driver and TEC-Controller on one board. More or less keeping the modulation input of the LD-Driver low while aktiv.
In this case I would suggest having a modulation in/out pin on the TEC-PCB which would either switch the out pin low or the in pin thru to the out pin via a relais contact thus making it suitable for any kind of LD-Driver with a modulation input.