Over-current Protection Circuit

[Penri]

Terry

I have a fairly busy morning this morning at WMT but we could talk a little later this afternoon, at around 16:00 onwards, I’ll call you if that’s ok.

Penri

[TerryJC]

OK.

[TerryJC]

I'm back on this activity after the work on the Lady Hanham hardware.

Back in October, I posted some diagrams taken from a Spice clone called QUCS (viewtopic.php?p=5312#p5312) and implemented this as shown, but had problems with breakthrough of the PWM on the DC O/P.

After a longish break I decided to implement just the integrator on breadboard with jumpers and I get exactly the same problem.

Here is the circuit as implemented on the breadboard:

Non-inverting_Integrator_Circuit_(as_built).png (19.97 KiB) Viewed 199 times

Any ideas as to why this circuit doesn't replicate the Simulator output?

If I can't crack this, I guess one way forward would be to buffer the PWM signal with a standard inverting Op Amp and then integrate that using the classic inverting Integrator.

[Penri]

Terry

It should act like a pretty good low pass filter, beginning to roll off at around 10Hz with a cut-off frequency of 50Hz.

Can you provide i/p and o/p screen captures of what you get, please.

Penri

[TerryJC]

See below. Traces at 10%, 40% and 50% PWM.

PWM_10PC.JPG (208.22 KiB) Viewed 193 times

PWM_40PC.JPG (203.37 KiB) Viewed 193 times

PWM_50PC.JPG (196.43 KiB) Viewed 193 times

One thing that I've realised; the circuit doesn't seem to draw any current from the Bench PSU (although to be fair it's a 30 A supply and the quiescent current in the LM358 is 700 uA, so it probably wouldn't.

Also. I now recall that we discussed this circuit back in October when you suggested the addition of an inductor, but that was to cure the effects of undershoot. That particular discussion went down a rabbit hole and the problem now seems a bit different.

I hope that all this makes some kind of sense.

BTW the penny dropped just now. My idea to use an inverting amp won't work because we would need a negative supply.

[Penri]

Terry

These don't look right; I can sort of rationalise the first two as the o/p DC level is increasing as the mark space ration changes but the last trace is totally wrong. When I say sort of rationalise I still don't understand the o/p perturbations on the rising and falling edges. If almost looks as if the op/amp isn't in circuit.

Can you try something, replace the active integrator in your circuit with a simple RC filter, a 100k resistor in line and 10u to deck would be fine, and see what difference that makes. You should get a flattish sawtooth.

Penri

[TerryJC]

Penri,

I'll have to do that tomorrow now.

[TerryJC]

Can you try something, replace the active integrator in your circuit with a simple RC filter, a 100k resistor in line and 10u to deck would be fine, and see what difference that makes. You should get a flattish sawtooth.

As discussed, I connected a completely standalone circuit. For the avoidance of doubt here it is:

Passive_Integrator_Circuit_(as_built).png (35.73 KiB) Viewed 187 times

• With R = 100 k and C = 10 uF, the O/P is DC that increases with PWM Percent.
• With R = 10 k and C = 1 uF, the O/P is also DC that increases with PWM Percent.
• With R = 1 k and C = 0.1 uF, the O/P is a shallow sawtooth superimposed on DC that increases with PWM Percent.

This is not too surprising because the Time Constant of the RC network in the three versions above are 1000 s, 10 s and 100 us respectively. The Period of the PWM is also 100 us.

So according to that, all I need to do is to slap a comparator onto the O/P of network 2 above and job done. It seems too simple.

[Penri]

A) I’m glad it worked
B) The active circuit should been at least as good at filtering as the 100k / 10uF combination but with a higher dc o/p level

I’m annoyed that I don’t know why the active circuit didn’t behave as it should, I have the components on their way to build myself one.

Terry can you work with what you have, if the output from the passive R/C combination to too small to allow decent discrimination by the comparator you could put in an op amp based gain stage to boost the signal level.

Penri

[TerryJC]

Terry can you work with what you have, if the output from the passive R/C combination to too small to allow decent discrimination by the comparator you could put in an op amp based gain stage to boost the signal level.

That's what I'd intended to do unless you came up with anything else. I'll add the comparator to the breadboard and if that works, transfer the components to the Veroboard.

I should be OK with voltage level because I can increase the value of R Out to get the range I need.