LM311 comparator setting

[Al (astral highway)]

Hey folks, I want to sample the current in a circuit at a high sample rate via a current transformer and burden resistor , full wave rectify the output and pass it to an op amp as a comparator.

The sample rate will be set by an RC time constant and will be probably 5 cycles How do I calibrate the threshold setting to trigger a pulse from the comparator? This pulse will pass through some more logic and shut down a gate driver above a certain preset current in the load. I might have two or three settings once I have established the character of the IGBTs at the sharp end.

I have a good circuit , it’s just the calcs I need to get hold of. I know I could do this empirically but I am looking for a short-cut !!

[kalee20]

You want the bridge rectifier to be BETWEEN the Current Transformer and the burden resistor, so you don't have diode drops to worry about affecting accuracy.

I'm not sure about the rest of it - having difficulty understanding! Can you put up a sketch of the circuit? Are you using the LM311 strobe pin?

[Al (astral highway)]

Originally Posted by kalee20

Can you put up a sketch of the circuit? Are you using the LM311 strobe pin?

Hi Peter, I'm out and about at the moment, I'll try and get something posted up this evening.

[G6Tanuki]

I'd have the comparator sampling the DC from the bridge continually - then use a bit of 4000-series CMOS to gate the comparator's output with your 'sample pulse' R-C-oscillator for your output.

You should even be able to do both the comparator-gating and RC-oscillator sample-pulse-generator in one 4001 chip and still have a gate or two left over.

[Handwavily glosses over the need to have a Schmitt Trigger in there somewhere to ensure the output of the comparator is truly logic-0 or 1 because CMOS can behave strangely and consume lots of power if the gates are biased into their linear region]

[MrBungle]

LM311 oscillates around the trigger point as do all comparators. Make sure you add some hysteresis.

[Al (astral highway)]

Originally Posted by kalee20

You want the bridge rectifier to be BETWEEN the Current Transformer and the burden resistor, so you don't have diode drops to worry about affecting accuracy.

Hey so I drew this before your comment about the bridge before the burden resistor... here is a simple sketch.

The actual build will also have some visual indicator of overcurrent.
The current transformer is clamped to a big resonant inductor at rectified mains voltage. Average current is only 8A but pulses are potentially huge.

The shutdown logic disables the IGBT drivers.

[Ed_Dinning]

Hi Al, if you Google Comparators it should take you to the likes of the Nat Semi site where there is usually a downloadable app note that will tell you all about the wrinkles to use with comparators. If set up correctly you could have a switch for current limit levels. the addition of hysteresis (positive feedback) greatly reduces false triggering in the presence of noise.

Ed

[Al (astral highway)]

Thanks, Ed, good tip !!!

[Argus25]

Hi Al,
I could give some advice on this but not everyone might agree. If you have such a slow sample rate , there probably isn't any need for a fast comparator which as noted would need hysteresis added, typically a few meg ohms from output to positive input.

I would recommend a low slew rate op amp instead, which will be intrinsically noise proof and not give you any high frequency issues. Actually a vintage 741 works well if you don't need rail to rail output swing. Another good OP amp with a low slew rate and rail to rail input and outputs is the OP295 or 495(quad).

When one wants to make a "comparator" the first inclination is to reach for a comparator IC, but these can be difficult to tame and have high frequency stability issues that can be awkward. So a low slew rate OP amp can be a much more stable approach unless you need a very rapid response.

[Al (astral highway)]

Hey Hugo, great advice, thank you’

I will be firing the device with an interrupter at between 40-200pps , with 40-250uS on time .
I think i’d like to pick up overcurrent within 10 mS, will a bog standard OP do the trick?

[Argus25]

Al, 741 too slow, OP295 good to about 75KHz, maybe just not enough. From the figures you gave, just use a standard LM358 would be fine.

I looked up some more OP amp specs the LF351 would be fine, plenty fast enough for the application.

[MrBungle]

Just beware that standard opamps have a recovery time in addition to the slew rate when driven into saturation. This is individual device specific. Unless they are RRIO they won’t get anywhere near the supply rails either. Comparator with frequency compensation network is much better.

[Argus25]

Originally Posted by MrBungle

Just beware that standard opamps have a recovery time in addition to the slew rate when driven into saturation. This is individual device specific. Unless they are RRIO they won’t get anywhere near the supply rails either. Comparator with frequency compensation network is much better.

Mostly I agree, but for a lot of applications where the delays cause by saturation doesn't matter, the op amp with a low slew rate is better behaved in experimental circuits.

[Al (astral highway)]

Just to clarify , the operation frequency is still around 325KhZ. The impulse generator is just a form of on-off time and pulse width controller...

[kalee20]

I've used an op-amp as a comparator (spare section of AD712). I've also used a comparator as an op-amp (spare section of LM239).

All worked OK, though the applications weren't requiring super speed or precision.

What's the difference anyway? Well, they are optimised for different things. The op-amp may have picoamp input currents, but may not be able to withstand more than a few hundred millivolts between its inputs. The comparator spend most of its time either saturated hi or lo, so will have no internal stability capacitors which could add delays, but may do silly things during the transition, like draw a gulp of current, or oscillate.

Argus's suggestion of an LM358 is good! It copes with inputs right down to the negative supply rail; it's tame; it's cheap; it's low power consumption. In Astral's application speed is not an issue (witness the big capacitors on the input).

As to the original question, I'm still a bit confused as to the need. The circuit as sketched out, will change state after a long time with a small overcurrent, and more quickly with a large overcurrent. But generally, you want an instantaneous response to a large overcurrent (the small overcurrent can have its long delay to avoid 'nuisance tripping). This would require two comparators, with different trip thresholds.

[Al (astral highway)]

Originally Posted by kalee20

I'
Argus's suggestion of an LM358 is good! It copes with inputs right down to the negative supply rail; it's tame; it's cheap; it's low power consumption. In Astral's application speed is not an issue (witness the big capacitors on the input).

As to the original question, I'm still a bit confused as to the need. The circuit as sketched out, will change state after a long time with a small overcurrent, and more quickly with a large overcurrent.

Which is a good thing, right? We don't want it clamping for every small overcurrent.

Peter, the idea is that different IGBT's can handle different pulse currents, apparently way above spec for short runs. The idea is to find the threshold where maximum energy goes into the resonant circuit without busting the half-bridge. Whats' good for an IGBT brick will bust a smaller one. Some people are running these things at 500A peak pulses on a full bridge.

You'll be interested to see that I'm using one of the amorphous alloy toroids you sent me a few years ago, one of the grey ones!