opto coupler with or without base connection

[Simon]

I'm in the process of making up an interface module between a control system and a very noisy motor driver. This will basically consist of an optocoupler. In the past the motor driver designer put a CNY17 opto inside the motor driver but this seemed to pik up noise from the driver and self turn on. So far we have thought that due to the very large current and electromagnetic emissions from the unit the opto input was picking enough of this up to turn the LED in the opto on or that is was capacitively coupling across.

But I have another theory, the CNY17 has a base pin for the output transistor, now I'm not sure why you would want the base of an opto available but i suppose there are some instance where it would be useful, in this case very much not. So what if the base pin is picking up enough "crap" to spuriously turn the opto coupler on. Perhaps I should be using an opto coupler without a base pin for the output transistor ?

[Zero999]

There are various potential uses for the base pin on an opto-coupler. One is speed. A Schotky diode could be connected between the base and collector, forming a Baker clamp. This stops the transistor from operating in the saturation region and speeds up the switch off time.

If you believe the base pin is picking up noise, you could connect it to the emitter via a resistor, say 10k to 100k.

What frequency is the opto-coupler supposed to be switching at? If it's fairly low, try a capacitor between the base and emitter. It will slow things down a lot but will short the emitter to base at high frequencies and reduce pick-up.

[Simon]

Looking around i have found mention or using the base to reduce the sensitivity. The frequency is low 100Hz to 1KHz and it's to transfer PWM so providing high and low transitions are roughly equal or so small they have no overall effect there won't be a problem.

I was thinking of picking an opto without an exposed base, we are looking to have this reintegrated into the controller so understanding it might help making it into one unit rather than a controller and dangling "box of tricks"

[T3sl4co1l]

Any guess on noise magnitude and spectrum?  Can't filter it?

6N136 is quite nice -- good and fast.  Except that it has that pesky base lead, which picks up huge amounts of noise.  (The datasheets will lie to you about this.  The Fairchild datasheet, I believe, says it's good for quite a high CM dV/dt, but at the massive voltage step of 5V or something!  Other ones have more realistic values, like < 1kV/us for a 100 or 1000V step.)

Fortunately there is help: the SFH6345 is identical, but the base pin is kept internal.  It's just as fast, but now it's good to 10kV/us at 1kV step!

Any of the fancy ones, like 6N137 or HCPLxxx, with the amp/detection logic inside, are good as well, especially if you need the speed.

You might also consider magnetic isolators by then (e.g., ADUM2201).  Beware that some of these are not guaranteed for DC operation (they communicate via pulses); the cheaper ones (e.g. Silicon Labs) don't make such a claim, and I wouldn't recommend they be used on DC-dependent codes (e.g., gate drivers).  Digital data with error detection/correction should be fine.

My experience with ordinary things like 4N35 or CNY17 has been: as long as you're using it at a modestly low impedance (say 1k or less -- need not be in saturating mode), with suitable filtering, it's too slow to care about common mode noise.  Two pin devices may be better or worse, since: although there's no base pin to pick up noise (a plus), you can't pull the dirty little secret of adding B-E resistance (47k or 100k being best IME) to speed it up, shunt noise and reduce CTR.

Tim

[Simon]

Well I do know that a CNY17 will work if not put inside the motor controller that drives up to 60A so huge current spikes, so i have a safe option for now to keep everyone happy but long term I'd like to be able to propose putting the opto back into the controller but done properly this time. The input to the controller is just a uC pin with a 2.2K pull up so no huge current demand but I would like to strongly clamp it to ground to stop noise coupling through.

There are also different CTR options for the CNY17, I suppose this is like a "gain" so more gain harder clamp but more sensitivity to noise.

Will a base pin allow me to get rid of more noise than a non base pin version will pick up ?

[T3sl4co1l]

Well I do know that a CNY17 will work if not put inside the motor controller

Alright.

that drives up to 60A so huge current spikes,

Whoa whoa, that doesn't necessarily follow.  Why does it spike so?

so i have a safe option for now to keep everyone happy but long term I'd like to be able to propose putting the opto back into the controller but done properly this time.

So, long term, you can change the design (or at least make recommendations)?  That sounds like a better question to me.  And has knock-on benefits of fixing EMC problems (that's Elephant-in-the-room Magnetic Compatibility, if you were wondering).

The input to the controller is just a uC pin with a 2.2K pull up so no huge current demand but I would like to strongly clamp it to ground to stop noise coupling through.

What VCC?

No filtering to a digital input is asking for trouble.  If it's sampled frequently and filtered internally, that can sometimes be worked around, a little.  But external analog filtering (and probably protection besides) is always better.

An isolation barrier isn't exactly front panel material: there, you have to worry about squishy, electrostatically charged meatbags poking at anything conductive.  But under high dV/dt conditions, it can be almost as bad as ESD, and filtering (and even protection -- smaller clamp diodes/TVS) is looking very welcome.

There are also different CTR options for the CNY17, I suppose this is like a "gain" so more gain harder clamp but more sensitivity to noise.

Will a base pin allow me to get rid of more noise than a non base pin version will pick up ?

I wouldn't worry about the CTR options, aside from which one is cheaper or more available, if it matters at all.  Higher CTR would allow you to add relatively more B-E loading, though.

Tim

[Simon]

Well I am told that while the average consuption can go up to 60A the actual current flowing around the motor windings (and therefor in the output stage of the controller) will probably be short pulses at a higher current.

The input is 5V, I can't do much for filtering other than put a small capacitor across the opto output and make sure every peice of wire is a twisted pair to minimize pickup.

[Simon]

well I'm ordering some CNY17-4, fair child do a version with and without the base.

[f5r5e5d]

there are many dozens of opto parts for reasons of needing differing specs

you can get built in Schmidt Triggers, Emi shielding for extreme dv/dt

and as mentioned quite a few new isolation techs have become commodity parts in the last decade

if making one-offs just pay up for the guaranteed functionality on the spec sheet - probably not worth it to tweak for lower cost parts until you're designing for thousands quanity production runs

[SeanB]

What about using TOSLINK modules, that way you have the LED inside the motor and a fibre optic plastic cable to the controller, no worries about noise pick up.