'Bother!' said optocouplers / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

'Bother!' said optocouplers

Move Thread

LAN_403

LithiumLord

Mon Jul 27 2009, 04:07PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Well, I finally found the reason of weird interference I kept picking up in my coil from time to time that got way stronger with the improved feedback. Apparently, the 'reason' was hidden in the optocouplers, and it has nothing to do with CMR. The situation goes like this - the drivers are inverse, meaning no LED current means the switch is conducting. Now the problem only happens on the on-state, shutting my fets off and clamping my power while the field is high. Any minor ringing may cause this as they are kinda sensitive to small currents - a stray spike in the bridge may start the oscillations. Now the question - how to fight it? I have the resistors on the motherboard presently, in the same box as the active clamp and PSU which is separated from the HV box. The interference is coming from the twisted pairs linking the LEDs to the socket the screened control cable comes into. I have three different ideas: either clamp the coupler diodes with antiparallel ones and add the negative threshold or just add the threshold or just add some caps. About the caps I just have no idea what value I need and if it will work at least, the other two methods are questionable though - I have no idea which is better among them. Could anybody give me a hand, as at least I want to finally make sure the new clamp method will work for ~2kVA coils? (at most - just watch the heavy sparks fly and blast things around :D)

edit: I guess I figured out what to do (none of the above), will implement it if nothing better in this thread will follow.

...

Mon Jul 27 2009, 06:25PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Have you considered that it is not spikes turning on/off the led (which I would fix with a small low pass filter) but rather just the di/dt from the bridge turning on/off? Most optocouplers aren't rated for that kind of abuse (few are in fact), so it might be worth checking out. If it were me I would make my own coupler with a led/photodiode probably using one of the u-shaped ir transmitter/reciever pairs used for non contact posisition sensing.

Dr. Dark Current

Mon Jul 27 2009, 06:35PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

... wrote ...

If it were me I would make my own coupler with a led/photodiode probably using one of the u-shaped ir transmitter/reciever pairs used for non contact posisition sensing.

I have a feeling these would be unusably slow...

Conundrum

Mon Jul 27 2009, 07:41PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

parallel resistors should fix that..

i used around 1K for the LED on my uni project, and 10K on the phototransistor.
-A

LithiumLord

Mon Jul 27 2009, 08:45PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

... - nope. I'm sure that that's not a CMR issue as the switches can only handle the same value at max, I checked the sheets. Also another sign this was not that issue is that you've probably seen the YouTube video with the coil performing properly, and as I tried now the presence of strikering that shields the primary and therefore reduces the ringing moves the problem threshold up. Also running the device open-box and moving my hand around the wires causes the same. Even more - the CMR-based switching has the phase impossible in this construct - the state causing the gate to go high also would cause the opto to activate the driver - and the driver is already active. And even more - the low-side switches exhibit the same problem.

@Conundrum - thanks a lot, I'll try it out then. Myself I also wanted to (and will) add ferrite rings on each pair to ensure no stray currents pass it, and add a small add-on: a diode/resistor pair bypassed with a cap to make the optocoupler ground be like 0.5v displaced. Guess those three tricks combined will remove the noise issue.

Dennis Rogers

Mon Jul 27 2009, 08:56PM

Registered Member #1837 Joined: Tue Dec 02 2008, 02:20PM
Location: NYC
Posts: 65

If you're sure it's the twisted pair did you ground the shield? Usually there is foil or copper around the twisted pair that can/should be grounded. If your fear is capacitive coupling/inductance in the pair the shielding is meant to divert that when grounded.

LithiumLord

Mon Jul 27 2009, 09:58PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

It's not a twisted pair you could think of - the controller outs a strip of hand-made 5-wire screened cable with a dense screen grounded at the side of the supply, the five wires (each switch and com one) all are rigged to a connector that outs 4 strips of twin wires twisted together, they should work but now I see they definitely need filtering. I guess putting a ferrite ring will make sure they carry the same currents and trim any noise therefore, the resistor will clamp the spikes and the -0.5v bias will ensure there's nothing to trigger the LED.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob