Why is this opto-thyristor not turning off (commutating)?

Thread Starter

BigHairyAnimal

Joined Dec 23, 2016
17
Hello All About Circuiverse,

I'm working on some line voltage motor control and need to drive an AC relay coils from an isolated DC source. I know that there are other options (SSR's, electromechanical relays, etc.) but I am looking for small and inexpensive as part of a commercial product. I thought I had found the perfect solution in the IXYS CPC1972 (datasheet here). From the datasheet: "The CPC1972 is an AC Solid State Switch using optical coupling with dual power silicon controlled rectifier (SCR) outputs to produce an alternative to optocoupler and Triac circuits."

upload_2018-11-23_13-49-31.png

The supposed advantage of antiparallel SCRs over a triac is that they don't have a limited dv/dt for commutation and thus don't need snubbing - an essential characteristic when driving inductive loads. However, I just received the parts and tried to switch the 120VAC coil of a G7L relay (data sheet here). Once turned on, I can remove the input LED current, and the load stays on! It won't commutate and turn off! It will turn-off a few seconds later sometimes but it's very inconsistent.

In my troubleshooting, I disassembled one of the relays and found a simple rectification circuit ahead of the relay coil (capacitor across line + bridge rectifier). That's not too unusual since the manufacturer can just use a DC coil and drive it with the ac signal. I checked the current waveform to ensure it was actually crossing zero and it was (waveform below).


This breaks my mental model of SCRs and seems to defeat the whole purpose of this device. Please help me to understand and remedy the issue.
 

Reloadron

Joined Jan 15, 2015
4,573
What exactly is the waveform? I ask because the waveform is an awfully small signal level. I also agree that a full circuit diagram would likely help a great deal.

Ron
 

Alec_t

Joined Sep 17, 2013
10,429
If the coil is fed via a bridge with a reservoir capacitor it's possible that at the voltage zero points there is sufficient residual current to keep the SCRs conducting, if they have a very low holding current.
 

Thread Starter

BigHairyAnimal

Joined Dec 23, 2016
17
If the coil is fed via a bridge with a reservoir capacitor it's possible that at the voltage zero points there is sufficient residual current to keep the SCRs conducting, if they have a very low holding current.
Hi @Alec_t - But if each SCR is only conducting only one half cycle, and the waveform shows the current going negative, wouldn't that mean that one of the SCR's is reverse-biased and would naturally turn off before the next cycle?
 

crutschow

Joined Mar 14, 2008
23,487
But if each SCR is only conducting only one half cycle, and the waveform shows the current going negative, wouldn't that mean that one of the SCR's is reverse-biased and would naturally turn off before the next cycle?
That would seem to be true, which makes it a puzzle as to what exactly is happening. :confused:
 

Alec_t

Joined Sep 17, 2013
10,429
That G7L relay is a somewhat unusual animal in that, as per the datasheet, it is designed to withstand temporary loss of mains power for its coil. It's nominal coil current is about 20mA, so if the internal capacitor has a reasonably high capacitance you can expect the relay to hold in for a significant time after the opto's SCRs turn off.
 

ebp

Joined Feb 8, 2018
2,332
It looks very much like the SCRs in the coupler, in spite of their fairly high dV/dt rating, are being triggered by rate of rise of the voltage. There may be no choice but to use a snubber.

What you actually need to look at is the voltage across the coupler output.
 

Thread Starter

BigHairyAnimal

Joined Dec 23, 2016
17
That G7L relay is a somewhat unusual animal in that, as per the datasheet, it is designed to withstand temporary loss of mains power for its coil. It's nominal coil current is about 20mA, so if the internal capacitor has a reasonably high capacitance you can expect the relay to hold in for a significant time after the opto's SCRs turn off.
@Alec_t - Interesting thought. In looking at it, the capacitor is on the AC side of the bridge rectifier, which is opposite from what I would expect (maybe a power factor thing?). I also tried turning off the relay with just a plain old switch and it turns off as soon as I release the switch. I'm perplexed.

 

Thread Starter

BigHairyAnimal

Joined Dec 23, 2016
17
It looks very much like the SCRs in the coupler, in spite of their fairly high dV/dt rating, are being triggered by rate of rise of the voltage. There may be no choice but to use a snubber.

What you actually need to look at is the voltage across the coupler output.
Hi @ebp - this is where my understanding of thyristor behavior gets less solid. As I understand there are two dv/dt ratings.
  1. Static dv/dt - if this is exceeded the thyristor may turn itself on.
  2. Commutating dv/dt - if this is exceeded the thyristor cannot turn itself off. This exists for triacs but should not exist for SCRs (methinks).
Thinking that through then, even though this coupler has a very high static dv/dt, perhaps it's being commutated (turned off) successfully then turned immediately back on because of the voltage? I pulled some more waveforms to help ...

This waveform captures the coupler turning off (nearly 10 seconds after I removed the LED current). CH1 is voltage across the coupler output.


This waveform shows the behavior of the coupler when the LED current is removed. CH2 shows the LED trigger current going away and CH1 shows the voltage across the coupler output. When I look very closely I can see that the amplitude of CH1 goes up very slightly when I release the trigger button. Otherwise, it's unchanged.


A closeup of the +/- transition point for the voltage across the optocoupler output. Nothing crazy here.


More info and still doesn't make sense. Your continued collective thinking is much appreciated.
 

Alec_t

Joined Sep 17, 2013
10,429
In looking at it, the capacitor is on the AC side of the bridge rectifier, which is opposite from what I would expect (maybe a power factor thing?).
From the post #14 pic, it looks as though that blue cap is being used as a volt-dropper. What are the markings on it?
 

Thread Starter

BigHairyAnimal

Joined Dec 23, 2016
17
Update!
  1. I got ahold of IXYS IC (different than IXYS, apparently) and they informed me that the datasheet representation of the CPC1972 is not representative ... and that the CPC1963 is actually the one with discrete SCRs in it. Very frustrating, since the datasheet for the CPC1972 clearly shows and states two distinct SCRs. Ugh. I ordered the CPC1963 to test.
  2. I got ahold of OMRON Industrial (different division entirely then OMRON Components), and they linked me to a different datasheet for the same relay. This datasheet shows a schematic of the AC coil circuit on the 2nd page of the pdf. The schematic shows a component on the line side of the rectifier. I had originally assumed the component to be a capacitor, which seems to be wrong - finally some new information! The guy on the phone mentioned it is a varistor but that symbol is unfamiliar to me and not showing up on google.
Schematic from datasheet:
upload_2018-11-26_14-41-42.png

Thoughts on what that component is and why it's there? If it's a varistor - is it dropping more voltage as the waveform peaks? That would make sense, given the square-wave-esque current going into it (image in post #15).
 

ebp

Joined Feb 8, 2018
2,332
The markings (0471 K7) suggest it is a 470 V ±5% 7 mm varistor. I have no idea who it might be made by. It is probably there to protect the bridge rectifier from overvoltage transients.

The square wave current is due to inductance in the coil and the bridge rectifier. It appears that the inductance is high enough that the current changes little in the course of each half cycle, so the current for each successive half cycle essentially starts at the same absolute value of current that the previous half cycle ended with.

I'm beginning to wonder if dV/dt simply isn't visible on your scope image - lost to inadequate sample rate. You may need to trigger on the edge of the current waveform and turn up both horizontal and vertical to get some decent resolution around the transition. "Infinite persistence" may help, though it will probably result in smeared edges.
 
Top