Electronics > Projects, Designs, and Technical Stuff
relay recommendation for insulation test scanner card
(1/1)
rounin:
I've been needing to measure the insulation leakage current for an electrical connector.
I've rigged up a little insulation test setup with a HP 3457A, a Fluke 415B power supply, a 100k ohm resistor, and a safety crowbar gas discharge tube. I'm getting voltages in the 1uV - 30mV range across the resistor (~300GOhm - 1GOhm), more typically 5uV - 50uV. Test voltage is 30V - 500V, surviving 1000V would be nice.
I'd like to make an 8 channel relay scanner card so I can do all the combination of pins in the connector. Maybe first by throwing some DIP switches, later by fully automating w/ GPIB, the input scanner card on the 3457a, and some GPIO to control the relays. I want to be able to apply voltage to and measure current out of any combination of pins, I'll probably have 16 SPST relays.
Do I need anything special for the relays? Like low EMF? The signal is down in the uV range, but since the sense resistor is local and not going through the relay so I don't know if it matters.
I was considering these ones, high voltage rated reed relays. The relays are not directly switching the full voltage since the DUT is between them, but in case of a burn-through I was looking for relays that could handle the full voltage.
Reed:
* Meder SHV05-1A85-78D4K
* Meder LI05-1A85
* Pickering 100-1-A-5/2D
LI05-1A85 looks nice because it has very low leakage, 10Tohm. I'm leaning towards this one, although it is not low emf and is not shielded.
It doesn't need to switch super fast, doing 5 samples per test point at 100 PLC, so maybe will switch every 5s-10s. i'd like to be able to switch at a few Hz.
* Should I use a reed relay? Which do you like for 1000V 0A?
* Should I get a special low EMF relay? Does thermocouple effects matter?
* Should I use a more normal mechanical relay? Maybe latching? Favorites?
Circuit sketch attached. The TP are the pins of the connector, so the sketch is measuring the resistance of pin 1 to pin 2.
ninux:
Hi, I'm not very much of an expert on relay technology but there are two things I would have an eye on when choosing your relay, as I had issues with this in past projects.
1) When dealing with HV in a lab / test environment or anywhere with potential of electric shock or dangerous failures, try to use a force guided relay. With this, you can check the relay status during operation. This is also very helpful for automated testing as you will have reliable feedback signals. Here a generic example for such a device: https://image.schrackcdn.com/katalogseiten/k_schueen6_85-88_en.pdf
2) Most relays rely on a minimal current to enable good contact and it is different for the specific contact material. As an example, I used a relay similar to the above mentioned force-guided relay, where I used one of the contacts as a feedback for the supervision controller. However, after some testing and discussion with the manufacturer, I learned that the minimal current for a reliable contact is 20 mA. But this was for a LV feedback, I don't know if this is completely different for HV. A workaround to this problem can be an extra capacitor to implement a high current when closing the relay.
Maybe have a look at daves video "#544 - Fluke 5450A Resistance Calibrator Teardown", there are some relays in the resistance calibration circuitry. I guess they have excellent contacts (http://www.farnell.com/datasheets/1486738.pdf).
rounin:
Hmm. those coto ones in the 5450A look nice, 10^14 ohm! I poked around the coto catalog some, and wan't able to find a good balance of shielded and ~10$ ea. I didn't really see any instrumentation sized force guided relays I liked. I guess using a 2-form-a or changeover isn't quite the same thing from a safety perspective since a 2-form-a could jam independently.
I'm thinking about using these to switch each pin, 500V hot switch 1000V standoff
https://www.pickeringrelay.com/product/100-1-a-5-4d/
And this to switch the upstream input voltage, 1000V switch, 1500V standoff (or might use the 3kv version, but I really only see this growing to a 700V-1000V tester)
https://www.pickeringrelay.com/product/104-2-a-5-2d/
Can switch off the input voltage when above 500V using the 104 series, since the 100 HV part can withstand 1000V but only switch 500V hot. Then could reconfigure the per-pin switches cold and turn the input back on.
Both are shielded so I can pack dense, and are min 1Tohm leakage - better leakage than the SHV series, can pack denser than the LI series.
rounin:
I probably just need something quality wise in between the 5450A's and a generic clunker. Don't need to be scientific level accurate, just be able to say "its more than a few Gohm" with confidence. The spec to pass will probably be in the hundreds of MOhm.
rounin:
--- Quote from: ninux on July 29, 2020, 07:07:17 am --- I learned that the minimal current for a reliable contact is 20 mA. But this was for a LV feedback, I don't know if this is completely different for HV. A workaround to this problem can be an extra capacitor to implement a high current when closing the relay.
--- End quote ---
Some of the Pickering relays are qualified for "cold" low current switching (model 100 with switch 2, unfortunately I want switch model 4 for HV). This definitely seems to be a thing to watch out for... I am a little worried. I could get the cold switch model and always turn off input voltage. Risk is a relay could arc over if the connector material breaks down since that model is only 220V standoff.... Not hazardous to a operator I guess, but would be annoying.
Navigation
[0] Message Index
Go to full version