Author Topic: passing signals over galvanic isolation without an opto coupler  (Read 2639 times)

0 Members and 1 Guest are viewing this topic.

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
passing signals over galvanic isolation without an opto coupler
« on: October 17, 2019, 07:54:05 am »
I am working around an application where PWM signals need passing through a galvanic isolation barrier. Opto couplers seem to have piss poor leakage at higher temperatures and the manufacturers short of lying won't put all the specs on the data sheet. A Vishay manufactured CNY17 that is rated for an ambient of 100C only has specifics given up to 75C.

Is there another way? Various data protocols have dedicated isolated parts but I need something that is generic enough for PWM.
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 4136
  • Country: gb
Re: passing signals over galvanic isolation without an opto coupler
« Reply #1 on: October 17, 2019, 07:57:55 am »
Might be possible to use a pulse transformer if you don't need 0-100% duty cycle

 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #2 on: October 17, 2019, 08:25:00 am »
Um, well we don't usually use 0 or 100%. We usually go from 10% to 90% to distinguish between a desired drive an a lack of signal. How well do these work in general digital electronics passing 5V PWM at a few hundred Hz to a kHz?
 

Offline mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
Re: passing signals over galvanic isolation without an opto coupler
« Reply #3 on: October 17, 2019, 08:27:31 am »
There are plenty of magnetic isolator products out there ( NVE and Analog Devices come to mind)
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #4 on: October 17, 2019, 08:29:21 am »
There are plenty of magnetic isolator products out there ( NVE and Analog Devices come to mind)


Are they all called pulse transformers? Basically what sort of search terms lead to what I am looking for, what do people call these things?
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 4136
  • Country: gb
Re: passing signals over galvanic isolation without an opto coupler
« Reply #5 on: October 17, 2019, 08:31:39 am »
Simple answer for a varying frequency and duty cycle is that I don't know, you'd need to look at the data sheets but  I think 'a few hundred Hz' might be an issue for pulse transformers (the ones I've used were ferrite cored) so maybe something like the P1200 line isolation transformer from a modem would be suitable?

https://www.farnell.com/datasheets/1801043.pdf

 
« Last Edit: October 17, 2019, 08:34:05 am by CJay »
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 21226
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #7 on: October 17, 2019, 08:55:22 am »
I wonder how good these things are with EMC
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3382
  • Country: gb
Re: passing signals over galvanic isolation without an opto coupler
« Reply #8 on: October 17, 2019, 09:17:06 am »
A Vishay manufactured CNY17 that is rated for an ambient of 100C only has specifics given up to 75C.

The datasheet seems to show all the usual temperature sensitive characteristics up to 110C, which parameters are missing?

Is C-E leakage really a problem for a PWM application?  For the CNY17 it's still only around 1uA at 100C so unless you want to run at very low collector currents it's not going to have a great impact on your logic threshold noise margins.  If required you could remove any shift in operating point by following the opto with a data slicer.
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #9 on: October 17, 2019, 09:26:13 am »
that is a different datasheet from the one i got off RS. Overall looking at opto couplers I have not found any that really shine at high temperature with specifications being vague or missing. Basically I think we need something else. Opto's are good but things have moved on and so have the demands of industry.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: passing signals over galvanic isolation without an opto coupler
« Reply #10 on: October 17, 2019, 09:39:01 am »
Only problem with these digital isolators is the time quantization, but for khz range PWM I don't think the 325 ps jitter for Silabs matters.
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3382
  • Country: gb
Re: passing signals over galvanic isolation without an opto coupler
« Reply #11 on: October 17, 2019, 09:53:15 am »
that is a different datasheet from the one i got off RS. Overall looking at opto couplers I have not found any that really shine at high temperature with specifications being vague or missing. Basically I think we need something else. Opto's are good but things have moved on and so have the demands of industry.

The datasheet on RS relates to slightly different part number (CNY17., CNY17G vs CNY17)

That a pretty vague statement, what demands do they not meet?  If it's something concrete like reliability then seeking a better solution with the trade-off of increased complexity and/or expense is perfectly justified, but it sounds like you want to try something else out of interest.
« Last Edit: October 17, 2019, 09:58:24 am by mikerj »
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #12 on: October 17, 2019, 11:39:05 am »
Only problem with these digital isolators is the time quantization, but for khz range PWM I don't think the 325 ps jitter for Silabs matters.

I think we can just about call that of no significant influence, worse case is with PCM where ns is as small as i would be concerned about and with PWM I'm only worrying about µs at a few hundred Hz
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #13 on: October 17, 2019, 11:44:33 am »
that is a different datasheet from the one i got off RS. Overall looking at opto couplers I have not found any that really shine at high temperature with specifications being vague or missing. Basically I think we need something else. Opto's are good but things have moved on and so have the demands of industry.

The datasheet on RS relates to slightly different part number (CNY17., CNY17G vs CNY17)

That a pretty vague statement, what demands do they not meet?  If it's something concrete like reliability then seeking a better solution with the trade-off of increased complexity and/or expense is perfectly justified, but it sounds like you want to try something else out of interest.

Not out of interest. I am not comfortable with opto's after a few failures and looking at opto couplers as a whole (I went through a few Datasheets) I would prefer not to use them. The leakage over 80C changes from 2nA to 200nA. They are ideal to cheaply solve some problems but I do not feel comfortable with them in the application I am involved with and with every other part rated to 125C the opto's are now the weakest link.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7678
  • Country: ca
Re: passing signals over galvanic isolation without an opto coupler
« Reply #14 on: October 17, 2019, 12:38:45 pm »
Optocouplers aren't good with age or temperature, almost none of them are AEQ.

IC's using digital isolation with tiny air-core transformers are sexy but a challenge for EMC. Their high carrier frequency is not mitigated in any app notes or datasheets, as far as AD iCouplers. It's a great place to get bit in EMC testing with them.

For a single analog, using a transformer and PWM works for lowest cost. It depends on your isolation voltage, regulatory requirements and accuracy.
 

Offline ConKbot

  • Super Contributor
  • ***
  • Posts: 1400
Re: passing signals over galvanic isolation without an opto coupler
« Reply #15 on: October 17, 2019, 02:25:46 pm »
Optocouplers aren't good with age or temperature, almost none of them are AEQ.

IC's using digital isolation with tiny air-core transformers are sexy but a challenge for EMC. Their high carrier frequency is not mitigated in any app notes or datasheets, as far as AD iCouplers. It's a great place to get bit in EMC testing with them.

For a single analog, using a transformer and PWM works for lowest cost. It depends on your isolation voltage, regulatory requirements and accuracy.
I like the AD appnote suggesting you interleave groundlayers so the parasitic capacitance provides a return path for the EMI. I'm not familiar with regulations and industry practices, but interleaved ground planes if you need significant isolation or a proper reinforced barrier seems a tad suspect.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 21226
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
Re: passing signals over galvanic isolation without an opto coupler
« Reply #16 on: October 17, 2019, 02:31:12 pm »
I am not comfortable with opto's after a few failures and looking at opto couplers as a whole (I went through a few Datasheets) I would prefer not to use them. The leakage over 80C changes from 2nA to 200nA. They are ideal to cheaply solve some problems but I do not feel comfortable with them in the application I am involved with and with every other part rated to 125C the opto's are now the weakest link.

Optos have a long-term degredation in their current transfer ratio. I would have thought leakage was only relevant in a DC coupled circuit.

You could probably get around both issues by modulating the DC signal onto an AC carrier (at say 1MHz), and having an AGC in the receiver.

Without knowing the application, threats, and constraints, it is difficult to offer a solid recommendation.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Re: passing signals over galvanic isolation without an opto coupler
« Reply #17 on: October 17, 2019, 02:35:51 pm »
Some digital isolators use inductive coupling. Some use capacitive coupling. The ones using capacitive coupling seem to cause a lot less problems during EMC testing.

While most opto-couplers have issues with time and temperature, some are designed for demanding medical isolation requirements. The fading output of the LED over time means their analogue performance will change with time, but if you have a digital application that won't matter much. You need to look around if you have demanding requirements.
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15797
  • Country: fr
Re: passing signals over galvanic isolation without an opto coupler
« Reply #18 on: October 17, 2019, 02:49:22 pm »
Never had any problem with the AD digital isolators. The ones giving me occasional issues with EMI were the ones embedding a small isolated DC/DC converter, which are pretty noisy. Other than these ones, I found those isolators pretty robust and trouble-free.

AFAIK, if you want to go for capacitive-coupled ones, go for the TI ones.
SiLabs also makes some that are quite good (Si8xxxx). They are based on yet another approach. No microtransformers, no capacitive coupling, but some kind of internal RF carrier.

 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #19 on: October 19, 2019, 12:05:06 pm »
I used a TI isolated CAN transciever that past militarry spec EMC and that uses capacitive coupling.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: passing signals over galvanic isolation without an opto coupler
« Reply #20 on: October 19, 2019, 02:48:59 pm »
I like the AD appnote suggesting you interleave groundlayers so the parasitic capacitance provides a return path for the EMI.
It adds capacitance between the planes on the two sides, which might help a tiny bit with EMI but makes isolation worse and will defeat the purpose of the IC in some cases. Instead of less than 10 pF between either side you get more than 100 pf.

Has anyone ever put an isolator on a double sided PC with simple split power planes on either side and 100 nF bypass capacitors, and actually measured the EMI with proper equipment to see if the field intensities are even relevant for regulations in the realistic worst case situation?
 

Offline profdc9

  • Frequent Contributor
  • **
  • Posts: 334
  • Country: us
Re: passing signals over galvanic isolation without an opto coupler
« Reply #21 on: October 19, 2019, 02:57:29 pm »
For low frequencies, an audio isolation transformer would probably work.  It can handle frequencies above 200 Hz to 10000 Hz or more.  Note if you need really high voltage isolation then you probably need a more customized solution.   Some Teflon wires can have insulation strength 3 to 5 kV.

For frequencies 50 kHz - 1 MHz, you can use a MnZn ferrite core with high permeability.   The higher the frequency, the fewer turns you should use.

For frequencies 1 MHz to 100 MHz, you can use a NiZn ferrite core, type 43 or 61.

The input and output needs to be capacitively coupled, and it's advisable to detect the output using a Schmitt trigger inverter or comparator with some hysteresis.
« Last Edit: October 19, 2019, 02:59:17 pm by profdc9 »
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: passing signals over galvanic isolation without an opto coupler
« Reply #22 on: October 19, 2019, 03:21:26 pm »
If you pass PWM over transformer, note that you probably won't get low frequency (signal) content, and certainly won't get DC.  What you need is a coupling capacitor on one side, and a DC restorer on the other.  This references the signal to its peak value, rather than its mean.  Follow it up with a Schmitt trigger and you have clean digital PWM back (which can be sent to a 1-bit DAC and filtered to recover the original signal).

Sigma-delta is better than PWM in most cases, and reduces to PWM in the extreme cases; AFAIK there are ready-made isolators available, based on this approach.

Can also do push-pull on both sides of a transformer, so that the DC voltage at the center-tap is equal -- a DC transformer.  Feed a biased signal into the center tap, and there you go (or, if you're using analog switches rather than unidirectional transistors, the signal can be bipolar, no DC offset required).  This has a few downsides largely due to transformer non-idealities.

Tim
« Last Edit: October 19, 2019, 03:24:02 pm by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: passing signals over galvanic isolation without an opto coupler
« Reply #23 on: October 19, 2019, 03:31:14 pm »
Has anyone ever put an isolator on a double sided PC with simple split power planes on either side and 100 nF bypass capacitors, and actually measured the EMI with proper equipment to see if the field intensities are even relevant for regulations in the realistic worst case situation?

I don't know what you muean by simple split planes. I put the capacitive CAN transciever on a 4 layer PCB with 330nF each side and it passed.
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 11336
  • Country: us
  • $
Re: passing signals over galvanic isolation without an opto coupler
« Reply #24 on: October 19, 2019, 04:19:08 pm »
for digital isolators and EMC, look at galvanoresistive isolators, but they are more expensive.. I had it in mind to use them before I got tired of PPM designs.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf