Analog OptoCoupler 1-10MHz

Discussion in 'The Projects Forum' started by HMartins, Apr 4, 2013.

  1. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    I am looking after an analog optocoupler that can run at 10MHz (as much as possible but 10MHz could be OK), below 10MHz if there is no other way.

    This stuff is not easy to find also because I also need it's SPice model.

    I found NEC (Renesas) PS8502 +- 0.35uS propagation delay but I can't it's PSpice model.

    http://documentation.renesas.com/doc/YOUSYS/document/PN10657EJ04V0DS.pdf

    1KV isolation wold be enough.

    I need some advice on this matter.

    Martins
     
  2. Papabravo

    Expert

    Feb 24, 2006
    10,144
    1,791
    It looks like a perfectly ordinary digital optocoupler to me. What do they mean when they talk about an "analog" output? It is just an ordinary photo transistor. It will not replicate a signal on its input. I'm confused.

    Maybe you could show us a schematic of how you plan to apply this part, and then tell us what you expect it to do.
     
  3. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    What do they mean when they talk about an "analog" output?
    Well, some optocoupler have buffers (too much) inside. They seem to behave as a TTL type device. As example, we can't just input a triangular wave and get a triangular wave at the output no matter how we polarize it's output. In some cases we can't polarize at all. They have +V, -V and Out (apart from the LED diode, of course).
     
  4. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    This is the sort of thing I am thinking about:

    [​IMG]
     
  5. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,388
    1,605
    10MHz is probably too high for anopto coupler. Note your sample design is driven by a potentiometer, which is running essentially at DC.
     
  6. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    OK. I just got good simulation for Avago CNR200. Looks like I can get around 1MHz but I still have to calm down the opamp (ADA4891) driver who is too fast to include LED + Phototransistor in feedback loop, forcing it to oscillate.

    Delay for CNR200 seems to be around 0.5 to 1uS. I'l check specifications against PSpice model behavior.

    http://www.avagotech.com/docs/AV02-1134EN

    Anyway, this is muuuuuuch better than before.
     
  7. crutschow

    Expert

    Mar 14, 2008
    13,014
    3,234
    What is the purpose of this circuit?
     
  8. t06afre

    AAC Fanatic!

    May 11, 2009
    5,939
    1,222
  9. Papabravo

    Expert

    Feb 24, 2006
    10,144
    1,791
    An optocoupler, by its very nature, is a current-in to current-out device. The parameter that characterizes this relationship is called the Current Transfer Ratio (CTR). Values from 20% to 600% are fairly common. This parameter does degrade with age. I suppose it might be possible to convert a voltage waveform to a current. Run it to the optocoupler. On the output you would need a current to voltage converter to reconstruct the original waveform.

    There are a number of problems with this approach, not the least of which is the fundamental physics of moving charges back and forth in the diode and the photo transistor. If you can get anywhere near the 10 MHz. goal I will be most surprised.
     
  10. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    crutschow [​IMG]

    I need to watch some osciloscope curves in two different circuits 300 to 800 volts away from earth voltage and from each other.

    My oscilloscope is quite old (Kikusui 5530GR) and I am broke. I can't buy special probes.

    The floating circuits will be powered by two separate 9V batteries.


    t06afre [​IMG]

    Meanwhile I'll investigate about HCPL-4562 or HCNW4562. Thanks for suggestion.


    Papabravo [​IMG]

    CTR seems not to be a problem. I will avoid distortion (some is acceptable) inserting one of the photodiodes into one of the OPAMPs feedback loop living the other for the earth related side. I will make sure both photodiode circuits have some load so that the sample photodiode connected to the feedback loop has the same behavior as much as possible as the other one (the usual).
     
  11. vk6zgo

    Active Member

    Jul 21, 2012
    677
    85
    Way back in the 1980s I had occasion to use a unit which was intended to isolate a TV video signal on a run of cable.

    It was optically linked,but that said,I haven't a clue what device or devices were used inside it.
    It was rated for a fairly high voltage ,but was overkill for our situation,where we were just trying to get rid of common mode crud on the video.(we eventually found the problem was an earthing fault in the building).

    We needed DC to 5MHz,but if I remember correctly,it was rated DC to 10MHz.
     
  12. Papabravo

    Expert

    Feb 24, 2006
    10,144
    1,791
    If the response of an optocoupler had a linear range your approach might make some sense. What I don't understand is how you expect a non-linear device to have a linear response. I would think about using transformers as isolation devices. You can find torroidal cores that have an adequate response in the 10 MHz. range.
     
  13. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    Combining a transformer (tinny) and CNR200 opto-cpupler, I got 10nS setup for a 1KHz square wave, 20ns after a slight ringing.

    Of course, several other OpAmp are involved.

    I am not sure what happens with a real world transformer, but this is encouraging. :)

    Total OpAmps per non differential channel, 11 .. 12 (depending on out buffer).

    PSU ICs not included.
     
  14. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    Patabravo:
    If the response of an optocoupler had a linear range your approach might make some sense. What I don't understand is how you expect a non-linear device to have a linear response. I would think about using transformers as isolation devices. You can find torroidal cores that have an adequate response in the 10 MHz. range.
    First, I don't need full linearity, just linearity.

    It is a common technique. We use two couplers: one for a local feedbacl loop, the other for the other side. The linearity correction introduced by the feedback loop corrects the second coupler. Now, the circuit connected to the out of both couplers must be as identical as possible. Thread 4# shows an image.

    Meanwhile I realized there was a second comments page (I am not yet familiar with this interface) and that you gently referred the use of transformers.

    Well, just transformers is not enough because I need DC coupling. I managed a way to do it merging opto and transformer in a single channel. Opto goes by a LFP, transformer by HPF. Alter careful balance of both (say) sub-channels, it seems (simulation) to work quite nicely.

    I will now investigate if I can find precise PSpice model for the transformer. As example, this could be a transformer:

    http://www.farnell.com/datasheets/1644448.pdf
     
  15. HMartins

    Thread Starter New Member

    May 10, 2011
    21
    2
    Well, that could be done (analog) several ways, one of them modulating everything in frequency modulation. Most broadcast video recorders worked that way modulating the complete video signal without touching color sub-carrier. VTRs like this ...

    http://www.youtube.com/watch?v=LRJGJKM5aXE&feature=related

    ... worked like that.

    They modulated, as much as I remember, the all video signal (composite) with a 7 to 9MHz carrier. I can investigate further, I still have a maintenance manual of an Ampex VPR-2B machine. That signal could go through an optocoupler without much trouble because precise waveform reconstruction of modulated signal itself wad no special meaning. What mattered was the signal quality after demodulation and (another very delicate matter) time-base correction. This last was not a problem in ground lifting composite video isolator.

    Ground lifting composite video isolator could, eventually, work that way. Some microwave beams worked also that way still up-converting to a much higher frequency.
     
Loading...