Optoisolating an I²C bus? any ideas?

neil · Joined: 08 Sep 2003 Posts: 128

Hi all. I'll give you some details of my project first...

I am designing a PIC-based thyristor trigger unit, which uses direct coupled mains onto the B0 interrupt pin to provide a nice zero crossing waveform. As the PIC is not isolated from the mains, I am powering it from a transformerless ('X' capacitor) power supply (might as well!) and want to optoisolate the I/O, which will be I²C.

I have seen Jerry Steele's idea at: <a href="http://www.national.com/appinfo/tempsensors/files/temphb.pdf" TARGET="_blank">http://www.national.com/appinfo/tempsensors/files/temphb.pdf</a>
(page 22 of pdf). I had a similar idea of inverse parallel opto's myself. I built his circuit using SFH6325 optos and 1N6263 schottkies, but it doesn't work very well. I am seeing long rise and fall times at 100KHz and suspect that the pullup resistors aren't low enough to drive the LEDs properly, as these devices have rise and fall times of 0.2 and 0.5µS. The other side of the isolation barrier is also not being pulled down properly and is sitting at a low state of +2V, not the 0.3-0.4V expected due to the schottky diodes.
___________________

What I am asking of you is: 1. Can you give me a quick reality check, based on what I've told you, or 2. Even better - Do you know of a better way? Has anyone brought out an optoisolator for I²C which does it for me?

Thanks in advance for your ideas.
Neil.
___________________________
This message was ported from CCS's old forum
Original Post ID: 144517576

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

:=Hi all. I'll give you some details of my project first...
:=
:=I am designing a PIC-based thyristor trigger unit, which uses direct coupled mains onto the B0 interrupt pin to provide a nice zero crossing waveform. As the PIC is not isolated from the mains, I am powering it from a transformerless ('X' capacitor) power supply (might as well!) and want to optoisolate the I/O, which will be I²C.
:=
:=I have seen Jerry Steele's idea at: <a href="http://www.national.com/appinfo/tempsensors/files/temphb.pdf" TARGET="_blank"> <a href="http://www.national.com/appinfo/tempsensors/files/temphb.pdf" TARGET="_blank">http://www.national.com/appinfo/tempsensors/files/temphb.pdf</a></a>
:=(page 22 of pdf). I had a similar idea of inverse parallel opto's myself. I built his circuit using SFH6325 optos and 1N6263 schottkies, but it doesn't work very well. I am seeing long rise and fall times at 100KHz and suspect that the pullup resistors aren't low enough to drive the LEDs properly, as these devices have rise and fall times of 0.2 and 0.5µS. The other side of the isolation barrier is also not being pulled down properly and is sitting at a low state of +2V, not the 0.3-0.4V expected due to the schottky diodes.
:=___________________
:=
:=What I am asking of you is: 1. Can you give me a quick reality check, based on what I've told you, or 2. Even better - Do you know of a better way? Has anyone brought out an optoisolator for I²C which does it for me?
:=
:=Thanks in advance for your ideas.
:=Neil.

Optoisolators are the obvious choice, but there are alternatives. Look at IsoLoop digital isolators by NVE (Digi-Key# 390-1007-ND US$5 for example). I have not used them, but they look neat and claim to be real fast.

___________________________
This message was ported from CCS's old forum
Original Post ID: 144517579
_________________
The search for better is endless. Instead simply find very good and get the job done.

Neutone · Joined: 08 Sep 2003 Posts: 839 Location: Houston

:=Hi all. I'll give you some details of my project first...
:=
:=I am designing a PIC-based thyristor trigger unit, which uses direct coupled mains onto the B0 interrupt pin to provide a nice zero crossing waveform. As the PIC is not isolated from the mains, I am powering it from a transformerless ('X' capacitor) power supply (might as well!) and want to optoisolate the I/O, which will be I²C.
:=
:=I have seen Jerry Steele's idea at: <a href="http://www.national.com/appinfo/tempsensors/files/temphb.pdf" TARGET="_blank"> <a href="http://www.national.com/appinfo/tempsensors/files/temphb.pdf" TARGET="_blank">http://www.national.com/appinfo/tempsensors/files/temphb.pdf</a></a>
:=(page 22 of pdf). I had a similar idea of inverse parallel opto's myself. I built his circuit using SFH6325 optos and 1N6263 schottkies, but it doesn't work very well. I am seeing long rise and fall times at 100KHz and suspect that the pullup resistors aren't low enough to drive the LEDs properly, as these devices have rise and fall times of 0.2 and 0.5µS. The other side of the isolation barrier is also not being pulled down properly and is sitting at a low state of +2V, not the 0.3-0.4V expected due to the schottky diodes.
:=___________________
:=
:=What I am asking of you is: 1. Can you give me a quick reality check, based on what I've told you, or 2. Even better - Do you know of a better way? Has anyone brought out an optoisolator for I²C which does it for me?
:=
:=Thanks in advance for your ideas.
:=Neil.

You have to drive the LED's hard for a fast interface. A 4.3K series resistor is not going to operate fast. The speed you can operate is related to the drive strength on the optocouplers.
___________________________
This message was ported from CCS's old forum
Original Post ID: 144517583

Yashu · Guest

Yashu · Guest

http://www.nve.com/products/isolators/ten.php

Yashu · Guest

http://www.nve.com/products/isolators/ten.php

Kenny · Joined: 07 Sep 2003 Posts: 173 Location: Australia

Reposting with modification. My reply on previous board is not visible on this one. Also this post got lost too. New board has some teething problems. I was registered but today I had to re-register.

Your post jogged my memory, I had seen something like the circuit in the link before in a magazine. It was Elektor July/August 1998 and it also a National Semiconductor application note according to the article, but I couldn't find it on their site.
The circuit is pretty much identical, but instead they use 6N139 Darlington optos and 3k9 resistors. Darlington optos need less current in the led than standard optos and this is important in this circuit.
The Schottky diodes are BAT85.
To make it completely symmetrical they have shown bidirectional for SCL as well, presumably for multimaster use.
Of course, the usual pullup resistors for SDA and SCL are needed on both sides of the barrier. In the circuit 3k3 resistore are shown. This resistance is shunted by the current through the 3k9 and led, but total current is within i2c spec.
It's one of those circuits that I filed away and have never used so I can't comment on it's performance.

A trawl of the net with 'i2c isolation' revealed this link:
http://www.microrobotics.co.uk/pdf/opto-i2c.pdf
For extra drive they use a bus extender on one side and a 2N2907 transistor on the other, and 6N137 optos.

HTH
Kenny

neil · Joined: 08 Sep 2003 Posts: 128

Hmm, there are some teething problems with the site! Still, it will be good when we're all used to it. I have used other forums which appear to use the same engine, and they are great.

Thanks to Kenny! I have not had much luck on the search engines looking for isolation. I kept getting all sorts but what I was looking for!

That link to Micro-robotics was useful. I have found app notes by Philips from that schematic by looking at the 82B96 extender. It would be a good idea to use one of them anyway, as my I²C link is over a twisted pair, rather than on a PCB.

Cheers,
Neil.