Slightly OT - 2kV voltage isolation?

[Grahame Marsh]

I'm looking to pass a, i2c bidirectional data signal between two systems
that have about 2kV voltage difference between their respective
"grounds". I can find several device on, say, Mouser's website but I'm
getting lost with all the insulation terms. There is a variety of test
voltage conditions like "Vrms for 1 minute" and "galvantic isolation in
kV". These voltages are typically 4 - 6kV. But I also find "Maximum
working insulation VIORM 566 voltage" - so does this mean a normal
working voltage difference? And so well less than 2kV.

Sorry, but I need an idiot's (that's me) guide to what isolation voltage
I should be looking for to isolate these systems that live 2kV apart? I
know there are solutions that use multiple optoisolators and drivers but
I was looking for a single chip solution. Any helpers please?

(The application is between a microcontroller and the cathode circuit of
a CRT so it is almost on topic.)

Grahame

[Dan Harboe Burer]

What about this type of opto?

Dan

-----Oprindelig meddelelse-----
From: Grahame Marsh
Sent: Saturday, November 30, 2013 7:13 PM
To: neoni...@googlegroups.com
Subject: [neonixie-l] Slightly OT - 2kV voltage isolation?

--
You received this message because you are subscribed to the Google Groups
"neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to neonixie-l+...@googlegroups.com.
To post to this group, send an email to neoni...@googlegroups.com.
To view this discussion on the web, visit
https://groups.google.com/d/msgid/neonixie-l/529A2AD3.8030705%40googlemail.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Grahame Marsh]

Thanks Dan,

I can find optos like this but you need 3 or 4 of them and a driver chip
to set up the two bidirectional lines for an i2c link. Mouse sell
single chip that aren't necessarily opto, some are capacitive for
example, specifically design for i2c links like this one

http://www.mouser.com/ds/2/405/sllseb6b-261049.pdf

where I get lost is understanding what working voltage it will isolate,
I need 2kV, as against some test code voltage.

Cheers Grahame

[David Forbes]

Grahame,

A quick Google search turned up this HP (Agavo) app note. It may or may
not help. There are amusing graphs of volts vs hours of lifetime. 2KV is
usually at 10 hours.

http://www.avagotech.com/docs/5989-0802EN

--
David Forbes, Tucson AZ

[Grahame Marsh]

On 01/12/2013 01:22, David Forbes wrote:
> On 11/30/13 11:13 AM, Grahame Marsh wrote:
>>
>> I'm looking to pass a, i2c bidirectional data signal between two systems

>> that have about 2kV voltage difference between their respective ...

>>
>
> Grahame,
>
> A quick Google search turned up this HP (Agavo) app note. It may or
> may not help. There are amusing graphs of volts vs hours of lifetime.
> 2KV is usually at 10 hours.
>
> http://www.avagotech.com/docs/5989-0802EN
>

David

Thank you for the paper it has helped a lot - if I have read it
correctly, the best performance is a catagory 3 opto coupler can stand
off 1kVDC for > 100 000 hrs. So it looks like my search for a part ends
'cause there is no such part... back to the drawing board.

Cheers Grahame

[Matthew Smith]

Quoth Grahame Marsh at 2013-12-01 19:18 ...

> Thank you for the paper it has helped a lot - if I have read it
> correctly, the best performance is a catagory 3 opto coupler can stand
> off 1kVDC for > 100 000 hrs. So it looks like my search for a part ends
> 'cause there is no such part... back to the drawing board.

I've looked at comms with HV systems before. I2C it ain't - you'd need
to do protocol conversions - but I figured TOSLINK would be a reasonably
cheap option, considering availability of parts.

Currently trying to figure how to get a microcontroller to drive a
regulator for the filament voltage of an X-ray tube (easy) but fully
isolated (not so easy with up to > 40kV) plus monitoring the current.
Discounted Bluetooth as too fiddly/unreliable - TOSLINK with half-metre
optical fibres would isolate to about any voltage I'd want. UART comms,
rather than I2C, though.

--
Matthew Smith

Business: http://www.smiffytech.com
Blog: http://www.smiffysplace.com
Linkedin: http://www.linkedin.com/in/smiffy
Flickr: http://www.flickr.com/photos/msmiffy
Twitter: http://twitter.com/smiffy

ABN 16 391 203 815

[Grahame Marsh]

On 01/12/2013 09:32, Matthew Smith wrote:
> Quoth Grahame Marsh at 2013-12-01 19:18 ...
>
>> Thank you for the paper it has helped a lot - if I have read it
>> correctly, the best performance is a catagory 3 opto coupler can stand
>> off 1kVDC for > 100 000 hrs. So it looks like my search for a part ends
>> 'cause there is no such part... back to the drawing board.
>
> I've looked at comms with HV systems before. I2C it ain't - you'd need
> to do protocol conversions - but I figured TOSLINK would be a
> reasonably cheap option, considering availability of parts.
>
> Currently trying to figure how to get a microcontroller to drive a
> regulator for the filament voltage of an X-ray tube (easy) but fully
> isolated (not so easy with up to > 40kV) plus monitoring the current.
> Discounted Bluetooth as too fiddly/unreliable - TOSLINK with
> half-metre optical fibres would isolate to about any voltage I'd want.
> UART comms, rather than I2C, though.
>

Hi Matthew

I had thought about, but not looked, at other protocols - I was looking
for the $1 part that would do the job!

Standing off 40kV is much more fun...

Cheers Grahame

[John Rehwinkel]

>
> Standing off 40kV is much more fun…

Yeah, one monitor I had used fiber optics for that trick,
>
> Cheers Grahame

>
> --
> You received this message because you are subscribed to the Google Groups "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
> To post to this group, send an email to neoni...@googlegroups.com.

> To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/529B0F00.2030407%40googlemail.com.

[GastonP]

Hi Grahame

Texas Instruments has a line of digital isolators specifically for this kind of service. Perhaps some of them can be of use to you...

http://www.ti.com/lsds/ti/interface/digital-isolator-products.page

Gastón

[Michel van der Meij]

Hi Grahame,

I don't know if this is of any help but if you want to go for the $1 solution I assume you like to test things with parts you already have lying around. I was thinking, you could use 2 optocouplers in series so that each carries 1kV. The only "problem" is that you need to have a power supply available between the 2 optocouplers that can carry a 1kV working voltage. For testing you could use a 9V battery, or a DC/DC converter as in attached drawing. The two 10M resistors make sure each opto carries half the total isolation voltage.

 

Cheers,

Michel

 

 

[Matthew Smith]

Quoth Michel van der Meij at 2013-12-02 08:16 ...
> ...The only "problem" is that you need to have a

> power supply available between the 2 optocouplers that can carry a 1kV
> working voltage. For testing you could use a 9V battery, or a DC/DC
> converter as in attached drawing.

You can also use cascaded (isolated) DC/DC converters - every
transformer providing isolation. (I believe that transformer cascading
is one method of getting very high voltages in the first place.)

The high-isolation stuff I've been looking at would probably just have a
couple of AA cells per stage - cheap and easy, as this isn't a system
that would be run more than occasionally.

[Grahame Marsh]

Hi

Thanks for all the comments I have now found this $5 part

With creepage and clearance of greater than 13mm, the ACNV2601 is designed to provide high isolation voltage (7500 Vrms). It can withstand a continuous high working voltage of 2262 Vpeak and a surge voltage of 12,000 Vpeak

http://www.avagotech.com/pages/en/optocouplers_plastic/plastic_digital_optocoupler/10_mbd_logic_gate/acnv2601-000e

The datasheet says VIORM = 2262V

So, unless I have (again) misunderstood the isolation voltages, this does the job with 10% headroom.  Change the protocol from i2c and I'm there.  I would appreciate correction if I'm wrong.

Grahame

On 01/12/2013 21:46, Michel van der Meij wrote:

Hi Grahame,

I don't know if this is of any help but if you want to go for the $1 solution I assume you like to test things with parts you already have lying around. I was thinking, you could use 2 optocouplers in series so that each carries 1kV. The only "problem" is that you need to have a power supply available between the 2 optocouplers that can carry a 1kV working voltage. For testing you could use a 9V battery, or a DC/DC converter as in attached drawing. The two 10M resistors make sure each opto carries half the total isolation voltage.

 

Cheers,

Michel

 

 

on Dec 01, 2013, Grahame Marsh <graham...@googlemail.com> wrote:

On 01/12/2013 09:32, Matthew Smith wrote:
> Quoth Grahame Marsh at 2013-12-01 19:18 ...
>
>> Thank you for the paper it has helped a lot - if I have read it
>> correctly, the best performance is a catagory 3 opto coupler can stand
>> off 1kVDC for > 100 000 hrs. So it looks like my search for a part ends
>> 'cause there is no such part... back to the drawing board.
>

[Oscilloclock]

Delightfully on topic for me! This makes the second part that I'm using perhaps well outside its design limits. (The first was Michel's pointing out that my bias on the DAC's analogue ground is outside parameters.)

I hadn't thought too much about the 6N137 isolation wear
problem, beyond just choosing a manufacturer whose specs seemed to have the
highest isolation rating in terms of voltage. I did notice in passing that
they also specified 1 minute at that voltage - and just assumed this is the
UL rating methodology (by which they certify and/or compare different
units), and doesn't reflect actual operating conditions.

I did read somewhere that the isolation survives far better under a unipolar
and constant potential than a fluctuating, bipolar voltage (crossing zero).
At least in this case, we don't have pulses or fluctuation of any kind
involved! My models built so far haven't broken down - yet...

But Grahame - fantastic that you found this part and I'll definitely be trying it out at the next opportunity!

Aaron

[Ulysses Balis]

Why not just make your own optocoupler?  Fill a lexan or acrylic 10 cm tube with clear, slow-setting epoxy and then affix an IR led on one and a darlington phototransistor on the other end, with these two components serving as end plugs for the epoxy.  Then paint the unit black with a low-voc paint.  The resultant construct will offer over 100KV of permanent breakdown voltage immunity.  You'll need to make several of these for I2C functionality.  Alternatively, consider a micro power wireless local area network, with an rf link serving as the  data channel.  Place two bi-directional I2C transponders in a Faraday-shielded container, with 1 cm ceramic mounting posts for each transceiver board, along with HV glass passivated feed-throughs to make a permanently isolated HV hardened / isolated construct.  Optionally, fill the whole cavity with Sylgard 184 (and let cure for 4 hours at 60C) when assembled for the ultimate in dielectric breakdown immunity.

-ub

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscribe@googlegroups.com.

To post to this group, send an email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/529AF7FA.5000100%40googlemail.com.

[Ulysses Balis]

Alternatively, for a simpler approach, take a look at this 12 kV single component isolation solution:

http://www.micropac.com/images/HVOC/66353%20Datasheet.pdf

It's even space-qualified.

[Ulysses Balis]

Finally, instead of building the tube assembly that I originally described, you could simply select this integrated 40kV-rated ceramic tube construct (which has a superb 10 MHz signal bandwidth):

http://www.micropac.com/images/DataSheet/66147.pdf

They're a bit pricey at 22 bucks each, but at least you would have the peace of mind of knowing that they would never fail.

Cheers,

-ub 

[GastonP]

Last famous words :)

On Monday, December 2, 2013 12:14:39 AM UTC-3, Ulysses_Balis wrote:

<snip> that they would never fail.<snip>

[Tom Harris]

Use SPI protocol, then you only need 3 optos with simple drivers, no bidirectional stuff. For isolation, use cheap jacketed polymer fibre optic, about $1 a metre, with matching detectors & transmitters $2 each or so.

Tom Harris <celep...@gmail.com>

Grahame

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscribe@googlegroups.com.
To post to this group, send an email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/529A2AD3.8030705%40googlemail.com.

[Matthew Smith]

Quoth Tom Harris at 2013-12-03 08:10 ...

> Use SPI protocol, then you only need 3 optos with simple drivers, no
> bidirectional stuff. For isolation, use cheap jacketed polymer fibre
> optic, about $1 a metre, with matching detectors & transmitters $2 each
> or so.

Or four? MISO, MOSI, SCK, CS. (Three if unidirectional.)

[Charles MacDonald]

On 13-12-02 04:40 PM, Tom Harris wrote:
> Use SPI protocol, then you only need 3 optos with simple drivers, no
> bidirectional stuff. For isolation, use cheap jacketed polymer fibre
> optic, about $1 a metre, with matching detectors & transmitters $2 each
> or so.

But how do you determine the "safe" working voltage for that?

Which brings up my other question from this discussion, what are the
failure mechanisms that would cause the Optocoupler to hold of 2 Kv for
a minute, but fail after an unspecified number of hours?

Is this dirt accumulation caused by the fields?, is it breakdown of the
plastics? is it too many electrons sitting for too long that they
finally make a run for it? Is it arcing?



--
Charles MacDonald Stittsville Ontario
cm...@zeusprune.ca Just Beyond the Fringe
http://Charles.MacDonald.org/tubes
No Microsoft Products were used in sending this e-mail.

[Grahame Marsh]

>
> Which brings up my other question from this discussion, what are the
> failure mechanisms that would cause the Optocoupler to hold of 2 Kv
> for a minute, but fail after an unspecified number of hours?
>
> Is this dirt accumulation caused by the fields?, is it breakdown of
> the plastics? is it too many electrons sitting for too long that they
> finally make a run for it? Is it arcing

The answer is, of course, Fowler-Nordheim tunnelling and Poole-Frenkel
hopping, as anyone nows, errrrr....

The position I have now, which appears to be workable, is to use the
acnv2601 which will stand of Viorm of over 2.2kV. To do a straight
highspeed, bidirectional interface I can use 8 (!!!) for i2c, 4 for SPI
but 2 connecting using UARTs should work and only need 2 optos. Ok, the
device I want to talk to is still i2c but I can reform the i2c
conversation using a 8pin microcontroller.

One technology I was pointed at in a PM was using pulse transformers. I
read one manufacture's datasheet that said the high voltage impulse
tests should not be made repeatedly as the plastics degrade during the
test. I would guess plastic -> carbon -> conductor -> arcing -> fried
low voltage electronics. If the presence of the EHT causes gradually
degradation of the plastic in the same way plastics degrade in heat or
UV or other radiation then it is bad news at the end of the day.
Another guess is that one factor is the separation distance between the
light source and sensor. It's why 1/2 metre of fibre optic works so
well when you want to stand off 40kV. The acnv2601 datasheets makes a
lot out of the 13mm separation distance between the led and
phototransistor. Most other optos (eg 6n137) the distances are much
smaller. And in those big EHT switching opto diodes, very much bigger.

The problem with a lot of manufacture's datasheets is they tell you what
impulse voltage their devices will standoff but not what they will stand
off continuously. If you are connection two low voltage systems
together (low being, say, < 500V) then the question of wear probably
doesn't arise, but you get your protection from high voltage spikes.

The answer to your question is, of course, Fowler-Nordheim tunnelling
and Poole-Frenkel hopping, as anyone on the Clapham Omnibus knows,
errrrr....

Not a physicist or a chemist! I'm just glad I have a part I can use.

Grahame

[John Rehwinkel]

> The position I have now, which appears to be workable, is to use the acnv2601 which will stand of Viorm of over 2.2kV. To do a straight highspeed, bidirectional interface I can use 8 (!!!) for i2c, 4 for SPI but 2 connecting using UARTs should work and only need 2 optos. Ok, the device I want to talk to is still i2c but I can reform the i2c conversation using a 8pin microcontroller.

That's probably the way to go. There are I2C optos, like this IXYS/Clare part:

http://www.clare.com/home/pdfs.nsf/www/CPC5902.pdf/$file/CPC5902.pdf

But, like the others, it doesn't hold off enough voltage. There are things like the OPI1268 that hold off more voltage, but the Avago part is probably in the sweet spot.

- John