Best type of analog isolation for protection for high voltage surges
Best type of analog isolation for protection for high voltage surges
(OP)
Hi all,
My first thread so hopefully I am posting in the right forum/format/etc.
Looking for advice on the best isolation device to use for analog signals. Heres the problem:
The circuit (this part of it anyway) is a device for measuring the cathode current of a high voltage vacuum tube (we manufacture these believe it or not). The anode is at 37,000-40,000V and occasionally (far less occasionally than we would like) this high voltage will make its way back to the cathode in the form of a brief arc. The cathode circuit now is very simple: It consists of an op-amp (tl082) with a 10:1 gain to measure voltage drop accross a 100ohm resistor to ground... The only other thing here is all the stuff on there to try and dissapate the arc energy. Basically this includes capacitors and TVS diodes (which incidentally I have found respond much faster and are generally better than gas discharge tubes and MOVs in case anyone is wondering). That said dispite my best efforts the op-amp is still not surviving so I think I should use some sort of isolation circuit. This is the much shortened version of the story... Before getting to this point I have tried many different ideas to make this circuit work as is and while much progress has been made its still not bullet proof.
Anyway... Heres the requirments:
Needs to measure DC only (should make is easier)
Needs to measure 0-10mA
Needs to be pretty linear (I will be measuring values from near 0 up to 5mA+ and need an accuracy of around +/-10 microamps)
I used an opto-coupler in another part of the circuit (the part that controls the grid 1 of the tube) and it works great but since the control signal (that biases the tube on) goes there and its relying on the cathode to know how much its turned on linearity and difting with age don't matter much... It just goes up and down based on the measured value at the cathode. I tried to use this same optocoupler in the measuring (cathode) side but it was not linear enough and I am told they age pretty badly and this causes the current ratio to go down over time.
Can anyone suggest a methode of isolation thats simple and reasonably cheap (looking to avoid A/D-digital isolation-D/A type solutions)?
Anyone used an AD210? You think think these would work? (dispite their outragous price tag)?
I am guessing I am seeing about 560V of common mode voltage though I don't really care to put a scope on there for obvious reason...
Thanks.
My first thread so hopefully I am posting in the right forum/format/etc.
Looking for advice on the best isolation device to use for analog signals. Heres the problem:
The circuit (this part of it anyway) is a device for measuring the cathode current of a high voltage vacuum tube (we manufacture these believe it or not). The anode is at 37,000-40,000V and occasionally (far less occasionally than we would like) this high voltage will make its way back to the cathode in the form of a brief arc. The cathode circuit now is very simple: It consists of an op-amp (tl082) with a 10:1 gain to measure voltage drop accross a 100ohm resistor to ground... The only other thing here is all the stuff on there to try and dissapate the arc energy. Basically this includes capacitors and TVS diodes (which incidentally I have found respond much faster and are generally better than gas discharge tubes and MOVs in case anyone is wondering). That said dispite my best efforts the op-amp is still not surviving so I think I should use some sort of isolation circuit. This is the much shortened version of the story... Before getting to this point I have tried many different ideas to make this circuit work as is and while much progress has been made its still not bullet proof.
Anyway... Heres the requirments:
Needs to measure DC only (should make is easier)
Needs to measure 0-10mA
Needs to be pretty linear (I will be measuring values from near 0 up to 5mA+ and need an accuracy of around +/-10 microamps)
I used an opto-coupler in another part of the circuit (the part that controls the grid 1 of the tube) and it works great but since the control signal (that biases the tube on) goes there and its relying on the cathode to know how much its turned on linearity and difting with age don't matter much... It just goes up and down based on the measured value at the cathode. I tried to use this same optocoupler in the measuring (cathode) side but it was not linear enough and I am told they age pretty badly and this causes the current ratio to go down over time.
Can anyone suggest a methode of isolation thats simple and reasonably cheap (looking to avoid A/D-digital isolation-D/A type solutions)?
Anyone used an AD210? You think think these would work? (dispite their outragous price tag)?
I am guessing I am seeing about 560V of common mode voltage though I don't really care to put a scope on there for obvious reason...
Thanks.





RE: Best type of analog isolation for protection for high voltage surges
The point is that even non-linear optocouplers can be restricted to a reasonably linear region. Maintaining calibration of DC signal levels might be an issue for your application. If that's an issue then V to F and F to V is the most obvious solution.
There are some mighty long optocouplers to meet any reasonably voltage isolation requirement.
RE: Best type of analog isolation for protection for high voltage surges
RE: Best type of analog isolation for protection for high voltage surges
RE: Best type of analog isolation for protection for high voltage surges
http://www.avagotech.com/docs/AV02-0951EN
Benta.
RE: Best type of analog isolation for protection for high voltage surges
BrianG: I have not tried spark gaps. I know they were common for use in CRT video amps because they had no leakage and very little capacitance (this made them preferable in applications were there was a high speed video signal) but everything I've READ indicates they are inferior in pretty much every other way to regular TVSs diodes... Of course this is just going on what I've read. I know the TVSs provided notably better results then MOVs that I initially used.
VE1BL: First... Is VE1BL a ham radio call sign? Maybe canadian or something? If so awesome! I am KC8KQO. Second: That was one of my first ideas... Maintaining the DC level will be a problem but it would be possible to have the equipment auto-zero each time it was used but taking a reading of the current before anything else is turned on and setting this as the zero point. Extra programming for me but has an added benifit of letting the equipment know if a card is malfunctioning (if no current is detected initially it would know somethings up). This is pretty high up on my list of things to try. My first prototype used and optocoupler (without biasing) and the idea was the have the equipment run the value though a function that corrected for the non-linearity but my little tiny PLC could not hack the math without a co-proccessor module and its a 4 slot "expandable micro brick" PLC and all 4 slots are already used for the analog channels.
LiteYear: Kind of... I tried adding a higher value resistor in front of the 100 ohm (10K or something)... This was not a real solution because the voltage drop would be high enough that the voltage on the cathode would exceed allowable (making the g1 bias range wrong). It was a good test but it didn't work. Hard to tell if I saw any improvment at all though my guess is that there was some... Just not enough.
Benta:
Looks interesting... For general analog applications they recommend another part: HCPL-7510... Might give one fo those 2 a try... Gonna have to see how the price compares to the AD210.
RE: Best type of analog isolation for protection for high voltage surges
Did some expirments with some opto-couplers and found that the DC level thing will be a big issue. The ones I have lying around are 1:1 ratio, 15KV (IIRC) and rated for up to 40mA. They appeared to be most linear in 20-30mA range but they were not stable... They tended to drift around by a hundred micro amps or so... After about 10 minutes they were MORE stable but still not great and anyway 10 minutes is too long... Incidentally they were not really close to 1:1... More like .9:1 on the low end and almost 2:1 on the high end...
I apparently had not had enough caffine this morning because I thought they were suggesting a different part in that datasheet Benta posted... I glanced at the PDF file name and assumed it would be the same as the part number hence my confusion. Anyway...
Looks like those may work so I ordered some to play with... My only concern is that I may damage them by having too high voltage between Vin+ and Vin- (GND1 and VDD1 will be powered by an isolated supply with no referance to the HV PS or anything else so hopefully nothing will try and sneak though that way)... I plan to minimize this voltage by using a carbon resistor for my 100 ohm (or whatever value I end up using) and bypassing it with a couple capacitors and a diode. Anything else I should do to maximize my chances of success?
I of course will employ the same mitigation techniques I am employing now for the amplifier on the output... No sense is getting ride of any of that since I have plenty of space on the board and parts are cheap.
I am so looking forward to a new rev... Not only to get this project out of my hair but because the board shop sends me microwavable popcorn, t-shirts and other free stuff with my order... LOL
RE: Best type of analog isolation for protection for high voltage surges
You may still decide that an opto device is an easier route if you can make it linear.
RE: Best type of analog isolation for protection for high voltage surges
RE: Best type of analog isolation for protection for high voltage surges
I would think it would be hard to get consistent and accurate results in the micro-amp range... Probably possible but I bet it'd be a pain.
BrianG: Yes and unfortunately a brief arc means that even a fairly low inductance return path can see hundreds of volts on the IC... And with hundreds of volts it only takes a small amount of difference in the return path at each pin to make pin to pin voltages well in excess of the rail voltage... I got these HCPL-7510s in and started to throw together a circuit but got inturrupted several times so I don't think I'll get it done today... Always next week I guess.
Anyway... Thanks for all the help I'll post some results of how this thing works once I get something up and running.
RE: Best type of analog isolation for protection for high voltage surges
RE: Best type of analog isolation for protection for high voltage surges
I've used one interesting method to provide isolated power - common mode power inductors usually used to filter lines. I picked out one with the lowest current and best transfer was at about 40khz.
What about trying duty cycle. A PWM chip like a TL494 can convert current to percent duty cycle. That can easily be transfered with an opto and averaged to produce a voltage proportional to current. I like these chips that have a voltage reference, opamp, and driver all in one package.
RE: Best type of analog isolation for protection for high voltage surges
Finally fixed the problem... I tried one those HCPL-7510 parts but could not make it last. Problem was that the measuring side could not be isolated from the other side... The measuring side is measuring return to chassis ground so it HAS to be at earth ground. Unfortunately I realized the PLC side had to be at earth ground too (all of the analog channels share a ground and some of the other stuff on those analog channels needed and earth ground so isolating the whole thing was out). So instead I tried seperating the grounds by providing seperate grounds... The measuring side got a nice wide ground strap going straight back to the HV PS... The PLC side used the original ground on the card. It worked pretty well but not perfectly and when it did go out the HCPL-7510 was more of a pain to replace so I went back to the original design... Tried a BUNCH of stuff... various capacitors, TVS diodes, low impedance ground path... Eventually I put some 220 ohm carbon resistors in series with the power supply rails to try and make this a less appitizing return path for energy and that finally (presumably in combination with everything else)solved the problem... I ran about 50 hours of testing myself and turned it over to production... It now has about 250 hours of production use under brutal conditions (I know this because I've seen significant arcing just while I've been walking by and in at least one case they attempted to process a tube that was at air causing all of the other elements of the tubes to see a bunch of sparking at the max current of the HV PS)... So far not a single failure of that circuit or anything else in the unit... Not even a hiccup.
Company president has already suggested we built another unit to increase capacity and I am gonna spend some time applying these techniques to some other circuits in the building that fail regularily and were designed many years ago (in some cases before I was born... Not even kidding... We have a test thats old enough to collect social security and still in use).
RE: Best type of analog isolation for protection for high voltage surges