How to achieve proper input isolation???

[MikeSD]

input isolation

I have a problem and am having difficulty with the solution. I have a design, for which I need to adapt to a new project. I cannot change the internals of the design.

The inputs to the design consist of 28VDC (3 lines), 28VDC RETURN, and 3 discrete signals.

The new project requires lightning protection (on every input line) and I have installed transorbs from each of the inputs to chassis ground. The problem lies in the fact that I also have an input insulation requirement that isn't being met with these transorbs. The requirement is 100 Meg Ohm from pin to chassis.

The problem exists because the input discretes have 10K pull-ups to 28VDC, inside the module that I can't change. with these pull-ups, all of my 5 inputs are placed in parallel, and even with the low leakage of these transorbs, this parallel combination gives pin to chassis resistance of about 30 MOhm.

I have tried placing a series diode inline with these inputs but that only improves the isolation but doesn't get me all the way there. And, I don't think stacking more diodes is the proper solution.

What are some good solutions to get the insulation resistance needed? I have considered opto-isolators but since there is only one source and since that is the source of my problem, I don't think this will work. Beside, I can't use opto-isolators on the power lines.

Any suggestions?

 

[newelltech]

transorbs

If I understand your circuit, the basic problem is that you have 3 Transorbs protecting the input lines to chassis ground. Each Transorb provides 100 MOhm, but 3 in parallel only give you 30 MOhm?

My first suggestion is to look carefully at the Transorb specs. What voltage are you testing at? How close is it that to the 'knee' of the transorb. Maybe selecting a Transorb with the next higher voltage rating would give you better isolation.

Also at 100 MOhm, your test setup will affect the results - i.e. dirt, humidity, etc.
Have you measured the isolation without any transorbs?

Do the Transorbs vary from unit to unit? As a last resort you could pick out the best ones for some more improvement.

Also you may want to research Surgectors. I did a quick search for low leakage transient suppressors and found some Surgectors with a leakage of < 50nA so at 100 volts that would give you well over 300 MOhms for each one. I don't know if they would work for you otherwise.

Hope this helps a little.

 

[artem]

ligthning protection

put optotransistor between input of you system and transorb

 

[VVV]

how to measure transorb

I don't understand why you installed the transorbs from each line to the chassis. I think you should have installed one from each line to ground. After all, you need protection from overvoltages between these lines. Then you can install something else from ground to chassis, like a single transorb that would give you the required isolation. Or perhaps you csould even use a gas arrester (which will have lower leakage) if you can afford a fairly high voltage between the lines and chassis.

How important is the isolation resistance? IS HI-POT a concern, or simply resistance to chassis? As far as I know, you can remove any protective devices when you are doing the HI-POT.

 

[MikeSD]

transorbs specs

put optotransistor between input of you system and transorb

Opto-isolators requrire different supplies on either side, to achieve isolation. I don't have that and opto-isolators won't work for this. Besides, I said I considered that but opto-isolators can't be used on the power lines, in any case.

 

[artem]

testing of transorbs

"Because that is the requirement. The requirement is 100MOhm between pin and chassis."

measured under what condition ? What is the measurement voltage to measure input impedance?
If you are referring standard it should be stated there.
If measurement voltage is known and it is less than minimum detectable voltage on discrete input for logic level 1, you can just put circuit which opens only when input source voltage is above that measurement voltage. Put it between input source with transorb and your modified (as due to level shifting volatge will be decreased) discrete input .


Regarding optocouplers :
if you are allowed to draw few milliamps from input source , and can separate input return line and as far as you said that - there is pullup on input side - connect phototransistor to that pullup and ground , and put led through current limiting resistor to input source and another side to its separate return line .

 

[MikeSD]

low capacitance transorb

My first suggestion is to look carefully at the Transorb specs. What voltage are you testing at? How close is it that to the 'knee' of the transorb. Maybe selecting a Transorb with the next higher voltage rating would give you better isolation.

That was my first correction. Our tests are done at 45 volts and the transorb I was using had a reverse standoff of 48. So I selected another (58V) that was better but still not good enough. The reverse leakage spec was 20uA at 58V. Not good enough. I can't select others because I will lose the protection I need and have a rather high lightning current pulse to deal with. I have a 320A peak pulse current to deal with. I'm currently using MDE 30KP58CA diodes.

Also you may want to research Surgectors. I did a quick search for low leakage transient suppressors and found some Surgectors with a leakage of < 50nA so at 100 volts that would give you well over 300 MOhms for each one. I don't know if they would work for you otherwise.

I will investigate that. I'm not familar with those.

Update: I have been doing a little checking and this might be an option. My lightning requirements are Lightning Induced Pin Injection of 1600V @ 320A. If I can find a suitable surgector, this might work.

I don't understand why you installed the transorbs from each line to the chassis. I think you should have installed one from each line to ground. After all, you need protection from overvoltages between these lines.

Because that is the requirement. The requirement is 100MOhm between pin and chassis. I think the transorb must be between signal and chassis (as in our design), or I run the risk of the surge entering through the return path and finding a path to chassis on it's own. Placing a transorb between the return and earth ground, doesn't help my insulation test problem. Still have parallel paths.

 

[funster]

transorb capacitance

you can use optic coupler or isolating transformer

between source and the input.

best regards

I have a problem and am having difficulty with the solution. I have a design, for which I need to adapt to a new project. I cannot change the internals of the design.

The inputs to the design consist of 28VDC (3 lines), 28VDC RETURN, and 3 discrete signals.

The new project requires lightning protection (on every input line) and I have installed transorbs from each of the inputs to chassis ground. The problem lies in the fact that I also have an input insulation requirement that isn't being met with these transorbs. The requirement is 100 Meg Ohm from pin to chassis.

The problem exists because the input discretes have 10K pull-ups to 28VDC, inside the module that I can't change. with these pull-ups, all of my 5 inputs are placed in parallel, and even with the low leakage of these transorbs, this parallel combination gives pin to chassis resistance of about 30 MOhm.

I have tried placing a series diode inline with these inputs but that only improves the isolation but doesn't get me all the way there. And, I don't think stacking more diodes is the proper solution.

What are some good solutions to get the insulation resistance needed? I have considered opto-isolators but since there is only one source and since that is the source of my problem, I don't think this will work. Beside, I can't use opto-isolators on the power lines.

Any suggestions?

 

[newelltech]

transorb testing

you can use optic coupler or isolating transformer

between source and the input.

best regards

He's already answered why he can't do it this way a couple of times. Doesn't anyone read the replies anymore???????

 

[JohnJ]

opto-isolated discrete input

Someone mentioned gas tubes. We had a similar requirement for Medical equipment and found the "gas discharge" type surge protectors have excellent specs. They are relatively cheap and have only glass case leakage. I saw some as low as 50V. They also are not consumed by surges as transorbs are. One requirement is very low capacitance and they are excellent for that.

Medical has to have a highly isolated "patient ground". The problem is they want this stuff to ignore electrosurgical work, which (by a foggy memory) something around 5000V. Try protecting you EKG or EEG microvolt inputs against that!

Just to reinforce what others have said; 100meg ohms is very hard to measure. If you just breath on the insulator you will fail from the humidity. Even a PC Board will have trouble with that if ever left in high humidity for any time. The list goes on and on.