Protecting LM324 from High Voltage Sparks

[Ravaner]

Hello. I use an OpAmp LM324 as follower behind a voltage divider connected to a high voltage in the 8 kV. That works, but in some cases a spark can occur between this HV and ground and in that case systematically the LM324 is destroyed (distance between spark place and AmpOp is about 4" and ground line is common). Is there a good protection to avoid this risk ? For instance a Zener diode parallelized with the OpAmp input line ?

 

[yungman]

You can use small transorb for protection. I don't remember the part number anymore, but they come is small package of two back to back or something. It is very common and I believe we got it from DigiKey. I also use schottky diode sometimes. Main thing is to make sure it does not change the reading because leakage current might induce error on your divider. But I had no problem getting accurate reading for 15KV before.

 

[Ravaner]

Many thanks for answer. I'm discovering these kind of component. Effectively I've to check that read value (which is used in a regulation loop) will not be disturbed by leak current.

 

[skeptic2]

In order to really correct this problem you should try to discover what exactly is happening to the op amp that is destroying it. For instance, at the moment of the spark what happens at each input relative to the V+ and ground connections? Try to see if you are getting voltage spikes on any of the pins, and you may need a storage scope to see it.

 

[yungman]

In order to really correct this problem you should try to discover what exactly is happening to the op amp that is destroying it. For instance, at the moment of the spark what happens at each input relative to the V+ and ground connections? Try to see if you are getting voltage spikes on any of the pins, and you may need a storage scope to see it.

It is always on the input pins. I did a lot of these kind of protection and never once it is not the input. Ultimately it's the current that burn, power pin always has cap to buffer, output can handle some current. We automatically put transorb at the input of these kind of circuit.

One thing to be careful is to make sure there is enough distance from the HV source to the input pin and the resistor is rated for the voltage. Creepage distance is very important and the cleaness of the PCB is very important.

 

[jim hardy]

What is causing the spark?

I once had a similar problem - the high voltage test point was right next to a resistor connected to an opamp a few inches away. When you put a test prod on the test point it was prone to slip and cause a short, wrecking the opamp.
we had the board re-designed to move the test point and made the test point a different style less prone to slip.

You might need to insulate something with a glob of RTV sicone .

Be aware that garden variety resistors have a maximum voltage rating typically 400 volts.
High value resistors can reach that limit before reaching their power limit.
Have you checked the resistors in your high voltage divider? There are special HV resistors made for that application.

Check carefully. Mother Nature will find ALL the weak spots in your design.
Hopefully in your shop not the customer's.

 

[yungman]

Creepage is to most likely cause of problem. Creepage is the current travel over the surface, it is much worst than air gap. It depend on humidity, how clean is the surface. We troubleshoot creepage by turning off lights and induce arc and look for a lighting shot! That is the reason I mentioned about using the right resistors. Caddock or Vershai??(spelling) make high voltage resistors. If using normal 2W carbon, you have to be careful of the pcb surface under the resistor. I always route channel on the pcb to increase the creepage path.

 

[skeptic2]

I don't know what the failure mode is and I'm just making this up as an example. Suppose when a spark occurs it creates a low impedance path from the high voltage source to ground and it raises the ground voltage above that of the op amp inputs. Many op amps use PNPs on the input pins and if the ground connection were made positive with respect to the input, it might cause the base collector junction to become forward biased and destroy the transistor. Though the problem may be with the ground, it destroys the input pin.

 

[yungman]

I worked many years in high voltage environment. It is very hard to pin point arcing problem. Using test equipments to look is not exactly a good idea, we had so many dead Fluke meters, computers, scopes to show for. We were joking that we actually went out and bought 4 or 5 cheaper desk top to hook up to the system instead of the more expensive computers. We called them the "fuse". One arc was all it take to pop a scope, a computer. I won't suggest hooking up a scope probe to look for arc!

Our usual way was to turn all lights off, induce an arc just by arcing in something, other potential arcs might follow. I actually gave this the name "sympathetic arc". One place arc, the other might follow! Then you look for the flash and the path of the flash. You'll be really surprised how the arc path goes. It is quite mysterious! I think is even more so than RF. No matter how hard RF, it is still in electronics. With arcing, you have other totally unrelated factors affect the circuit. Like in HV environment, you cannot have any conductor that is not tie to some known potential, it will get charged up and arc! It is tricky.

As I said, we always protect the input and I have never run into problem. We rather put in all the protection and have less problem to start with. And still it can make your hair turn grey when you get to 12+KV.