high voltage low power resistor 1

[zappedagain]

I have a RC compensation network on a high voltage op amp (Cirrus PA78). It consists of a 3.3K resistor in series with a 33pF capacitor that cycles at 1000 Hz maximum with a 210V maximum swing. The dynamic power dissipation in the resistor is f*C*V^2, or 1.45 mW. If I put 210V across a 3.3K resistor the static dissipation is 13W! There is no staic dissipation in this application though because this resistor is in series with a capacitor.

So I need a high voltage resistor but it doesn't need to be able to dissipate a lot of power; a bit of a contradiction. If I pick a resistor with a limiting element voltage higher than 210V and a power rating better than 1.45mW will it be reliable? Something like the Rohm KTR family seems to fit this.

Thanks,

John D

 

[Comcokid]

Your issue is very similar to a RC snubber on some switching power supplies.

The 1206 resistors I typically use (vishay/dale) are rated 200V, 0805 are 150V. In this situation I would use two 0805 or two 1206 resistors in series.

 

[itsmoked]

Follow Comcokid's suggestion. You just need to stack them up in series to exceed your needed voltage rating.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[itsmoked]

Also for thru hole you just step up the power rating as the operating voltage climbs with power rating. That's why you'll see 1/2W resistors where maybe only 1/16W would be expected to dissipate.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Is the 210v across the series RC? If so, then how much peak voltage does the resistor itself actually see?

The capacitor value 33pF is pretty darn low, almost within "stray" territory. And 1kHz isn't exactly RF (but beware harmonics).

 

[ScottyUK]

High voltage resistors are available with better stability under high electric fields than standard types. For some reason they are often made with a light blue glaze beneath the colour code.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[zappedagain]

Thanks, all. My indirect question is "if I stay under the limiting element voltage and under the power rating will I have a reliable resistor?" By the answers so far it looks like yes.

VE1BLL has a good question. As the 210V is across the RC, the actual voltage is dependent upon the signal rise time. My worst case is about 9 uS, so i=C*dv/dt (33pF*210V/9uS) gives 0.72mA, and i^2*R is 1.7mW, very similar to the actual dynamic power in the capacitor.

I have a space constrained design, so I'm trying to keep the solution as small as possible. it looks like one 2010 can do it, or two 0805 resistors depending up on the PCB space available.

John D

 

[OperaHouse]

I had an application that need high voltage in a small package. Picked a RCD RSF1A which had a 2KV pulse rating in a package not much larger than an older 1/2W. I guess that model is still available or one like it ( I bought a lifetime supply). Oh, and they are blue. See lots of applications using three series SMD resistors to get equivalent flash over voltage.

 

[VEBill]

"...gives 0.72mA..."

So the peak voltage across the resistor itself about 2.4 volts (?).

 

[zappedagain]

Ah, and now onto the interesting part. When I built my prototype I used two 1.6K 0805 resistors in series (not pretty but it works). When a technician built a second one he used a single 3.3K 0805 resistor. This resistor eventually turned a very interesting brown color and failed as an open circuit.

If my worst case current was 0.72mA the max voltage was 2.4V (<< 150V) and the max power was 1.7mW (<< 125mW), why did this resistor fail? It has zero static power dissipation (open circuit at DC due to the series capacitor), and low dynamic power dissipation. Why did it fail?

A capacitor failure could have brought this on, but the series cap is rated for 3000V (AVX 1808HA330KAT1A) so that seems unlikely. I'll need to test that though.

Thanks for the help,

John D

 

[zappedagain]

The cap is good; I measured 43 pF, less than 10 nA leakage current (at low voltage).

 

[VEBill]

Time to recheck assumptions and launch an investigation.


Or - forget about it and just use two resistors in series from now on. Used the time saved to drink beer.

 

[zappedagain]

Now that deserves a star!