I am attenuating an audio frequency (50 KHz) signal that is pulsed at 666 W for 300us with a period of 300ms. I estimated the power dissipated averaged over a period to be approximately:

\$\frac{300 \times 10^{-6}}{350 \times 10^{-3}}\times 666 = 0.5709 \approx 600\text{mW}\$

In the past, I have had no issue with going smaller than the peak power I expect to see on the power rating for the resistor as long as the average over the (reasonably small) period was less than power rating for the resistor. The resistors warms up, but they are ok long term.

Now I am working with some mini circuits attenuators (VAT-XX) that are rated for 1W with various attenuations (1dB - 20dB). Running this signal through these attenuators results in "POP POP POP" and some dead attenuators. Before the failure, measuring the DC resistance gave some reasonable value around \$300\Omega\$ for the chain of attenuators. The DC resistance is now in the 10s of mega-ohms. The failure happened about 5 or 6 periods after I turned on the source and the attenuators did not heat up. The popping continued after the first pop and has a period that audibly matches the signal period.

I am skeptical that the failure is due to exceeding the power rating and I think that it is possible that the attenuators were damaged by the voltage during the pulses, which is ~850V peak-to-peak (perhaps the coax dialectric was broken down).

I would appreciate any helpful suggestions on why the attenuators failed.

  • \$\begingroup\$ After the devices fail they measure as open, that does not indicate a dielectric failure as the dielectric is there to isolate. If that would fail it might fail and short circuit. I agree with Neil_UK's answer that the resistors are damaged as that can result in a open circuit. \$\endgroup\$ – Bimpelrekkie Jul 18 '18 at 19:29
  • \$\begingroup\$ audio frequency (50 kHz)? That's just batty. \$\endgroup\$ – Brian Drummond Jul 18 '18 at 21:18

You have exceeded the pulse power handling of the attenuators.

Some resistors, like bulk water, or wirewound, dissipate the heat in a very large volume of material, so can have very high pulse power ratings.

Some, like thin film, have a very small volume of resistive material deposited on a more-or-less thermal conductor. Unfortunately, the thin film requires heat to be conducted away fast enough to keep its temperature reasonable.

The difference between 666W and 600mW may be key to your problem. I often exceed average power, but by a factor of a few, or maybe 10. More than a ratio of 10, you really need to look into the pulse energy rating of your resistors. You are trying for a factor of 1000!

Now some attenuators, and I know this from when I blew them up as a junior engineer, have a profile that concentrates the heat in just one tiny part of the film, and they drop like flies with the merest pulse overload.

You need to multiply 666W with 300uS, and find an attenuator rated for that number of Joules, as well as the average power.

| improve this answer | |
  • \$\begingroup\$ Thanks for the insight. As it happens, I am a junior engineer who is currently learning the same lessons. I assume that to get a pulse energy rating I will either need to conduct my own tests or contact Mini Circuits as the product page and datasheet are no help here. Do you have advice for communicating with manufacturers? \$\endgroup\$ – user146139 Jul 18 '18 at 20:00
  • 1
    \$\begingroup\$ you might want to start with taking an attenuator apart, and posting a photograph of the failed pad, could be very instructive. Manufacturers are always more responsive when you have evidence for what you;re talking about. \$\endgroup\$ – Neil_UK Jul 18 '18 at 20:43

Examine the specific heat of bulk carbon mixed with clay, used in some attenuators. And examine the specific heat of various metal films, used in some attenuators.

And consider the thermal timeconstant of copper (a useful value to memorize).

One cubic meter of copper has timeconstant of 9,600 seconds, measured from face to opposite face, that cube inserted into a thermally conductive chain identical cubes.

10cm cubes of copper have 100X faster Tau, or 96 seconds.

1cm cubes of copper have 100X faster still, or 0.96 seconds.

1mm cubes (about what you'll have in an RF attenuator, for bulk resistors, have Tau of 9.6 milliSeconds.

100micron cubes (0.1mm) have Tau of 96 microSeconds.

10micron cubes have Tau of 960 nanoseconds.

1micron cubes have Tau of 9.6 nanoseconds.

Do you feel lucky today?

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy