Solving the Mystery of Unstable Amplifier Performance

AI Thread Summary
The discussion centers on an RF device using the AMMC-6220 in-line amplifier, which experiences gain spikes at specific high-frequency bands that vary by unit. Lowering the amplifier's voltage eliminates these spikes, while increasing it exacerbates the issue, despite the bias voltage being within specifications. Adjusting the input match also influences the spikes, raising concerns about the overall stability of the system, as the AMMC-6220 is designed to be unconditionally stable. Participants suggest investigating the total system's stability, checking for oscillations with a spectrum analyzer, and ensuring proper isolation of power supplies and components. The original poster remains perplexed and seeks further advice on troubleshooting the problem.
lennybogzy
Messages
94
Reaction score
0
Gentlemen and ladies,

I have an RF device and an in-line amplifier (specifically an AMMC-6220 - http://www.datasheetcatalog.com/datasheets_pdf/A/M/M/C/AMMC-6220.shtml). Certain units develop gain spikes at the high-freq band. They are in very specific frequency locations, and vary in amplitude from unit to unit.

Lowering the voltage to the amp eradicates the spikes. Raising it has the opposite effect. The bias voltage is well within spec.

Another thing: playing with the input match to the amplifier has an effect on the spikes. However, the AMMC-6220 is (supposed to be) an unconditionally stable amplifier. This shouldn’t happen, assuming that what I’m seeing is oscillation.

I’m stumped, and can’t do much except keep replacing this relatively expensive amplifier. The manufacturer doesn’t know what’s going on, nor do engineers a lot smarter than I. I can’t even seem to isolate the problem to a bad lot of amps. Any advice, theories, ideas, words of emotional support, would be appreciated.
 

Attachments

  • gainspikes.JPG
    gainspikes.JPG
    34.9 KB · Views: 513
Last edited:
Engineering news on Phys.org
lennybogzy said:
Gentlemen and ladies,

I have an RF device and an in-line amplifier (specifically an AMMC-62220). Certain units develop gain spikes at the high-freq band. They are in very specific frequency locations, and vary in amplitude from unit to unit.

Lowering the voltage to the amp eradicates the spikes. Raising it has the opposite effect. The bias voltage is well within spec.

Another thing: playing with the input match to the amplifier has an effect on the spikes. However, the AMMC-6220 is (supposed to be) an unconditionally stable amplifier. This shouldn’t happen, assuming that what I’m seeing is oscillation.

I’m stumped, and can’t do much except keep replacing this relatively expensive amplifier. The manufacturer doesn’t know what’s going on, nor do engineers a lot smarter than I. I can’t even seem to isolate the problem to a bad lot of amps. Any advice, theories, ideas, words of emotional support, would be appreciated.

Could you post a link to that AMMC-62220 amplifier? I'm having trouble finding it with Google. What is the RF device that feeds the amplifier? Where does the output of the amp go? If you go straight into the input of the spectrum analyzer with coax and no other feeds to other devices, do you still get the same results?
 
Thanks for the reply and I apologize for the typo. It's an AMMC-6220 and the datasheet can be found here http://www.datasheetcatalog.com/datasheets_pdf/A/M/M/C/AMMC-6220.shtml

The problem is the amp is part of a series of cascading amps, so I would have to tap directly onto the trace after this amp. There are a few attenuators, an equalizer, some diodes, some other amps and various filters along the way. I will have to give more details on Monday.
 
Any way you can get an S-parameter sweep?
 
attached to original post
 
The amp may be unconditionally stable but your total system may not be.

I can't see the frequency range on your plot, but I'd hook up a DC to 30ghz spectrum analyzer and look for an oscillation that depends on the bias voltage.

The second spike on your chart looks to me like what happens when the network analyzer sweeps past a resonance.

You need to set up your analyzer to do a gain-phase plot and start looking for ways that the output is leaking back to the input or bias supply.
 
lennybogzy said:
attached to original post

My bad, didn't see that. Is this a sweep of the rf device and the amp? If that's the case, is your rf device stable? Also, how are the other S-parameters impacted?

I'm going to agree with Antiphon, and I definitely would check on the isolation of your power supply/amps/inputs.
 

Hot Threads

Recent Insights

Back
Top