Power Amplifiers Blow-up in a Bi-Directional System - RF Cafe
Post subject: Power amplifiers blow-up in a bi-directional
system Posted: Thu Dec 08, 2005 7:22 am
The transistor of the
final stage of the amplifiers in a bi-directional amplifier system
often blows-up. The system is the usual bi-directional/repeater
system: amplifier in each direction, combining diplexers at each
end, middle filters for the isolation in the cross-over band. The
amplifiers have high gain – above 90dB. The diplexers and the middle
filters supposedly give enough isolation so that there is no closed
loop around the amplifier-diplexer loop. The pass band of the diplexers
and the other filters is 5MHz and the distance between the 2 bands
is also 5MHz. This is anywhere in the 400-500MHz or 800-960MHz.
When the power supply is switched on no oscillations occur. One
can measure the gain in each direction on a network analyzer. But
when you start measuring P1dB or IP3 at some power level one of
the final amplifier stages in one or the other direction blows-up.
The amplifiers are definitely unconditionally stable. In other systems
with the same final stages, but in which the gain is lower and the
filters can give much better isolation this never happens. Yes,
the isolation in the high gain system is a bit too close to the
gain but there are never oscillations. We’ve come up with the idea
that the distortion products cross couple in the opposite bands
and somehow an instantaneous oscillation happens through the distortion
products and that blows the final stages. It seams that the blow-up
happens only when certain level of power is applied at switch-on
of the signal generator or when the power is being increased and
the generator switches through a internal attenuator power level
which is seen on the spectrum analyzer as a broadband pulse.
Comments and advices please!
Post subject: Posted: Thu Dec 08, 2005 9:36 am
looked at the current? Approcahing the P1dB point will increase
the DC current. Perhaps the part is overheating? Another possibilty
is tranisents on the bias line. I would check to see if there are
Posted: Thu Dec 08, 2005 9:50 am
The amplifiers do not blow
up if the loop is open – lets say you disconnect the amplifiers
in the other direction. The blowing-up final stages have been tested
with input powers higher than the power they deliver in full compression
including at frequencies where the filter (diplexer) at the output
is rejecting (or fully reflecting back). It’s definitely something
that goes around the loop of amplifiers-diplexers but we could not
model it or fully explain how it happens.
Post subject: Posted: Thu Dec 08, 2005 3:16 pm
What is your margin on Isolation? (Both gain summed minus both isolations
in one channel).
Are those class A or cass AB amplifiers?
Post subject: Posted: Thu Dec 08,
2005 3:22 pm
Class A. Isolation supposedly at least 10dB.
Post subject: bidirectional amplifierPosted:
Thu Dec 08, 2005 4:10 pm
What is the rejection of the diplexer
and middle filter for the other band? Are the amplifiers running
saturated? Are the bias lines sufficiently isolated?
Post subject: Posted: Sat Dec 10, 2005 7:16 am
Joined: Mon Jun 27, 2005 2:02 pm
description of your problem, it seems that you are right in your
idea that the problem can be caused by intermodulation products.
If this is the problem, I suggest to add additional filter stage
to further supress these products, because from your description
it seems that you are working on the edge and the difference between
this system to the other systems with lower gain is small. You can
also alternatively/in addition put a low value attenuator between
the middle stages to the final stage in each direction, that will
reduce a bit the gain and provide further isolation.
keep us posted if this helps! I will be keen to help you more.
Post subject: Posted: Sat Dec 10, 2005 1:31 pm
My thinking is basically the same and actually
your suggestions have been tried already. The attenuator helps but
the we can not deliver the promised gain; more filtering helps also
but then the system is bulkier and more expensive. But again, that’s
the only way, so now the right people have to be convinced that
that’s the only way, which is the most difficult part of the job.
Post subject: Posted: Sat Dec 10, 2005
Joined: Mon Jun 27, 2005 2:02
Hi again johnny,
Yes convincing the managers who often aren't engineers is sometimes
harder than to solve the problem. I am happy that you found the
solution to the problem. I guess that there is no simple solution
to such problem but addition of components as attenuators and additional
filtering which could somewhat degrade the performance. Good luck
in the next steps!