[hpsdr] Predistortion and different types of amplifiers?

[Steven B. Dick]

In my previous life working with digital predistortion, I found a great deal
of difference in performance with different types of amplifiers.  The
performance often degraded as a function of the number of stages in the
amplifier - more stages resulted in more degradation.  I suspect the
degradation is possibly due to several reasons:
1. The amplifier's amplitude transfer characteristics may be more abruptly
non-linear as you start getting into compression.  That makes it more
difficult for the predistortion to correct, as a relatively small correction
change can cause a large output change around the nonlinear region.
2. There may be very large phase changes with more stages, again making it
more difficult for predistortion to correct.

A rule of thumb for required bandwidth is that the bandwidth should be at
least five times larger than the bandwidth you are trying to correct.  Ten
times more would be better.  Another rule of thumb, with regard to phase
corrections is that you need oversampling by at least a factor or 8, or
phase corrections will not work very well. In fact they could make things
worse.   

I don't know the specifics of the algorithm details, but it appears to work
extremely well for the high voltage LDMOS amp, looking at the graphs.  If a
particular amplifier is working poorly, it may be worth disabling the phase
correction part of the algorithm and only use amplitude correction and see
if performance is any better. 

Regards, "Digital Steve", K1RF  


Message: 5
Date: Tue, 11 Feb 2014 13:07:19 +0100
From: "Kjell Karlsen" <la2ni at online.no>
To: hpsdr at lists.openhpsdr.org
Subject: Re: [hpsdr] Predistortion and different types of amplifiers?
Message-ID: <op.xa39ahqinknqug at kjellkarlsen-pc.lan>
Content-Type: text/plain; charset=utf-8; format=flowed; delsp=yes


Hi.

I can confirm what Helmut says, the high voltage LDMOS transistors works
extremely well with predistortion. It is much more difficult to get PAs
using RD100HHF1 to work with PD. I have tested several Munin boards and the
best improvement I can get is 15-20 dB. Increasing the Vdd to 24V does not
help. And at higher frequencies the less improvements, sometimes I can get
no reaction on 10 and 6 meter.

I see that some of the testers have achieved better results with ANAN-100 so
I do not understand why the two amplifiers behave different. The PA in
Apollo using RD16HHF1 and the one in ANAN-10 using RD15HVF1 works perfectly
with PD. I have also tested with lower bias current and even at 100mA each,
the improvements are good.

As the transistors are from the same family, it is difficult to explain this
difference. Also the difference between Munin and ANAN-100 is strange. I
have used other material in the output transformer that gives better
efficiency but it is hard to believe this is the reason. So there is a lot
of testing left to solve this problem. I am open for ideas!

73, Kjell



P? Tue, 11 Feb 2014 12:14:41 +0100, skrev Helmut, DC6NY <dc6ny at gmx.de>:


 1392236450.0