Hi Tom,
replying to your message from last year:
I am working on a solid state amp project. After a few false starts,
and damaged parts, I am trying a new approach and not pushing the
parts anywhere their limits.
As they say, you can very well push them to their limits - but not
beyond them! ;-)
The trick is to know where exactly each of the limits is. That's why
most people prefer to keep some safety distance to them.
I have a 300W boards purchased online from rfsource. They accept
either BLF278 or MRF151G parts. The boards arrived (from Greece) and
appear well made.
I'm looking at that company's website. I assume you bought the PHV30S
boards?
A BLF transistor was installed on the first board with a 48V supply
and Idq of 1A. The board would put out about 100W and then the drive
power needed to push it further would rise dramatically. The
current would not rise with the extra input power and power out
divided by DC power in looked fine.
That sounds like the gate bias is being pushed down by the RF drive.
Since I can't find a schematic for that board, I don't know if there is
a problem with the design, or perhaps you made some mistake during assembly.
> The board has a 3db pad in
front of the amp and I could feel the resistors getting hot so I
suspected the input matching as it simply uses a 9:1 transformer.
It's entirely normal that the power resistors in an attenuator circuit
get hot, when doing their job!
After a lot of experimenting with input matching yesterday, I could
get the first board to look good. Efficiency was typically about 50%
at 150W and got better as frequency increased from 1.8 to 28Mhz.
Output power and gain looked OK. However, the same approach did not
work on the other BLF278 board.
Check the DC bias as you apply drive. If it goes down, there is a problem.
There were two MRF151G's left from a prior experiment and I put them
on two other boards. They performed noticeably different than the
BLF parts, but similar to each other. At 150W, the gain was over
20db (including the pad) but the efficiency was very poor. It was
just under 50% on 160, but dropped steadily as frequency rose to
about 20% on 10. I presume this will get worse when I have real
antennas and not a dummy load.
How are you doing the necessary ultra low impedance ground connections
between the MOSFET's flange and all the different crucial ground spots
on the board? Even one nanohenry of stray inductance in those
connections is a problem! I understand that with the boards you bought,
the idea is to connect the flange to the PCB just through the heatsink
and the mounting bolts. This is a poor method. I would absolutely add
some copper straps from the flange bolts, soldered to the PCB ground.
Does that PCB have a ground plane on the backside, or is it single sided?
Many people believe that in a push pull circuit the grounds are
unimportant, because the signal is balanced. This would be true if the
circuit operated in class A! But in class B, or AB, or C, the signal is
NOT balanced at all. At any given time during most of the RF cycle, the
drain current flows only through one device (one half of your Gemini
MOSFET) and through ground. So the correct ultra low inductance
connections between the flange and all relevant parts of the circuit is
as important in push pull circuits as it is in single ended ones!
The gate circuitry has some more balance, because of the relatively high
capacitance of the gates, often further aided by low value swamping
resistors. But still it can often help to have a low impedance between
the driver circuit center point, and the MOSFET's flange.
Maybe the reason for the unpredictable performance of your boards is
simply a varying impedance in the flange-to-board connections, depending
on exactly how you mount and connect the MOSFET!
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|