[Amps] SS Amp questions

[Manfred Mornhinweg]

Tom,

> Unfortunately, multiple
> design, manufacturing, and quality issues have repeatedly put the
> product on the shelf for 6 other contests. 

Well, it's an MFJ, after all... I have also bought a few MJF producst 
over the years, because they just have an attractive combination of 
price and features. But the quality is typically poor. For example my 
MFJ-784B filter was so badly soldered that it arrived with half a dozen 
components flying around freely inside the box, and many others had bad 
solder joints.

> It has been back to MFJ
> multiple times for repair, including one set of blown finals.  While
> the contest duty cycles are high, I do not push the amp over 1KW, the
> antennas are well matched, and the bandswitching is automatic so the
> amp has not been abused.

Did you find out why they blew? If not, my best guess would be simple, 
plain overheating. I haven't had a chance to inspect an ALS-1300. I have 
only seen it in the advertising photos in QST. But based on those 
pictures, and on my own experiences with solid state amplifiers, I think 
that the FETs are running far above their safe temperature limit in that 
amp.

> My question is how many of the transistors need to be matched? 

I would set these rules:

- Any transistors that go in parallel need to be matched.
- Any transistors that work in push pull, too.
- Any transistors that share a single bias adjustment, too.
- Any transistors that work in separate modules, with separate bias 
settings, and that contribute power to a combiner, do not strictly need 
to be matched in most cases, but ideally they should be matched too!
- If the circuit design forces a good degree of load sharing, the 
transistor matching requirement can be relaxed, often to the point tat 
no matching is required at all.

Then there is another matter, simplifying things: Transistors that have 
a high level of homogeneity from the factory might not need to be 
measured and matched in groups. They might be close enough just as they 
are. Specially when load sharing is included in the design. This is what 
I'm doing in my amplifier (which might reach publishable maturity 
sometime in 2050 or so...).

> I noticed that all 8 of
> the original transistors are hand marked with a "17."  I am guessing
> that is the measured gain and that is how they are matched.

Mosfets would more likely need matching for cutoff voltage than for gain.

> Also, even if matched, there will be some variation.  What is the
> best way to adjust bias?  Do I just match the quiescent currents?

If you have separate bias controls, set them so that all quiescent 
currents are the same. And the total value of quiescent current should 
ideally be determined while measuring the IMD. But often the difference 
isn't dramatic, so the important thing is that there is some quiescent 
current, and that it is small compared to the max current. A typical 
value would be like 100 or 200mA total. It must be big enough so that it 
will never become zero (under any thermal condition), and low enough to 
never cause thermal runaway.

> Should I be watching the waveforms at full power and then adjusting?

No. If you want to watch waveforms, do it at low power. But it's hard to 
see what you need to see.

> Should I be adjusting based on a spectrum analyzer and two tone test?

That's much better.

> Right now, the amp appears functional and will make power, so the
> diagnosis was correct.   However, it is not yet usable as the current
> draws between modules is different and there is a big imbalance in
> output of the two modules.  I am good with replacing the other two
> transistors on the board, but not too keen on buying 8 new matched
> ones.  Either way, I will need advice on the biasing process
> 
> Any suggestions appreciated.

Leave the other 4 transistors alone. Try to measure how much difference 
there is between the 4 in the module that failed. Replace all four only 
if the ones you have now are too mismatched. At least that's the 
cheapest solution. The best, of course, is to get a full matched set of 
all 8, but given your previous experience with that amp, it's not up to 
your operating requirements, so it won't be worth it.

I would be interested in knowing the size of the heat spreaders of that 
amp (if any!), and of its heatsinks. If you can measure that (including 
the thickness of each section), I could try to calculate how hot those 
fets are running at 1kW. When doing such thermal calculations, many 
people, even professional engineers, forget such "minor details" as the 
thermal resistance along the heatsink's fins.

Manfred

========================
Visit my hobby homepage!
http://ludens.cl
========================