[Amps] More pictures...

[Manfred Mornhinweg]

Hi Martin,

> For those who wants the pictures of the EB-104 construction I prepared a 
> PDF guide from the pictures grabbed out from the faulty site. Download 
> here:
> 
> http://www.ok1rr.com/download.php?view.167

Thanks for that! I downloaded it, and had a thorough look at it. Well, I 
had thought that there would be some good information about some modern 
legal limit SS amp design, and then discovered that this is "only" the 
assembly procedure for a design that must be roughly 20 years old, and 
has been available on the web almost "forever"!  Well, that's my fault, 
I should have read closer.

Still, a few comments about the job shown in those photos:

It would be better to FIRST bolt the transistors to the heatsink, with 
thermal compound and all, and THEN solder them. This technique results 
in minimal chance for undue mechanical stress on the connections. When 
doing it in the opposite order, as shown in these photos, inevitably the 
transistors will end up with more mechanical stress on them, which might 
lead to failures in extreme cases.

Judging from the photos, the solder didn't flow very well under the 
transistor tabs. I normally tin them lightly before installing them. 
This results in much better soldering, closer to the transistor body, 
which is important specially in the source tabs, to minimize the 
inductance. True, with 50V MOSFETs running at HF this is not too 
critical. But with 12V transistors running VHF and higher, it can make 
the difference between an amplifier and an attenuator!

The soldering of the power resistors also looks really poor! I suggest 
to improve this, before having the resistors fall off when unsoldering 
themselves from their own heat! The rule used by old-time radio 
technicians, who often mounted components a bit like that, was: First 
you wrap the leads two turns around the support (or the other resistor's 
leads, in this case). THEN you solder. Tin-lead solder is a very soft 
material that easily creeps. It is not mechanically strong enough to 
support a large component like these resistors over a long time, without 
some additional help like the wraps. And that's specially true when the 
component gets hot in use! I have had my share of power resistors 
falling off, when I mounted them in this way! Also, these resistors 
might dissipate a good portion of the heat through the leads, if they 
are properly mounted (short leads to something large, like the board). 
This additional dissipation capability is often useful.

Finally, a comment about the use of thermal compound, and this comment 
is just my opinion, not a guaranteed truth: It seems to me that when 
spreading out the thermal compound to a thin layer, and then mounting 
the part, it's just too easy to get areas which aren't properly filled 
by the compound, just as happened here. One problem is getting the layer 
even enough (almost impossible), another is avoiding the inclusion of 
air bubbles, which might never come out. For that reason, I think that 
my method of applying the thermal compound is better: I apply a rather 
large blob of it in the middle of the surface, and small clean beads of 
it around the screw holes. Then I mount the part, and torque it down 
properly, so that the compound squeezes out of the joint as much as 
possible. If necessary, I warm up the assembly to get the compound to a 
lower viscosity, and then re-torque the bolts. This method never leaves 
any air bubbles or unfilled areas between the part and the heatsink. Its 
drawback might be that the compound layer ends up thicker, but I don't 
think that really happens, judging from the low thermal resistance I 
obtain from these joints. And it should not be done if the compound is 
too stiff, because the flange of the transistor might be deformed. But 
then, such stiff compound should not be used anyway, because it will 
never form a thin layer! The Arctic Silver compound surely is fluid 
enough to use this method.

Well... I'm still on the lookout post for some legal limit SS PA design 
that has advantages over the age-old and well-established ones, which 
typically use 50V MOSFETs of ancient Motorola lineage, in class AB. The 
advantages could be in obtaining high efficiency by running class E or 
class F with envelope restoration, or at least in doing away with the 
power supply by running directly off the line, and using low cost 
transistors. If nobody comes up with something like this soon enough, I 
guess I will! We need to innovate!

But right now I'm playing with PICs rather than RF MOSFETs. There's less 
to blow up with them. When I get fed up with those little cute digital 
toys, I might return to the big and serious world of PA development! :-)

Manfred.

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