Hello N1BUG, Hello Paul, by way of comparison my home made PCB is
attached right by the FET's by copper wire 90 degree angles, soldered
to ground plane of PCB and ground plane of the main amp double sided
board. Teflon co-ax (probably overkill, but i like soldering Teflon
stuff as it doesn't melt!) lead between output socket of main amp
board to input socket of LPF is about 2 inces long. If I get chance
tonight I'll scope the drains with amp direct to dummy load, no LPF as
a comparison you could look at.
I would tack those 1uF ceramics on and see what happens. My output
transformer is not as pretty as yours and no huge care was taken
winding it, in fact it looks pretty ugly ;) I used pure copper, multi
strand heavy speaker cable for the primary and normal UK 15 Amp single
core mains wire for the secondary. Apart from short gate leads I can't
imagine that at 136khz PCB design is that critical. If it were me I'd
shorten the gate leads even if the PCB was just hanging on the other
leads with a hefty wire grounding it to the main board, and shorten
the co-ax to the LPF, but mostly I'd try those decoupling caps! They
made the biggest difference for me, including helping stop any odd
oscillations.
I remember trying a different driver board and using about 4 inch long
leads to the gates. the patterns were terrible and initially I blamed
the different driver design. I then mimicked the short leads of the
W1VD one and the results were fine. *BUT* the poor waveforms were also
apparent on the gates, not just the drains. Try those de-coupling caps! :)
I can't see if you used sockets for the IC's, but if you rebuild I
suggest you do use them as a FET bang can sometimes take out the
driver IC and once the other one, too
Monday, April 9, 2018, 2:22:06 PM, you wrote:
> Hello Chris,
> There are plenty of possible causes here I think. When it comes
> right down to it, I'm not at all happy with the end result of my
> construction on this amp. If it ends up not working, it's because
> I'm not a good enough builder. Despite building for 35 years
> including many QRO tube amplifiers, I felt this pushed my skills to
> the limit... or beyond? It's not the fault of the design of course,
> it's me. This requires a very different skill set from what I've
> developed through the years.
> The driver board you see is my fourth attempt yet I'm still very
> unhappy with my layout and on-board routing of connections.
> I probably made a huge error in design and mounting of the driver
> board. Everyone seems to agree my gate leads are too long. This may
> take a while. If I'm starting over I'll have a few more goes at
> trying to come up with a more sensible layout and on-board routing.
> I'll probably have to design the new board so it can be soldered to
> the ground plane for mounting. The only other way I can think of
> would involve complete deconstruction of the amp to drill and tap
> new holes in the heat sink for relocated board mounting standoffs.
> I literally spent two weeks and likely 50 attempts at winding the
> output transformer, still not happy with it. I'm not sure the
> primary winding is acceptably tight.
> Today I'll try swapping out the cables connecting amp to LPF and LPF
> to dummy load. No reason to suspect bad cables, but I've seen
> strange stuff happen. The cable between amp and LPF is a bit long at
> 4 feet, maybe it doesn't like that. I'll have a look on the PSU side
> of that choke with the scope too. Good idea. Could be some RF there
> causing the PSU to go insane.
> I haven't added bypass caps on the driver board yet. I'm out of .1
> uf caps. I have some 1 uf ceramic but that may be a bit much? I'd
> probably get shorter leads by tacking them on the back side of the
> board. Bear in mind I'm testing at 13V which gives about 65W output
> so there's not massive amounts of RF around yet.
> Still puzzling over why my gate waveforms look OK (I think?) but
> drains don't. To my usually wrong way of thinking that suggests the
> ringing problem is caused by some defect in the drain circuit or
> outside the amp (cables, LPF). I think I'd like to try and eliminate
> those possibilities before I go back and completely redesign the
> driver board and mounting to shorten gate leads. No doubt it'll
> still come to that but I can hope not a bit longer. :)
> Paul
--
Best regards,
Chris mailto:[email protected]
|
