Jay,

At least now I've seen some real evidence the gate drive overlap
issue and ringing are related, if not cause and effect. Agreed mine
is slow to turn on as well as off, though arguably being slow to
turn on might not be such a bad thing when one is fighting overlap.
;-) I think we can safely say now that my driver board simply
doesn't work. See below.

My twisted pairs are now about 1.75 inches of #18. 'Traces' on the
board are built up solder rows that are roughly the cross sectional
area of #12 wire. There are ground buses like this down each long
edge of the board, terminating in solder lugs which ground it to the
chassis when the screws are tightened. Currently only the two near
the IR2110 are grounded in this manner since I stood the board on
end to reduce gate twisted pair length. I think you are right that
the board layout and construction is at fault. Originally pins 2 and
5 of the IR2110 were grounded through a thick solder trail to the
ground bus on just one edge of the board. Pin 13 was run to the
ground bus on the opposite edge. This evening I modified it so that
pins 2, 5, and 13 are all tied together as close to the chip as
possible and to a 'trace' that goes all the way across the width of
the board to the ground buses along each long edge, taking the
shortest possible route. Doing this made the drain ringing
significantly *worse*! There's not much room left on that board for
adding ground plane pads.

I'm clearing this off the bench as I have some other things that
need to get done. I will have to consider the future of the project
at this point. Thanks for all your help Jay. Thanks to everyone else
who contributed also. Whatever the future holds I have learned a lot
from this endeavor.

Paul




On 04/13/2018 05:57 PM, [email protected] wrote:
> Paul
> 
> So ... we're still stuck at solving the gate drive overlap issue. Your's 
> seems to be slow to turn on
> and to turn off.
> 
> Think I questioned back a while ago the driver circuit board 'trace' widths. 
> Since you're using perf
> board I wonder if there are long runs using small gauge wires for the +V and 
> ground connections. If
> you look at my driver board, the top (component side) is a complete copper 
> ground plane ... which
> makes for a very low impedance ground. Screws through the four mounting holes 
> makes an excellent
> connection between the board ground plane and the amplifier chassis. Perhaps 
> you can come close to
> replicating this with pieces of copper flashing on the bottom of your perf 
> board. The IR2110 is
> charging and discharging a realtively high gate capacitance ... so if the 
> IR2110 connections to the
> + and - 'rails' and ground aren't reasonably stout the charging and 
> discharging process could be
> slowed ... maybe enough to cause the problem. For the record, I used #20 for 
> the twisted pairs ...
> but the ground connection also exists through the chassis.
> 
> Jay W1VD
> 
> 
> ----- Original Message ----- 
> From: "N1BUG" <[email protected]>
> To: <[email protected]>
> Sent: Friday, April 13, 2018 1:09 PM
> Subject: Re: LF: W1VD amp help - finally some progress!
> 
> 
>> Today I verified board connections again, including power to the
>> driver IC (tnx Luis). With the scope I checked +5v and +12v power at
>> IC pins while operating to be sure it is clean. All looked OK.
>>
>> I replaced the timing capacitors with recommended ones (tnx Jay). No
>> change.
>>
>> I replaced resistors in the timing circuit with closely matched ones
>> of a different type or at least different manufacturer. No change.
>>
>> I added bypass capacitors at all +5v and +12v pins on the ICs. No
>> change.
>>
>> I replaced the 10 ohm resistor / diode on each gate lead with 20 ohm
>> / diode. Finally some change! First peak of the ringing is down from
>> 62V to 45V. This is with 12.5V drain supply. Current consumption and
>> power output are down about 5% but efficiency remains the same as
>> before.
>>
>> Looking very carefully at my scope patterns and the one Jay provided
>> a few days ago, I still believe my gate voltage drops much more
>> slowly than his. I don't know why. If my gate voltage would decay as
>> fast as Jay's, I think the FETs would not cross conduct.
>>
>> I don't know what to do next. Is there any way I can speed up the
>> gate turn off? Can it be the 1N4148 diodes? Mine are of unknown age
>> from a large batch I got quite some years ago.
>>
>> Paul
>>