Hello Paul,
I was on holiday last week and didn't pay much attention to what passed in the
reflector.
So my appoligies if I tell things that have been told earlier.
The "both sides conducting" issue is typical for the power MOSFETs as the turn
off delay is much larger than the turn on delay. Exactly for that reason most
class D push-pull PA's have a gate resistor to delay the turn on a bit and a
"reverse" diode to bypass the resistor at turn off (so the turn off is not
extra delayed).
If you see the one side turning on befor the other side is turned off it migh
be worth to increase the gate resistors (from 10 Ohm to 15, 18 or even 22 Ohm).
This will increase the turn on delay and maybe solve the issue.
73, Rik ON7YD - OR7T
________________________________________
Van: [email protected] <[email protected]>
namens N1BUG <[email protected]>
Verzonden: woensdag 11 april 2018 16:10
Aan: [email protected]
Onderwerp: Re: LF: W1VD amp help - more waveforms
Hi Rik,
Of course that is true and I can see a difference in the gate
waveform with Vdd on and off. But it doesn't explain why under same
conditions (Vdd off) Jay and I see quite different gate waveforms.
I just did another test, gate waveforms with Vdd on.
http://www.n1bug.com/gates12-wdv13.jpg
Depending how I interpret this, I can almost believe I see both
sides conducting for a very brief instant.
Looking at this critically, I can see that one starts to turn off
before the other starts to turn on. So there is a delay being
generated by the driver circuit. But it's not clear that one is
fully stopped conducting before the other starts. Seems as if the
decay time is not fast enough. Maybe it's the length of those gate
leads messing things up after all...
Jay's article says the 10 ohm resistors and 1N4148 diodes slow gate
turn on and speed up gate turn off. Given that both my turn on and
turn off seem slower than Jay's, can it be my 1N4148's are somehow
not the same as his?
Paul
On 04/11/2018 09:14 AM, Rik Strobbe wrote:
> Hi Paul,
>
> about testing without drain voltage:
> Be aware that the input (gate) capacitance will be a lot larger at 0V Uds.
> This might affect the waveform at the gate!
>
> 73, Rik ON7YD - OR7T
>
> ________________________________________
> Van: [email protected] <[email protected]>
> namens N1BUG <[email protected]>
> Verzonden: woensdag 11 april 2018 14:59
> Aan: [email protected]
> Onderwerp: Re: LF: W1VD amp help - more waveforms
>
> Chris,
>
> I think it was after looking at your driver board photos that I
> decided to order these particular caps. They look exactly like
> yours. ;-)
>
> I haven't tried the bypass caps yet. Given we're now testing without
> generating any RF (no drain voltage on the FETs) I'm a bit dubious
> but I will try them. It's four caps, right? Pin 14 of the 74F74 to
> ground, pins 9, 3, and 6 of the IR2210 to ground.
>
> The other day Jay asked about power to the IR2110 and trace
> thickness... which might affect drive to the FETs but probably not
> the timing issue we're now chasing. However, for sake of
> completeness in this thread it's a 25 amp supply!!! through 3 feet
> of #18 wire to the board. The weakest link is about 3 inches of #22
> wire from the board +12V input terminal to pins 3 and 6 of the
> IR2110. Anticipating a possible question... no it's not the same
> supply I'm using for the FET drains, just a spare bench supply that
> was conveniently located for testing.
>
> Paul
>
>
> On 04/11/2018 07:40 AM, Chris Wilson wrote:
>>
>>
>> Hello Paul, effectively the same as I am using, so I think you can
>> safely discount those. Have you tried de-coupling caps yet, especially
>> on the driver IC? I think the hump on the gate waveform is the driver
>> overcoming the gate capacitance of the FET's.
>>
>> Wednesday, April 11, 2018, 12:30:38 PM, you wrote:
>>
>>> Jay,
>>
>>> The caps I used are polyester. Here's the datasheet:
>>
>>> https://www.mouser.com/datasheet/2/212/KEM_F3101_R82-1103738.pdf
>>
>>> It mentions timing applications, but I am no capacitor expert!
>>> Comments appreciated. Mine are 5% tolerance marked 1nJ100.
>>
>>> I always measure resistors because I don't trust my eyes. The ones
>>> used here measured about 14.98k on my Fluke DMM.
>>
>>> My proto board construction technique uses 'rows' of solder which
>>> undoubtedly results in higher than usual stray C. However I would
>>> expect it to be a few pf, not something that would throw the circuit
>>> off this much.
>>
>>> Paul
|
