PWM modulator problem

[Bronek M0DAF]

I have built a PWM modulator using Steves PCB for the PWM generator, with my own layout for the output stage, but true to his circuit and components. The HT is 130 V and I have set the pulse width to give me 50 V for carrier via the output filter which is designed for 5 ohms. All the tests are conducted with a 5 ohm dummy load connected.

I wonder of anyone can advise me regarding a problem I am having with what appears to be instability associated with the PWM output stage, before I attack it with my soldering iron! Although the PWM generator produces a clean output on its own (the output of the generator and the gate waveform are both OK with no HT applied to the FETs) as soon as HT is applied I get what I consider to be a very unsatisfactory gate waveform. Although the transitions appear near vertical on the scope, they consist of high frequency spikes, both on the upward and downward transitions. Apart from that the waveform has a flat top. There is ringing associated with the transitions, but I feel certain that it is not all ringing. This behaviour is reflected in the drain voltage trace, and it looks as if the high frequency components are getting back into PWM generator circuitry, because there are glitches on the PWM generator output. Similarly, the DC output from the filter into the dummy load looks rough, as I would expect from the uncertainty in the position of the on/off transitions in the gate voltage. It looks as if the charge pump is working (as evidenced by the amplitude of the pulses referenced to ground).

I have photos of all the scope traces to which I refer but do not know how to include with this post. If anyone would like to see them, I can e-mail, or please tell me how to post.

Thank you

Bronek M0DAF

I have the following photos
Gate waveform without HT applied, 12 V peak
Gate waveform with HT applied
Drain waveform with HT applied (same scale as for gate waveform display)
Waveform at attenuator (feeding the optoisolator) without HT applied to FETs
Waveform at attenuator (feeding the optoisolator) with HT applied to FETs
Waveform at R415 on PWM board with HT applied to FETs

[kf1z]

Your photos have to be stored on another site ( photo sharing, or personal web site space etc..)
Then you post a link to them here.
You can't upload photos here.

Well,
The "spikes" sound normal to me.
But, would be nice to see the output.

Now, I have NEVER tried to look at the drains with HV applied.. only after the filter... There is a reason... Steve will say what that was..
( I think his words were "can't", because of how that portion is set up... different ground, etc....)

You are GOING to see some nasties, I believe, on the drains. I could be wrong....

What is inportant is after the filter... at the load.


Depending on how , and exactly WHERE all your grouns are connected, that could cause problems too.

But also, put the freq counter on your PWM output of the board, and make sure it's in the right frequency range.

I remember a time I had a wrong cap in the PWM timing circuit, and the freq was too low,,,
THAT made for some REALLY nast scope patterns!


Anyhow..
Email me the pics, if you have no where to post them, and I'll temporarily put them up for you.

kf1z@myfairpoint.net

Bruce
kf1z

[kf1z]

A couple good pics of your layout could be real handy too!

Bruce

[kf1z]

Here is a PDF of Bronek M0DAF's, scope pictures and information about his issue.

http://www.greenmountainradio.com/M0DAF_scope1.pdf

[Gilly]

Hi Bronek and Hi there Bruce....

Congratulations on a fine looking PWM / Power Supply box...looks really great....

I'm not a technically qualified person but to me it all looks very normal....and compares with what I have built as well....
I must say that the first PWM I built...... I too had no idea as to what were good bad or whatever waveforms and found the whole lot rather confusing....
Like what Bruce said what follows after the LPF is whats important and a picture of that would be great.....

If you think of whats happening at the Source of the FETs ....
Sharp on/off ... large energy from the first inductor of the LPF having to go somewhere during the off cycle and probably though the FETs capacitance or whatever ... all of this maybe causing some small distortion and harmonics as a result....

I would check the output after the LPF and do an audio sweep and would say it would all look good hopefully....

Wayne

VK3ALK

[Bronek M0DAF]

Dear Wayne,

Thank you for the kind comments about the construction. There is one design fault: the wires to the back panel which carries the PWM output stage are so short that I cannot easily service without having to disconnect countless circuits! Still one lives and learns, but this is especially awkward when diagnosing problems. Overall the layout is much more dense than I have seen from others who have built this circuit.

My concern over the switching behaviour actually arose from looking at the output of the LPF into a dummy load consisting of 20 x 100 ohm 100 W resistors in parallel on heatsinks. My output (50 V) has a rather rough 0.6 V approx. 200 Hz ripple on it with bursts of HF after each on or off transition has occurred. The latter correspond to what is superimposed on the PWM generator output shown in my scope traces. It may be getting through the LPF, but I rather think that it is being induced directly into the output wiring. I wondered whether this HF component was created by some kind of instability in the PWM output stage, e.g., the result of a slow transition in gate voltage. Hence the reason for getting preoccupied with what is happening in the PWM output stage. The ripple is more of a mystery because it is quite stable with regard to frequency and I have not yet found a place in the layout where my intervention changes anything about it. The shape of this waveform is a skewed half cycle of a sine wave, with a faster increase in voltage than decay. The outline of the waveform is a fine staircase, which again leads me to connect to the switching which is taking place. It has not been possible to photograph this on account of jitter on the waveform as displayed on my conventional scope. Although I am using a 6 pole LPF, the inductors and capacitors are not that different from those of Steve’s 4 pole version, so I am doubting whether it is some unfortunate combination of PWM output stage parameters and filter parameters.

I look forward to hearing from anyone who has experienced anything similar (and solved it!).

Thank you

Bronek M0DAF

[kf1z]

On your filter..

Is this about what you have for values?

L1 - 27.45uH
L2 - 102.48 uH
L3 - 75.02 uH

C1 - 3uF
C2 - 4.1 uF
C3 - 1.1 uF

[Bronek M0DAF]

Yes, spot on! We clearly use the same software!

The inductors are on hiflux cores and inductances measured to confirm. The capacitors are made up from HF film components which have very low inductance and work very well up to 3.5 MHz according to manufacturer data. There are several in parallel to create the desired values. All low inductance layout.

Regards

Bronek M0DAF

[Gilly]

Hi Bronek

I'm just heading off to work so just a quick post....

There is a bit of reading there....
You may have tried this but I always start with the easy things first...

What happens at reduced HV say in the tune up position......I have a test power supply controlled by a variac thats easy to connect using clip leads....
For testing purposes also have 2 Dummy Loads ....one at 7 Ohms and one around 12 Ohms.....


Wayne

[Bronek M0DAF]

Dear Wayne,

I have managed to capture the waveform at the output of the LPF which I can send you by e-mail if you wish, but I suspect that it doesn't really help.

When I reduce the voltage to 'tune' the amplitude of the offending ripple gets less but it also changes its appearance slightly with an occasional cycle of ripple missing. I had done that test many times before but not looked at the outcome as critically as I have just done. So it does look like some sort of instability, possibly caused by an earth problem or loop. I'll keep fiddling with it to see if I can learn any more.

With best wishes and thank you

Bronek M0DAF

[Diego-CX4DI]

Dear Bronek,

Regarding the "Gate waveform with HT applied referenced to ground". In this case the "ground" could be the "floating ground" not the chassis ground. Is very important how you connect the oscilloscope terminals. ( in the the oscilloscope AC power cable the ground cable could be unconnected).

best regards,
Diego
CX4DI
cx4di@adinet.com.uy

[Gilly]

Hi Bronek

I wonder how you went with the problem or where you are up to now....
Tried to simulate your problems on my spare PWM ( have 4 in total ) but no luck....
Maybe moving away the Output filter or checking the grounding of the its capacitors.....

Anyway would be good to know what stage your at...


Wayne

[Bronek M0DAF]

Dear Wayne,

Thank you for your continued interest in my ‘problem’. I decided to put it on hold to clear the head while I build a DDS VFO, and I am horrified that it is such a long time since I visited this Forum!

However, this diversion may have been a blessing in disguise; after completing the DDS with a line driver to give TTL levels, I decided to look at a 7.23 MHz output (2 x f) on the scope. I was horrified at what apparently happened to the square wave when I hung the scope probe on the end of the line. It made a lot of difference to the shape of the square wave whether I used the probe, or whether the line went straight into the scope and was terminated there and, indeed, I am now looking at a perfect and clean square wave!

So this has made me wonder as to how much of what I see with the PWM is due to some obscure loops which exist between the scope and the modulator, and possibly also due to incorrect termination at the scope input. Furthermore, if I make a small pick up coil (3 turns x 1 cm dia) and hang that off the probe, by holding near the respective inductors, I can see an intense and clean PWM waveform at the input to the filter and this dies away completely as I go down the filter chain. So, something looks as if it is working OK.

So the question seems to have changed and I am wondering whether “what I see is what I get”. I think that the best way to settle this will be to finish the RF deck and see what comes out of the antenna socket.

Thank you once again.

With best wishes

Bronek M0DAF

[DMOD]

Furthermore, if I make a small pick up coil (3 turns x 1 cm dia) and hang that off the probe, by holding near the respective inductors, I can see an intense and clean PWM waveform at the input to the filter and this dies away completely as I go down the filter chain. So, something looks as if it is working OK.

It sounds like you are ok and I would put a dummy load on the RF out, modulate it, and look at the resulting waveform. The floating ground in the Modulator can cause some interesting probe waveforms.

I do know in my last conversation with Steve, he did change the gate resistors in the PWM drive from 4.7 to 5.6 Ohms, IE, the R503 to R507 resistors that shunt the diodes.

Phil - AC0OB

[DMOD]

Furthermore, if I make a small pick up coil (3 turns x 1 cm dia) and hang that off the probe, by holding near the respective inductors, I can see an intense and clean PWM waveform at the input to the filter and this dies away completely as I go down the filter chain. So, something looks as if it is working OK.

It sounds like you are ok and I would put a dummy load on the RF out, modulate it, and look at the resulting waveform. The floating ground in the Modulator can cause some interesting probe waveforms if your scope isn't fully isolated.

I do know in my last conversation with Steve, he did change the gate resistors values in the PWM drive from 4.7 to 5.6 Ohms, IE, the R503 to R507 resistors that shunt the diodes.

Phil - AC0OB