What is causing this audio highs attenuation/ distortion problem?

[K1JJ]

** This thread is a continuation of the 4-1000A rig built here:

http://amfone.net/Amforum/index.php?topic=33816.0

---------------------------------------------------------------------

I wonder if anyone has come across this problem or can suggest where to look?

This is Fabio II, a 4-1000A plate modulated by a pair.  If I sweep it with a sinewave from 30 Hz to 2500 Hz, the waveform looks perfect from 1% to 120% positive modulation. The crossover point looks clean and hits -100% nicely.  At 3KV, 100% modulated,  the modulator current is 400ma and the regulated modulator  screen current JUST begins to flow.  

But, when I sweep it above 2500Hz, it takes progressively more plate current as the frequency is increased and has a problem going more than 50% modulation without severe crossover distortion. At 5,000 Hz, 400 ma of plate current is barely 40% modulation.  It takes more and more current to do the same thing at a higher freq..

* But at 20% modulation, the waveform looks perfect from 30 Hz to 6,000hz.  This frequency limitation problem seems to start when serious audio power is taken.

The solid state audio driver signal looks clean going into the modulator grids, so it is after this point. I looked at the RF screens  during modulation and they appear to look OK and clean for audio.  

I have used this 1KW RCA mod xfmr in a few rigs and it has performed well in the past to 6 KHz. I am using electronically regulated mod screens and reg grids.

I am using .001 screen bypass for each RF screen.  Also have .002 for the plate choke bypass and .002 for the plate coupling caps. Maybe 500pf wud be better, but I would think that 3500Hz should not be affected so badly with .002.

Tmw I'll look at the modulator audio output at the tube plates and xfmr output.


Any ideas what could be causing this problem?

T

[W7TFO]

HI Tom,

I'll take a stab at it.

>>>The higher the audio frequency, the higher the duty cycle for any given waveform.

>>>Don't forget your modulators pull grid current..gots to be able to supply that.

>>>There might be a bad turn or two in your mod iron.

Somewhere to look and/or think.

73DG

[steve_qix]

Hi Tom,

You said crossover distortion, and I just want to verify that you are indeed getting crossover distortion and not some other kind of distortion.

If it is crossover distortion, that indicates a change in the operating point of the modulator.  Do you have a lot of inductance in the cathode of the modulators?  This could possibly cause a problem at high frequencies which might manifest itself as crossover distortion.

[K1JJ]

Geez... I hope it's not the mod xfmr, Dennis.  That wud be a deal breaker.  What baffles me is that it will do 140%+ easily and cleanly at the lower frequencies.  (30Hz to 2500 Hz)     Would a shorted turn affect the highs so drastically and leave the lows alone?


Steve, the modulator filaments (cathodes) are connected to the fil xfmr and the CT to ground on key up.  I think I used .01's for fil to ground bypasses, but will have to look.  I know some say they don't bother with bypass caps there.

Do you think .002 for both plate bypass and plate coupling is too high?  I used to use 500 pf there, but these were available in the junk box.

The distortion I see (at 4Khz) begins at about 30% modulation and starts at the crossover point and then gets a ripply second harmonic look to it.  I've seen tubes that were biased too hard or with plate voltage too low do this, but I am idling them at 200ma.    I use diodes in the fil center tap as bias, so they are not sagging.

The triangle wave looks perfect at 400 Hz up to 120% modulation.  It's just when I go above 2500Hz when it all  falls apart.

If I run the audio with the D-esser working hard at about 3500 Hz and up, the audio sounds FB, but who wants to run a band-aid like that?


T

[KA2DZT]

The .002 bypass cap seems too big.  The coupling cap or caps, how many are there? By coupling you're referring to the DC blocking cap I assume.  I would reduce them.  For a plate choke bypass cap .001 should be enough.

What size bypass caps are you using on the PA screens??  More than .001-.002 total and that can be a problem.  Add together the screen bypass caps for both tubes.  More than .002? since both screens are tied together.

I think you're using a separate PA screen supply and a screen choke to modulate the screens.  The bypass caps between the choke and the tubes should be less than .002 total.  Another problem could be the screen choke itself.  May be too much capacitance to ground or even too much capacitance through the choke back to the supply where it gets grounded by any filter caps in the supply.

If you modulate the PA screens through a dropping resistor off the modulated B+ you don't have this problem.  This may be something to try if you have the right size resistor, probably a few hundred watters are needed.

Fred

[steve_qix]

Very important question - what is the DC resistance (voltage / current) of the final RF amplifier?  I'm thinking it is low, but please give us the number.

The two .002 uF caps in parallel (total of .004uF - RF bypass and DC blocking together) could be an issue.  I say in parallel because the DC blocking cap usually looks into a DC short circuit to ground as a protection device (in case the DC blocking cap fails shorted) so the blocking cap appears in parallel with the RF bypass as far as the modulator is concerned.

Ok, if that is true your .004uF cap has a capacitive reactance of 3980 ohms at 10kHz, essentially across the mod transformer.  Maybe this is low and maybe it isn't.  Depends on the DC resistance of that RF amp.

Another issue *could* be, if the final RF amp's DC resistance is very low, that the mod transformer is designed to look into a much higher resistance and may have too much inductance for the DC load.  This can be verified by running the RF amplifier such that it is representing the DC resistance for which the mod transformer was actually designed, and rerunning the test.

Just ideas... no real solution yet.

[N2DTS]

I tend to go small with bypass caps.

I have an issue with one of my rigs, it tests fine with a sine wave out past 8 KHz, sine wave injected at the same point as the mic, just at line level. With me speaking into the mic, I can not get past 2500 Hz, it rolls off badly even if I boost the highs.
The exact same audio chain works fine with other rigs.

I can boost the highs as much as I want, but they do not pass through the transmitter.

[w1vtp]

Why would a symmetrical source have a higher positive modulation level than the negative side?  Is it the result of a negative diode limiter circuit?

Male voice = asymmetrical modulation

Sine wave = symmetrical modulation

Al

[KA2DZT]

Tom

I re-read your info, I see that you are using .001 caps for the PA screens.  They should be OK, but is there any other circuits between the screen choke and the tubes??  Like meter circuits or over current sensors or what, if anything??

Fred

[K1JJ]

Good suggestions guys, thanks!

Steve, at 3500V, I am running the rig about like I have in the past and this mod iron swept FB.  I figure about 5000 ohms plate impedance.  The mod xfmr came out of a BC xmitter using a pair of 4-400s X 4-400s. So this is reasonably close in impedance the way I am running it now.    I will try substituting in 500 pf for both plate caps and see what happens.  Tron always suggested 500 pf for these class rigs, but that was 10Khz - I would have thought the effects of .002 wud not show up until maybe 5Khz at least.

You may have hit upon something else...  to the best of my memory, I thought the rig swept FB when I first built it. It would have been a red flag right away, since I always sweep them looking for perfection.  Since then I added Hall effect shutdown circuitry for both the final and modulators in the cathodes and screens. This problem may have started then and I did not sweep it afterwards, just went on to another project.  I have long leads to bring the Hall effect devices out to a central point for convenience.    I probably should have heavily bypassed the filaments after this addition.

Question:  If I have long fil CT leads can I simply bypass the fils with 1uf or even electrolytic caps to eliminate the long inductance path?   These leads also involve the metering.


Fred:  I measured the screen modulation choke to have 180 pf to ground.  There is a  .001 at each screen tube pin.  On the other side of the choke it shud not matter, cuz there is the power supply.  I have the Hall effect and metering on the PS side of the screens.

Jeff/ W2NBC suggested maybe an audio parasitic.  I already have parasitic chokes in the modulator tube plate leads. Maybe I shud add some to the grids too. But I will bypass the fils as mentioned above.  The GFZ solid state audio driver input to the mod grids looks clean when the distortion appears, so hopefully it is not that.

I'll run some of these tests today and see if things improve.  My gut feeling is there is some lump capacitance shunt or series inductance that once cut out will let the highs pass cleanly.  Hopefully not mod xfmr damage. The fact that the rig sweeps very well below 2500 is encouraging.

T

[steve_qix]

Question:  If I have long fil CT leads can I simply bypass the fils with 1uf or even electrolytic caps to eliminate the long inductance path?   These leads also involve the metering.

*ANY* high inductance in series with anything (grids (DC side) screens, cathodes, plates, etc.) will impact the high frequency response.  I had the problem 40 years ago with my pair of 304-TLs modulated by 833As.  I had a relay coil in series with something (I don't remember - probably the cathodes), but it added inductance and caused a lot of high frequency distortion.  A bypass cap across the inductance fixed the problem.  1uF may not be enough. 

You'll have to calculate the capacitive reactance with respect to the impedance of the circuit being bypassed and see if it's sufficiently low for the frequencies in question.

[Opcom]

My transformer is also an RCA 1KW unit. Older though. The response is best around 1000Hz and diminishes a lot going up in frequency. The TX it is probably from used audio feedback to fix it. I've swept it but never recorded the result.

[KA2DZT]

Bypass the PA filament leads with .01 ceramic caps, you can use a few on each side plus the CT.  Can't hurt to also use one .001 ceramic on each side plus the CT.  Never use lytics they have too much inductance.  All the caps should be at the sockets.

Anything on the PS side of the screen choke shouldn't matter.

Make sure your regulated screen and grid supplies are not going haywire in the presents of strong RF.  I have seen this happen with some Lambda supplies.

Bypassing the mod filament leads probably needs more capacitance.  I don't remember bypassing my 811s fil leads.  Mod filament leads may not need bypassing but it can't hurt.

Fred

[K7MCG]

Check the regulation bandwidth of your regulated power supplies.  They may be providing low source impedance only at lower audio frequencies.

--Chuck K7MCG

[K1JJ]

Well, I did all the suggested stuff.  Measured the regulated supplies, bypassed the cathodes, disconnected the Hall effect shutdowns and some more stuff. NO difference. The rig still gets a very bad crossover / harmonic distortion above about 2500 Hz.


And wouldn't ya know it - the one thang I assumed  was rock solid - the solid state audio driver....   On the scope, the SS driver audio signal coming into the 4X1 grids looked fine at 5Kc and higher. But I didn't measure the DC offset that it used to bias the grids.  Under full modulation below 2500 Hz it was rock soild at 21 Volts DC no matter how heavily modulated - which is what we want.  But as I progressively went above 2500 Hz, this 21 V bias started to creep higher and higher until it went above 44 VDC!   So it appears it may be an audio parasitic that is charging up the SS driver or something related to audio instability getting into the DC bias of the audio driver.  

Steve / QIX was correct on his first post about crossover distortion being created by a DC offset of the modulator. And Jeff/ NBC  was on the right track saying it might be an audio parasitic.  More detective work is needed now.  I added some ferrite cores to the SS driver leads, but no change.  I must admit the SS driver leads are long, about 5' long. Frank warned me about this before, but it seemed to work until now.  I just don't have any room in the mod cabinet to put it, so it sits on a shelf.

More observations:

1) Disconnecting the audio negative feedback from the mod xfmr to the SS input has doesn't affect the problem.

2) Running the rig at full power or minimum power on low tap does not seem to affect the levels where the DC offset starts. It's like the RF does not affect the DC offset, rather it's based solely on audio frequency.  Maybe that's why the ferrite beads had no effect.

Just wanted to post this as I  zero into what may be the problem.  The 5 to 8Khz KC audio actually looks good up until about 30% modulation, but then when the DC offset moves up, it's like biasing the modulator into class C and the waveform goes to hell.   Got some hope and getting closer.

More ideas?

T

[K1JJ]

A small breakthrough:

The highs get better and the DC offset drifts less when I ramp up the  +- 300V supply on the SS driver.  I brought it to +- 325VDC and now up to 4KHZ is clean.  Still see DC drift > 4Khz.    I don't dare bring it higher with fear of blowing the MOSFETS.


This is the SS driver - look at post 3 for a schematic:
http://amfone.net/Amforum/index.php?topic=23632.0

It uses 11N90 MOSFETS. I'll axe Frank how high it can be raised.  But there must be some other reason for the DC drift based on my installation / instability.


Steve, in a post in that thread you said:

"I'm absolutely *AMAZED* you don't get a parasitic with the gate zeners connected directly from gate to source without a small resistor (usually a few hundred ohms is plenty) in series with the gate, between the gate and everything else (including the zener).  I have seen parasitics without the resistors (which is why I use them in all analog designs like the class H modulators).

A tube arc-back to the grid can cascade through the circuit.  I would humbly suggest 3 things:  a) strategically placed transzorbs,   b) back biased diodes going from the output to each rail and   c) fuses (as suggested on the circuit)."


Does this sounds like what is happening?


T

[N2DTS]

Do the electrons flowing from the cathode to the plate that get stuck on the grid have anyplace to go?

It sure sounds like a charge is building up on the grids, charging up the driver.
At lower frequencies maybe they have time to leak off, but at higher frequencies they do not, like an R/C circuit?

[K1JJ]

Do the electrons flowing from the cathode to the plate that get stuck on the grid have anyplace to go?

It sure sounds like a charge is building up on the grids, charging up the driver.
At lower frequencies maybe they have time to leak off, but at higher frequencies they do not, like an R/C circuit?

Good point, Brett -

I just checked and see I have 100K from each grid to ground that the SS driver feeds into. There is also a .001 from grid to ground there.  Maybe those need to be experimented with.  Maybe the resistor can be brought down to 22K to bleed off the grids better.  Or maybe it will get worse by creating a grid leak bias effect based on drive?  Or maybe the 100K is already doing that?  :-)

I will give it a go after supper and see if it has any effect...

T

[K1JJ]

OK, getting closer...

Adding resistor shunts to the modulator grid inputs had no effect.

What did help a lot was to snip out the .001 disc bypass caps that went from modulator grids to ground. I have a bad habit of sprinkling them around thinking I will kill all RF where it shud not be.

The highs are now clean up to about 4500 HZ.

Then I found that simply by idling the modulators higher via less diodes in the fil CT, I can hold off the effects of the SS driver bias creeping up.  Now the highs are clean to about 5000Hz, but I still see the effects of the parasitic.

I looked closely at the audio signal at the grids and for the first time I see the audio parasitic mixed in with the cross-over area.  It looks to be maybe  15 Khz or so - just a guess.   I need to find a way to kill it.  Gonna try parasitic suppressors in series with the grid leads next...

The audio below 4.5 kHz is now rock solid and looks pristine up to 150% positive. Just need a little more stability above 4.5 Khz..

Any more ideas are appreciated, as they have helped me immensely so far.

T

[K1JJ]

For those of you following this exciting saga, we have some late breaking news...

Part of the problem is now identified...

One of the SS audio driver channels is putting out spikes when goosed with highs.   When running alone without the transmitter, S’s cause  random spikes twice the height of normal..  Probably a bad FET in there.  The other channel acts normally.  Tron told me last night I had some highs spitting. He always knows.

Well, anyway, I’ll get the scope in there tmw and try to find the leaky FET and replace. Hopefully this will solve the high freq distortion problem, maybe not.

In this shakedown I also found a few weak points that caused arc overs. One spot showed soot where a bolt was arcing to the cabinet. Plus a small input tuning band switch had a melted wafer. This was causing havoc by shorting the RF grids to ground... yikes!

I haven’t had a flashover in the last day despite a lot of testing into the dummy load.  I'm pretty sure the mod xfmr is OK... whew.

More tmw.

T

[N2DTS]

Fun with home brew, eh?

Are the modulators running AB1 or are they taking current?

I ask because many circuits have a resistor in series with the grids, maybe to kill parasitic's?

Higher resting current is better for crossover distortion, maybe that was just masking the real problem?

[K1JJ]

Fun with home brew, eh?

Are the modulators running AB1 or are they taking current?

I ask because many circuits have a resistor in series with the grids, maybe to kill parasitic's?

Higher resting current is better for crossover distortion, maybe that was just masking the real problem?

Hola Brett -

Yep, who knows if the overall design doesn't have a flaw, thus the challenge of a homebrew vs: commercial rig. But the commercial rigs also have human designers... :-)

Normally I run the modulators in AB1, but don't have a grid limiting resistor. The grids are directly driven by 11N90 MOSFETS.   I see maybe 10 ma of screen current, even though it is regulated stiffly. The grids draw no current in normal operation.

I think this "parasitic" problem was a masquerading fooler and simply a driver MOSFET that was breaking down.  Whenever it heard high freqs it went into arcing or feedback internally. I think the transmitter was OK all along. But that's what it takes sometimes.

In hindsight I should have done the old divide and conquer and looked at the input system - one at a time.

It's probably a bad output FET according to Frank.  We'll see tmw and hopefully the rig will sweep without problems up to 7Khz.

Yes, you know all about troubleshooting from a look at your shack... ;-)

T

[N2DTS]

Well, I often have problems getting something to work as it should, mostly because I overlooked something, but I have had great luck with no problems once something is done.
I built my 813 rig 25 years ago and have yet to have a problem with it, nor have I had any crap outs in anything else I built.

I do not do a lot with solid state components at high voltages though.

Just yesterday I had the first problem with my 32V3 in 30 years, the plate current meter started sticking at 200 ma.

On the other hand, I never got my last push pull 812a rig to work and gave up.

[Opcom]

The site those things came from shows a new 1200V one with otherwise similar ratings "coming soon": STW12N120K5 1200V 12A.  It would be interesting to find out, if one of the FETs is bad, what damaged it. Also if you add more protection that would be interesting. Would it make sense to put clamps across each FET? x-employer used to sell these for lighting protection:
SMP100LC-400 or 1.5KE440A/CA (no affiliation with co.).

For now I'm sticking with the transformer drive but if I go direct to a large grid tube it will probably be 6080s. The grid of a 4-1000 is a dangerous place for SS to live.

[flintstone mop]

Thanks Tom
for the updates and reports on your progress. It is a learning experience and gets down to the nuts and voltz of what goes on in experimenting.
At least we learn something from your replies.
Some other threads go no where and the OP never returns with a victory or defeat reply...I guess they give up and put the project on a shelf.
Fred