WA1GFZ audio driver: a problem setting the NFB

[KK7UV]

For context, I am using this in my 813 rig project, so it's driving a pair of 813s as others have done.

Using a 50k resistor for 'local' feedback (off VO2 grid to wiper of NFB pot) I got 40mA total resting plate current in the modulator tubes with +3.6v of bias and +/- 280v rails.

I then removed the 50k resistor and hooked up the 'global' feedback - a 2Meg string of resistors from the plate cap of the VO1 tube to the wiper of the 2k NFB pot.   With the pot wiper at the ground end, key-down gives me the same 40mA of resting plate current.

I then injected a 1kHz tone at a level to give 50% modulation on the scope to establish a reference scope voltage for setting -6dB of NFB.

The problem comes when I then move the pot wiper away from ground, even just 1/8th turn, the resting modulator plate current starts to soar so I unkey the rig.  I wasn't expecting the pot setting to affect the resting plate current at all.

I've read back thru everything I have on this - including some recent tips from Tom and Frank - but can't understand what is happening.  At this point I can come up with two questions...

1.  going back to the 'local' feedback setup and setting the bias and resting plate current, does it matter where the NFB pot wiper is set?

2.  Was I actually seeing positive feeback off the VO1 tube when I brought the NFB pot wiper away from the ground end?  and is this what caused the resting plate current to soar?



   

[K1JJ]

Hi Steve,

The NFB pot should not affect the 813 tube idle currents.  

It sounds like you may be getting some unwanted audio positive feedback causing instability. I had those problems in the beginning, but it was because the GFZ driver was about 2' away from the modulators.  I finally fixed it with a bypass cap across the NFB ladder or input. You want to try small value bypass caps that do not affect the regular higher voice freqs too much.

I assume you are watching the audio and NFB leg with a scope. Do you see the oscillation when it takes off? At what freq?  Mine was very obvious when it happened, somewhere at about 7 KHz IIRC.  But now, I can do up to -12dB of NFB with no problems... stable.  I usually run about -6dB NFB these days on the 4X1s, 813s and series modulated 813 rigs..

T

[KK7UV]

Thanks for the reply Tom.

My driver board is inside the modulator chassis (shielded from the RF side) and just a couple inches from the modulator tube pins.  I used 4 inches of shielded leads to connect VO1 and VO2 to the grids at the sockets.

The oscillation starts to occur when I bring the 2k NFB pot up just a little - like 1/8th turn.  The scope indicates the oscillation is around 13.6 to 15.9 khz (6 to 7 cycles per horizontal 2.2 divisions, at 0.2mS/div).

I tried a .022uF cap across the 10k resistor at the bottom of the NFB string, seeing if it would bypass the oscillation freq.  I didn't appear to do much.  Perhaps a large value is needed. 

But more importantly right now,  I scoped the audio generator input on one scope channel and the bottom of the NFB string across the 10k resistor on the other scope channel and these signals appear to be in-phase - only injecting just enough signal to see the waveforms.  These should be anti-phase, correct?   I double-checked the connections - M1/VO1 goes to the modulator tube grid whose plate is connected to the NFB string back to the NFB pot wiper.   M2/VO2 goes to the other modulator tube grid.

Frank suggested I am getting some RF into the line as well.  This was also evident on the scope - seeing both my fundamental (7.288 MHz) and a smaller 65.59 MHz (nine times the fundamental) riding on the NFB line.  But is it actually on the NFB line or is it because my scope probe cable is  passing thru the hot zone near the RF and Modulator tubes?

So many things to ponder at this point.   

[KD6VXI]

Put a type 43 torriod choke as a common mode filter on the scope lead.

That will suppress from 75 meters to vhf and should help you with figuring out what is what.  Common mode current can and does cause a scope to go nuts.  My spec an isn't as bad, but same same.

In addition, if you've still ogt any neon bulbs a trick I used in the past was to put a NE2 on an insulated stick.  Put it near the tank in question.  If it's lit up orange, it's HF.  Purple is a VHF oscillation.
 
Hope that helps!

--Shane
WP2ASS / ex KD6VXI

[K8DI]

  Was I actually seeing positive feeback off the VO1 tube when I brought the NFB pot wiper away from the ground end?  and is this what caused the resting plate current to soar?    

Since it is dead simple to check this by reversing which plate cap you're connected to, why don't you do that instead of futzing with the scope probes, etc.?  If you had positive and moving it makes it negative feedback, it should settle right down..

Ed

[K1JJ]

Steve,

And how about trying a few ferrite toroid cores on the feedback leg and audio lead inputs to the modulators to stifle any RF getting in?

----

Frank suggested the following:  "I suspect the connection between the ladder network and the drive board is very susceptible to RF. My network is on the driver heat sink with very short leads. A large transmitter this is not possible. Bypass caps will add phase delay so you have to keep the value low. Also there needs to be a low Z ground between the driver and modulator."

----

It's been a few years now, but yes, maybe my oscillation was up around 13-15 KHz like yours is now.  It did seem to be RF getting in causing the audio feedback to lose its mind.  It's good that you have such well shielded and short connections overall.

Keep playing around with the suggestions here from the guys so far - as you are, and you will get it working FB. These kinds of [homebrew rig] problems sometimes take experiments and tests to get fixed. Every rig is different with unique layouts and components that makes fine-tuning a custom job in the end.


T

[KK7UV]

After many hours of troubleshooting this NFB issue and trying the suggestions, I have a couple more questions:

1.  Is NFB necessary for proper/reliable/safe operation of the GFZ driver?  What if I run the driver on the air without NFB?

2.  My GFZ driver board ground is presently connected to my modulator chassis as instructed.  However, my entire outboard audio chain is balanced with XLR connections all the way to the modulator chassis where pin 1 is chassis, pin 2 is signal plus, and pin 3 is signal minus.  To make the audio work I have to connect the signal minus from pin 3 to the driver board ground (and therefore chassis). I am not sure if this is proper way to go from balanced audio to the unbalanced GFZ driver.   Could this be causing an issue?
 

[w9jsw]

I have a XLR to 3.5mm mic plug. It connects signal minus to ground at the driven end which in my case is a Oxmoor distribution amp. At the board, the ring of the 3.5 is connected to the board ground.

IT has to have feedback to control the gain. I have been running my board with the temporary jumper in place for 2 years. Lazy, I guess.

[K8DI]

2.  My GFZ driver board ground is presently connected to my modulator chassis as instructed.  However, my entire outboard audio chain is balanced with XLR connections all the way to the modulator chassis where pin 1 is chassis, pin 2 is signal plus, and pin 3 is signal minus.  To make the audio work I have to connect the signal minus from pin 3 to the driver board ground (and therefore chassis). I am not sure if this is proper way to go from balanced audio to the unbalanced GFZ driver.   Could this be causing an issue?

Balanced to unbalanced done “right” depends on the type of output you’re converting.

 A typical moderate cost balanced output is happy just leaving pin 3 open. Somewhat better is take pin 3 to ground through a resistor equal to the impedance of the unbalanced input, which typically is on the order of 20k,

More expensive balanced drivers tend toward needing less attention as they are able to deal with the asymmetry, so just leaving pin three disconnected is more ok.

The only time you want to just ground pin 3 is with a  floating transformer output, because if you don’t, there’s no power transfer/no signal. The rest of the time just grounding pin three is to one degree or another shorting an amplifier to ground.


All this said, I doubt this has anything to do with the bias situation.  

Ed

.fixing balanced audio on stages and in studios since 1984…

[w9jsw]

FYI - My Oxmoor has a floating transformer output...

[K8DI]

FYI - My Oxmoor has a floating transformer output...

You replied while I was writing mine, no intent or even knowledge to call you out…

And, I am a HUGE proponent of transformers. One can argue that they’re expensive or not as wide bandwidth all day long, but when there’s audio to pass and you want galvanic isolation, there’s just no substitute…

Ed

[w9jsw]

K8DI,

We should have a QSO sometime. I don't think i have ever talked to you.

John

[KK7UV]

"It connects signal minus to ground at the driven end""At the board, the ring of the 3.5 is connected to the board ground."

-OK John, this is essentially what I have right now.  I'll leave it as is.

As you did, I might just plug the local feedback in and try it for a while as I figure things out.

In the photo you can see what happens when I advance the NFB pot just a little from ground - some small oscillations start riding in on the 1khz test tone.  I estimated this at about 13.6khz.  Advancing a little more NFB and these little cycles increase in amplitude and multiply toward the crest of the waveform.  A little more NFB still and the whole screen turns into top-to-bottom spikes and the modulator plate current takes off.

With no tone injection at all, I can advanced the NFB pot up about 1/4 turn before the modulator plate current wants to take off.

Frank suggested RF getting into the NFB.  I tried to tackle this by bypassing the 10k R21 with .002uF, and from there connecting to the pot wiper with a a 6-inch length of RG-174 with 5 turns thru a #31 ferrite bead), coax shield to chassis ground on each end.




  

 

[KC2ZFA]

parasitic-suppressing resistors at the 813 modulator plates ?

[K8DI]

Frank suggested RF getting into the NFB.  I tried to tackle this by bypassing the 10k R21 with .002uF, and from there connecting to the pot wiper with a a 6-inch length of RG-174 with 5 turns thru a #31 ferrite bead), coax shield to chassis ground on each end.

There's a workaround for testing and troubleshooting I've used a bit, that also allows testing modulators in conditions of heavy overdrive without the RF consequences:  A dummy load.  Not a 50 ohm RF resistor, but a collection of high power resistors as a load for the modulation transformer output.  A set of 100w metal resistors screwed down to an old heatsink, and then you can beat the heck out of your modulator with no RF, no negative modulation, no loss of load to kill the mod transformer (that occurs with over 100% negative).  This completely eliminates RF feedback from the testing, and allows you to connect one end of the mod transformer to ground so scopes and voltmeters can be used on the mod transformer output without concerns about differential or floating measurements, too.

Ed