What is causing this audio highs attenuation/ distortion problem?

[WD5JKO]

Looking at that audio driver, capacitor C12 is used to roll off the high frequency negative feedback. I'm sure this is for a good reason, but increasing the gain as the frequency rises as part of a NFB loop can be tricky. Could this be part of the issue? What is the NFB gain reduction in DB at 1 Khz, and then 5 Khz?

Jim
Wd5JKO

[K1JJ]

Looking at that audio driver, capacitor C12 is used to roll off the high frequency negative feedback. I'm sure this is for a good reason, but increasing the gain as the frequency rises as part of a NFB loop can be tricky. Could this be part of the issue? What is the NFB gain reduction in DB at 1 Khz, and then 5 Khz?

Jim
Wd5JKO

Thanks for the thoughtful replies, Brett, Fred, Opcom and Jim.

Yes, I try to make these threads a learning experience, even if it makes me look like a dummy sometimes... :-)

Jim:  I disconnected the negative feedback completely and the problem remained.  At this point in time it is definitely the SS driver. I have it on the bench alone and ONE channel  is showing a strange distortion, like the bias is going crazy whenever the audio gen goes above 3 KHz.   Below 3Khz it runs fine. Exactly what the transmitter has been doing.

I replaced the phase splitter MOSFET as well as the two output MOSFETS. No difference. Checked all the zeners and resistor values - all within tolerance.  The caps check good with a cap checker but at this point I think one may be leaking and causing the splitter to increase bias, causing the distortion. I will need to sub-replace them.  

Is there a way to test for leaky caps in circuit, like using a 1 meg VOM range??

This is actually good news cuz it looks like the transmitter is fine - and it's now down to finding the bad component in the driver.  The SS driver has worked well for several years now, so the design is good.
I'll bet it got damaged when I was doing the HV shakedown run a few days ago. The rig flashed over at least five times the other night on 75M. I fixed the flashover problems, but the damage was probably already done to the SS driver..

More troubleshooting today.  I'm sending voltage readings to Frank today and maybe we can zero in on the bad stage.

T

[N2DTS]

If its not M1/M2, then the next things that don't like high voltage are D3 and D4, and C2 and C3....

[N2DTS]

Oh, and if you do not have spares, maybe swap parts from one channel to the other and see if the trouble moves?

[K1JJ]

I don’t get it.  Today I fired it up and it’s working like it shud. Perfect sine wave  I put a heat gun to it and then froze all the components individually. No change.  Vibration  / slam test - no change.    It will be baaaack.

So maybe an intermittent somewhere?  The design was always sound.
 
Now I don’t see any change in the DC outputs when I go from full signal to none like before above 3 KHz, which is a good thang.  It stays at about 20V on both channels.  I was getting that big DC offset change before that was killing my modulator bias and causing class C distortion.
 
With the ss driver alone on the bench, I ran the sig gen from  30Hz  to 25Khz and it is solid now...  Triangle wave starts to deform below 30 Hz cuz it is capacitively coupled. Triangle looks very good up to 10 KHz and then starts to slope slightly.    The sinewave starts to fall apart below 20Hz. I will be fattening up the capacitor coupling caps to bring this lower.
 
Maybe I knocked loose a solder blob somewhere, I dunno...  Gonna get out the magnifying glass and inspect every solder connection carefully.  

I’ll let it run on the bench and beat it up all day to see what happens.  It would be nice if this really fixes the problem.
 
T

[K1JJ]

So far the SS driver is running on the bench FB with no crap outs.   We'll have to wait until the big rig is back on the air for more excitement.

While on the bench I padded the five 600V coupling caps from 2.2 to 5 uF each.  Big difference in the triangle and sine wave ability to hit 10 to 20 Hz cleanly.  A clean audio driver that will handle required load power is as important as a good mic, maybe more so.  I suppose now the limiting factor is the modulation transformer.


T

[W2VW]

  I suppose now the limiting factor is the modulation transformer.


T

Road trip? Bring a truck.

[K1JJ]

Road trip? Bring a truck.

Ya never know, Dave.  Vely appealing.  Finding the room to shoehorn it in is the problem, of course...  

Well, tonight I gots the noove to fire the rig up with the SS driver.  I worked it like a rented mule. It behaved like Tiny Tim - good as gold and better.  It swept like a textbook sinewave - flat from 10 Hz to about 6500Hz before started to roll off. I couldn't be more pleased.

Because I added bigger coupling caps to the audio driver, IE, less phase shift, I was able to increase the negative feedback a few more dB.   I hear a noticeable difference in quality with the NFB in vs: out.  The reduction in gain seems to be at least 10dB or so.

After running triangle tests at 50, 400Hz  and 1K, I was satisfied the overall system is playing well again. The only problem I see is the roll off at 6500 where a slight harmonic distortion begins. It seems to need more and more audio drive to hit 100% above 6500Hz. I'm wondering if it's the .002 coupling and .002 plate bypass caps I have in there.  With the tank impedance at about 5K, I can see a meaningful path to ground above 6500Hz.  Think I'll try 500pf for each.  I have a few caps somewhere around.  Could also be the mod transformer topping out too. I don't plan to run bandwidth more than 5-6Khz, but it is still important to have a clean headroom so the audio has room to roll off cleanly. Even if it's down 20 dB at 7500Hz, there is still a potential for splatter if the waveform turns to crap up there.

So, back to more tests...

T

Fabio II in test mode:

[steve_qix]

The combination of those 2 caps represents about little under 4000 ohms at 10kHz.

That's probably enough to mess with the high frequencies a bit!

[AB2EZ]

In this transmitter, the effect of the total screen bypass capacitance, on the rolloff of the higher modulation frequencies, is likely to be greater than the effect of the capacitance bypassing the modulated B+; but, as has been already stated, both are important.

Stu

[K1JJ]

In this transmitter, the effect of the total screen bypass capacitance, on the rolloff of the higher modulation frequencies, is likely to be greater than the effect of the capacitance bypassing the modulated B+; but, as has been already stated, both are important.

Stu

Interesting...

Stu, I reduced the total screen bypass capacitance of the two tubes to 1000 pf. (500 pf per tube)

Right now I'm reducing the plate bypass to about 500 pf and the same for the coupling.  I see that 1000 pf at 7000 Hz is several times the 5K tank impedance, so that should be good.

One thing that concerns me:  I just measured my plate choke and see it's about 303 uh.  At 7Khz audio freqs, this is under 20 ohms, which is good.   But at 1.9 Mhz RF, 303 uh is only 3.6 K. At 3.8 Mhz is it only 7.2K.    With a 5K plate impedance, the choke seems  too small in value.   However, on 75 and 160M, the amplifier tunes sharp and I feel no heat coming off this plate choke.  I always thought that we like to see  the plate choke inductive reactance to be X 5 to 10 times the tube impedance.  

Essentially, on 160M don't we have a 3.6K path across the tube, shunted to ground thru a 20 ohm capacitor to ground at the end?  I would think that would draw a lot of power off the 5K tank circuit.   Or is the power factor of inductance and capacitance the key here...What am I missing?


BTW, 500 pf at 1.8 Mhz is  167 ohms - at 3.9 Mhz this is 81 ohms.  Even this capacitive reactance at the bottom of the plate choke seems a little high for bypassing a 5K circuit, or is it?

T

[AB2EZ]

Tom

This is my current thinking/understanding of the issues you asked about:

1. With respect to the modulated B+

In this traditional type of modulator... with the output resistance of the modulator matched (via the modulation transformer) to the modulation resistance of the output RF stage... the equivalent circuit of the modulator itself is a voltage source in series with a resistor whose value is equal to the modulation resistance of the RF stage. The Norton equivalent of the combined circuit (modulator + the modulated RF stage) is a current source in parallel with a resistor whose value is 0.5 x the modulation resistance of the modulated RF stage. Therefore, assuming a 5k ohm modulation resistance,  we want the total capacitance to ground to have an impedance at 10kHz that is higher than -j2.5k ohms in order to minimize high frequency rolloff.

[Note: when using a backward-connected output transformer in conjunction with a modern, very low output impedance audio amplifier, the acceptable impedance of the total capacitance to ground can be less than 0.5 x the modulation resistance of the modulated RF stage]

2. With respect to the screen voltage:

Note that it is the modulation of the screen voltage that modulates the plate current. The modulated plate voltage has much less direct effect on the plate current. The real purpose of modulating the plate voltage (with a tetrode or a pentode) is to increase the efficiency of the output RF stage. In a self modulated configuration, the modulation of the plate voltage causes the screen voltage to be modulated... and it is the resulting modulated screen voltage that modulates the plate current.

For self modulated screen voltage using a choke between the screen power supply and the screen: you want the impedance of the bypass capacitor, at 10kHz (i.e. the high audio frequency screen-to-ground impedance), to be larger than the impedance of the choke at 50Hz (i.e. low audio frequency screen-to-ground impedance). For example, a 10 Henry choke has an impedance of j3142 ohms at 50Hz. In this example, you would want the total screen bypass capacitance to be smaller than .005uF.

When using a resistor from the modulated B+ to the screen (or from the unmodulated B+ to the screen), you want the impedance of the bypass capacitor at 10kHz to be larger than: the screen voltage at carrier / the screen current at carrier.

3. With respect to the plate choke

If the choke's impedance at the lowest RF frequency of interest is not high enough, this will have an effect on the efficiency of the RF stage that can be obtained at that RF frequency. The relevant impedance to compare to, for the purpose of evaluating the plate choke's efficacy in delivering a constant current to the RF stage during each RF cycle, is the impedance looking into the output "tank" circuit... which is typically 0.5 x the plate voltage / the plate current for a class C amplifier. That is: the question you are trying to answer is: how long is the "L/R" time constant, when R= 0.5 x the plate voltage/ the plate current. If this time constant is less than 1/the lowest RF frequency of interest, then the efficiency of the amplifier will begin to drop. Putting aside its role in providing a current source to the plate of the amplifier, the effect of this inductance on the tank circuit will be a small change in the setting of the tuning capacitor at resonance

[N2DTS]

I have a question along these lines.
If you use a big wire wound pot to drop the screen voltage from a fixed (but adjustable) source, should you bypass around the resistor for audio?

Being a wire wound resistor, it will have some inductance.

Would there be any interaction with a choke in series with the screen?
If using a resistor, do you need the choke?

[K1JJ]

Good question, Brett.

I suppose the sure way to do it is sweep it -  try bypassing and not bypassing experimentally.  Much will depend on the specific circuit, loads and interactions of the components.  I think fine tune tweaking ala Dean/ WA1KNX's screen technique is good.  Maybe Stu will have some comments...


Thanks for the info Stu.  I will rewind my plate choke with thinner #24 enamelled wire and bring it up to 1.5 mH.

Question:  I believe in testing a plate choke for series resonance installed IN the circuit.  What would be the proper procedure to do this... connections for the RF sig gen and scope?

I also have an MFJ 259B analyzer - could this be of help?

T

[AB2EZ]

Brett

This resistor will add impedance at all audio frequencies between the screen and ground (the output of the power supply looks like ground for audio frequencies of interest), which is what you want for self modulation. The resistor's inductance is also helpful, in this regard, but probably has a negligible effect at all audio frequencies of interest.

You should not put an audio bypass capacitor across the resistor.

If this resistor has a large enough value (i.e. equal to the impedance of the choke you would be using, at 50Hz audio frequency), then you probably can leave out the choke, and still get the self modulation of the screen that you need.

The key is that you want additional negative charge (electrons) to accumulate on the screen when the plate voltage is modulated downward ... therefore modulating the screen voltage downward as well. Likewise, when the plate voltage is modulated upward, negative charge will be removed from the screen (i.e. some of the electrons that have accumulated on the screen will leave the screen, and flow to the plate), and the screen voltage will be modulated upward as well (less negative charge on the screen). In order for this self-modulation of the screen to work properly, you don't want the varying (modulated) charge on the screen to discharge too quickly to ground via a resistor or an inductor (or both) between the screen and ground. Therefore, the sum of the impedances of the series inductor (choke) and the series resistor between the screen and ground (again, the power supply output looks like ground for audio frequencies) must be sufficiently high for all audio frequencies of interest.

Stu

I have a question along these lines.
If you use a big wire wound pot to drop the screen voltage from a fixed (but adjustable) source, should you bypass around the resistor for audio?

Being a wire wound resistor, it will have some inductance.

Would there be any interaction with a choke in series with the screen?
If using a resistor, do you need the choke?

[AB2EZ]

Tom

Check out the various articles on the web regarding the design of RF plate chokes for high power amplifiers. You may decide that designing/constructing a viable plate choke with a higher inductance (1500uH v. 303uH) will be difficult, because of resonances, that creep into the design, that are associated with distributed capacitance.

You might be better off sticking with what you have.

Stu

Good question, Brett.

I suppose the sure way to do it is sweep it -  try bypassing and not bypassing experimentally.  Much will depend on the specific circuit, loads and interactions of the components.  I think fine tune tweaking ala Dean/ WA1KNX's screen technique is good.  Maybe Stu will have some comments...


Thanks for the info Stu.  I will rewind my plate choke with thinner #24 enamelled wire and bring it up to 1.5 mH.

Question:  I believe in testing a plate choke for series resonance installed IN the circuit.  What would be the proper procedure to do this... connections for the RF sig gen and scope?

I also have an MFJ 259B analyzer - could this be of help?

T

[N2DTS]

When I built my 4x150 rf deck, I used a plate choke out of an old FT102, and when I added a 2nd tube, the choke wanted to melt down one of its pie windings (just one of 5).
I switched to a very old one (4 pie) and it works fine.
Its likely just pot luck with all the variables...

Thanks for the info Stu.

[K1JJ]

Breakthru!  Just what I had hoped for...


For the last test I ran yesterday, both plate bypass and coupling caps were 2000 pf and I saw problems starting at 6000 Hz. It took an excessive 800 mA of modulator current to hit 100%.

I changed the plate choke bypass cap from 2000 pf to 500 pf.  The coupling cap is still 2000 pf.

Now with the 500 pf in place, I can modulate cleanly to 7,000 Hz. At  7500 Hz it takes only 600 mA. to hit 100%.  The waveform starts to get harmonic distortion starting at about 7000.  This is quite a nice step in the right direction..  (It takes only 400 ma to hit 100% at 1Khz)

Now I will change out the 2000 pf plate coupling cap for 500 pf and see what happens.  I'm hoping it will modulate cleanly to 8000 Hz.   At least the modulator is not straining like it was above 4500 Hz before.

I think the modulator was getting shunted by the 2000 pf, thus the screen, no matter how clean it was, was not getting its cues from the plate due to its distortion.

I used to run 500 pf in both spots in the past, but I had such nice 2000 pf caps available, I just couldn't resist using them.

Bottom line is use the minimum values you can in those positions for AM plate modulation. For a linear amp, it doesn't matter and might as well go large.  I haven't tried it up on 160M yet, but the Tron has recommended 500 pf to cover 160M in the past for a 4-1000A final.  Of course, if the tube impedance is lower, it will require larger values.


T

[N2DTS]

I wonder what the plate cap will do.
I have 1000 pf at 15kv in the 3x4D32 rig, 1200 volts, 300 ma, 4000 ohms.

I also have a 1000 PF in the 4x250 rig, about the same plate impedance.

I thought larger was better there for some reason.

[AB2EZ]

Brett

At audio frequencies: the path through the pi network's inductor, and from there to ground via the safety choke (as an example), has essentially zero impedance. Therefore the plate coupling capacitor is (at audio frequencies) essentially a bypass capacitor to ground. The same is true for most other types of output network configurations.

Stu

[K1JJ]

Brett

At audio frequencies: the path through the pi network's inductor, and from there to ground via the safety choke (as an example), has essentially zero impedance. Therefore the plate coupling capacitor is (at audio frequencies) essentially a bypass capacitor to ground. The same is true for most other types of output network configurations.

Stu

Yes, the path through the tank inductor, thru the loading cap certainly is low impedance.  Especially the last safety RF choke shunting the tank output to ground.  Using a small coupling cap will minimize all that.


Question:  If an RF stage is series modulated thru its cathode, (like a tube PDM rig) does the plate coupling and RF choke bypass caps have the same shunting effect on audio as modulating thru the plate choke?    

In the early stages of testing my tube PDM rig, I reduced the coupling cap from 3900 pF to 500 pF and thought I saw no difference. But this is when I had filter problems, so not sure what wud happen if I tried it again.

T

Check out the new coupling caps mounted on a Plexiglas shelf. They are actually two 200 pF in parallel = 400 pF @ 14KV.   They are super high quality that I picked up at Deerfield for $5.  I used another set for the plate bypass.  Never thought I would use them anywhere.  Plate modulation demands large, good quality coupling caps or they will sing or even explode sometimes.

I still need to finish cutting that bolt and run a new sweep test tmw..

[w8khk]

Check out the various articles on the web regarding the design of RF plate chokes for high power amplifiers. You may decide that designing/constructing a viable plate choke with a higher inductance (1500uH v. 303uH) will be difficult, because of resonances, that creep into the design, that are associated with distributed capacitance.

There is an article on pages 30-33 of the May 1954 issue of QST by Vernon Chambers, W1JEQ,  entitled "R.F. Chokes for High-Power Parallel Feed" An answer to the Multiband Choke Problem.  This article details the series and parallel resonance issues and provides some design and testing insight.  I studied this article prior to winding the choke for my 80M - 10M dual  4-400A amplifier in 1968.  The article is downloadable from the ARRL Periodicals Archive.

[AB2EZ]

Tom

I think the answer is no.

Looking into the RF stage from the cathode (cathode and grid voltage being modulated up and down, at audio frequencies, by the same amount with respect to ground), I think the tube looks like a resistive load to ground (from the perspective of the modulator), whose value is the normal modulation resistance of the RF stage. Audio frequency bypasses from the RF stage's plate(s) to ground (e.g. via the plate coupling capacitor) should be irrelevant... since the audio frequency path from the RF stage's plate to ground is already essentially 0 ohms (via the plate choke and the plate B+ supply's output).

In this case, you would be concerned about any RF bypass capacitors from the RF tube's  (tubes') cathode(s) to ground...unless they are part of the PDM output filter.

Stu


[/quote]


Question:  If an RF stage is series modulated thru its cathode, (like a tube PDM rig) does the plate coupling and RF choke bypass caps have the same shunting effect on audio as modulating thru the plate choke?    

T


[/quote]

[KA2DZT]

Tom

I agree with Stu,  the plate choke bypass cap doesn't matter with a PDM type rig.

I did mention, two pages back, the caps should be reduced. Seems like it is having a positive effect on the audio.

OK that's great,  now here's another one of my award winning tips.  Before you cut that long bolt, shut the rig off.

Fred

[K1JJ]

Yep, youse was right, Fred.  Those caps are certainly a bottleneck.  I got an email from Jeff/ NBC who said he runs 250 pf in both positions. His sweeps up to 12K.  

OK on the PDM series modulated audio NOT being affected by the caps, Stu, I will keep what I have then.


ANOTHER BREAKTHRU:

Today running 400 pF for both coupling and bypass, I saw another improvement.  Now the modulator plate current remains down at 400 ma @ 100% modulation even at 7000 Hz!   The top end is about 8K before I see some harmonic distortion creep in when I go above 100%. The modulator current hit 600 ma at 100% when doing 8,000 Hz.  This tells me there still is a limitation in there - maybe the mod xfmr.  Maybe I'll disconnect the modulator lead and measure the total capacitance going into the RF final. That should tell me where it's coming from. Hopefully I will see no more than 1000 pF or less total into the rig.  (Coupling + bypass + stray capacitance)

The difference on voice is quite dramatic for highs. The S's don't spit anymore and the CH's are much cleaner.  I rolled off my hi-cut to about 6500 Hz and it limits well.  

The triangle looks pretty decent at 50 Hz and near perfect at 100 Hz.   The triangle at 4K is now very good - it used to have a slope up there before.

I added in some more negative feedback and see no tendency to take off.  The RF final seems stable with all the changes and haven't had a flashover for days now.

So gonna button it up for now and run it on the air to see what breaks.  Need to install a Hall effect shutdown circuit in  the Dual Quads tube PDM rig, so that will be the next focus.

Thanks for all the help, guys. It probably prevented me from getting frustrated and tearing it down. I was tempted to build up a big PDM tube rig, but now that Fabio II is running so well, I will stick with it.

Seez ya on the airwaves.


T