Distortion in Plate Modulated Transmitters

[W2XR]

Gentlemen,

Assuming you have a plate modulated transmitter with a properly designed linear/low-distortion audio chain (speech amplifier, audio driver, modulator, and no audio processing), and with adequate headroom and audio power capability relative to the modulated RF amplifier DC input, where is the majority of the distortion in the modulated output of the transmitter generated?

Is it in the driver/modulator stage, or in the modulated class C final amplifier?

This query also assumes that the RF final amplifier is correctly designed and set up with the proper operating parameters; e.g., bias point, grid drive, antenna loading, etc., and that the modulator-to-class C load impedance matching is optimal. It also assumes, of course,  that the transmitter is not modulated to the point of carrier cut-off, and that carrier shift due to power supply regulation is less than 2%.

I know that in a plate modulated transmitter, the distortion rapidly increases as you approach 100% modulation, but again, is this largely due to the inherent non-linearity of the modulator/driver stage, or the inherent non-linearity of the modulated amplifier?

I guess I could have taken data on the distortion of my audio chain, and load the modulation transformer with a purely resistive load equivalent to the class C load impedance, and connect my distortion analyzer across the resistor through a suitable attenuating network, while driving the modulator to an audio output level that is required for 90% modulation of my transmitter. Reconnecting the modulator to the final amplifier, I could then look at the distortion of the transmitter output by connecting my distortion analyzer to the "measure fidelity" output on my modulation monitor, while modulating the transmitter at 90%. The results would probably roughly correspond to where the bulk of the distortion is coming from. But driving the modulator into a resistive load is not a real world test, when compared to the modulator looking into the complex non-linear impedance presented by the final amplifier, the modulation reactor, the DC blocking capacitor, etc.

I also suspect that most of the distortion we hear when listening to our transmitters' modulated RF output in a decent wide-band crystal receiver, such as that found in a high-quality commercial broadcast modulation monitor, is primarliy IMD, and not lower-order harmonic distortion. Can anyone verify or elaborate on this? IMD is much more offensive sounding than that of low-order harmonic distortion, and somewhat more difficult to measure. Of note, the FCC never specified a maximum IMD level for AM broadcast stations; they only specified the maximum RMS percentage of THD vs. frequency and modulation percentage throughout the entire station audio chain and transmitter/antenna system.

I realize this is a simple question (or maybe not!!), and one that I have pondered for quite some time; perhaps someone has an informed answer based upon their testing and experience, etc.

Don/K4KYV, I am waiting for your reply!!!!

Thanks!

73,

Bruce

[N2DTS]

I would think its all over the place, depending on what people are running and how well it is adjusted.
Some modulator tubes are high distortion, most can be made high distortion if you dont build and adjust it right.

Then you have mod transformer mis matches, not enough class c grid bias/drive, screen grid tubes with their screens being under or over modulated, etc...

I had a push pull 812 rig (no screens) modulated by 811a's driven by an 8 ohm source, and that was very clean.
Better yet would have been AB1 modulators generating low distortion, like an old tube hifi amp was made.

If everything is good, distortion will be very low I would think, and if its very low, why think about where the majority is? A teeny bit here, a teeny bit there....

Brett

[W2XR]

Hi Brett,

Thanks for your reply!

I understand what you are saying, and I don't disagree. Note that I made the following point in my posting; "properly designed linear/low-distortion audio chain (speech amplifier, audio driver, modulator, and no audio processing), and with adequate headroom and audio power capability". By my logic, I want to minimize the variable of any audio distortion being introduced by this subsystem of the transmitter, otherwise it becomes impossible to determine where the majority of the distortion is originating from.

I also made the point that this "also assumes that the RF final amplifier is correctly designed and set up with the proper operating parameters; e.g., bias point, grid drive, antenna loading, etc., and that the modulator-to-class C load impedance matching is optimal. It also assumes, of course,  that the transmitter is not modulated to the point of carrier cut-off, and that carrier shift due to power supply regulation is less than 2%."

However, my gut feeling is that the bulk of the distortion is not in the actual audio section of the rig, again assuming a highly linear audio system, but in the actual modulation process within the plate modulated final amplifier. I am looking for validation of this from someone who has perhaps done testing or analysis in this regard.

A class C amplifier is of course the only class of amplifier that can be linearly plate modulated. As we all know, there is no such thing as a perfectly linear transfer function in any analog input vs. output relationship; in this case, the nonlinearity would be any departure from the ideally perfectly linear relationship between Ep (modulated B+) and the RF output of the amplifier stage. Unfortunately, I never seen the actual (not theoretical or ideal) graphical relationship between Ep (modulated B+) and the RF output of the plate modulated class C amplifier stage, so I have no idea as to how linear it is in practice.

Any further thoughts on this from the team?

73,

Bruce

[N2DTS]

I would think triodes would modulate cleaner, without any screen interaction.
But maybe different  triodes running at different operating points modulate better or worse than others?

Brett

[k4kyv]

You can tell if the final amplifier is modulating linearly by using the trapezoid pattern on the scope.  Feed the horizontal plates with a sample of the audio output of the modulator and the vertical plates with a  sample of the rf.  It is best to feed the plates directly without using the internal amplifiers in the scope, since the horizontal and vertical amplifiers in a scope are not perfect and will inevitably exhibit some degree of non-linearity.  Also, you will probably have to take some measures to eliminate phase shift in the audio sample.

Once you have a linear trapezoid pattern with no phase shift, a perfectly linear class-C final stage will show perfectly straight sides on the triangle or trapezoid of the scope pattern.  Nonlinearity will show up as curvature on the sides of the triangle or trapezoid.  Examples of trapezoid patterns showing various types of nonlinearity and distortion can be found in older ARRL handbooks.

One warning: A problem with running the trapezoid pattern all the time is that it will eventually burn a vertical line on the face of scope tube corresponding to the bright display of the unmodulated carrier.  I use the envelope pattern most of the time, and the T/R function automatically disables the scope during stand-by periods. I use the trapezoid only occasionally to check linearity of the final from time to time.

A well adjusted class C amplifier should show nearly perfect linearity.  Distortion is more likely to come from the modulator and/or driver.  Class B amplifiers inherently have more THD than class A ones.  The ultimate low-distortion modulator would run push-pull class A.  A close second would be class AB1.  But these are low in efficiency compared to class B or AB2.

With a class B or AB2 modulator, the weak link in the chain is the driver stage.  The grids of the modulator are driven positive over just a portion of the audio cycle, and the grid current is not a linear function of grid voltage, so the load impedance on the driver stage varies non-linearly over the audio cycle.  That is why a low internal resistance is necessary for the driver stage if the modulator is driven into the grid current region.  The driver transformer will be a source of distortion unless there is extremely tight coupling between primary and secondary.  This requires special winding techniques to produce a good driver transformer, much the same as the requirements for a top-quality output transformer in a hi-fi amplifier.  Two hi-fi output transformers back to back , or a re-purposed multi-match modulation transformer probably won't cut it.

There are some solid state circuits out there that are far superior to any tube type driver, with extremely low internal resistance.  Steve, WA1QIX published a circuit a number of years ago.

With a tube type driver, low plate resistance triodes such as the 845, 6B4G and 2A3 are best.  Beam tetrodes are the absolute worst unless a considerable amount of negative feedback is used.  The driver transformer should have as much step-down as possible but still drive the grids hard enough to achieve full modulation.  Ideally, as the audio drive is increased, the driver stage and class B modulator should both go into saturation at exactly the same time, and for maximum efficiency, the modulator should be driven to a point just below saturation right at maximum voice peaks.  Some improvement in distortion may be had by not driving the modulator stage quite so close to saturation on peaks, but this will be at the expense of efficiency, and will result in  the tubes heating up more for a given peak audio output voltage.

[W3RSW]

Bruce,

Assuming you have a plate modulated transmitter with a properly designed linear/low-distortion audio chain (speech amplifier, audio driver, modulator, and no audio processing),

....

I thought I'd just add a thought for "beginners" edification and delight. 

As you know, every stage in a real audio amplifier is actually an audio processor. Each has a bandpass determined by R/C coupling or transformers, etc.   Each has generated distortion from the tube or transistor non-linear characteristics, feedback phase at frequency distortion around the single stage's components, power supply stability and 'cleanliness', etc.. 

The overall distortion and bandpass of the multi-stage amplifier are cumulative products of all the stages.  Seldom does an amplifier even approach 'a straight wire with gain.'

So even with the best of all triode tube circuits, with no overall feedback, and driving a stable load you'll get, what? ,.... say 5%, maybe even as low as 1 or 2% distortion.   Then you have to define distortion.  IMD, Harmonic, and more recent "varities."
This is before any modulation or class C stages.

Fun Question. Looking at the discussion with interest.

[KM1H]

This could wind up as a complete tutorial from mike to mod iron!

Ive never been totally satisfied with the audio stages in any of my manufactured or HB seups and to be honest Im still confused. Having QST, CQ, HRM, and the various Handbooks just makes matters worse as everyone claims their design is the best.

As far as drivers Ive found 12B4's to be the best Ive tried but thats really because I refuse to waste good 2A3, 6A3's, etc on a ham project. The few NIB versions Ive had stashed away for decades are for various high end consoles Ive restored and listen to often.

Some of the best communications receiver radio Ive heard comes from the old Super Pros where they use a triode connected 6F6 to drive a triode connected pair and with the IF cranked wide open to 16kc. Has anyone tested the distortion on one?  Jays BA tests show the HRO-60 to be mighty good. If the audio is seperated from the IF and detector it should be outstanding?? 

Carl
KM1H

[k4kyv]

As far as drivers Ive found 12B4's to be the best Ive tried but thats really because I refuse to waste good 2A3, 6A3's, etc on a ham project. The few NIB versions Ive had stashed away for decades are for various high end consoles Ive restored and listen to often.

I don't consider that a waste.  The 2A3 (and later, 6A3 and 6B4G) have been recommended by RCA for use as class B drivers ever since those tube types were released in the 1930's, because of their low plate resistance.  Many of the circuits in amateur radio literature from pre-WW2 up through the much-touted "death" of AM in the late 60's, used 2A3's.  I believe they were also used as modulator drivers in some of the lower power broadcast transmitters.

In the ham publications they were shown being used not only as audio drivers, but in voltage regulator and electronic cathode keying circuits as well.  I have even seen circuits using them as final rf amps in low power transmitters.  I would now consider that a waste.  For voltage regulator and DC amplifier applications, I have a pile of 6AS7G's, which is a better tube for those applications anyway.  Unfortunately, they are not recommended for running at the full 300 plate volts that the 2A3 is rated for.

What I really consider a waste of good 2A3's is when audiophools use them in single-ended class A power amplifiers, convinced that the even harmonic distortion adds "tessitura".

After I built my first class B driver I started accumulating 2A3's whenever I could find them at hamfests, etc.  You used to be able to find them for $1 to $5 each.  I use a quad of 2A3's to drive my 810 modulators and still have a good collection of spares.

A friend of mine went SK several years ago.  He had quite a large collection of tubes, including about a dozen NIB spare 2A3's for his BC-610.  His survivors had let a few people in to look over his stuff as they were trying to sell it. Most everything got sold including the BC-610, but they hadn't yet tried to do anything with any of the tubes and parts.  I was interested in the 2A3's, but looking through the stuff we discovered that some douche-bag had managed to rip off every single one of them, without touching any of the other tubes.

[flintstone mop]

THanks Don,
The trapazoid pattern is a very good indicator for what is going out to the ether.

Fred

[KE6DF]

"As far as drivers Ive found 12B4's to be the best Ive tried"

Has anyone used a 6080 to drive class B modulator tubes through a transformer?

I know they are often used as cathode follower drivers, but it seems like they would be good driver tubes in a transformer coupled design also.

Another tube I have a supply of that might work is 6FM7's. These are dual dis-simlar triodes. One triode is hi-mu and the other low-mu.

They are 12 pin compactrons.

Very roughly you could consider them a 6c5 and a 6a3 in one envelope.

I've seen some articles on the web about audio people using them, but the prices are still low.

Has anyone used them, or heard of anyone using them, as a combined voltage amp/driver in modulators?

You could also use two of them as a combined phase inverter/driver.

I kind of like the idea of using compactrons as they are cheap (except for the big sweep tubes).

[Bill, KD0HG]

Consider using triode-connected power pentodes for drivers.

I'm not at home right now with tube manuals available, but the 6L6GC, 6550, KT-66, KT-88 & etc. all make excellent low Rp triode drivers.

I have always believed that plate modulated power triodes in class C are inherently more linear than a plate modulated tetrode or pentode. As long as they have adequate drive..The key thing is the class C final *must* precisely double its plate current when its plate voltage is doubled. You sometimes have to putz around with power tetrodes to make this true, by adjusting screen voltage, by partially modulating the screen, or letting the screen self-modulate behind an audio choke.(Depends on the tube)

Also bear in mind that not all distortion is 'bad'. Odd-order products such as you get from clipping sound harsh and wretched, a modest amount of even order products don't necessarily sound bad at all.

[Steve - WB3HUZ]

http://www.amwindow.org/tech/htm/scrnmod.htm

[W2XR]

Consider using triode-connected power pentodes for drivers.

I'm not at home right now with tube manuals available, but the 6L6GC, 6550, KT-66, KT-88 & etc. all make excellent low Rp triode drivers.

I have always believed that plate modulated power triodes in class C are inherently more linear than a plate modulated tetrode or pentode. As long as they have adequate drive..The key thing is the class C final *must* precisely double its plate current when its plate voltage is doubled. You sometimes have to putz around with power tetrodes to make this true, by adjusting screen voltage, by partially modulating the screen, or letting the screen self-modulate behind an audio choke.(Depends on the tube)

Also bear in mind that not all distortion is 'bad'. Odd-order products such as you get from clipping sound harsh and wretched, a modest amount of even order products don't necessarily sound bad at all.

Thanks to all who replied!

I have looked at the modulated output of the rig with the trap pattern during the entire history of my transmitter, and the final is indeed modulating very linearly. I now suspect the residual distortion I am hearing may well be normal and that I am perhaps being too much of a perfectionist in terms of what I would ideally like to hear.

I have listened to Steve/QIXs class E rig right off of the on-air monitor in his superb modulation monitor, and the audio quality is literally squeaky clean, with no noticable artifacts, and incredible transparancy, etc. To me, it represents the benchmark as to what a good sounding AM transmitter should sound like. It may be a bit unfair to compare the audio quality of a well-designed and implemented class E rig vs. a well-designed and implemented high-power (1000-watt class) plate modulated transmitter, and I think it would be impossible to expect a plate modulated rig to sound as clean and open as a class-E device.

I believe I have carried the design of my homebrew plate modulated rig as far as I can, in terms of guilding the lily for lowest transmitted distortion and extended audio bandwidth, and I'm probably at that point on the curve of diminishing returns. Originally, I used a pair of triode-connected EL-34s with lots of negative feedback as the audio driver stage, with back-to-back high quality output transformers as an ersatz driver transformer, and this worked well. Later, I scrapped that and went with a pair of push-pull class A 845s, and the driver transformer from a Gates BC-1F, along with 6 dB of voltage feedback from the 833A plates back to the push-pull voltage amp stage of the driver. At that time, I added a pair of parallel-connected 6AS7Gs as a shunt regulator for the class B 833A grid bias supply, but this did not make any really appreciable improvement in the overall distortion of the rig.

The final amplifier is a pair of 4-400As (tetrodes, of course) with a 30 hy screen reactor. I agree with Bill's thoughts insofar as triodes being somewhat more linear in terms of their modulation characteristics vs. tetrodes, and this may well be responsible for the non-linearity I am hearing. Going to a triode PA is major reconfiguration of the rig, and not something I am willing to consider at this time.

I think that what I am hearing (which apparently cannot be heard on-air, unless everyone is being exceptionally kind) is whatever non-linearity exists in the tetrode class C modulated final amplifier stage, and this non-linearity is so insignificant that it cannot be seen in the trap pattern.

Just my thoughts. Again, thanks for the ideas, guys!

73,

Bruce

[KM1H]

I

don't consider that a waste.  The 2A3 (and later, 6A3 and 6B4G) have been recommended by RCA for use as class B drivers ever since those tube types were released in the 1930's, because of their low plate resistance. 

Im aware of their use over the ages Don, I have all those QST's and the 2A3 was just one of many featured in articles. By "a waste" I refer to my own situation where my available stock is limited and I feel that under typical ham conditions their perceived performance is a waste. Other tubes do about as well and the difference is only for bragging rights on a distortion analyzer. I'll keep the 2A3's for my 1931 Scott and others.

Id really like to see a distortion benchmark comparison among a wide variety of drivers. I suspect that a few multigrid designs will be right up with the best.

Carl
KM1H



Carl
KM1H

[w1vtp]

From Bruce's comments:

"....I think that what I am hearing (which apparently cannot be heard on-air, unless everyone is being exceptionally kind) is whatever non-linearity exists in the tetrode class C modulated final amplifier stage, and this non-linearity is so insignificant that it cannot be seen in the trap pattern. ...."

What are you monitoring your sig with?  Could it be that you are getting these artifact from your monitoring system?  As far as folks being kind -- have someone like Bill IFR or someone with a known good recording system record you and listen for yourself.  I could do it with the Flex but I'm sure you can find someone else.  Just have them email it to you so you don't have some over-the-air artifact confusing the issue.

Al

[WA1GFZ]

Hey Bruce I'll talk to you even if you are a bit distorted. Better to have a distorted signal than a distorted mind.

[W2XR]

From Bruce's comments:

"....I think that what I am hearing (which apparently cannot be heard on-air, unless everyone is being exceptionally kind) is whatever non-linearity exists in the tetrode class C modulated final amplifier stage, and this non-linearity is so insignificant that it cannot be seen in the trap pattern. ...."

What are you monitoring your sig with?  Could it be that you are getting these artifact from your monitoring system?  As far as folks being kind -- have someone like Bill IFR or someone with a known good recording system record you and listen for yourself.  I could do it with the Flex but I'm sure you can find someone else.  Just have them email it to you so you don't have some over-the-air artifact confusing the issue.

Al

Al,

Thanks for the question!

I am using a Belar AMM-3A; this is a commercial modulation monitor designed for use with standard AM or shortwave broadcast stations. It drives an Altec-Lansing triode-connected push-pull audio amplifier into either a pair of AKG 'phones or my Altec 15-inch coaxial monitor speaker.

As I stated earlier, I think I may be being overcritical and unrealistic on myself and my equipment; i.e.,  in what I am expecting from the technology of the rig in terms of exceptionally low distortion and transparency.

73,

Bruce

[W2XR]

I don't consider that a waste. The 2A3 (and later, 6A3 and 6B4G) have been recommended by RCA for use as class B drivers ever since those tube types were released in the 1930's, because of their low plate resistance.  Many of the circuits in amateur radio literature from pre-WW2 up through the much-touted "death" of AM in the late 60's, used 2A3's.  I believe they were also used as modulator drivers in some of the lower power broadcast transmitters.

I seconded that, you can’t beat a pair of 2A3’s, 6A3’or 6B4G’s for audio drivers in an AM transmitter. They are low umhos triodes and you can even get by with directly coupling 6J5’s, 6SN7’s or a 6N7, etc. to their grids without even using an interstage transformer. I use a pair of 6A3’s to drive my 810’s and you can’t get much cleaner non-distorted audio using any other pair of driver tubes.

Well, except for maybe WE300B's

I once considered using a pair of 300Bs in class A push-pull for my audio driver; they would have replaced the pair of triode-connected push-pull class AB1 EL-34s I had been using at that time.

I use class A 300Bs in my homebrew stereo system, so I am quite familiar with the tube and it's application. It is certainly one of the most linear audio power triodes of all time. In my opinion, I think it may be a waste to use them to swing the grids of a pair of 810s or 833As delivering 500 watts of audio, as you need more audio power and headroom than push-pull class A or AB1 300Bs can deliver, to keep the distortion introduced by the driver stage to a minimum. They would be great for driving a lower power modulator, where the grid driving power requirements are significantly less.

I went with push-pull class A 845s with 1250 volts on the plates instead. Much more headroom, audio power, superbly linear tubes, and a perfect plate-to-plate load impedance match to the Gates BC-1F audio driver transformer I had, when compared to either 300Bs or triode-connected EL-34s. I think a pair of 845s is ideal for driving the difficult load presented by a pair of class B 833As, and I added 6 dB of voltage feedback between the 833As and the audio driver when I installed the 845s.

73,

Bruce

[W2XR]

Hey Bruce I'll talk to you even if you are a bit distorted. Better to have a distorted signal than a distorted mind.

Hey Frank,

I'll talk to you any time too, regardless of how good or bad the rig sounds! Then again, I'd rather talk with you in person; sorry I missed you at Deerfield last month.

I should be back on the air in a week or so. I am doing some modifications in the shack, and I recently sent a load of sheetmetal for new projects and radio restorations out to the paint shop for painting, and my VU meter panel is in that batch and gone for now. I can't watch the audio level to the rig without the meter panel.

Best 73,

Bruce

[w1vtp]

From Bruce's comments:

"....I think that what I am hearing (which apparently cannot be heard on-air, unless everyone is being exceptionally kind) is whatever non-linearity exists in the tetrode class C modulated final amplifier stage, and this non-linearity is so insignificant that it cannot be seen in the trap pattern. ...."

What are you monitoring your sig with?  Could it be that you are getting these artifact from your monitoring system?  As far as folks being kind -- have someone like Bill IFR or someone with a known good recording system record you and listen for yourself.  I could do it with the Flex but I'm sure you can find someone else.  Just have them email it to you so you don't have some over-the-air artifact confusing the issue.

Al

Al,

Thanks for the question!

I am using a Belar AMM-3A; this is a commercial modulation monitor designed for use with standard AM or shortwave broadcast stations. It drives an Altec-Lansing triode-connected push-pull audio amplifier into either a pair of AKG 'phones or my Altec 15-inch coaxial monitor speaker.

As I stated earlier, I think I may be being overcritical on myself and my equipment; i.e.,  in what I am expecting from the rig in terms of exceptionally low distortion and transparency.

73,

Bruce

Bruce -- but you are hearing something.  For the sake of "science" it might be time to run the xmtr into a good load, couple it off to a spectrum analyzer and run some IMD tests.  At some point in time you probably need to decide what is acceptable for measurable distortion.  Hearing distortion is so subjective.  I could imagine running the output into high power 30 dB attenuator to remove any "floating" RF and running that directly into a good spectrum analyzer.

Anyway, it does sound like an interesting pursuit I don't quite agree with the plate modulated vs class E comparison but we all have to agree that a properly built and adjusted "E" transmitter lends it self to very low distortion and very wide frequency response.  But if one were to be realistic about expectations re a plate mod transmitter it should be fine for the discriminating ear.  Now -- at this point in time I'm dying of curiosity on what you have there.  Even a schematic along with your plate iron etc.  It may be out of the scope of this thread.  If so, I'd love to correspond with you via email

Fun subject and I'm sure your transmitter sounds great.  

Al

[N2DTS]

I have 2 modulators for my rig, a pair of 4cx250b's in AB1 and push pull paralell 100th's.
I hear more distortion in the mod monitor using the 100th mod deck, so I never use it.
I can switch between mod decks in seconds, and had people listen and they tell me they sound the same, even though I can clearly hear some sort of distortion on the 100th deck.

I think what you hear in the mod monitor is much worse then what the guy at the other end hears.

Brett
 

[Opcom]

One warning: A problem with running the trapezoid pattern all the time is that it will eventually burn a vertical line on the face of scope tube corresponding to the bright display of the unmodulated carrier.  I use the envelope pattern most of the time, and the T/R function automatically disables the scope during stand-by periods. I use the trapezoid only occasionally to check linearity of the final from time to time.

A worthwhile technique here for those of us encumbered with CRTs would be to rectify a small amount of the audio signal and use it to progressively intensify an otherwise cutoff CRT beam. This will provide a uniform brightness per square area of pattern, if the amplitude of the audio sample is adjusted properly. It avoid the situation where the point (0% mod) of the trapezoid is bright and the wide end (100% mod) is dimmer. The sampling can also be taken from the RF output but it is more complex to handle since a DC bias shift of 30-100V may be required depending on the CRT characteristics and operating (accelerating) voltages. I would like to discuss further at some time.

Bruce,

Assuming you have a plate modulated transmitter with a properly designed linear/low-distortion audio chain (speech amplifier, audio driver, modulator, and no audio processing),

....

As you know, every stage in a real audio amplifier is actually an audio processor. Each has a bandpass determined by R/C coupling or transformers, etc.   Each has generated distortion from the tube or transistor non-linear characteristics, feedback phase at frequency distortion around the single stage's components, power supply stability and 'cleanliness', etc..  

The overall distortion and bandpass of the multi-stage amplifier are cumulative products of all the stages.  Seldom does an amplifier even approach 'a straight wire with gain.'

One way to look at a TX is as an analog computer. The whole TX is in fact an analog computer. A rather crappy one at that. But the better it can be..  Despite the truth that very good results can be had with simple circuits, simpler is not always better due to the variables above. Some means of correction will help alot.

---

To track down distortion that can already be seen on a scope, there are alot of possible test points.

1.) I find it helpful to check each stage at a time, including across transformers (phase shift as frequency changes) and it is also useful when using a trapezoid display to try the audio from many stages.

2.) One approach to finding the "problem stage" in an audio chain is to use the scope connected to the input and output of each stage as either an XY display to look for a clean diagonal line or as two vertical channels and subtract them, adjusting the gain for the smallest wave on the horizontal trace. If the wave does not go to zero with scope gain adjustment, there is an imperfection between the probes.

3.) another way: If the PA is modulating cleanly as evidenced by a good trapezioid, the audio wave can still be very distorted because the display is not dependent on waveform, only the linearity between AF and RF inputs. Tracking the audio back while comparing it to the RF, the bad stage will be where the trapezoid first looks incorrect.

The problem of having feedback in the system can complicate the effort of trying to find the poorly performing stage. Well anyway there are alot of ways to look at it and even more ways to poke around inside it.

To answer your question directly, in my case the biggest issue is the modulator power stage. It's the hardest to correct. I have a modulator crossover distortion issue with a pair of zero-bias 3-500Z's. Resting current 200 mA, plate volts 3000. Go figure! I will eventually get it right.

[W2XR]

I think it may be a waste to use them to swing the grids of a pair of 810s or 833As delivering 500 watts of audio, as you need more audio power and headroom than push-pull class A or AB1 300Bs can deliver, to keep the distortion introduced by the driver stage to a minimum. They would be great for driving a lower power modulator, where the grid driving power requirements are significantly less.

I went with push-pull class A 845s with 1250 volts on the plates instead. Much more headroom, audio power, superbly linear tubes, and a perfect plate-to-plate load impedance match to the Gates BC-1F audio driver transformer I had, when compared to either 300Bs or triode-connected EL-34s. I think a pair of 845s is ideal for driving the difficult load presented by a pair of class B 833As, and I added 6 dB of voltage feedback between the 833As and the audio driver when I installed the 845s.

It depends on how much modulator power you need. Say you use a pair of 810's to modulate a single 4-400 when they normally modulate two 4-400's. If you only use one 4-400 then your modulator has plenty of headroom right? So it doesn't require driving the 810's very hard. 810's only require an average of 10 to 13 watts max for full drive at 725 watts of audio power above 2000Vdc on the plates. My 810 plate supply runs at 2200Vdc and I only have one 4-400 RF final. Its a CG so I get a bit more power than a standard 4-400A, about 500 watts max. If I drive the 810's up all the way, well.... you know what happens. I get well over 150% modulation. I did that on purpose so I could keep the entire audio chain turned down well below the saturation point starting with the mic amp tube all the way to the 810 finals. I don’t overdrive anything so distortion pretty much doesn’t even exist. If I’d used driver tubes other than the 6A3’s they wouldn’t sound near as good or be near as clean.
 
To me its always better to overbuild the modulator for a high level plate modulated AM transmitter. It solves problems and it will sound much better in the end. You can then go ultra modulation with it too if you want

I'm sorry, I don't know your name!

A good point; if you are not in need of the kind of audio power a pair of class B 810s or 833As can easily provide (>500watts, particularly 833As), the demands on the audio driver stage can be reduced accordingly. When modulating say a single 4-400A at perhaps 500 watts DC input, the modulator need only deliver about 250 watts of distortion-free audio for 100% modulation. A pair of class AB1 (not class A) 300Bs would be a reasonably good choice here, but not the most cost-effective one, particularly if you elect to use original WE-300Bs!

A fun thread, indeed!

73,

Bruce

[W2XR]

I have 2 modulators for my rig, a pair of 4cx250b's in AB1 and push pull paralell 100th's.
I hear more distortion in the mod monitor using the 100th mod deck, so I never use it.
I can switch between mod decks in seconds, and had people listen and they tell me they sound the same, even though I can clearly hear some sort of distortion on the 100th deck.

I think what you hear in the mod monitor is much worse then what the guy at the other end hears.

Brett
 

Hi Brett,

What class of operation are the 100THs biased up for?

You make a very good point about what you can hear in your on-air monitor, vs. what others cannot hear. Once our signals take a trip through the aether, they usually sound pretty damn good at the other end of the link. The atmospheric noise and the other fellows receiver can mask a multitude of sins. However, when listening to your own voice right off of the modulation monitor, suddenly the emperor has no clothes; it's very revealing.

When I was in college, I had an engineering job at a local AM station, where they ran a Gates BC-1F. I was 18, and my critical listening skills were not at all sophisticated at the time. I recall the BC-1F sounded spectacular, like a big jukebox, very full and rich sounding. I loved the sound of that rig.

I would like to have a critical listening session with that very same BC-1F now, 37 years later. I wonder if I would hear the same low-level artifacts in that transmitter that I hear in my rig. The entire audio line-up in my transmitter is absolutely identical to the BC-1F, yet the BC-1F used triodes (833As) in the RF final amplfier. I use tetrodes (4-400As). As I indicated earlier, I am beginning to question the source of the distortion as perhaps being in the tetrode HPA; a very small and probably quite normal non-linearity in the HPA.

I have heard numerous recordings of my rig made on-air, and as expected, I can never hear the distortion artifacts in those recordings. I can only hear them when listening to myself through the mod monitor audio output driving a pair of good quality headphones through a known clean triode amplifier.

Bone conduction issues anyone??

The thoughts presented here by the members of the board are fascinating to read.

73,

Bruce

[N2DTS]

The 100th's are running class b.
The problem is in the driving power I think, the 20 watt LS series driver transformer might be too light for four 100th's.
That or the ab1 deck is much cleaner than anything running class b.

Brett