High Level v Low Level: re audio quality

[G8VOQ]

Let's imagine that we have two audio amplifiers one delivering 75W, the other 7.5W, and when fed into a matched resistive load, each would show the same level of distortion.

Let the 75W audio amplifier plate modulate a class C PA amplifier. Let the output be 150W of AM. (High level modulation).

Let the 7.5W amplifier plate modulate a class C PA and then let that 15W of AM be amplified by a linear amplifier outputting 150W of AM. (Low level modulation).

Would the 150W AM output from the low level modulation method be as good as using the high level modulation method? If so, would the linear amplifier need to be a particular class? (A, AB1, AB2 or B).

I know maybe most use high level modulation and there could be good reasons for that,  but I'm just interested in the issue respecting audio quality.Thanks.

[G8VOQ]

We have two plate modulated stages, one outputting 150W the other 15W. In theory, I would say that if the AM quality generated in the PA stage, was the same for each PA stage, linear amplification would,  in the low level scenario, introduce a bit more distortion.

[W7TFO]

I think you are correct in the surmization that this basic topology inherently had more audio distortion than plate or high-level modulation .  However, things were done to remedy that problem.

In the usual pre-war broadcast transmitters using low-level (grid) modulation followed by a linear stage or two, they used global rectified RF feedback to the first audio stage to get the noise and distortion down a lot. 

Audio fidelity was only limited by the bandwidth of the input stages, and the modulation percentage was usually over 95% capable.

If a Doherty linear amp (like Western Electric and Continental Electronics) was used in the final efficiency was over 65%.

They actually sounded very good with this topology, and audio passband out to 16 kHz with THD under 3% was the norm. 

Tough to get those kind of specs with high-level class B audio systems full of phase-shifting iron without spending a lot on that iron...

73DG

[k4kyv]

First of all, using an amplifier delivering 75w of audio power  to modulate a class C final amplifier running 150w carrier power, would not be capable of modulating 100%.  In theory, a class C amplifier running measured 150w output would run at least 200w DC input.  In actual practice, after considering tank circuit losses, etc, the DC input would probably run closer to 250 watts. The amount of sine wave audio required to 100% modulate a final, is one-half the DC input.  So, to assure 100% modulation capability, you would need a modulator rated at more like 125 watts output.  Another loss often overlooked is losses in the modulation transformer.  A good modulation transformer is probably at best 90% efficient.

Assuming you have enough audio to fully modulate the final, a well designed high level modulation system should be  roughly comparable to using a linear amplifier fed by a fully modulated lower level stage.  After all, the higher power audio amplifier used as the modulator stage is nothing but a linear amplifier operating at audio frequencies.  The distortion will be no more nor any less than that of a linear amplifier operating at rf. Additional distortion may be generated in the high level system if the modulation linearity of the class C final is not  perfect, and it rarely is.

One of the main advantages of high level plate modulation is ease of adjustment.  Once the final is loaded into the antenna, and if everything is running within ratings, the tube is not weak in cathode emission, the bias and grid current are correct, you should  get good results when modulating it with an adequate modulator.  Antenna loading should make negligible effect on the modulated waveform, and you should have considerable leeway with grid current.

OTOH, with low level modulation, and this includes grid modulation as well as linear amplification, grid drive and antenna coupling become critical for maintaining 100% modulation capability and modulation linearity.  If the final is not loaded to the antenna exactly right or the grid drive is not exactly at the critical current reading, the positive peaks may be clipped or the final may modulate non-linearly, as indicated by an increase or a decrease in DC plate current as the final is modulated. Furthermore, antenna loading and grid drive adjustments are affected by one another.

As for overall efficiency, the plate modulated transmitter is not that much more efficient than the low level modulated stage, once modulator filament power, plate loss and transformer losses, along with the power consumed by the audio driver stage are taken into consideration.  The ratio of power drawn from the mains to power delivered to the antenna is close to the same with either system.  


For voice modulation, the amount of audio required for 100% is peak power. The average power may be much less, not even 50% of peak power.  Therefore, the tubes and modulator power supply may be run to deliver the higher peak power while keeping the average power level within tube and power supply ratings.  One way to do this is to  run the plate voltage somewhat higher than specified for a given output.  Of course, if the modulator is designed to deliver 100% peak modulation, you want to limit the audio signal to voice waveforms, and not try to run a sine wave at 100% modulation.

[G8VOQ]

Yes, I got my figures wrong. For 150W output that was in fact radiated, the amount of power in the sidebands would be measured at 50W in the case of a single pure sinusoidal tone, and 100W would be carrier. But the AF power supplied to get that condition would be more, because of losses. And this represents 100% modulation. Speech results in the average power radiated in the sidebands being less than 50W.

Seems that setting up a low level arrangement is a bit more tricky and in that sense audio quality is at risk, due to setting up, rather than inherently part of a low level arrangement.

I've never built a linear amplifier, but would one particular feeding arrangement or linear amplifier arrangement be better than another another with AM?  What about the classes of amplification, how do the various classes impact, particularly on audio quality?

BTW, I understand "high-level" simply refers to modulation taking place at the final PA stage feeding the antenna, and says nothing about the actual modulation method, be it grid leak, suppressor grid, cathode, plate etc.

[k4kyv]

High level means anode modulation, or perhaps an alternative term would be anode power supply modulation, since it is the DC feeding the plate of the final that is being modulated.

Low level, or efficiency modulation, includes linear amplification, control, screen or suppressor grid modulation, or any combination thereof. Plate modulation of a screen grid tube requires that the screen grid be modulated along with the  plate, since screen voltage has a greater effect on plate current than plate voltage. That alone could be a topic of discussion, since there are various methods of achieving the screen modulation and disagreement over the percentage of screen modulation tracking the plate modulation.

Linear amplification and control grid modulation are very close to the same thing.  You could think of linear amplification as grid drive modulation.  That is, the rf excitation to the grid is modulated and the grid bias voltage is held constant, while with control grid modulation, the rf drive is held constant and the grid bias voltage is modulated. In both cases, both the grid bias voltage and rf driving voltage must be as well regulated as possible. Variations of grid bias with linear amplification, and variations of rf drive with grid bias modulation will produce similar distortion.

The best solution for the above problem would be to use class AB1 linear amplification, but the rf output and PA efficiency tend to be lower in that  class of service.

I have always preferred plate modulation, and in 51 years on the air, have rarely used anything else.

[G8VOQ]

Well, I also understood high level to be plate modulation and the rest low level.  Then reading the RSGB handbook recently I read "The modulation systems so far considered are known as high-level systems because the modulation is applied to the rf stage working at the highest power level" - (Radio Communication Handbook,5th EDN, p9.7)

I think perhaps there is a difference in US/UK terminology here. I thought myself it does make better sense calling high level modulation taking place at the highest level in the RF chain, i.e. at the final PA. And not because I am English. :c)

The book goes on to say low level modulation system is used in SSB transmission.

In my scenario where AM is generated by plate modulation then subsequently amplified in a linear, that would be called low-level modulation in the UK, if the handbook got it right as to proper UK meaning. Modulation in that scenario does not take place in the linear amplifier. (Yet employing a linear invokes a sense of efficiency modulation?).

Edit: I see KYV says in relation to this issue, high level is the term associated with "DC feeding the plate of the final that is being modulated". That is true, because of the word "final".

In my scenario using a linear amplifier I think all would agree it's low level modulation.

[G8VOQ]

http://www.w8ji.com/amplitude_modulation.htm

Low level modulation is more convoluted/complicated than I thought.

I note this:

"A rig certainly does NOT need to be plate modulated to sound perfect, and as a matter of fact most amateur plate modulated transmitters have terrible distortion as a percentage of modulation. It's just that most people can't actually hear the distortion, they listen to and enjoy the frequency response and might actually "like" a little distortion, and they confuse distortion with good sound. Contrary to popular myth, there is no difference in the sound of any AM transmitter when amplified in a properly tuned and operated linear amplifier. This is because a properly tuned and operated linear, be it a Heath SB220 or anything else, has much less modulation distortion than the typical boatanchor rig. The real problem with a linear is NOT the sound. The real problem is heat caused by poor carrier efficiency."

I wonder, should it say that the real problem is heat caused by tube plate efficiency?

Anyway, I see now how the big problem is efficiency on the final PA with low level modulation. That's the main point of trouble.

[w3jn]

There is no inherent "best sounding" method of generating AM.  There are class-C plate modulated rigs that sound wonderful and some sound awful.  There are stock riceboxes with low level modulation that will make your ears bleed and there are those that sound fabulous.  And there are class-E PDM rigs that sound great and a few, whose owners are a bit overenthusiastic in pursuing peaks over 125%, that sound really gritty on a standard AM diode detectos.

Some methods are easier to generate good-sounding AM than others, however.  Low level modulation (ie feeding audio to a balanced modulator at the +6 to +12 dBm level) is by far the easiest - you don't hafta worry about high level audio amplifiers, audio transformers, and other souces of distortion.  High level modulation (either plate or screen) is more difficult, as distortion will creep in, transformers saturate with high levels of DC current, and negative feedback (the easiest way to reduce distortion) becomes problematic unless the mod tranny has a tertiary tap to provide it.  Other methods of high-level modulation such as drain or collector modulation can work very well but you still need a high level audio stage.  PDM is quite complex and requires good filters on the output of the PDM modulator stage to en sure you're not transmitting spurs.

The obvious disadvantage as identified by your research is that linearly amplifying low level AM is inherently inefficient.  Linear amps meant for SSB and pressed into AM service need to be carefully driven so as not to overtax the power supply or the amp's capacity to extract heat.   As an example, the Collins 30L-1 KW amp is a reliable stalwart for SSB service but don't expect much over 100 watts output on AM for any period of time.  The PS and fan aren't up to the task of much more.   

As a further example you can convert a SSB rig to very high quality AM, but a 100 watt class transmitter isn't going to be good for much over 15-20 watts.  I converted one of those crappy Heathkit HW-12 monobanders to AM by disabling the ALC circuit, removing and bypassing the xtal filter, and feeding the audio right into the balanced modulator.  It would faithfully pass 10 KHz square waves(!) but the pair of 6GE5s ran out of headroom past 20 watts.   Also I wasn't too happy about chasing the modulator balance around as it tended to drift and the carrier would creep up.

And finally let's consider the receiving end.  Few older receivers can handle much over 80% modulation without the diode detector distorting the audio.  There are some newer receivers that are much better, but inject static, QRM, and multipath fading into the mix, the pursuit of very low distortion AM transmitter on HF when most guys' receivers aren't up to the task is rather a fools' errand.  That is not to say that the pursuit of decent audio is worthless - absolutely not - but very often "better" is the enemy of "good enough".

[KM1H]

If you want great sounding linear AM start with a clean phasing transmitter running DSBAM and then use any of the very low distortion amplfier tubes. With -40dB and better IMD and driven properly you will be clean as well as having decent audio bandwidth. The end result is less physical space with actual power consumed and heat being somewhat of a tradeoff.

A SB-220 run in the 1800V CW position is about as clean as it gets with 3-500Z's. If you have the equipment you can experiment with the bias for minimum IMD which can be in the high -40's.

Carl

[k4kyv]

The efficiency of a linear amplifier is exactly the same whether it is amplifying SSB or DSB AM with carrier.  Efficiency is a function of the amplitude of the signal up to the point of saturation.  Theoretically, the efficiency should be more than 60% when the signal level is just under the saturation point.

At zero signal level, the efficiency is zero, since the tube draws some static plate current but the output is zero.  At intermediate points between zero signal and the saturation point, efficiency varies in direct proportional to the amplitude of the signal.

The reason that the AM linear gets a bad reputation for "inefficiency" is because with the unmodulated carrier, which should reside midway between zero signal and the saturation point, the efficiency is approximately half the maximum peak efficiency.  If the peak efficiency is 60% as mentioned above, the unmodulated carrier efficiency is about 30%.  This is most obvious with full carrier AM because of the 100% duty cycle of the carrier, whereas with SSB, voice peaks that hit the midway point are of short duration and cause less heating of the final.

With SSB, the average signal level with the typical human voice (without a lot of processing or overdriving) is about 30% of the amplitude of the maximum peaks.  This means that the efficiency of a SSB linear running a clean signal averages something on the order of 18% most of the time even though it may run 60% or more on the maximum peaks. By the same token, the AM linear peaks close to 60% efficient on positive modulation peaks. 

With a sine wave tone modulating the carrier 100%, the tube will run cooler than when there is no modulation of the carrier. A properly operating AM  linear draws steady plate current regardless of modulation, therefore the DC input is invariable.  With 100% sine wave modulation, we see a 50% increase in total rf output, accounting for the upper and lower sideband energy in addition to the carrier. Since the DC input is the same regardless of modulation, that extra 50% has to come from somewhere, so that means the final runs at higher efficiency to generate that extra power.

[KM1H]

VOX is probably the easiest way to reduce the heat but it annoys some who could probably use the annoyance I remember when VOX AM was very popular in roundtables.

There have been several bias shift circuits published to improve the carrier efficiency with no audio. Some add distortion

[k4kyv]

I find VOX on AM to be extremely annoying, particularly if the time constant is set so that the carrier drops out between words in a sentence.  Better to use PTT.  You can still work fast break-in type QSOs with PTT. My T/R switch is attached to the end of a spiral microphone cord salvaged from a junked 2-way radio hand mic.

[KM1H]

Unfortunately the windbags dont use either which makes AM very annoying in a roundtable.

It seems that the lower you go in frequency there are more gasbags

[Steve - WB3HUZ]

Me too.

http://www.youtube.com/watch?v=3O8GFtspk9U&feature=related

I find VOX on AM to be extremely annoying, particularly if the time constant is set so that the carrier drops out between words in a sentence.  Better to use PTT.  You can still work fast break-in type QSOs with PTT. My T/R switch is attached to the end of a spiral microphone cord salvaged from a junked 2-way radio hand mic.

[KM1H]

I had a nice 3 way on 160 last eve and all were on VOX. I was playing with the HT-32B/Alpha 76PA 3 holer into the new antenna and looking for any arcs or other funny stuff with a 400W carrier and a 2.5:1 VSWR at around 1950 (Its real low down low where I usually hide chasing DX). The 450' of 3/4" CATV hardline had been left at the 180' level with just a splice block with one end capped for about 15 years. I guess those O ring seals really work.

There was someone on 1885 with a nice S-9+ carrier around 5:45PM but the audio was too weak to copy except the word CQ once in awhile. By the time I got tuned up he was gone.

[flintstone mop]

Me too.

http://www.youtube.com/watch?v=3O8GFtspk9U&feature=related

I find VOX on AM to be extremely annoying, particularly if the time constant is set so that the carrier drops out between words in a sentence.  Better to use PTT.  You can still work fast break-in type QSOs with PTT. My T/R switch is attached to the end of a spiral microphone cord salvaged from a junked 2-way radio hand mic.

GOOD ONE STEVE