Separate Power Supply for Modulator?

[KE6DF]

What are the plusses and minuses of using one power supply for a class C final RF state, and a separate power supply for the plate modulator?

I talking about a transmitter in the 300 Watt range.

It seems like separate supplies are nice in that you can run the modulator off a different voltage than the final stage. Separate supplies give some flexability in using one modulator with multiple RF decks -- if you have multiple tube types.

On the other hand, if both the modulator and final are run off one supply, the class C state presents a constant load to the supply leading to better regulation for the class B modulator (which is a highly variable load).

What do most of you do?

[KM1H]

What do most of you do?

All the above.

If both output tubes take the same voltage then they share a supply that is filtered better than a seperate one for the Class C stage. At the 300W level that is easy to do as long as the iron has the current capacity.

Even at higher power its sometimes easier to use a higher voltage rated tube in the modulator such as 572B's instead of 811A's when running 813's. A heavy duty xfmr delivering 2000-2500VDC is often easier to find than a pair of different ratings.

It all depends upon what you are trying to accomplish, whats on hand, available space, and how much you really want to spend. I usually start with a big zero in the latter column and then try and back into it

Carl
KM1H

[Opcom]

I prefer a separate modulator supply, especially on bigger stuff. At 300W (carrier) I would definitely consider it, but both supplies must have very good regulation, or they should track one another %-voltage wise under the varied operating conditions presented by both the modulator and the RF PA. The problem arises because of PA current differences according to tuning and loading. The tracking will be correct only for one set of tune/load conditions.

There is an effect on modulation percentage and carrier shift caused by power supply regulation imperfections when the separate PA and modulator B+ regulations do not track or when the common supply sags under load. I wrote an article on power supplies that addresses this as one of its last points and the excel sheet can be downloaded as well if anyone is interested in that.
http://www.bunkerofdoom.com/lit/psu_reg_mod/index.html

I also want to say that the RF PA stage, when modulated, demands much more current from the RF PA power supply. When the modulator doubles the RF PA plate voltage, the RF PA power supply has to supply twice the peak current.

carrier:
PA supply = 3000V
PA current  = 300mA
DC IN = 900W

peak envelope
PA supply = 3000V
modulator adds 3000V
PA current = 600mA
DC IN (instantaneous) = 3600W

Tihs has to be drawn from the filter capacitor, and that won't be recharged till the next mains cycle.

Assuming an extremely well regulated common supply, The only reason I would go to the expense of separate supplies is so I can adjust the voltages independently. The effect of separately adjustable supplies to fine tune the rig so that at 100% modulation the carrier shift is reduced as much as possible. For a 300W rig, a pair of 5A variacs might do.

There are advantages to separate adjustment of the RF PA and the modulator plate voltages. I submit these theoretical asumptions which assume perfection. We don't get perfection because of nonlinearities and the like, but it works.
This can be applied to any size of transmitter. I am using larger numbers than for a 300W transmitter because of what I have been working with and I am less likely to make confusing mistakes if I use numbers like those in the equipment.

From the standpoint of the HV supply, the RF PA presents (or should present) a certain resistance which will be the same for any frequency assuming the stage is tuned properly.

For any level of HV, the DC load resistance presented by the properly tuned RF PA is the same. (assume bias and drive are properly adjusted)

Therefore the load impedance of the modulator will always be the same when the stage is operated a certain way, for example it is loaded so that at 3000 volts it draws 300mA.

For a modulator operating in a steady state condition such as "maximum undistorted power", the same is true. For argument sake, assume the modulator draws also 300mA at 3000VDC.

So
1.) Here we have a RF PA stage with 900W input and in class C at 70% efficiency, 630W output.
2.) Here we have a modulator with 900W input and at 50% efficiency, 450W output.

From the above two statements, one would wonder why separate supplies would be wanted, because it is a perfect 100% modulation scheme. The reason is that by doing nothing other than varying the B+ with one or both, the power input and output of either or both stages can be scaled to accommodate slightly deviant or special operating conditions.

If I want to try 120% positive modulation, I could either:
1.) Lower the RF PA B+ to bring the DC input into the range where the modulator could supply enough power for that.
2.) Raise the modulator B+ to increase the modulator's power output capability while keeping the RF PA untouched.

The reason for separately adjustable voltages is so that the desired ratio of modulation transformer secondary-side audio voltage versus RF PA plate voltage can be had. If the modulator is oversized power-wise, this can be taken pretty far. I've figured these things out from working on the "Tucker KW" home brew which can be set for 200W-1500W DC input and has a modulator capable of about 1400W RMS. And no a 100W carrier with two 700W sidebands on it does not sound good -it sounds like a super bowl CB base station! But a 300W carrier with two 700W sidebands makes the bird swing and the rig sing! There are all kinds of practical experiments to do with separately adjustable B+ voltages and that's why I prefer them.

So my humble reccommendation is to either have a really stiff common supply and overbuild the modulator so it delivers more than enough voltage swing (maybe p-p of 1.4 times the B+ value) and adjust the audio drive to control modulation regardless of RF PA tuning and loading, or, go with separately adjustable supplies and really get into the control of everything.

[N2DTS]

All the rigs I have built used seperate supplies up till the 3 x 4D32 rig which uses one.
I cant say I notice any difference.

I run it at 1200 volts, the RF  deck pulls about 300 ma, the modulator does what 811's do, not much resting, but plenty on peaks.

I would like to offload some current from the one supply, so will eventualy build a seperate supply for the modulator.
That is one advantage, you can use smaller parts.

Brett
N2DTS

[steve_qix]

You summed it up in your original post 

I would personally never build a transmitter with a seperate modulator supply for the modulator.  I've never seen a broadcast transmitter with a seperate supply for the modulator and RF. 

The RF amplifier's constant load on the supply really improves the regulation, and of course with a common supply, and 100% grid leak bias on the RF amplifier, there is no chance of the modulator operating into "nothing" (unless the RF amplifier tube heater burns out - use parallel tubes  ).

I am pretty sure a common supply would occupy less room (and weight) than seperate supplies.

On the issue (or advantage) of being able to switch between various RF decks - I do this here as well.  So, I just switch the output of the modulator/power supply combination to the RF deck in question, and all is just fine !

Anyway, just my experience 

Regards,

Steve

[WD5JKO]

  I saw something elsewhere on this site that is relevant here. Many transmitters (Johnson gear, Heathkit, Gonset, etc,) use common supplies with a mod transformer that has a turns ratio that prevents over modulation beyond 100% positive. I think that is a 2:1 stepdown ratio (??). In a situation like this, we can use separate supplies staggered such that we can modulate upward to 125+ %.

   One way to do this is with a common transformer and dual rectifier and filter circuits. Maintain the existing choke input for the RF stage, and add solid state rectifiers and capacitor input for the modulator. On the modulator supply make sure the capacitance is large, such as 100uf or more. The modulator idle current should drag the mod B+ down to about 1.2X the transformer RMS voltage, and the RF B+ with choke input will be about 0.9 X transformer RMS. This should allow for nice 125% upward modulation capability.

    This idea highlights the need for step-start surge reduction, and then complicates push to talk B+. Nothing simple is ever easy.

my 2 cents,
Jim
WD5JKO

[WU2D]

Using separate supplies gives you more flexibility in impedance matching. Most of the time on the small rigs that I have played with, you are usually using a push pull modulator and a couple of tubes in parallel in the RF section, for example a pair of 6146's modulated by 6146's in P-P. Unless you have a modulation transformer that steps down (like on the Apache for instance), you are going to want more voltage on the modulators than on the RF tubes. 

Mike WU2D

[k4kyv]

The RF amplifier's constant load on the supply really improves the regulation, and of course with a common supply, and 100% grid leak bias on the RF amplifier, there is no chance of the modulator operating into "nothing" (unless the RF amplifier tube heater burns out - use parallel tubes. 

The regulation in question is in regard to the modulator plate voltage.

This allows you to use a higher value bleeder resistor since a heavy bleeder is not needed to maintain regulation.

But if you also use the rig for CW, the heavy bleeder is necessary on the RF power supply to maintain a good keyed waveform.

Separate supplies allow you to adjust the modulation percentage capability of the transmitter with a fixed ratio modulation transformer.  With about a 1.6:1 turns ratio and common power supply, modulation percentage is limited to about 100% with no extra headroom in the positive direction.  To gain 125% positive modulation capability using the common power supply, the transformer ratio must be about 1.3:1 or less.  But then the plate-to-plate load impedance may be lower than optimum for the modulator tubes.  One solution would be to run a quad of modulator tubes in pushpull-parallel.  But then, that increases the requirements at the audio driver stage.

With separate power supplies, more careful design is required to maintain good DC voltage regulation, but the voltage to the modulator can be increased above what is on the class-C final, to allow for a higher percentage modulation capability; the increased voltage allows for greater peak audio power capability, at a higher plate-to-plate impedance load on the modulator tubes, which may be closer to optimum for those tubes.

On my HF-300 rig I run 2000 volts on the rf final and 2600 volts on the modulator.  The mod xfmr turns ratio is 1.55:1.  This allows for almost 150% positive modulation capability.  Even if I don't run that much peak audio, the extra headroom allows for a cleaner signal than would a modulator that is approaching the saturation point on voice peaks.

I run two 50k 200 watt bleeder resistors in parallel in the modulator power supply, with 28 mfd of filter capacitance, choke input with a swinging choke approximately 8/30 henries over the current range I pull  from the supply.  This maintains good regulation, both static (as read on the analogue voltmeter with mechanical movement) and dynamic (as indicated by an instantaneous voltage indicator such as an oscilloscope or l.e.d. voltmeter).

My 8005 rig uses a common power supply, with a Gates BC1-T modulation transformer that has approximately 1.35:1 turns ratio.  It is capable of just a hair over 100% modulation in the positive direction, just like the BC1-T I use on 160m.

In over 40 years of running plate modulation, I have never lost a modulation transformer due to losing the class-C plate voltage.  For one thing, the load is still there on the positive half of the audio cycle.  And without negative cycle  loading, the modulator in all plate modulated transmitters operates without a load every time the transmitter is inadvertently overmodulated in the negative direction - something that happens in practically all transmitters, even those in well maintained broadcast stations.

[Opcom]

speaking of possibly losing RF PA high voltage, if the modulator has a separate supply and it is higher than the PA supply, a diode could be put in so that if the PA high voltage went, the modulator supply could supply some current. Might trip the modulator overcurent, which would not be a bad thing either.

[w4fms]

This has been an interesting topic.   I'm in the process of restoring a GK 500C which obviously uses seperate supplies for RF deck and modulator.  The original HV supply has an oil filled 4uf cap for the 1800v on 4-400 while the modulator uses only 6uF for the 1000v on the 811s.  The 4uF cap is a little rusty so I was going to replace it anyway.

Carl, WA2UJX had mentioned to me that he used electrolytics mounted to plexiglass to beef up his supplies which I thought is a great idea.  Using this method, it would be easy to get at least 30-50uF of capacitance in the same space.

Considering the RF deck running in Class C has more of a static load on the HV supply, perhaps its ok to leave will enough alone and beef up only the modulator supply?

Is there a hard and fast rule as so how much capacitance should be in both of these supplies?  Any recommendations?

Frank, W4FMS

 

[KE6DF]

While the class C final does have a constant load from producing the carrier even under no modulation, the current required for the final still peaks up quite a big under 100+% modulation. I assume the current drawn on voice peaks would be at least double the unmodulated current -- probably more.

So, to me 4 uf sounds awful light. I would think it might result in hum problems and dynamic voltage regulation problems.

In the good old days, 4 uf was a good sized cap, but these days it just doesn't cut it IMHO.

I would think that if you are going to the trouble of changing caps for the modulator, you would be best off adding a much larger cap for the class C stage too.

Dave

[k4kyv]

Gates used only 8 mfd (single L-C section) in their later BC-1 series transmitters.  But I prefer to use about 25 mfd.  That has enough storage capacity to maintain good dynamic (instantaneous) regulation without the necessity to resort to step-start (using choke input).

They can get away with the 8 mfd because of the constant load on the power supply from the class-C final, and adequately sized power supply components.  The modulation reactor contributes enough additional ripple filtering to the power supply to reduce the hum to an acceptable level.

When I first tried my Gates BC1-T on CW, with the mod reactor shorted, the hum level was way too high.  And the plate voltage jumped all over the place, between 4000 volts and 2600 volts.  I modified the circuit to switch the mod reactor in series with the power supply choke in cw mode, and the voltage stayed within 100 volts of 2600, key up or key down, but checking with a scope, I found that the instantaneous voltage was jumping between about 1500 and 4000.  By increasing the filter capacitance to 25mfd and reducing the bleeder resistance to 50K, the instantaneous voltage variation dropped to a reasonable value, and the keyed waveform was no longer hideous looking.

[Opcom]

% ripple is also important. The more C up to a point the better. Don't want anyone teling you you got a hum.

There was a school of thought that said the PA should ave 2% or less, but you could get away with 5% in the modulator if it is push-pull. But it may not be so because the 5% ripple will modulate the modulator very slightly.. I'm running 32uF in the mod and in the PA HV supplies which is overkill but there's no audible hum as the ripple is 0.5%.

If you have the L figure and know the load resistance presented by the transmitter's stages, you can figure the ripple for any C.