screen modulation again...

[N2DTS]

Tested ok on 40 and 80 meters, likes to run at 200 watts carrier, will do more but that generates a lot of heat.
1500 volts, 400 ma cathode current, 300 watts carrier 1500 watts pep.
1500 volts, 300 ma cathode current, 200 watts out 900 watts pep.

Sounds good to me and looks good on the scope.
40 meters seemed dead today.

[AB2EZ]

Brett

The efficiency, at carrier, seems to be amazingly high... given the headroom required for (more than) 100% positive modulation.

1500Vdc and 300ma average cathode current implies less than 450 watts of electrical plate input power at carrier.

200W of RF output power, at carrier, implies that the efficiency (including the losses in the output tank circuit), at carrier, is greater than (200W/450W) x 100% = 44%

That's amazingly high for a real screen modulated stage... capable of more than 100% positive peak modulation... although it is theoretically possible with the tubes biased deep into class C operation.

Stu

[N2DTS]

Sru,
Not sure things are so efficient, 450 watts in for 200 out...
My meters have home made shunts so may not be exact.

I just looked at my notes, I have at one point 2000 volts and 500 ma for 1000 watts in an 350 watts carrier out, 1400 watts pep. Tube temps got up to 170F.

1500 volts and 300 ma for 200 watts out seems a lot better, lots of power wasted to get another 150 watts.
It seems to run more efficient at lower power levels.

The rig is flat from 20 Hz to 22 khz, and sounds very clean.
 

[AB2EZ]

Brett

In a screen modulated class C RF amplifier:

If there is sufficient headroom* to modulate from carrier level up to 100% positive modulation, then the efficiency at 100% modulation will be twice the efficiency at carrier.

*Since the plate B+ voltage is not modulated, the loading must be adjusted so that, at carrier, the amplitude of the fundamental RF frequency component of the plate voltage swing is no more than half of the fixed plate B+. Otherwise, one will not be able to produce 100% positive modulation peaks. This implies that the plate power input, at carrier is, at least 2x the RF power output at carrier... and more like 2.5 to 3x in a real amplifier.

An efficiency of 45% at carrier implies an efficiency of 90% at 100% modulation peaks... and proportionately higher efficiency at modulation peaks that are higher than 100%

I would expect the efficiency at carrier to be lower than 45% in a real class C screen modulated transmitter... particularly if you can obtain somewhat higher than 100% positive modulation peaks.

Something between 35% and 40% would be less surprising.

Stu

[N2DTS]

http://n2dts.smugmug.com/Ham-radio/i-JZGmdCv/X2

[N2DTS]

I am working in integrating the new RF deck and modulator into the station, I put the modulator above the 813 rf deck with the 4x4x150 rf deck above that.
I need to run line level audio to it, make up some interconnect cables in such a way as I can run one or the other RF decks, plate or screen modulation.
Once I get it done, I can measure the power input to power output ratios and write things down.
I did test the distortion, but did not write things down, but I did notice it could be adjusted to give very low distortion levels per the sdr-iq, at 150 watts carrier and 90% modulation, the harmonics were way down.

I also noticed some difference between 80 and 40 meters, 80 seems to work better, with a positive carrier shift under heavy modulation at higher power levels, and on 40 it tends to shift the carrier downward past about 250 watts carrier and lots of audio.
At 100% modulation I get no carrier shift on any band.
It also tunes somewhat different on 80, loading up to max current works well, on 40 loading should go past peak and on its way down (overloaded) for the best peak power.
Tank circuit Q might have something to do with it, I have not tried different coils yet.

No matter how I set things (within reason) it always seems to sound very good.

[AB2EZ]

Brett

The definition of "loading" has always struck me as being more likely to confuse than to be helpful... but we are stuck with it.

Even though I understand the details (i.e. the equations), I had to edit the paragraphs below several times... to get them right.

Increasing the "loading" (as loading is defined) implies a decrease in the RF load impedance looking into the tank circuit... i.e. from plate to ground. Increasing the "loading" (as loading is defined) also corresponds to a decrease in the capacitance of the "loading" capacitor in a pi network*.

In your experiments with your new rig... the reason you need to increase the "loading" (i.e. less "loading" capacitance, and less RF load impedance looking into the pi network, at resonance)... when you are tuning up the rig at carrier level... is because you need to ensure that the amplitude of the RF signal from plate-to-ground (at resonance) is less than half of the value of the plate B+ supply voltage.

If you reduce the "loading" from the above setting... producing more RF plate voltage swing at resonance (dip), and, therefore, more RF output at carrier... you won't have enough headroom for 100% positive modulation peaks. I.e. as the screen voltage is increased to try to obtain 100% positive modulation, both the average and RF components of the plate current will increase to twice their carrier-level values. As a result, with a fixed RF load impedance from plate-to-cathode, the plate-to-cathode voltage will swing down to zero for a portion of each RF cycle. This can result in "downward" shift of the average RF output power when you apply sine wave modulation, as well as distortion on positive modulation peaks.

This is the same reason why the "loading" of a linear amplifier should not be adjusted for maximum output at carrier level.

*Note: In a pi network, increasing the value of the "loading" capacitor... which is across the output of the pi network (i.e. less loading capacitive reactance)... results in a higher value of pi network input impedance (at resonance). This is because the impedance, looking into the input of a pi network, at resonance, is (to a good approximation): Z_in = Q² x R_load = [(2pi x f x L)/R_load]² x R_load, which is proportional to L² / R_load ... where R_load is the real part of the total impedance across the output of the pi network, and L is the inductance of the pi network inductor. More "loading" capacitance results in a lower value of R_load. This adds (in my opinion) to the confusion associated with the definition of loading.

Stu

[N2DTS]

Less is more! (loading)

The old 2x4x150 deck on 40 meters acted normal, and so does the new one, but the new one on 80 seems to work fine when peaked to maximum output power with the loading control for some reason.
That surprised me, but if I peak it, it modulates over 100%, has positive carrier shift, and sounds good!
That is at the 200 watt level.
Plate tuning cap 80% meshed, so I am likely a bit short in inductance, 4 tubes are low impedance I guess.

Many tests to follow, I need to try some different plug in coils and see how that changes things.
Grid drive does not seem to matter much, tuning is not hard, I just tune to max power, then adjust the loading control past peak power a bit and away we go.
The variac input voltage to the modulator is a bit of a dial a power output control within a good range if the loading is 'heavy'.
In all cases, it seems to sound good, it always seems to sound cleaner then the plate modulated rigs.

I suspect the 4x150/4cx250 tubes screen modulate very well, the screen seems to have a lot of control.
In the AB1 modulator, you can set them for 90 ma resting per tube with 350 volts on the screens, and run the plate voltage from 1000 to 2000 volts and the resting current does not change, and there is no need to adjust the bias. Not sure you could do that with any other tubes.

[AB2EZ]

Brett

With the new rig on 80 meters... and the loading control adjusted for maximum RF power output, at carrier level.... what is the percentage of mesh of the loading capacitor (roughly)?

Stu

[N2DTS]

80 or 85%.
Not sure what value the cap is, two sections, both in use, maybe its a bc610 tuning cap.
150pf? per section, total of 300, so figure about 260 or 270 pf?

1500 to 1700 volts on the plates, 300 ma, 200 watts out, about 650 or 700 volts into the modulator, I did not note the readings of the screen volts and current (output) 200 or 300 volts, 40 ma?

I started out at 1500 volts, tried 2000, and settled on about 1750 as the sweet spot I think.

[AB2EZ]

Brett

In class C operation... where the plate current is a sequence of pulses... each of which is significantly narrower than half the duration of an RF cycle:

The amplitude of the component of that plate current, at the fundamental RF frequency, is (approximately) 2x the average plate current.

If the average plate current (at carrier) is 300mA, then the amplitude of the component of the plate current, at the fundamental RF frequency, is (approximately) 600mA.

If one wants the plate voltage swing, at carrier, to be half of the 1500V B+ plate supply voltage... then the required RF load impedance (looking into the pi network) at the fundamental frequency is: 0.5 x 1500V / 0.6A = 1250 ohms.

If the Q of the pi network is 10, then the required load across the output of the pi network... consisting of the parallel combination of the loading capacitor and the antenna system is:

1250 ohms/(10² +1) -jX ohms = 12.4 ohms - jX ohms... where X is not critical.

If the Q of the pi network is 5, then the required load across the output of the pi network... consisting of the parallel combination of the loading capacitor and the antenna system is:

1250 ohms/(5² +1) -jY ohms = 48 ohms - jY ohms... where Y is not critical.

But

300pF (maximum) at 3.885MHz => -j136 ohms

With 50 ohms looking into the antenna system, even at the maximum value of 300pF, this loading capacitor would have very little effect on 80 meters.

I.e. -j136 ohms in parallel with 50 ohms = 44 ohms (resistive) in series with -j16 ohms (capacitive).

The effect of the series capacitive reactance (-j16 ohms) is to reduce the effective series inductance of the tank circuit.

Therefore:

Regardless of the setting of the loading capacitor (somewhere between 0pF and 300pF), the impact of that setting on the actual "loading" is going to be small at 3.885MHz.

If the Q of the pi-network is 10 with a 12.4 ohm load at the output, then even setting the loading capacitor fully meshed (300pF) will result in an impedance looking into the pi network that produces an RF plate voltage swing, at carrier, which is only about 12.4/44 x 50% = 14% of the plate B+ supply voltage (neglecting the effect of the -j16 ohms of series reactance).

If the Q of the pi-network is 5 with a 48 ohm load at the output, even setting the loading capacitor fully meshed (300pF) will result in an impedance looking into the pi network that produces an RF plate voltage swing, at carrier, which is about 48/44 x 50% = 55% of the plate B+ supply voltage (neglecting the effect of -j16 ohms of series reactance).

I was surprised that the setting of the loading capacitor that produces maximum RF output power, at carrier, is not fully meshed.... when operating at 3.885MHz with a loading capacitor whose maximum value is 300pF. But, talking into account the effect of the series capacitive reactance at the output of the pi network... that arises from the parallel loading capacitance... the optimal loading capacitance could be less than fully meshed. I.e, with this value of pi network physical inductor, and at this operating frequency...  there may be a value of loading capacitance beyond which the total pi network series reactance (the physical inductor's reactance minus the series capacitive reactance of the output load) may be decreasing faster than the effective pi network output load resistance is decreasing. Recall that the input impedance, at resonance, of the pi netwrork is proportional to L x L / R, where L is the effective series inductance of the pi network (taking into account the series capacitive reactance of the output load), and R is the series resistance of the output load.


Stu
 

[N2DTS]

Stu, the plate tune cap is 300 pf, the loading cap is a bit small on 80 meters I think at about 1700 pf.


I did some checks after I got everything in place:

700 volts into modulator:

  plate voltage   plate current   watts out   screen voltage  screen current
     1500                400 ma           200              170                  8 ma

     1750                 420 ma         210              170                  7 ma

     2000                 420 ma         250              170                  5 ma

800 volts into modulator:

      1750                 540 ma         320             200                   8 ma

      2000                 600 ma         450             210                   9 ma


That last run had things getting quite hot, 1200 watts in, 450 out, 750 watts of heat.

I think I will run the rig at 1750 volts, and 210 watts out.
If I tune the final for maximum power at carrier level, I get crazy modulation, 210 watts of carrier, positive carrier shift on the mod monitor and 1000 watts pep on the watt meter!

[N2DTS]

In place:

[AB2EZ]

Brett

Your latest measured values (below) are consistent with what I would expect from theory.

At 1500V plate B+ supply voltage and 400mA of average plate current, the plate input power is 600W at carrier. The plate efficiency is 200W (RF output power) / 600W (electrical input power) x 100% = 33%. This is what I would expect for a class C amplifier, with the RF plate voltage swing equal to a little less than half of the plate B+ supply voltage.

The amplitude of the fundamental frequency component of the plate current is approximately 2 x 400mA = 800mA

The required RF input impedance of the pi network, at resonance (dip), to produce this swing, is about 750V / 0.8A = 937 ohms

If the Q of the loaded pi network is 10, then the load impedance at the output of the pi network has to be 937 ohms / [102 + 1] -jX ohms = 9.3 ohms -jX ohms, where X is not critical.

If the loading capacitor is fully meshed, i.e. 1700pF, then it has an impedance at 3.885MHz of -j24 ohms.

The parallel combination of the loading capacitor (-j24 ohms) and the antenna system input impedance (50 ohms) results in an impedance of: 9.4 ohms -j19.6 ohms.

This implies that, on 80m, (if the loaded Q is 10) the loading capacitor should be fully meshed to produce the correct RF load between plate and ground (i.e. the amplitude of the voltage swing, at carrier, is a little less than half of the B+)

Using the free application "PI-EL", I played around with various values of Q (as Q is defined in the PI-EL application).

At 3.885MHz, with an impedance (looking into the pi-network) of 937 ohms, and with the pi-network looking into an antenna system whose impedance is 50 ohms, and with a pi-network Q of 10... PI-EL computed the required tuning capacitance (which includes the output capacitance of the four tubes) as 365pF, and a required loading capacitance as 1362pF. The corresponding physical pi-network inductor is 5.44uH

With the Q increased to 12, PI-EL computed the required tuning capacitance as 434pF, and the required loading capacitance as 1700pF. The corresponding physical pi-network inductor is 4.63uH.

PI-EL appears to consistently calculate Q to be 20% higher (e.g. 12 v. 10) than I calculate Q to be (in the same circuit). This appears to be a result of PI-EL using a different definition of Q than I am using.

This is why you are adjusting the loading capacitor for maximum RF output, at carrier, on 80m. I.e. maximum RF output on 80m corresponds to a nearly fully meshed tuning capacitor, and a fully meshed loading capacitor... which (using the definition of "loading") means that, on 80m, the amplifier is still properly "over loaded", even with the loading capacitor fully meshed.


Stu

Stu, the plate tune cap is 300 pf, the loading cap is a bit small on 80 meters I think at about 1700 pf.


I did some checks after I got everything in place:

700 volts into modulator:

  plate voltage   plate current   watts out   screen voltage  screen current
     1500                400 ma           200              170                  8 ma

     1750                 420 ma         210              170                  7 ma

     2000                 420 ma         250              170                  5 ma

800 volts into modulator:

      1750                 540 ma         320             200                   8 ma

      2000                 600 ma         450             210                   9 ma


That last run had things getting quite hot, 1200 watts in, 450 out, 750 watts of heat.

I think I will run the rig at 1750 volts, and 210 watts out.
If I tune the final for maximum power at carrier level, I get crazy modulation, 210 watts of carrier, positive carrier shift on the mod monitor and 1000 watts pep on the watt meter!

[N2DTS]

The loading cap is not quite all the way open, I have about 10% left, but depending on how I run the deck, I think it could be all the way open...

It sure does run very well as it is on 80, and maybe it would work as well on 40 with less coil.
I have a bunch of coils to try, and its easy to bend up more.

Here is how things are set for 80 meters:

http://n2dts.smugmug.com/Ham-radio/i-JZGmdCv/A

[N2DTS]

Last night I tried different coils and jumped between 80 and 40 meters to make up my tuning chart mounted to the front panel.
Very different results with various coils, and I think I need more coil or more capacitance on the plate tuning, its sort of at the edge.
I guess four tubes is a lot of current and not a lot of voltage so the tank circuit is a bit different from my usual stuff.

Without a tuning chart, I could see the need for a modulator bypass (tune) switch, to put full screen voltage on the tubes for tune up at maximum power, then switch in the modulator.

Another odd thing I noticed is that I can adjust things to give very different results.
Depending on how you tune things up, I can get lots of negative carrier shift, or lots of positive carrier shift, or no carrier shift at all.
Changing the phase of the audio can give lots of peak power with negative carrier shift, or less peak power with more positive carrier shift. Both sound good.
With plate modulation, carrier shift is mostly power supply drag down from a common supply, or a bias/drive problem. Not so with screen modulation.
I can adjust things so I get crazy amounts of peak power output, and normal operation has the peak power a bit over four times the carrier power, 250 watts carrier, 1100 watts PEP.
That is with no carrier shift, 1750 volts, 400 ma.

I think I will make a second modulator with improvements.
Instead of the input transformer, I will just use a 600 volt 1uf cap, that voltage rating should protect the audio equipment in the event of some failure.
I will also use a 1500 volt oil filled cap as the output cap, up from the 1000 volt mylar cap I have in there now, include a bypass/tune switch, and try and come up with a better and more robust system to replace the trim pots.
Like a 500K pot on a vernier drive with two 250K resistors on each side of it.

[N2DTS]

20 Hz, as shown through the receiver IF output:

http://n2dts.smugmug.com/Ham-radio/i-V4FTg9h/A

And 100% modulation pre fine tuning

[N2DTS]

Doing tests, at 250 watts carrier, 400 Hz, I could not get the distortion products (2nd audio harmonic) below 30 db down at 95% modulation.
The waveform looks very good, the rig sounds very good to me, so why the 2nd harmonic?

I think it can go quite low at lower power outputs, around 150 watts, but at 250 watts I tried adjustments to everything without much improvement.
Plenty of positive modulation and peak power.

Adjustments made:
Grid drive and bias levels, voltage input to the modulator, the modulator adjustment pots, plate tuning and loading, plate voltage.

Rig into dummy load, sdr-iq running sdr-radio, mod monitor and O scope on receiver working, audio generator into the modulator directly (has input transformer 600 to 600 ohms).

[w4bfs]

Doing tests, at 250 watts carrier, 400 Hz, I could not get the distortion products (2nd audio harmonic) below 30 db down at 95% modulation.
The waveform looks very good, the rig sounds very good to me, so why the 2nd harmonic?

It turns out that human hearing cannot hear even order harmonics until reaching a large amount of them .... much more sensitive to odd order harmonics ... I don't know why

This has been a very interesting thread and has me considering the implecations .....

[N2DTS]

Past tests have shown that my plate modulated rigs were quite low in distortion, but they do not sound as clean as the screen modulator.

I really like the idea that I can run 20 Hz at 100% modulation into the transmitter and not worry about mod iron blowing, no odd stuff that happens in transformers.
In the future, I will document the distortion levels as shown on the sdr of various rf deck tubes, 4D32 and 813's.

In normal use, I can put plenty of audio on the carrier and the signal does not look wide at all on the sdr, it looks kind of narrow.
With the sine wave input, the carrier shifts way up and the peak power seems to go much higher then 4 times the carrier power.
 

Doing tests, at 250 watts carrier, 400 Hz, I could not get the distortion products (2nd audio harmonic) below 30 db down at 95% modulation.
The waveform looks very good, the rig sounds very good to me, so why the 2nd harmonic?

It turns out that human hearing cannot hear even order harmonics until reaching a large amount of them .... much more sensitive to odd order harmonics ... I don't know why

This has been a very interesting thread and has me considering the implecations .....

[N2DTS]

Today, I hooked up the 4x100TH modulator deck to the 4x4x150 rf deck and did some tests.
At 700 watts carrier output, the RF deck runs cooler then at 250 watts screen modulated.

1800 volts at 500 ma input (900 watts) gives 650 watts out.
At 700 watts carrier the rf deck is running much cooler then with screen modulation.
I can fully modulate the carrier with the 100th mod deck at 2200 volts easy.
As a test, I took it up to 1000 watts carrier output, 1800 volts and 800 ma, that is about half the plate dissipation of the tubes.
My power supply is limited by the chokes, 500ma ccs ratings.

A pair of 4x150's would make a nice 500 watt (carrier) rig at low voltages, 1500 to 1800 volts.
 

[WD5JKO]

  Brett,

   I wonder if you have one of those "Kill A Watt" power line meters that can measure actual power feeding a transmitter (volts X amps X power factor)?

http://www.homedepot.com/p/Unbranded-Kill-A-Watt-Electricity-Monitor-P4400/202196386

I have a 120 volt version here that I find quite handy. There are 230 volt versions, but I don't know how available or expensive they are.

Screen grid modulation is inherently inefficient, much the same way AM linear amplification is. In both cases the efficiency increases with modulation.

This thread has interested a lot of readers. Hams using the 4CX250B and the variants on AM are not many. You have been telling us for years how two of them as class Ab1 audio modulators work. Just vary the plate voltage....the idle current is constant. Neat!

So here is something to contemplate. Compare a 400 watt output plate modulated transmitter against a 400 watt output screen modulated transmitter. I am referring to carrier power with headroom for at least 100% upward modulation. For sure the screen modulated RF transmitter will run hotter at about 1/2 the efficiency. But then back-out the power draw from the modulator. I contend that from power cord AC line watts to coax RF output at 50 ohm watts, the two setups compare about the same in terms of overall efficiency. It takes 125 watts just to light up four 100TH's. So two power supplies, bleeder power, transformer losses, etc all add up. I'd be surprised if either setup resulted in much more then 20% efficiency overall.

Jim
Wd5JKO

[N2DTS]

I think, overall its a wash, but each system has its advantages.
Screen modulation, no mod iron, no mod iron limitations, no separate power supply and modulator, all that saves a lot of space, weight an parts. The screen modulated RF deck does not have to deal with high peak voltages so the parts only need to be good for the plate voltage.

The downside is you need a big tube to run any sort of power.
The four 4x150's loaf along at 250 to 300 watts of carrier in screen modulation.
They also loaf along (are hardly working at all) at 700 watts carrier in plate modulation.
Say 700 watts out, 900 to 1000 watts in, 300 or 400 dissipation in the tubes out of a total of 1000 watts.

Do you need to run that kind of power? I do not think you buy much improvement over 300 watts really.

I really like the 4x150 / 4cx250b type tubes. A pair modulated by a pair makes a nice legal limit TX that is running at about 1/2 power with plenty of audio power in AB1.
Run them easy and a fan on low speed cools them enough without any racket.

A pair can run an easy 200 watts screen modulated, plus, if you build it right, you can use both setups.

I bet for a 200 watt carrier, both setups draw the same amount of power from the AC lines, not that it matters really.
If you add the power and filaments of the modulator, it offsets the screen only setup.

I do not have 220 in the shack, but ran two 120 volt 20 amp feeds on different phazes from the breaker box, the rf can go on one, the modulator on the other.

If I limit things to 15 amps, that is 120 volts at 30 amps.
3600 watts total, I never trip breakers unless I do something really stupid.

On the plus side, the 4x150 type tubes are easy on filament power, 6 volts (not 6.3) at 2.6 amps each, I run them off 5 volt transformers as my line voltage is high.
15.6 watts per tube, 62.4 watts for 1000 watts of plate dissipation.

Another thing I like about the 4x150 tubes is they work at low voltages, from 1000 to 2000 volts is good, you coul likely run them at 750 or 800 volts ok. They take very little drive, four takes less drive then a pair of 813's or three 4D32's.

Amazing little tubes.



 

[pa0ast]

May be this is an interesting Schematic,   screengrid modulation and a Dorethy pa.
Anton

[pa0ast]

Sorry forgot the picture