Constant magnetisation modulation

[PA0NVD]

I did build a small AM transmitter using a 6AK5 oscillator with a ceramic resonator VXO, an EL84 or 6CH6 modulator and an EL84 or 6CH6 final, and a 12AX7 mic amp. Just a small experiment to keep me out of the bar using Heising modulation. Worked very nice, no visible distorsion. But, it needs a lot of Iron in order not to saturate and compress the positive peaks.
So I thought, why not a balanced transformer instead of the choke with the center at positive, one side to the modulator and the other side to the PA. Really based at Constant current (Heising), but transformed the constant current in constant magnetisation of the core (just as heising) but the DC magnetism is almost cancelled. Amazing results, about three times less Iron required (I used an old EL95 Balanced amp output transformer). If the Iron is too small, it results in odd harmonics, not the bad sounding even harmonics. Looks like a soft limiter effect. I even used a few small power transformers with two 115 Volt windings. Worked very nice, but you have to take a little more care that the winding can support the higher voltages and you loose some low. I used 3 different transformers, also one too small and they all worked perfectly from about 100 Hz upto at leats 8-10 kHz Only the smallest one started to saturate with low frequencies and high modulation and some compression of the modulation was visible. But it was just a 6VA transformer!!.
So look into your stock for transformers with a balanced winding, fun experiment.
Kind regards
Nico

[WD5JKO]

  Niko,

    Your results are similar to mine on my Gonset G50. To boost the efficiency you might try some sliding bias on the Class A tube such that it idles lower, and then slides to the correct operating point at maximum modulation.

    My G50 circuit had some screen grid compensation to the 6146 RF final tube. I also had a form of negative peak limiting. With a hard limit to the screen voltage at around +10v on the way down, this forms a soft limit as seen on the plate. It worked quite well. To get the positive peaks on voice modulation to hit 100% I used the old Heising trick to drop the modulated B+, and to bypass the dropping resistor with a cap. Could also phase the transformer secondary with the modulated B+ to get a little auto transformer boost.

Also look at these possibilities:
http://www.radiomuseum.org/forum/audio_transformer_substitutes.html

Jim
Wd5JKO

[PA0NVD]

Hello Jim
Thanks a lot for the reply and the information, may keep me out of the bar again...
I like to try as well series modulation with a power FET in the cathode circuit of the RF PA. I saw a lot of circuits using two tubes, but a FET will be more easy and can handle very high currents. Than you exchange the total efficiency of a class A modulator with the PA efficiency for the 25% max carrier. So overall the difference may be not too great and no Iron at all.
Do you have information if this has been tried?

[WD5JKO]

Here is a FET series modulator for a vacuum tube final:

http://amfone.net/Amforum/index.php?topic=27856.0

Jim
Wd5JKO

[KD6VXI]

I have built that hv series modulator.   Works great.

--Shane
KD6VXI

[PA0NVD]

Very interesting. I like to try it in the cathode circuit in order to avoid HV ( I am a little sensitive for HV)
Attached photo's of the small experiment. AM carrier 6,7 Watt, CW full break-in, 9,5 Watt

[WD5JKO]

I like to try it in the cathode circuit in order to avoid HV ( I am a little sensitive for HV)

If it is series modulation, it is essentially the same as to whether the modulator is on top or on the bottom. To limit the magnitude of the high voltage with respect to ground, you could have two supplies referenced to ground. For example, one could be +400v and the other -400v. Transformer coupling might be necessary on the bottom active element.

That is a dandy little rig you have built. Nice!

Here is another series modulated rig with the modulator on the bottom:
http://www.amwindow.org/tech/htm/wb9eckseriesmod.htm


Jim
Wd5JKO

[KD6VXI]

Nevermind.   Upon looking at the hi res schematic,  they are all the same.

--Shane
KD6VXI

[KA2DZT]

Those resistors are not critical as to their value.  Usually 100 ohms is used in this type circuit.  I can't see why the values are different for the two tubes.  Looks like an error.

Fred

[Opcom]

Coming late to this discussion I want to throw some transformer stuff out there, regarding balanced signals only (no unbalanced DC). Transformer LF response is interesting.

Menno van der Veen  wrote that the power handling of a transformer has a relationship to the frequency such that when the frequency is divided by 1.414 (sqrt 2)  the power handling is 0.5x  His paper is attached.

[KD6VXI]

Those resistors are not critical as to their value.  Usually 100 ohms is used in this type circuit.  I can't see why the values are different for the two tubes.  Looks like an error.

Fred

The hi res download has all of them at 100 ohms.

I didn't think it was critical,  but thanks for chiming in.   I do believe the local res schematic has an error.

--Shane
KD6VXI

[PA0NVD]

Hello all
I did try a combination of series modulation in the cathode and grid modulation in order to avoid high DC at the series modulator. The grid part is done by connecting the grid resistor to ground, not to the cathode circuit. The trick to get low distrorsion is partly in the grid drive impedance. I drive at approx 5 kOhm and 35 Vtt for an EL 84, Not too far from normal except the 5 kOhm. A higher impedance / higher drive will result in some compression of the positive modulation. I did not investigate yet why, but I assume the AF at the grid will start to modulate as well due to the grid current.
The results are very promising, and I am very positive in my assumption that it may be done as well at much higher power levels. I have 2 Watts carrier and 8 Watts PEP at 350 VDC at the anode.
The voltages at the modulator are very low, the drain has a voltage of approx 25 Vtt, That will not hurt me too much... .

[PA0NVD]

I did a few more tests, and indeed the distorsion is due to the modulation of the RF drive signal. If the cathode voltage rises, the grid current becomes less and the loading of the driver becomes less. That results in an INcrease of the RF at the grid while the output should DEcrease. When I lower the dive impedance with a low G1 resistor to ground, this effect is much less and the modulation distorsion disappears. Than it seems that the curve of the FET is quite nicely compensated with the modulation characteristic of  G1 plus a part series modulation plus a small part G2 modulation. (G2 is also referred to ground not to the cathode circuit).
But you need ti fiddel around a bit with the source resistor, the G1 resistor and the bias to get a very nice modulation. Seems worth the trouble, with a FET of approx 800 Volts and a few amps you can modulate kilowatts with a low FET dissipation.
WHO IS GONNA TRY THIS?
It results is a safe circuit, because, when the FET is pinched off or open, the cathode voltage is very limited. Both the drive and the tube are cutt-off. In my little circuit the cathode voltage will not rise more than approx 40 Volts with the cathode open.

[WD5JKO]

Nico,

  I wonder if an RF tube for the RF PA would be a better choice over the EL84? Perhaps a 5763, or two in parallel would do better. Some of the high Gm audio tubes don't like G1 current to flow. The EH 7591's are an example, but the 50 year old 7591's seem to be fine with a little grid current.

   With a big percentage of grid modulation, you have moved from series modulation at high RF efficiency into efficiency modulation. The output is quite a bit lower, and tuning sensitivity are bigger factors now. Still, your approach has an distinct high performance simplicity to it. Thanks for posting this.

Jim
Wd5JKO

[Opcom]

http://www.montagar.com/~patj/amxmtr_dc_classamod.html

[PA0NVD]

hallo Jim
Thanks for the reply. I guess you are right.
I agree, there are better valves than an EL84, but see this as just a small experiment and I had them laying around. I tried also 6HC6 tubes with approx the same result. a little less. But that is because the 6CH6 tubes I have here are worn out. The chassis is made of PCB material, so nothing real permanent....
And after all, the EL84 performs quite fine. When used in CW, the output is approx 12 Watts with an efficiency just under 70% When used in AM, I lower the HV a little and there is hardly any difference between the plate modulation (constant magnetisation) and the cathode modulation. Same output, 9 Watts . The main reason is that, when the FET pulls the cathode down (max power), the tube is in class C with quite some grid current resulting in a good efficiency. Perhaps an EL84 doesn't like that too much, but it was his choice to become one....And stupid to end up in the hands of this radio ham....
Off course the plate modulation goes to approx 30 W PEP and the cathode modulation is 9 Watts PEP, but that is true for all other modulation systems other than plate modulation and real series modulation.
Your point tuning sensitivity is absolutely true, but hardly a problem. The only sensitive point seems the drive level. A wrong drive level results in some distorsion.

[W4DNR]

I did a few more tests, and indeed the distorsion is due to the modulation of the RF drive signal. If the cathode voltage rises, the grid current becomes less and the loading of the driver becomes less.     

If the drive were link coupled between the grid and the cathode, would the drive be more constant ? 

Don W4DNR

[PA0NVD]

Hi Don
I think that link coupling doesn't make a difference. At the receiving side of the link you have to regenerate the voltage to drive the tube with a transformer or resonant circuit and you have your high impedance back. So there again a couplig C to the grid and a quite low resistance to ground as grid leak.