Ameco AC-1 ECO: Plate modulated

[AB2EZ]

As an offshoot of another recent thread:

Attached are:

 a) A schematic showing how I plate modulated an Ameco AC-1 (electron coupled oscillator) that I owned a few years ago
 b) An MPEG of the associated demodulated audio from my off-air monitor

The AC-1 put out about 5 watts of carrier. Even with plate modulation, there was no noticeable FMing of the transmitter's output.

Stu

[N2DTS]

Sounds very good.

[Rob K2CU]

How far could you downward modulate it before the oscillator cut out?

[AB2EZ]

Rob

Good question!

I don't recall, and I didn't keep any notes on this in my archives.

Speculatively: it must of been more than 75% because I'm sure that I was monitoring the RF envelope on my digital scope, with sine wave modulation... and I wouldn't have considered it "working" if the positive and negative peaks (the modulation linearity was very good) were less than 75% with a sine wave as the audio input.

Stu

[k7mdo]

This method is similar to what I have thought about in the past... I have two of the 20 watt Heathkit monobloc audio amps and one I have restored to working and the other is beyond repair but has the output transformer which I could put back to back with the working monobloc?

The transformer, in the original circuit, is hooked up PP with 6L6's but I don't know the impedance... thinking it must be around 5-10k with 4/8/16 ohm outputs.

I have renewed hopes that this might work!

Tom

[AB2EZ]

I miss my AC-1, so I just ordered an AC-1 "clone" kit from Mike Vogel (see his E-mail to me, attached below)

I plan to AM modulate it, just as I did with my previous AC-1.

I will explore the issue that Rob raised... and experiment with some simple "keep alive" circuits... particularly to keep the screen voltage from dropping too close to zero volts. The screen of the tube, in conjunction with the grid and the cathode, acts as the "effective plate" of the oscillator portion of the ECO.

"Dear Stu,
 
I have a couple of kit's available. Price for the full complete kit with all new part's, new tubes, ready installed varicaps, 1 xtal of your choice (3560 or 7030 or 7040 kHz) and one clear coil-form with wire is 189.00 Euro. S+H is 36.00 Euro.
The price is somewhat high due to the high quality professionally made chassis and transformer. I also have to pay high tax here in Germany for selling kit's.
After receipt of money in Euro I will send out the kit to your address.
Payment: Please PayPal in Euro to mike-vogel@gmx.de
 
vy73
 
Mike
DL3ECN   www.tuberig.com "

[Steve - K4HX]

I recall when WA1HLR was messing around with modulating a self-excited triode oscillator, he found that negative peak limiting was required if you wanted to modulate heavily and maintain any semblance of frequency stability. I think he used the 3-diode circuit.

[AB2EZ]

Steve

Good input.

I plan to play around with various combinations of a negative peak limiter on the modulated plate voltage, and/or a negative peak limiter on the modulated screen voltage... to see which works better with respect to: modulation fidelity, ability to accommodate negative peaks, and oscillator stability.

Stu

[N0WEK]

This may be a bit of a thread hijack but I'm planning to plate modulate my Johnson Adventurer transmitter with a NOS Triad TY-66A transistor modulation transformer.

It's rated at 40 watts and is designed to take 200 ma DC through the secondary. It's got 3K, 4K and 6k outputs (the Adventurer needs the 4K) and a 6 ohm CT primary which was designed for Push-Pull transistors but will work fine when fed with one of the tube amps I've got.

Of course the Adventurer is a MOPA design so I won't have to worry about modulating the Osc.. The transmitter even has an octal socket in the back to hook the transformer to. Somebody rewired it to bring out the screen separately for a two tube screen modulator too, using a 12ax7 and 12au7.

If you come across one of these transformers, mine came of Epay cheap, they should work great for all those "novice" transmitters out there.

[AB2EZ]

The is some very well written information about ECO's here:

http://faculty.frostburg.edu/phys/latta/ee/6cl6xmtr/6cl6schematic.html

Note that for CW operation, the screen is connected to a regulated voltage supply.

Also note, of course, that one could use a Zener diode instead of a vacuum tube voltage regulator; and one could use a solid state full wave rectifier (e.g. a bridge rectifier) and a simple capacitor-input filter (no choke) for the B+ supply.

Stu

[AB2EZ]

I just put my new Ameco AC clone on the air on 40 meters (actually into a 50 ohm dummy load). I haven't done any in-depth experiments... but so far it modulated beautifully with 100% positive and negative peaks using a sine wave audio input. Also, with voice input, the modulated envelope of the RF output tracks perfectly with the waveform of the audio input signal to the modulator (using a 2-channel digital scope, triggered off of the audio input waveform).

The output power at carrier is about 5W. I think I can get a little more by playing with the output pi network to reduce the Q. Right now, the maximum output occurs with the loading capacitor set to minimum. This implies that I have a little too much inductance in the coil of the pi network.

I don't see any evidence of oscillator instability on negative modulation peaks. The sidebands are symmetrical. I need to attach my scope to the screen voltage waveform to see how far down the screen voltage is going on negative plate modulation peaks.

The AC-1 clone, as in the case of the original AC-1, uses a 15k ohm resistor between the plate and the screen. The screen bypass capacitor, the plate choke bypass capacitor, and the plate coupling capacitor are all .001uF. I'm modulating it using a backward connected Hammond output transformer in series with the B+ supply. The input to the low impedance side of the transformer is being produced by a modern solid state audio amplifier.

It took a few weeks for the AC-1 clone kit to arrive from Germany. I received it two days ago. Naturally, with its small number of large, traditional components, and point-to-point wiring, it was easy to build.

Stu

[AB2EZ]

As an update:

To accommodate the long key down (at carrier level) associated with AM operation, I changed the 6V6 ECO tube to a 6L6. The 6L6 is rated for higher plate and screen dissipation. It draws more filament current, but the transformer in the AC-1 clone doesn't appear to have any problems (running cool). If needed, I could have unplugged the 6X5 tube rectifier, and plugged in a pair of 1N4007 diodes... but so far, I haven't bothered to do that.

I measured the screen voltage waveform on the 6L6 ECO, under modulation. The voltage was going all the way down to zero (or even negative) on negative modulation peaks... which is obviously not a good thing to do. So... I added a simple single diode-based negative peak limiter to the screen circuitry... to limit the minimum screen voltage to 20 volts (10% of its level at carrier). Based on measurements with my frequency counter, at low audio modulation frequencies, the ECO provides a stable oscillation frequency under full modulation, and the negative modulation peaks still are close to 100%.

As an aside... although it is perhaps no surprise... the cathode current of the ECO is the sum of a DC and a sine wave. Therefore, the ECO is running in class A, with 300V B+ and about 75ma of DC cathode current (of which about 6.7 mA is DC screen current). Therefore, the modulation resistance is about 300V/0.075A = 4000 ohms. Since the tube is operating in class A, the desired RF load impedance (looking into the pi network at optimal loading) is around 300V/0.068A =  4412 ohms. The impedance of the 7uH pi network coil at 7.3MHz is around j320 ohms. The impedance looking into the pi network is 320 ohms x 320 ohms / R, where R is the resistive (real) part of the impedance of the parallel combination of the 50 ohm antenna impedance and the impedance of the loading capacitor.  Therefore, the resistive part of the parallel combination of the 50 ohm load and the loading capacitor needs to be around 25 ohms. This means that the loading capacitor must have an impedance of around -j50 ohms at optimal loading. At 7.3MHz, this corresponds to a loading capacitance of 436pF. The variable loading capacitor consists of two (2) 365pF sections in parallel... so this works out fine.

All is well, and I have made a few contacts on 7285 and 7290 (the frequencies of the 40m AM crystals I have).

Stu

[Steve - K4HX]

Very cool Stu. I wonder how many recently upgraded Novices tried to modulate their AC-1 back in the day.

[W3GMS]

I just put my new Ameco AC clone on the air on 40 meters (actually into a 50 ohm dummy load). I haven't done any in-depth experiments... but so far it modulated beautifully with 100% positive and negative peaks using a sine wave audio input. Also, with voice input, the modulated envelope of the RF output tracks perfectly with the waveform of the audio input signal to the modulator (using a 2-channel digital scope, triggered off of the audio input waveform).

The output power at carrier is about 5W. I think I can get a little more by playing with the output pi network to reduce the Q. Right now, the maximum output occurs with the loading capacitor set to minimum. This implies that I have a little too much inductance in the coil of the pi network.

I don't see any evidence of oscillator instability on negative modulation peaks. The sidebands are symmetrical. I need to attach my scope to the screen voltage waveform to see how far down the screen voltage is going on negative plate modulation peaks.

The AC-1 clone, as in the case of the original AC-1, uses a 15k ohm resistor between the plate and the screen. The screen bypass capacitor, the plate choke bypass capacitor, and the plate coupling capacitor are all .001uF. I'm modulating it using a backward connected Hammond output transformer in series with the B+ supply. The input to the low impedance side of the transformer is being produced by a modern solid state audio transformer.

It took a few weeks for the AC-1 clone kit to arrive from Germany. I received it two days ago. Naturally, with its small number of large, traditional components, and point-to-point wiring, it was easy to build.

Stu

Hi Stu,
Thanks for your update on the AC-1 clone from Mike in Germany.  Mine just came last week and in another week or so I will put it together.  It certainly is a great quality kit.  
Joe, GMS

[WD5JKO]

I measured the screen voltage waveform on the 6L6 ECO, under modulation. The voltage was going all the way down to zero (or even negative) on negative modulation peaks... which is obviously not a good thing to do. So... I added a simple single diode-based negative peak limiter to the screen circuitry... to limit the minimum screen voltage to 20 volts (10% of its level at carrier).
Stu

  Stu,

   I have "prior art" on that trick.    I did it with a modulated single 6146 that was getting the screen excessively modulated such that the carrier would pinch off early. The thing that surprised me though was that a hard clamp of the screen voltage only made a small slope change in the plate swing as the screen was clamped on the downward direction. It seemed to be an effective, and low distortion way to make a negative peak limiter.

Jim
Wd5JKO

[N2DTS]

It sounded good on the air today Stu.

[AB2EZ]

Brett

Thanks for the comments (-:

Joe

If you have any questions... let me know.

Jim

Yes... after years of serving on corporate patent committees (i.e. determining which "inventions", disclosed to the committee by employees, should be submitted to the USPTO as candidates for being granted a patent), and after years of serving as an "expert consultant" in patent litigation cases... I came to the opinion that there is hardly any idea that is really new!


Stu

[AB2EZ]

I made some additional modifications ... and these seem to make the plate modulated AC-1 clone work really well on 40m.

Previously posted modification: substitute a 6L6(GB) for the 6V6 to accommodate higher plate and screen dissipation.

New modifications:

A. I am using a negative peak limiter on the modulated B+. The screen is fed with the stock 15k ohm dropping resistor from the modulated B+, with the stock 0.001uF RF bypass capacitor. This works as well (or better) than using a negative peak limiter on the screen.

B. (important) I removed the stock 22pF capacitor between the cathode and the grid of the tube. This capacitor is in parallel with the tube's parasitic grid-to-cathode capacitance (around 10pF). The total grid-to-cathode capacitance determines the RF current through the crystal. With the stock 22pF capacitor in place, the RF current (on 40m) is too high... and the crystal overheats. This results in a slow (several second) drift of the crystal frequency... of around 1kHz after key down. Removing this 22pF capacitor reduced the crystal current by a factor of at least 3 (1/9th the heating)... and the drift on key down is no longer noticeable.

Note: In the standard 6CL6-based crystal oscillator circuit that you find in Heathkit DX-20s (etc.) and Johnson Rangers (etc.) this crystal heating effect is not as big a problem... because the amplitude of the grid-to-cathode voltage during oscillation is significantly less for a 6CL6 ECO than it is for a 6V6 ECO or a 6L6 ECO. Therefore the current through the crystal (during oscillation) will be significantly less. Also the crystal current is half as much on 75m as it is on 40m because the impedance of the grid-to-cathode capacitance is twice as high on 75m.

Stu

[AB2EZ]

Another update:

I substituted a 6550 for the 6L6GC. At 300V B+, and 75mA of average plate + screen current, the output power at carrier is now 10 watts (v. 6W with a 6L6GC). The modulation fidelity is still excellent, with 125% positive peaks and 95% negative peaks. Note that the plate input power is around 300V x .068A (estimated average plate current) = 20.4W. With 10W of RF output power (at carrier), the efficiency is close to 50%... which is what it should be for a properly loaded class A oscillator/amplifier.

I have taken the 22pF cathode-to-grid capacitor out of the circuit (I'm depending upon the tube's parasitic cathode to grid capacitance)... to keep the crystal current low enough to avoid excessive crystal heating  and the associated frequency drift on key down. This seem to be a good idea with every tube type I've tried so far (6CL6 with socket adapter, 6V6, 6L6).

The external modulator that I am using provides both adjustable B+ and modulation, as well as 95% negative peak limiting. Therefore, I don't need the HV rectifier (6X5) tube... so I unplugged it. This helps to accommodate the extra filament current of the 6550 v. the stock 6V6.

Post script:

I noticed that with the 6550, running at 10W output, one my crystals... that appears to be particularly sensitive to heating effects... started drifting upward in frequency, quite a bit, upon going key down. I believe this is a result of the following:

A. The crystal itself is particularly sensitive to heating
B. The cathode to grid parasitic capacitance of a 6550 is about twice as large as it is for a 6L6GC. As noted in one of my earlier posts in this thread, this, by itself, would cause the crystal current to increase by a factor of 2, and the associated crystal heating to increase by around a factor of 4
C. There is parasitic capacitance between the plate of the tube and the grid. In an ECO, this is apparently causing the crystal to carry more current. If I detune the tank circuit, the crystal's frequency drift, on key down, is reduced considerably (because the amplitude of the RF voltage on the plate is smaller, and also the phase of the RF voltage on the plate with respect to the RF plate current changes from 0 degrees to a positive or negative value).

Bottom line: To accommodate my more temperature sensitive crystals, I will operate with the 6L6GC unless I can think of another simple way to reduce the crystal current.

Stu

[AB2EZ]

Last update:

I added neutralization to the AC-1 clone, and that made a significant improvement... both in terms of reducing residual FMing (which wasn't bad before), and reducing the crystal current. I can now use the 6550 to obtain 10W of carrier (see my prior post)

The way I neutralized the circuit was as follows:

Using a 1.25 inch long, 1 inch diameter, type 43 ferrite core, I wound an RF step down transformer with 18 turns on the primary and 1 turn on the secondary. The primary (18 turns) is connected across the tuning capacitor (i.e. plate-to-ground, on the DC blocked side of the plate blocking capacitor). One end of the 1 turn secondary is connected to ground. The other end of the 1 turn secondary is connected to a 15pF capacitor . The other end of the 15pF capacitor is connected to the grid of the tube. The transformer is connected to produce an inversion, as well as a step down. I was concerned that the transformer might saturate... but it appears to be working as an ideal transformer. Its presence has no noticeable effect on the setting of the tuning capacitor for resonance (i.e. the magnetizing inductance of the transformer is much larger than the inductance of the tank coil). The core does not get warm under long key down conditions.

Why does neutralization help:

Because of the "Miller effect", the plate-to-grid capacitance (0.6pF for a 6L6GC, and 0.8pF for a 6550) gets multiplied up by a large factor that depends upon the input-to-output voltage gain of the tube.... and adds to the effective grid-to-ground capacitance (i.e. across the crystal). Furthermore, since I am modulating the voltage gain (by modulating the screen and plate voltage), this results in a modulated value of this added capacitance that is effectively between the grid and ground.

The result is more crystal current (and crystal heating) due to the increased effective grid-to-ground capacitance, and more residual FMing due to the modulation of this effective capacitance.

Neutralization largely cancels (since it isn't perfect) these effects of the plate-to-grid capacitance.

As a side benefit, the primary of the transformer serves the purpose of a safety choke.

Stu

[WD5JKO]

  Stu,

  That neutralizing circuit is interesting. For the ferrite rod, did you mean type 43 material?

That rod made transformer should be fairly broadband. I wonder if it would work in a multi-band rig where there was a minimum inductance met for the lowest frequency band. While looking for Ferrite rod information, I came across this:

http://www.bytemark.com/products/rod_new.html#antennaapp

Looks like type 61 is a good choice for HF frequencies.

Why not a toroid?

Jim
Wd5JKO

[AB2EZ]

Jim

Yes, I meant to say type 43. I guess I have FT 243 crystals on my mind these days.

The core is 1 inch in diameter, 1.25 inches long, and has a 0.5 inch bore. [i.e. it is a toroidal core, not a rod]

These are the same cores as many of us use for transformers in class E rigs.

Please note that the peak voltage swing across the tuning capacitor is around 300V (at carrier) and the frequency of operation is 7.3MHz.

At lower frequencies and/or higher voltages one will require more core cross sectional area per turn and/or more turns on the primary winding in order to prevent core saturation and to ensure sufficient magnetizing inductance.

Stu

  Stu,

  That neutralizing circuit is interesting. For the ferrite rod, did you mean type 43 material?

That rod made transformer should be fairly broadband. I wonder if it would work in a multi-band rig where there was a minimum inductance met for the lowest frequency band. While looking for Ferrite rod information, I came across this:

http://www.bytemark.com/products/rod_new.html#antennaapp

Looks like type 61 is a good choice for HF frequencies.

Why not a toroid?

Jim
Wd5JKO