3CX3000 drive arrangements

[Opcom]

Has anyone ever grid-driven a 3CX3000 as a linear amplifier?

It takes 11.5W (carrier cond?) and would be OK for a smaller exciter. Like a DX-60 or similar fun items. The thought is that if it can be stable over a wide range, maybe using an attenuator and swamping resistors, the typical 100W peak exciter could be used and have more than enough power to let it reach ham radio levels. It would also make designing a solid state driver a lot more simple at a low power.

I'm aware all the designs by hams use cathode drive because it's easy and well-accepted, but it seems like a lower drive would be convenient for trying different transmitters.

The comparison between the Eimac and Svetlana data sheets are confusing because one says 11.5W and the other says 115W. Which is it?

Eimac:
85V peak=60V RMS. 0.13A DC grid current.. makes a 460 Ohm grid impedance, right?

[W3RSW]

When grid driving a triode at RF don't forget to neutralize it.
I'm not sure it's a zero or low bias triode anymore when grid driven either. At any rate the grid and plate circuits become a little more complicated than when cathode driven.

Also the flange and tube topography are optimized for very low impedance connection in cathode driven service. This is compromised somewhat in grid driven service.

Really when you look at the size of the tube and stuff like 50 amps just for the filament, I find it kind of stretching it to think that a 10 watt driver is all one would be inclined to use in "amateur" service, but the grid driven specs attached seem to limit power compared to cathode driven. Plate dissipation is about half, etc. so I wonder why?

http://www.g8wrb.org/data/Eimac/3CX3000A7.pdf

[AB2EZ]

Opcom:

If I assume that the input circuit, that you show in your post, acts as (roughly) a 2:1 voltage divider for the input RF signal...

Then you need about 170V of peak input signal to produce 85V of peak signal from grid-to-ground.

An RF signal having an amplitude of 170V, across a 50 ohm load corresponds to around 289 watts of average power dissipated in the load.

Check out W2DTC's (Ken's) 3CX3000 GG linear amplifier design. It appears to require only 100W of drive power (for 1500 watts output), and it's GG design is much better suited to this tube.

http://w2dtc.com/3cx3000f7.htm

If you stick with a grid-driven approach, with this tube... using an input matching network to step up the drive voltage... then you are going to have a "heck" of a time taming parasitic oscillations. The grid-to-plate capacitance of this tube is very high.

Rich:

I think that the specification sheet for the 3CX3000 gives an "apples and oranges" comparison between grid driven operation and grounded grid operation. I think that the grid-driven numbers correspond to operation at "carrier conditions", while the grounded grid numbers correspond to operation at "modulation crest conditions".

Best regards
Stu

[DMOD]

The comparison between the Eimac and Svetlana data sheets are confusing because one says 11.5W and the other says 115W. Which is it?

I would say the driving power for the control grid is 115 Watts or 51% of maximum.

BTW, The 3CX3000 was used in the original Collins design for their 5kW PWM transmitter.

Phil - AC0OB

[K1JJ]

With 380 watts devoted to the filament alone, this is a tube that should be run for a BIG signal.

When running a big signal, we need all the cleanliness we can get. Grounded grid config provides a built-in negative feedback that will help the IMD numbers.  If we run it grid driven, then adding NFB is a more complex circuit and unless done well, is prone to instability, especially on the higher bands. (in addition to a neutralization circuit - yikes)

I would run the tube in GG as it was intended for linear service.  This tube is clean (-40dB 3rd IMD) and stable as a rock that way.


More to consider:

The problem becomes setting up a driver that will provide -40dB 3rd IMD. Most of the riceboxes and ham linears are not that clean. I needed to build a class A 3CX-350J IPA driver with a 1 watt lab amp just to get a clean enough signal to complement my class AB1 8877 linear system.

IE, if a -30dB 3rd IMD driver drives a -40dB 3rd final amp - this equals  roughly a  -29dB 3rd final result.  NG.   The driver should be at least 10 dB better than the final to be nearly transparent and not add to the problem.   Predistortion technology would help here.


Remember the old rule... if you are X dB louder than the average QRO ham station, you'd better be X dB cleaner - just to blend in.  Otherwise, expect side channel complaints.

T

[Opcom]

When grid driving a triode at RF don't forget to neutralize it.
I'm not sure it's a zero or low bias triode anymore when grid driven either. At any rate the grid and plate circuits become a little more complicated than when cathode driven.

Also the flange and tube topography are optimized for very low impedance connection in cathode driven service. This is compromised somewhat in grid driven service.

Really when you look at the size of the tube and stuff like 50 amps just for the filament, I find it kind of stretching it to think that a 10 watt driver is all one would be inclined to use in "amateur" service, but the grid driven specs attached seem to limit power compared to cathode driven. Plate dissipation is about half, etc. so I wonder why?

http://www.g8wrb.org/data/Eimac/3CX3000A7.pdf

It seems the specs for grid driven service are for carrier conditions, but the cathode driven specs are for PEP. It has a proper socket, no need to use a flange. I have a hard time believing it is 10W, more like 115W? as in the Svetlana data sheet.

I was thinking of things like a DX-60 (with improved modulation) on AM. There are many ways to look at the possibilities when considering FCC service classes, that is, the 4-1000 only needs '0' watts for linear service, and the old Johnson Thunderbolt's drive requirement was a perfect match for chicken band radios. Amplifiers are like tools in this way as it depends on who is using it and what for, while the equipment is just an innocent piece of equipment.

[Opcom]

After reading this, I'm inclined to re-think the grid driven idea. I was under the impression that with sufficient swamping by a non-inductive load, an oscillation would not be able to start because the reactance of the plate to grid capacitance would be so much higher than the resistance of the grid to cathode circuit. I do not have good understanding on this for Triodes but it seems like a voltage divider to reduce the plate-grid feedback (FB voltage on the grid) to almost nothing.

However, after doing the math I see it is not a good fit with the 24pF capacitance.
At higher frequencies, Cgp Z is 200-400 Ohms, a no-go.
xc-Ohms F-mhz
   
6631   1
3316   2
1658   4
829   8
414   16
207   32


As to the attenuator circuit, it is only representative of a swamping method and some attenuation, going on the 11.5W drive assumption. The idea was along the lines of being able to reach 1500W output with a 50W PEP input such as from a DX-60.

About IMD, the exciter choices are TS430-S, and a Sunair GSB900DX. I imagine the Sunair is better.

Is there an advantage to using an ATU between the exciter and the amplifier? For harmonics yes, not so sure about IMD since that is closer in.

[W3RSW]

Depends on the exciter.  An old PI net tube amp "is" the tuner or matcher. As you've read,  Eimac recommends a Q=5 PI net input for the cathode driven case and I guess in the grid driven case the "old link coupled" parallel tuned input of classic transmitters is assumed.
10 watts of real driving power with all the voltage obtainable therein from a resonant LC input circuit does sound about right to feed a large triode with circuit losses accounted for, at the reduced power output specified.

[K1JJ]

About IMD, the exciter choices are TS430-S, and a Sunair GSB900DX. I imagine the Sunair is better.

Is there an advantage to using an ATU between the exciter and the amplifier? For harmonics yes, not so sure about IMD since that is closer in.

You could look it up, but I would think a TS-430's IMD is about -30dB 3rd, like most rice boxes.  The ones that run pure class A, like the expensive Yaesu 9000? stuff, are the exceptions.

This day and age an Apache labs SDR with predistortion will do -50dB ++ 3rd, so is the future.  Really nice driver.  Bottom line is unless the driver can do -40dB or better, it will be seriously degrading your hard work of building a clean 3CX-3000A7 linear.


As for an ATU (matcher) to reduce IMD, it will have no effect cuz the tuner is very broad and will have little effect on frequencies that are within 5 KHz of the center. IMD measures close-in distortion. You might need a minimum Q of 100 to help IMD, maybe more..      It will help, however, if the driver (like a SS ricebox) is sensitive to poor loads and you are feeding a 2:1 or greater into an amplifier. Then a 1:1 match will help the driver.  A tube driver with a pi-network would not need this, of course.

Go with the GG 3CX-3000A7 with a pi-network input cathode tuning and you will have an easy and simple life with it....   (50 ohms in, 50 ohms out, variable caps)

When you get your overall chain running, sample some RF into an SDR pan adapter and run some two-tones thru.  You will quickly see how difficult it is to break -30dB 3rd order unless you use an exceptional driver. The 3CX-3000A7, when loaded heavily, (C2 loading cap more unmeshed) will easily do -40dB.  But it will not show itself as clean until it sees a clean driver.  We need two to tango.

T

[AB2EZ]

Opcom

Adding to what Tom and Rich have already said:

There are at least two potential advantages to using an ATU between the exciter and the amplifier... although whether these potential advantages translate into worthwhile performance improvements depends upon the details of the applications... e.g. how sensitive the exciter is to the details of the load it is looking into:

1. (Obvious)

The ATU can fine tune the impedance match seen by the exciter, if you want the load on the exciter to be close to 50 + j0 ohms.

2. (Less obvious)

The load, looking into the grid of a grounded cathode amplifier, will be non-linear... if the amplifier is driven with a signal that drives the grid-to-cathode voltage positive (e.g. class AB2).

As a result, although the voltage from grid-to-cathode and the grid current will both be periodic (at the fundamental RF frequency)... they will both contain harmonics of the fundamental RF frequency.

If you don't want the exciter to see these harmonics... propagating back to the exciter from the amplifier (analogous to reflections from the amplifier's input; but actually generated by the non-linear load at the amplifier's input) ... then an ATU between the exciter and the amplifier can act as a low pass filter, and block them (with some amount of attenuation that depends on the design of the ATU). Like any 2-port passive filter, the ATU will reflect these harmonic components back to the amplifier... it will not absorb them.

To some extent, the input matching network will also perform the function of blocking harmonics of the RF input signal, generated by the non-linearity of the load at the amplifier's input, from propagating back to the driver.

Therefore the ATU and the input matching network will work together as a more elaborate input matching network.

With a grounded grid amplifier, operating in anything other than class A,  the load seen by the input matching network (and possibly the exciter) is also non-linear... because cathode current is compressed/cut off during part of each cycle.


Stu

About IMD, the exciter choices are TS430-S, and a Sunair GSB900DX. I imagine the Sunair is better.

Is there an advantage to using an ATU between the exciter and the amplifier? For harmonics yes, not so sure about IMD since that is closer in.

You could look it up, but I would think a TS-430's IMD is about -30dB 3rd, like most rice boxes.  The ones that run pure class A, like the expensive Yaesu 9000? stuff, are the exceptions.

This day and age an Apache labs SDR with predistortion will do -50dB ++ 3rd, so is the future.  Really nice driver.  Bottom line is unless the driver can do -40dB or better, it will be degrading your hard work of building a clean 3CX-3000A7 linear.


As for an ATU (matcher) to reduce IMD, it will have no effect cuz the tuner is very broad and will have little effect on frequencies that re within 5 KHz of the center. You might need a minimum Q of 100 to help IMD, maybe more..      It will help, however, if the driver (like a SS ricebox) is sensitive to poor loads and you are feeding a 2:1 or greater into an amplifier. Then a 1:1 match will help the driver.  A tube driver with a pi-network would not need this, of course.

Yes, go with the GG 3CX-3000A7 with a pi-network input cathode tuning and you will have an easy and simple life with it....

T

T

[K1JJ]

Stu,

Some good points.  Especially about creating a more elaborate input filter using the existing input circuit and ATU.

I use pi-networks on all my amplifier inputs. Thinking back about my IMD tests, I did notice a "sweet spot"  when tuning for best IMD.  It all depends on Q of the tube's input circuit and if one is even used. So, I must reverse my opinion that an ATU will not help on the input of an amplifer for IMD -  in certain situations.


A friend pointed out that the so-called "flywheel" effect can be improved with a tuner between the driver and cathode.   I assumed the cathode input circuit already has a Q of at least 3.  So how much effect  will the ATU have in IMD improvement by bringing the Q up even further,  I dunno.  (two tuned circuits in cascade)  Tests would show it. It is not easy to get a high Q when using a low impedance 50 ohm circuit.   I would think that there is a sharp diminishing return after the first tuned circuit in meaningful IMD improvement.  I cud easily add in a tuner and run some tone tests on the SDR spec analyzer to see.

T

[AB2EZ]

Tom

Agreed... I have never done any detailed measurements regarding these effects. When I operated my high power (a pair of Russian GS-35b triodes) GG amplifier, I used two boxes between the input and the Ranger exciter.

http://mysite.verizon.net/sdp2/id20.html

I installed a 200W rated 80m low pass filter at the input connector of the amplifier (to block harmonics from propagating back toward the Ranger), and I used a 300W rated antenna tuner between the Ranger and the low pass filter. The low pass filter was the only input matching circuit that I used.

http://www.arraysolutions.com/Products/wx0bbpf6.htm#200w

I didn't make any A/B comparisons regarding the benefits of using these... but I had them in my collection of stuff... and the Ranger seemed to work well, driving into 50 + j0 ohms.

Stu