G2DAF amplifier design

[KL7OF]

I acquired a homebrew 2x813 amplifier that has (2) 6202 rectifier tubes fed with a portion of the RF input drive ....the output of those tubes is fed to the screens of the 813's for screen voltage. My web research says that this might be the G2DAF circuit used to eliminate the need for a screen supply  or drop resistor and clamp tube etc...I'm testing it now.  The amplifier is very well built and I have used it on 75 AM but it seems anemic.  the gain seems low.   Has anyone had experience with this circuit?   I have experience with GG 813's and class c 813's and this design doesn't excite me....I will take some pictures because its a very nice build..similar to this......  http://www.ifwtech.co.uk/g3sek/misc/g2daf-1.gif

[w8khk]

I have had experience with that circuit, but not with 813s.  In 1967 I built an amplifier using a pair of Eimac 4-400A tubes in parallel, using a pair of 6AX4 rectifiers (TV damper tubes) in a voltage doubler to provide screen voltage.  

I used an Illumitronix (Air Dux) Pi 195-2 tank coil in the pi net output.  Zero bias on the control grid, but cutoff bias on standby.  No grid tuning network, just an array of non-inductive swamping resistors to load the exciter with about 100 ohms total.  I ran 4000 volts on the plates, and it did legal limit input with rather good efficiency.  The plate choke was a B&W 800, with Centralab 20DKT5 doorknobs for plate blocking and bypass, two in parallel at each location.  They were RF rated, 500 pF each at 20 KV.

I have heard recently that the G2DAF circuit is noted for poor IMD performance, but I never had any issues, nor did my father (W2DU) who used it for several years after I was discharged from active service in the USAF in 1970.  He operated it regularly on the RCA worldwide single sideband net, and I am sure the engineers there would have complained if the signal was not above reproach.

I drove it with a Heathkit SB101, he use the HW101, and we operated all bands from 75 through 10 meters.

I built the amplifier in my AF barracks a Davis Monthan AFB in Tucson, AZ.  It went together with no problems and never had a crap-out of any kind.  I used a UTC S-48 plate transformer and solid-state bridge rectifier, floating the center tap, with a 15 uF 5000 volt oil cap.  A 115 volt, 200 watt Kodak photoflood bulb was placed in series with the plate transformer primary for initial tune-up.  

[KL7OF]

Thanks, Rick  ..This amp has a 300 ohm swamping carbon and the drive splits there feeding the screens thru a .005 mica, the rectifier tubes and a 47 ohm resistor.  Do you know how much screen voltage your circuit developed?  I wonder how much the screen voltage varies under loads..?This is interesting...  Steve

[w8khk]

I do not recall what the screen voltage or current was.

I used four cheap 1 mA meters, with shunts and series resistors for current and voltage.  I monitored plate current, relative output level, screen voltage, and screen current, but I cannot remember any of the parameters.

I do recall it used a .01 uF disc ceramic from the RF at the control grids to the junction of the two rectifiers, an another .01 uF disc bypass at the screens.

I remember I only used about half of the drive capability of the SB-101 to tune to 1KW input power in CW mode.  But all the other voltage and current data are gone from these old memory banks.

I did monitor my signal output with a home-brew 5 inch scope, to be sure I was not over-driving.  The 4-400A tubes needed very little screen voltage and current running at 4 KV on the plates.  I have no idea what the 813 would need. 

I still have the complete G2DAF RF deck, but the power supply components have all been repurposed.  It would be rather difficult to put back into operation, and I have a home-brew 8877 linear, a Hallicrafters HT-32 fully restored, as well as an SB-220 and an Ameritron AL-80, so not much need for it anymore.

[KL7OF]

pics..... having some trouble posting

[KD6VXI]

VE7RF has built and used one before.

From the discussions I remember on the amps reflector, Rich Measures reflector, etc. it was a pain in the arse to get working correctly.  You either loved or hated them, but that was about it.

It would seem if you used a tetrode that pulled much screen current you where going to see a varying load on the driver.  Maybe that didn't matter with tube tx, but I don't think solid state would tolerate that much.


--Shane
KD6VXI

[KL7OF]

Yeah Shane, I read about the love/hate relationship too...This amp is really well built.The guy did a nice job but so far I remain unimpressed...I have been using a pi input tuner and driving with a rice box altho it works on 75 without the input tuner in line...I will play with it some more as I want to take some voltage measurements around the screen circuit..

[KL7OF]

http://www.redwaveradio.com/5_2498bad76bded39d_1.htm

Interesting comments on G2DAF design.

[KA7WOC]

Don't know if Brian told you or not as I couldn't hear you on 20.  When you've had enough fun go ahead ahead modify it with my blessings.
woc

[KL7OF]

Don't know if Brian told you or not as I couldn't hear you on 20.  When you've had enough fun go ahead ahead modify it with my blessings.
woc

It works pretty good on 40 sideband...The G2DAF circuit seems to be working but the output tank is lacking...Not enuf C or L on some bands.....Still tinkering and recording the results....Havin fun...

[KA7WOC]

Tried 20 or 15?  I think Leif may have mostly used it up there.

[w4bfs]

W5CBS used several tubes (4-400, 4cx1000,ect) in G2DAF rf derived screen voltage circuit.  He emphasized taking care with the RC time constant used on the screen bypass cap.  the time constant must be quick snuff to accurately follow the voice waveform with small error (say at least 10 times quicker than highest audio freq) and still be low snuff Xc at lowest rf op freq (say 10 times lower on 160 mars)  W5cbs used 27k resistor and 470pF bypass cap for a 13 microsec tc .  Most hams reporting poor results used too much screen bypass cap  

[KL7OF]

W5CBS used several tubes (4-400, 4cx1000,ect) in G2DAF rf derived screen voltage circuit.  He emphasized taking care with the RC time constant used on the screen bypass cap.  the time constant must be quick snuff to accurately follow the voice waveform with small error (say at least 10 times quicker than highest audio freq) and still be low snuff Xc at lowest rf op freq (say 10 times lower on 160 mars)  W5cbs used 27k resistor and 470pF bypass cap for a 13 microsec tc .  Most hams reporting poor results used too much screen bypass cap  

There is a mica to ground with a 2watt resistor across it for the screen bypass.......I'll have to check the values..  Thanks...

Looks like  the screen bypass is .01 and 22k ohm...After I get done testing with this bypass I'll try .00047 and 22 k ohm......

[w4bfs]

in addition it is not necessary to use vacuum diodes for the voltage multiplier.  Schottky diodes such as 11dq10 work well to 10 mtrs... putting them in series lowers capacity and raises piv as well.   I made a quadrupler for a 4-1000 amp and with noninductive 300 ohm input load driven by 1:4 unun with 1.7:1 vswr max 160-10 mars

[KL7OF]

W5CBS used several tubes (4-400, 4cx1000,ect) in G2DAF rf derived screen voltage circuit.  He emphasized taking care with the RC time constant used on the screen bypass cap.  the time constant must be quick snuff to accurately follow the voice waveform with small error (say at least 10 times quicker than highest audio freq) and still be low snuff Xc at lowest rf op freq (say 10 times lower on 160 mars)  W5cbs used 27k resistor and 470pF bypass cap for a 13 microsec tc .  Most hams reporting poor results used too much screen bypass cap   

There is a mica to ground with a 2watt resistor across it for the screen bypass.......I'll have to check the values..  Thanks...

Looks like  the screen bypass is .01 and 22k ohm...After I get done testing with this bypass I'll try .00047 and 22 k ohm......

I changed the screen bypass from 10,000 pf to 470 pf and the 22K resistor had gone high so I replaced that....On 40M single tone SSB 50W input it showed 7 ma screen current...now it shows 16 ma screen current...all the other parameters remain the same...including output power........I do see an increase in positive peaks..

[w4bfs]

that is good info to know ... have you looked on spectrum analyzer for unwanted artifacts?