Hi,there
I seen Screen Grid Drive topology from Berning EA-230, Milbert BAM-235 Amplifier which used 6SN7 drive through 6JN6 with Screen Grid input and short Control Grid to Cathode (or Ground on some schematic).
My question: If I want to use EL34 instead of 6JN6, How do I design for its Grid (G1, G2)?
Thanks for advance.
ANALOG GUY
I seen Screen Grid Drive topology from Berning EA-230, Milbert BAM-235 Amplifier which used 6SN7 drive through 6JN6 with Screen Grid input and short Control Grid to Cathode (or Ground on some schematic).
My question: If I want to use EL34 instead of 6JN6, How do I design for its Grid (G1, G2)?
Thanks for advance.
ANALOG GUY
Attachments
According to David Berning's writings, "I found that I could get the same tube linearity at one tenth the idle current normally used if I drove the tube from the screen as opposed to the grid. I further found that this only works well on TV sweep-type tubes and not very well on audio tubes, and it requires a much more powerful driver stage." Link here.
A good tube for screen driving is the 6L6GC or the beefed up version known as the 7027-A.....You need to calculate the screen current you will encounter at the max drive signal you plan on using...then calculate the appropriate driver impedance to handle the job...I see the schematic you posted has the cathode followers for driving... This is stuff right out of the 1940's and 1950's RF designing techniques...
The gm is low as heck for G2 drive.... Just like the McIntosh Unity coupled output stage, in that you will need to produce large voltage amplification in the driver stages then into a follower to feed into G2....
Keep in mind if you plan to drive the screen, then you need the proper curves to design this...The standard I-V curves won't help much...You need the plate characteristics for G2 modulation... Check out these curves....
http://www.triodeel.com/6l6gc_p7.gif
Look close and you will see they are for change in G2 while G1 is held to 0 volts...
The gm is low as heck for G2 drive.... Just like the McIntosh Unity coupled output stage, in that you will need to produce large voltage amplification in the driver stages then into a follower to feed into G2....
Keep in mind if you plan to drive the screen, then you need the proper curves to design this...The standard I-V curves won't help much...You need the plate characteristics for G2 modulation... Check out these curves....
http://www.triodeel.com/6l6gc_p7.gif
Look close and you will see they are for change in G2 while G1 is held to 0 volts...
SY said:
Well, I have to disagree....
There were many sucessfull RF power amps with G2 driving up to 450V with these 7027-A valves...safely operating within ratting....
One of my old time friend worked for RCA at that time..he said the 7027-A came off the same production line as the 6L6GC...It just tested stronger and therefore was sorted out and labeled 7027-A....
As for the 807 ....screen grid drive is limited to something lower like 250V ....or else they arc-over and pop....
Chris
Hi Chris,
Says who?
I can only find a static g2 voltage limit being stated in the spec sheets, which is 300V (not 250V) for tetrode op according to the STC sheet, by the way. And for triode op, the 807 screen is even allowed up to 400V (static, same spec sheet).
Don´t mix up static and dynamic operation. While for example 600V is given as an absolute (static) plate voltage limit for 807, the plate obviously can easily swing way beyond 1000V under signal conditions when working into an OPT, without any problem whatsoever.
The same principle holds true for the screen grid, most common example being UL operation, where (static) screen voltage is modulated down and up by a fraction of the plate voltage swing.
Now, for example think about 807 triode strapped operation. According to the spec sheet, max. Ea=Eg2=400V. To get some power out of it, allow Ea=Eg2 to swing +/-300V, which is not unreasonable.
This means Ea(peak)=Eg2(peak)=700V. And 807 g2 does neither arc-over, nor does the whole 807 pop under such valid operation conditions, except you have a faulty tube to start with.
Tom
Says who?
I can only find a static g2 voltage limit being stated in the spec sheets, which is 300V (not 250V) for tetrode op according to the STC sheet, by the way. And for triode op, the 807 screen is even allowed up to 400V (static, same spec sheet).
Don´t mix up static and dynamic operation. While for example 600V is given as an absolute (static) plate voltage limit for 807, the plate obviously can easily swing way beyond 1000V under signal conditions when working into an OPT, without any problem whatsoever.
The same principle holds true for the screen grid, most common example being UL operation, where (static) screen voltage is modulated down and up by a fraction of the plate voltage swing.
Now, for example think about 807 triode strapped operation. According to the spec sheet, max. Ea=Eg2=400V. To get some power out of it, allow Ea=Eg2 to swing +/-300V, which is not unreasonable.
This means Ea(peak)=Eg2(peak)=700V. And 807 g2 does neither arc-over, nor does the whole 807 pop under such valid operation conditions, except you have a faulty tube to start with.
Tom
Chris, the arc-over in 6146 tends to happen when the screen goes high and the plate goes low, as in screen-drive. If the two are driven together (e.g., in triode mode), no problem, which is why the triode voltage rating is 400 versus the 250V screen rating in pentode. I found this out the hard way in an old pentode amp of mine.
I've had a lot of reliability problems with 6146 that aren't forced air cooled when run near their ratings. Great ham radio tube, though.
I've had a lot of reliability problems with 6146 that aren't forced air cooled when run near their ratings. Great ham radio tube, though.
Obviously if one is going to drive g2, then one wants a tube with high gm2 at that g2, which points to the TV Sweep tubes. You can calc. the g2 gm of any sweep by taking its g1 gm and dividing by the g2/g1 Mu factor. (for some specified current for the stated g1 gm)
To translate that gm2 to some other current, use (Ip'/Ip)^.333
The greater linearity of the g2 is due to it conforming well to the theoretical 3/2 power law.
While a typical g1 behaves more like a 2 power law (square), due to grid wire proximity effects. {transform g1 gm by (Ip'/Ip)^.666 } Obviously these gm's don't track, so the stated g2/g1 Mu changes some (droops) with current. (approx. 1/3 power law droop)
Biggest operational problem using g2 drive (other than the HV driver stage) is the danger of breakdown and high screen current when the plate V goes low and g2 drive V goes high. So you don't want a low gm2 since that would require excessive drive V.
Now take a look at a new drive scenario.
1st, lets divide the drive between both g2 and g1 by ratioing the drives by the g2/g1 Mu. This splits the drives so that each grid is doing half the work of producing the plate current. This also means that the g2 drive V span can be halved, since g1 is doing the other half of the work. For a Mu 3 Sweep tube, the g2 voltage swing would be merely 3X the g1 voltage swing in this mode. (but since g1 swing is halved here, versus straight g1 drive, g2 is actually only swinging 1.5x the straight g1 drive V, driver problems evaporated!) Now the drives are still working in 3/2 power and 2 power, so we have compromised the linearity some from just g2 drive at 3/2 power. (but still better than g1 drive)
Then, 2nd, lets drive the g2 with a signal current source, rather than the usual V source. The g2 tends to intercept a constant fraction of the cathode current, at least until the plate V descends down to near g2 V range. So we get a flat current gain Beta out until high amplitudes where it droops. But it is a linear (1 power) nominal transfer, so better than traditional g2 drive. The combined g2 and g1 drive has halved the excursion on g2, so we are looking OK.
Now notice the resultant voltage swing on g2, it is a diode curve run backwards, or current driven. We are getting 2/3 power law V variation on g2 versus the linear current drive signal. So lets take a voltage divided g2 voltage (1/Mu, reasonably high R divider from -V1bias) down to g1 (which is operating in neg. V range, so no g1 current to interfere).
The 1/Mu 2/3 power V signal is fed into the 2 power g1, giving 1/3 power drive there. So the slowly drooping g2 Beta is being compensated by a slow 1/3 power boost on g1. This should give something approaching LINEAR drive altogether. Which beats even traditional g2 V drive.
Warning!!:
I haven't tried this yet. Great theory though.
Also note that the output Rp of this mode will be very high, so will depend on some type of local or global Fdbk to set the amplifier Zout.
To translate that gm2 to some other current, use (Ip'/Ip)^.333
The greater linearity of the g2 is due to it conforming well to the theoretical 3/2 power law.
While a typical g1 behaves more like a 2 power law (square), due to grid wire proximity effects. {transform g1 gm by (Ip'/Ip)^.666 } Obviously these gm's don't track, so the stated g2/g1 Mu changes some (droops) with current. (approx. 1/3 power law droop)
Biggest operational problem using g2 drive (other than the HV driver stage) is the danger of breakdown and high screen current when the plate V goes low and g2 drive V goes high. So you don't want a low gm2 since that would require excessive drive V.
Now take a look at a new drive scenario.
1st, lets divide the drive between both g2 and g1 by ratioing the drives by the g2/g1 Mu. This splits the drives so that each grid is doing half the work of producing the plate current. This also means that the g2 drive V span can be halved, since g1 is doing the other half of the work. For a Mu 3 Sweep tube, the g2 voltage swing would be merely 3X the g1 voltage swing in this mode. (but since g1 swing is halved here, versus straight g1 drive, g2 is actually only swinging 1.5x the straight g1 drive V, driver problems evaporated!) Now the drives are still working in 3/2 power and 2 power, so we have compromised the linearity some from just g2 drive at 3/2 power. (but still better than g1 drive)
Then, 2nd, lets drive the g2 with a signal current source, rather than the usual V source. The g2 tends to intercept a constant fraction of the cathode current, at least until the plate V descends down to near g2 V range. So we get a flat current gain Beta out until high amplitudes where it droops. But it is a linear (1 power) nominal transfer, so better than traditional g2 drive. The combined g2 and g1 drive has halved the excursion on g2, so we are looking OK.
Now notice the resultant voltage swing on g2, it is a diode curve run backwards, or current driven. We are getting 2/3 power law V variation on g2 versus the linear current drive signal. So lets take a voltage divided g2 voltage (1/Mu, reasonably high R divider from -V1bias) down to g1 (which is operating in neg. V range, so no g1 current to interfere).
The 1/Mu 2/3 power V signal is fed into the 2 power g1, giving 1/3 power drive there. So the slowly drooping g2 Beta is being compensated by a slow 1/3 power boost on g1. This should give something approaching LINEAR drive altogether. Which beats even traditional g2 V drive.
Warning!!:
I haven't tried this yet. Great theory though.
Also note that the output Rp of this mode will be very high, so will depend on some type of local or global Fdbk to set the amplifier Zout.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.