Not at the moment,
its just a thought if I can get the foundation of the amp..
Regards
M. Gregg
They look very similar from a Google about. Edcor are 25W Vs 30W for the Hammond.
You can get the Hammond ones from Bluebell Audio here in the UK and I have had excellent experience dealing with them
Plus after postage, taxes, messing about with the "post office" and the long wait, I think the Hammonds have it.
Of course the 10K transformers would probably be better but they are only 90mA which is a little low IMO for GM70. The trade off would be lower distortion and better damping factor all other things being equal, which they aren't of course.
Still excellent value but perhaps not the best looking trannys.
Cheers
Matt.
You can get the Hammond ones from Bluebell Audio here in the UK and I have had excellent experience dealing with them
Plus after postage, taxes, messing about with the "post office" and the long wait, I think the Hammonds have it.
Of course the 10K transformers would probably be better but they are only 90mA which is a little low IMO for GM70. The trade off would be lower distortion and better damping factor all other things being equal, which they aren't of course.
Still excellent value but perhaps not the best looking trannys.
Cheers
Matt.
Hi M Gregg, I think 6e5p with Rods shunt cascode takes some beating driving GM70. Ale has a good write up of the topology here:
The Shunt Cascode Driver | Bartola Valves
There is also a lengthy thread here about it somewhere.
I use a pair of 6K to 8R output trannys. They were a bargain unknown from eBay that seem to be custom Audio Note jobs. Measure very well and are heavy as @#$%. HT is 1KV.
Don't have a schematic at the mo as it's still on the breadboard. I can scribble down what it is now if you need but it ain't pretty
Cheers
Matt.
can a triode crctuit have more gain than its mu?
The "shunt cascode" is actually a normal (for SS) folded cascode, with a bipolar trans. as the cascode stage. The load resistance sets the gain, like for a pentode stage. (gm x Rl)
Seeing how it has nothing to do with a triode Mu gain function, I don't see the attraction for it. It's got non-linear 3/2 power law gain, just like a pentode.
A much more suitable way to "break" the Mu barrier is to use a pentode with a HV Mosfet follower after it, with a resistive divider from source terminal (to ground or +Vs) back to the pentode's screen grid. Then the Mosfet follower has higher gain at its source terminal than the intrinsic triode Mu of the pentode, and the whole thing acts like a lightly loaded triode, linearity wise. With very low output Z besides. Could even use a gyrator load above the pentode. A series screen grid stopper R included of course.
Seeing how it has nothing to do with a triode Mu gain function, I don't see the attraction for it. It's got non-linear 3/2 power law gain, just like a pentode.
A much more suitable way to "break" the Mu barrier is to use a pentode with a HV Mosfet follower after it, with a resistive divider from source terminal (to ground or +Vs) back to the pentode's screen grid. Then the Mosfet follower has higher gain at its source terminal than the intrinsic triode Mu of the pentode, and the whole thing acts like a lightly loaded triode, linearity wise. With very low output Z besides. Could even use a gyrator load above the pentode. A series screen grid stopper R included of course.
Last edited:
Depends, they both take some driving. I'm guessing you mean both in triode?
A good driver should drive both just fine.
Are you thinking A or A2? I like the ability for A2 even if it is rarely used.
I have been thinking back at my initial experiments with GM70. I was pretty excited when I got the output trannys so I lashed up a filament supply and used a stacked double SS bridge using a Antek 4T360 for around 1KV HT. Fixed bias from two back to back transformers giving around 0 to -160V. Toroidal mains tranny used cap coupled into the grid, driven by a Dennon mini system. Lol. I was blown away
Rods filament regs and switch mode supplies came next. Wow, massive improvement. Then onto the drivers. Still not settled.
This was my first SE amp. I always thought they were a bit wimpy, let alone inefficient. I had up to that point built several high powered push pull amps. Of course no DHT push pulls, that's next
Get a few 813, GM70 some sockets and have a play. Be careful, don't do it the way I did. Use a small C-L-C-L-C supply. Don't go for brute force as the amount of stored energy is a serious danger. Remember 50V will kill you in the right conditions, 1000V will kill you any time!!!!!
Cheers
Matt.
A good driver should drive both just fine.
Are you thinking A or A2? I like the ability for A2 even if it is rarely used.
I have been thinking back at my initial experiments with GM70. I was pretty excited when I got the output trannys
so I lashed up a filament supply and used a stacked double SS bridge using a Antek 4T360 for around 1KV HT. Fixed bias from two back to back transformers giving around 0 to -160V. Toroidal mains tranny used cap coupled into the grid, driven by a Dennon mini system. Lol. I was blown awayRods filament regs and switch mode supplies came next. Wow, massive improvement. Then onto the drivers. Still not settled.
This was my first SE amp. I always thought they were a bit wimpy, let alone inefficient. I had up to that point built several high powered push pull amps. Of course no DHT push pulls, that's next
Get a few 813, GM70 some sockets and have a play. Be careful, don't do it the way I did. Use a small C-L-C-L-C supply. Don't go for brute force as the amount of stored energy is a serious danger. Remember 50V will kill you in the right conditions, 1000V will kill you any time!!!!!
Cheers
Matt.
Here is a link to something close (2nd pic). I haven't been able to find my own old posting, using a Mosfet follower (instead of the 6SR7 follower shown here) and a resistive voltage divider from the follower stage back to the first pentode screen (6SJ7 here). (two resistors under the follower instead of one, with the tap connected back to the first tube's screen.)
http://www.diyaudio.com/forums/tubes-valves/119821-gain-stage-g2-feedback.html
So this diagram would get modified to a two resistor divider under the follower, with the tap going to the first tube's screen grid thru a grid stopper R. Output taken from the follower cathode (or Mosfet follower source). Bottom of the R divider could go to a constant positive DC voltage to assist the screen biasing. Possible gyrator load on the 1st pentode instead of the 56K R.
Since the g2 tracks proportionately to the plate voltage, it intecepts a constant fraction of the plate current, so it looks like a resistive impedance. (so preserving the R divider ratio, although altering its simple derivation) The R divider also (typically, depending on the constant bias V) will lower the screen grid voltage to well below the plate V, so drawing less screen current. The optional use of a gyrator plate load would keep the plate current constant, so the screen current would also remain constant. This removes the screen loading from the R attenuator altogether then (for AC).
http://www.diyaudio.com/forums/tubes-valves/119821-gain-stage-g2-feedback.html
So this diagram would get modified to a two resistor divider under the follower, with the tap going to the first tube's screen grid thru a grid stopper R. Output taken from the follower cathode (or Mosfet follower source). Bottom of the R divider could go to a constant positive DC voltage to assist the screen biasing. Possible gyrator load on the 1st pentode instead of the 56K R.
Since the g2 tracks proportionately to the plate voltage, it intecepts a constant fraction of the plate current, so it looks like a resistive impedance. (so preserving the R divider ratio, although altering its simple derivation) The R divider also (typically, depending on the constant bias V) will lower the screen grid voltage to well below the plate V, so drawing less screen current. The optional use of a gyrator plate load would keep the plate current constant, so the screen current would also remain constant. This removes the screen loading from the R attenuator altogether then (for AC).
Last edited:
Don, look at it closer!
The non-linearity you are expecting does not appear, if the amplifying triode is carefully chosen.
For example, the picture shows the made-for-cascode PC86 performance, and a shunt cascode load-line marked up. In this example, a driver circuit is shown that yields 160V pp very linearly.
I don't know whether the PC86's design had its transfer characteristic compensated deliberately to linearise it, but I do know that it works very well.
And I should mention that it sounds excellent!
In most cases you can build a single stage DHT driver, without a follower.
Shunt Cascode has particular advantages for a large-signal DHT drivers, compared to the usual triode stages, since the highly non-linear low-current region can be avoided completely.
Other triodes, and triode-connected pentodes work well, too, but you have to characterise them first, in order to find out.
I call it Shunt Cascode to better describe how it works. "Folded Cascode", which is mainly used in the IC design world, seems peculiarly meaningless.
.
re: Rod Coleman
"Don, look at it closer!" "The BLUE line is the Shunt Cascode load line, giving 160V pp. "
Wow! How did they do that? Sure would like to know the secret tube tech for doing that, could use it for a linear output tube maybe.
What bugs me still about the datasheet is that they still show an increasing gm with current through that "linear" range, but a straight line transfer would require constant gm. ?????
http://frank.pocnet.net/sheets/010/p/PC86.pdf
"Don, look at it closer!" "The BLUE line is the Shunt Cascode load line, giving 160V pp. "
Wow! How did they do that? Sure would like to know the secret tube tech for doing that, could use it for a linear output tube maybe.
What bugs me still about the datasheet is that they still show an increasing gm with current through that "linear" range, but a straight line transfer would require constant gm. ?????
http://frank.pocnet.net/sheets/010/p/PC86.pdf
Here is a GE datasheet for a 6LY8, which contains a frame grid pentode and a fairly constant u triode. Looking at the bottom of page 4 for the pentode or bottom of page 6 for the triode are similar curves with plate current versus grid voltage with a fixed plate voltage. These curves have a little "curve" in them still in the "straight" sections. And looking at the gm curves on the bottom of page 5 (pentode) or bottom of page 7 (triode), they have the same kind of increasing gm in that region as the PC86 datasheet shows.
I trust the GE datasheet since it looks realistic. Me thinks someone used a ruler on the Philips datasheet.....
Probably some tests are called for. 6BK7 and 6BQ7, both cascode rated tubes, do not show the straight section like the PC86 datasheet.
http://frank.pocnet.net/sheets/135/6/6LY8.pdf
A pair of tubes, like 12AT7, with constantly ramping gm curves can give linear gain in class A P-P, since the gm curves add together with one inverted (so reversed slope).
http://frank.pocnet.net/sheets/093/1/12AT7.pdf
I trust the GE datasheet since it looks realistic. Me thinks someone used a ruler on the Philips datasheet.....
Probably some tests are called for. 6BK7 and 6BQ7, both cascode rated tubes, do not show the straight section like the PC86 datasheet.
http://frank.pocnet.net/sheets/135/6/6LY8.pdf
A pair of tubes, like 12AT7, with constantly ramping gm curves can give linear gain in class A P-P, since the gm curves add together with one inverted (so reversed slope).
http://frank.pocnet.net/sheets/093/1/12AT7.pdf
Last edited:
I think someone at Philips used a ruler on that PC86 datasheet.
Here is a Mazda-Belvu datasheet for the EC88 which shows some curvature in that " super linear" region. (page 4) (Although looking at a grazing angle to the monitor, it looks like about 4 straight line sections are joined to make the "curve".)
And a Mazda-Belvu datasheet for the EC86, (page 4) which is also razor straight in that region like the Philips datasheet. But look down the straight part at a grazing angle to your monitor screen, and you can see the disjoint in the curve where the straightedge ruler was obviously used. They didn't get it lined up properly!
http://frank.pocnet.net/sheets/020/e/EC88.pdf
http://frank.pocnet.net/sheets/020/e/EC86.pdf
Here is another interesting one, note how TWO straight lines are used to draw this one (look at grazing angle with monitor), doubly kinked:
http://frank.pocnet.net/sheets/117/e/EC806S.pdf
Here is a Mazda-Belvu datasheet for the EC88 which shows some curvature in that " super linear" region. (page 4) (Although looking at a grazing angle to the monitor, it looks like about 4 straight line sections are joined to make the "curve".)
And a Mazda-Belvu datasheet for the EC86, (page 4) which is also razor straight in that region like the Philips datasheet. But look down the straight part at a grazing angle to your monitor screen, and you can see the disjoint in the curve where the straightedge ruler was obviously used. They didn't get it lined up properly!
http://frank.pocnet.net/sheets/020/e/EC88.pdf
http://frank.pocnet.net/sheets/020/e/EC86.pdf
Here is another interesting one, note how TWO straight lines are used to draw this one (look at grazing angle with monitor), doubly kinked:
http://frank.pocnet.net/sheets/117/e/EC806S.pdf
Last edited:
The 6BK7 is also worse, as evidenced by greater shifts in gm over similar current ranges.
While looking at those data sheets, see how µ also varies with anode current.
All the usual circuits dependent on µ are also subject to other real-world degradations - to which shunt cascode is immune.
For instance, the output impedance is referenced to ground (like the input), so all kinds of EMI and power-borne corruptions are rejected.
The current source further buffers the stage from the power supply.
And consider this: the instantaneous power supply current AND the steady state current are always constant, and no signal current flows in the power supply capacitor at all.
Even if the gm varies a little over the full swing of the 160V pp drive waveform (which is substantially ameliorated by the unbypassed cathode resistor) the shunt cascode gives outstanding performance. Ad taking account of the stage gain of 250 or more, you easily surpass the two stages of ordinary µ-amplifiers usually used for driver duty.
O, and did I mention that it sounds superb?
While looking at those data sheets, see how µ also varies with anode current.
All the usual circuits dependent on µ are also subject to other real-world degradations - to which shunt cascode is immune.
For instance, the output impedance is referenced to ground (like the input), so all kinds of EMI and power-borne corruptions are rejected.
The current source further buffers the stage from the power supply.
And consider this: the instantaneous power supply current AND the steady state current are always constant, and no signal current flows in the power supply capacitor at all.
Even if the gm varies a little over the full swing of the 160V pp drive waveform (which is substantially ameliorated by the unbypassed cathode resistor) the shunt cascode gives outstanding performance. Ad taking account of the stage gain of 250 or more, you easily surpass the two stages of ordinary µ-amplifiers usually used for driver duty.
O, and did I mention that it sounds superb?
Looking at the Mazda sheet for the PC86 (not the PC88, which was a shrink - to squeeze into TV tuners of the day).
If Philips used a ruler, they must have lent it to the French Engineers at Mazda Belvu:
http://www.mif.pg.gda.pl/homepages/frank/sheets/020/e/EC86.pdf
If Philips used a ruler, they must have lent it to the French Engineers at Mazda Belvu:
http://www.mif.pg.gda.pl/homepages/frank/sheets/020/e/EC86.pdf
Look down the Mazda Belvu curve at a grazing angle to the video monitor, and you can see that they didn't get the ruler aligned well, an obvious kink.
http://frank.pocnet.net/sheets/020/e/EC86.pdf
http://frank.pocnet.net/sheets/020/e/EC86.pdf
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.