Wrong question.
Check out Alex Kitic's RH807 and afterwards read http://www.diyaudio.com/forums/tube...ffective-global-fb-decent-damping-factor.html
Even Dave from Planet10 at one stage mentioned he did change his triode strapped pentode amp to a Kitic look alike.
One thing is certain: O H Schade's plate-to-grid feedback loop is found in a lot of good amps. (And the 6L6GC is one of the most linear tubes around but watch what voltage you operate it at, especially g2)
Enjoy
AM
Hi Amadeus,
RH807 isn´t acceptable and wrongly designed. For a Schade like this is, a pentode-driver or cascode(nothing else) with quite high current is mandantory acting like a current-source. This have been pointed out several times here. An E280F or D3a at around 20mA will do the job very well. Also a cascode like with 6H30 will do. But I admit 807 is really good for Schade. Send me a message and I will return a correct design if you want to. Pls also read the original paper written by Schade 1937 and RDH from 1954!
RH807 isn´t acceptable and wrongly designed. For a Schade like this is, a pentode-driver or cascode(nothing else) with quite high current is mandantory acting like a current-source. This have been pointed out several times here. An E280F or D3a at around 20mA will do the job very well. Also a cascode like with 6H30 will do. But I admit 807 is really good for Schade. Send me a message and I will return a correct design if you want to. Pls also read the original paper written by Schade 1937 and RDH from 1954!
Last edited:
A pure pentode with plate to grid feedback
Running a pentode in triode mode is like sticking a towel in the intake of an engine!
No it's not.
Don't ignore TV video tubes: EF184, EL821 series, and one notices on some data sheets the gain also given in triode mode and it's very reasonable when compared to standard triodes. A video tube operated as triode is about the straightest line on can get and at a limitless frequency range.
richy
OK so I have a question after reading Kuei Yang Wang's writings about gNFB around the OPT.
I have a GU50 powered SE amplifier (The Sweet Peach FU50) that I fancy redesigning to take advantage of a Schade feedback mode operation, so I can eliminate the NFB loop entirely. Currently the GU50 is triode strapped.
From what I can gather from revintage for drivers a triode does not work well with this Schade feedback method - but a pentode or a triode cascode does.
Well I only have triodes - matched triodes I have are the 6n6p which I think should be good for the job? Then I want to follow perhaps the UL cascode to make it work.
The GU50 is on about 435V and 55mA idle and is I think a regular pentode, so with a little work I should be aiming at a local feedback (Rk) 6n3p input stage, a local feedback (Rk) driver cascode and the GU50 pentode Schade connected to the cascode to give the OPT a nice low impedance drive.
What do you think - any ideas, hints, tips, no-nos, yes-yeses?
I have a GU50 powered SE amplifier (The Sweet Peach FU50) that I fancy redesigning to take advantage of a Schade feedback mode operation, so I can eliminate the NFB loop entirely. Currently the GU50 is triode strapped.
From what I can gather from revintage for drivers a triode does not work well with this Schade feedback method - but a pentode or a triode cascode does.
Well I only have triodes - matched triodes I have are the 6n6p which I think should be good for the job? Then I want to follow perhaps the UL cascode to make it work.
The GU50 is on about 435V and 55mA idle and is I think a regular pentode, so with a little work I should be aiming at a local feedback (Rk) 6n3p input stage, a local feedback (Rk) driver cascode and the GU50 pentode Schade connected to the cascode to give the OPT a nice low impedance drive.
What do you think - any ideas, hints, tips, no-nos, yes-yeses?
Shade driver has to have constant impedance over all frequency of interest.
Using only tubes to get there, better to shoot for near infinate impedance
than near zero. Cascode is probably the best Schade driver, cause you got
no significant screen current to muck up the equations.
But if you are willing to mix in a little sand: Triodlington and/or Mu follwer
configurations have a stable low impedance near zero. This too can be
used together with a fixed series resistance to drive Schade.
It is only unassisted triode plate resistance that can not be trusted to be
constant enough drive impedance. Yet I see this abused over and over by
folks who just don't seem to grock Schade. That two pentodes makes the
far better fake triode than Triode + Pentode . If you want Schade to fake
like a real triode, ditch the real triode! Or do something to flatten the real
triode's effective plate resistance against one extreme or the other...
------------
I'm investigating Schaded sand with 100% CB feedback (Schaded Mu=1)
as the heart of a concertina splitter. Mu=1 is neato, cause from the load
perspective, the active device appears a diode of variable voltage drop.
Push and pull irregularities crosstalk at low impedance across such a diode.
Or so theory goes...
Using only tubes to get there, better to shoot for near infinate impedance
than near zero. Cascode is probably the best Schade driver, cause you got
no significant screen current to muck up the equations.
But if you are willing to mix in a little sand: Triodlington and/or Mu follwer
configurations have a stable low impedance near zero. This too can be
used together with a fixed series resistance to drive Schade.
It is only unassisted triode plate resistance that can not be trusted to be
constant enough drive impedance. Yet I see this abused over and over by
folks who just don't seem to grock Schade. That two pentodes makes the
far better fake triode than Triode + Pentode . If you want Schade to fake
like a real triode, ditch the real triode! Or do something to flatten the real
triode's effective plate resistance against one extreme or the other...
------------
I'm investigating Schaded sand with 100% CB feedback (Schaded Mu=1)
as the heart of a concertina splitter. Mu=1 is neato, cause from the load
perspective, the active device appears a diode of variable voltage drop.
Push and pull irregularities crosstalk at low impedance across such a diode.
Or so theory goes...
Last edited:
That is the accepted truth and in most cases attempting to Schade a triode driving a pentode design doesn't work very well, in fact it sounds like &#*@! I have found exceptions to this though.
I built a little spud amp using one 6LR8 per channel. In every listening and bench measurement test the amp worked better with a Schade resistor from the output tube plate to the driver (triode) plate.
I have also tried Schadeing the Simple SE. It uses a CCS loaded 12AT7 driver and any of the popular output tubes. Here the results are mixed. Wiring the output tube as a triode almost always sounds the best, but sometimes lacks the power needed to "rock the house". UL with cathode feedback provides more power with a loss of triode detail. I have noticed that some combinations of tubes and OPT's respond quite well to Schadeing, but others don't. The 6L6GC and a 3000 ohm Transcendar (no UL tap) works very well, but the big Edcor and KT88's don't (other conditions the same).
You know, I never saw that thread before since it was posted before I joined diyAudio. I have been tinkering with several designs that use partial feedback and multiple grid drive to the output tube. Several designs have been "voted off the island" but a few remain. One of them, and the one that is currently in the "hot seat" is the amp that got designed in this thread:
http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp.html
After spending some time with Petes red board, I got the wild idea to wire the 6L6GC's as pentodes and go poking around with Schade resistors. Adding any significant NFB reduces the gain too much, so I fixed that by swapping the first 6SN7 to a 6SL7 and readjusting the current. Attempting to Schade this amp in the usual manner didn't work because the 6SN7's output resistance is too low. Adding resistors between the plate of the 6SN7 and the Schade point helps a lot. I decided to try taking the Schade resistors back to the first stage, and after building a 100 watt oscillator I discovered that you have to swap one set of ends. I have a whole bunch of experiments planned for this amp, but for now I am just listening to it. Thorsten had much to say in that thread, and it sounds like there were very few believers, but this thing sounds real nice. Other than my liberal sprinkling of sand (CCS under both LTP's and PowerDrive) my circuit is almost exactly like his (see the 6L6GC in AB2 thread).
I plan to try using this circuit to drive both G1 and G2 simultaneously. For now it is wired to use tubes with the same pinout as the 6L6GC (EL34, KT88, 6BQ6, 6DQ6), but I have a long list to try at different operating points, including the E130L.
I certainly created some noise here.
There is a lot of noise like a "triode" should not be used but it will work very well with a triode provided the triode has the right amount of transconducatance and enough current going through it. From what I understand reading about a CCS: that does not work since the short feedback loop is not having any effect on the input signal to the grid due to the fact that the previous stage has the CCS.
There are many "local feedback" designs (bot SE and PP) with 6V6 and 6L6 in the 1947 publication "Practical Amplifier Diamgrams, 45 proven circuits for the technician and experimenter" written by Jack Robin and Chester Lipman. This was downloaded from P Millet's website. (Technical books online about a thrid down).
Happy reading to ya all and peace.
AM
There is a lot of noise like a "triode" should not be used but it will work very well with a triode provided the triode has the right amount of transconducatance and enough current going through it. From what I understand reading about a CCS: that does not work since the short feedback loop is not having any effect on the input signal to the grid due to the fact that the previous stage has the CCS.
There are many "local feedback" designs (bot SE and PP) with 6V6 and 6L6 in the 1947 publication "Practical Amplifier Diamgrams, 45 proven circuits for the technician and experimenter" written by Jack Robin and Chester Lipman. This was downloaded from P Millet's website. (Technical books online about a thrid down).
Happy reading to ya all and peace.
AM
IIRC The RCA tube manual mentions that if you correctly bias the driver tube it will generate 2nd harmonics that will cancel the 2nd harmonic of the output tube. (The 6L6 was designed for PP operation and generates a lot of 2nd harmonic).
The ECC81 generates a lot of 2nd harmonics....wonder why it is liked a lot?
AM
The CCS just provides a constant current through the driver tube, which improves the linearity and increases the gain of most triode tubes. A CCS load generally does not work well with pentodes. A CCS load can affect how a triode behaves, so its use can affect how the Schade type feedback works, but does not preclude its use.
Generally a triodes plate resistance varies with tube current and applied signal. Depending on how it is applied the output impedance of the triode may vary with frequency. In an idealized Schade feedback network the driver tube will have an infinite output impedance and the feedback ratio is determined by the ratio of the Schade resistor, and the driver's plate load resistor. No tube has an infinite output impedance, but pentodes have a much higher and more constant output impedance than most triodes.
This does not mean that Schade style feedback will never work with a triode driver tube. It means that there are more variables at play and more experimentation is needed. Success is much more application dependent than a typical pentode - pentode circuit.
Many of my designs use a circuit I call PowerDrive. The driver is a CCS loaded triode buffered by a mosfet. This provides a gain approaching the Mu of the driver tube, with a very low output impedance (ohms). Using Schade in the typical fashion does not work at all since the output impedance is essentially zero. The Schade resistor sees a short, no feedback. Adding a resistor in series with the mosfet output provides an output impedance for the Schade resistor to work against, and provides a near perfect voltage divider. Add another buffer if you need low impedance drive, say for AB2. Of course this works well with triodes driving Schaded pentodes with a bit of sand in between.
In the case of Thorstens design the 6SL7 has a rather high output impedance. He did not post any component values, but I am using a lower than normal 75 K plate load. My Schade resistor is currently 750 K but I may experiment with both of these values in the future, since the DC voltages are not quite right yet. The 6SN7 has about 1/10 the plate resistance of the 6SL7, hence it did not respond well to Schadeing and required such a low value of Schade resistor that the resistor started smoking. OK, a bigger resistor could be used, but output power is being wasted by fighting the driver.
One of the other designs that I haven't ruled out yet uses a pentode LTP in the first stage followed by a triode driver. Another uses the same tubes in reversed positions. Schade feedback goes to the pentode in both cases. Each of the three designs has their advantages, but the 6SL7 - 6SN7 design works the best overall so far. It does not quite have the output voltage capability to drive screen driven sweep tubes, hence the dual grid drive experiments.
Does Schade feedback just manage the output tube or does it affect the driver tube too?
Perhaps to linearise an SE pentode output tube without Schade feedback requires a two-stage feedback loop (driver + output tube) from the output plate back into the cathode of the driver?
I.e. if you had a ECC82 triode driving a GU50 pentode a reasonable feedback would be from GU50 anode to ECC82 cathode?
Unfortunately you'd need a capacitor I think or the resistor would have to be too high a value to feedback very much and be prone to burning up. I.e. 2mA of feedback at 435V = 0.87W via at 220k resistor.
Then you'd need a smaller cathode resistor too and it would move the operating point. Anyone tried this?
Perhaps to linearise an SE pentode output tube without Schade feedback requires a two-stage feedback loop (driver + output tube) from the output plate back into the cathode of the driver?
I.e. if you had a ECC82 triode driving a GU50 pentode a reasonable feedback would be from GU50 anode to ECC82 cathode?
Unfortunately you'd need a capacitor I think or the resistor would have to be too high a value to feedback very much and be prone to burning up. I.e. 2mA of feedback at 435V = 0.87W via at 220k resistor.
Then you'd need a smaller cathode resistor too and it would move the operating point. Anyone tried this?
Hey Globu.....,
The point of using Schade is that you get triode performance with pentode effiency. We don´t want GFB as this usually degrades sound(my personal opinion).
So Schade is local feedback only over the output tube. On the downside is that it gets a little harder to drive it. For this to work we need a linear driver (not an ECC82).
In a two-stage amp we are also helped by distortion-cancellation. Probably we can expect less than 1% TDH at above 7W for the cascoded version above.
But back to linear triode-strapped pentodes. Theoretically I´m not impressed by what I´ve seen. EL156 and E130L that I both have in my stash does not look promising at all. The 813 I have looks great(curves to be found at Pete Millett) so I might have ago with them.
The point of using Schade is that you get triode performance with pentode effiency. We don´t want GFB as this usually degrades sound(my personal opinion).
So Schade is local feedback only over the output tube. On the downside is that it gets a little harder to drive it. For this to work we need a linear driver (not an ECC82
).In a two-stage amp we are also helped by distortion-cancellation. Probably we can expect less than 1% TDH at above 7W for the cascoded version above.
But back to linear triode-strapped pentodes. Theoretically I´m not impressed by what I´ve seen. EL156 and E130L that I both have in my stash does not look promising at all. The 813 I have looks great(curves to be found at Pete Millett) so I might have ago with them.
Last edited:
Hey Globu.....,
The point of using Schade is that you get triode performance with pentode effiency. We don´t want GFB as this usually degrades sound(my personal opinion).
So Schade is local feedback only over the output tube. On the downside is that it gets a little harder to drive it. For this to work we need a linear driver (not an ECC82
Ok - so the schade mixes in with the driver output impedance so I see that has to be constant to do that. I agree that GNFB is no good, do you know how much schade feedback changes the output impedance of the pentode? Is it lower than triode strapped?
I was considering a 6n6p UL cascode as a schade driver - any opinions?
Shade is a parallel feedback across an output tube. Exactly as an amp on an opamp in inverting connection. That means, it has the shortest FB signal path for lowest distortions in wider frequency band, lowest subtraction error, but decreases input resistance of an output stage significantly. So when you drive it by a stage that has linear transfer function, but non-linear output resistance (like in case of a cathode grounded triode) you are getting huge distortions as the result.
That's why we use pentodes (high linear resistance of the tube), or LTP pairs (high linear resistance of tubes), or cascode (the same reason). Such a way a feedback ratio is a ratio between resistance of feedback resistor (from anode to anode) and resistance of anode load resistor of the driver. CCS here does not work because output Z of the driver tube is back to the equation.
Another way is to use very low output Z driver and an additional resistor between driver's output and the point where input and feedback of an output stage meet.
Edit: I forgot to add that output pentode must have well regulated screen grid supply. I use particularly VR tubes that glow nice and are more temperature stable than Zener diodes. Like this one: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=180512428968
That's why we use pentodes (high linear resistance of the tube), or LTP pairs (high linear resistance of tubes), or cascode (the same reason). Such a way a feedback ratio is a ratio between resistance of feedback resistor (from anode to anode) and resistance of anode load resistor of the driver. CCS here does not work because output Z of the driver tube is back to the equation.
Another way is to use very low output Z driver and an additional resistor between driver's output and the point where input and feedback of an output stage meet.
Edit: I forgot to add that output pentode must have well regulated screen grid supply. I use particularly VR tubes that glow nice and are more temperature stable than Zener diodes. Like this one: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=180512428968
Last edited:
Yes of course I should have realised - as the output is always trying to push the input back, just like the opamp example.
Do you know how much schade feedback changes the output impedance of the pentode? In general is it lower than triode strapped?
I'm aiming for a low impedance output to drive the transformer without any GNFB (from the secondary or in fact anywhere). If required I'll put feedback around 2 stages (to lower the impedance enough) but I'd prefer if possible just local per-stage feedback.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.