The 21HB5 and 6HJ5 are at 50V/div horiz. and 50mA/div vert. The 38HE7 is a smaller tube and looks to be at 20V/div and 20mA/div.
Which the % of FB?
And in which way you have calculated it?
Last question : in which way you have get the curves ?
The standard circuit for Schade ( as Bartola) is a resistor from anode to g1 (directly), how the tester can handle this config? while the voltage on g1 will vary in negative voltage?
This question is also related to my Sofia tha hase three voltage generator, one of them is negative for g1
Probably with a cap for coupling?
I am working around Schade with tons of test.
I have opned a thread
Walter
Last edited:
The Tek 576 curve tracer here can cross continuously between + and - grid drive with a 20V overlap, It does have a button selection for primarily one polarity, up to 170V, + or - in the chosen direction. And, of course adjustable step ranging. (this 576 has been modified to handle up to +/- 200V thru the step amplifier. ) The step amplifier can also handle up to an amp either way, so works well for screen grid drive too. (can put a series floating supply in the step output to do higher grid2 V. ) There is a step current limit selector switch on the 576 too, normally used for SS devices, but helpful for protecting grids too. There is a separate Xantrex XT250 supply with current and V readout and I limiting for fixed screen V. It only outputs V2 when actually curve tracing.
For resistive Schade ( plate to grid1 ) Fdbk, I just use a high voltage step range thru a series resistor to simulate grid1 current drive. A large potentiometer is used for the Fdbk, so just gets adjusted for satisfactory curves. Usually comes out around the internal triode mu for best curves. Can use a cap isolator in the Fdbk, but can also just adjust the step drive offset V (to minus) to set the resulting tube grid1 bias.
For UnSet mode, similar except the step drive just drives the cathode directly. Plenty of step drive current available for operating the cathode.
Crazy drive is easy too, just bigger grid2 V steps with two extra grid2 to grid1 and grid1 to cathode resistors.
The Tek 576 has selectable plate V sweep ranges, I usually use the 400 (up to 450V max, via internal Variac ) range for most tubes. It does provide even a 1500V range, but I re-configured the rectifiers to change that to 750V max for safety.
There is also a selectible auxiliary input connector for Vertical display, which I have used with a Tek AM503 DC current probe for displaying screen current curves.
For resistive Schade ( plate to grid1 ) Fdbk, I just use a high voltage step range thru a series resistor to simulate grid1 current drive. A large potentiometer is used for the Fdbk, so just gets adjusted for satisfactory curves. Usually comes out around the internal triode mu for best curves. Can use a cap isolator in the Fdbk, but can also just adjust the step drive offset V (to minus) to set the resulting tube grid1 bias.
For UnSet mode, similar except the step drive just drives the cathode directly. Plenty of step drive current available for operating the cathode.
Crazy drive is easy too, just bigger grid2 V steps with two extra grid2 to grid1 and grid1 to cathode resistors.
The Tek 576 has selectable plate V sweep ranges, I usually use the 400 (up to 450V max, via internal Variac ) range for most tubes. It does provide even a 1500V range, but I re-configured the rectifiers to change that to 750V max for safety.
There is also a selectible auxiliary input connector for Vertical display, which I have used with a Tek AM503 DC current probe for displaying screen current curves.
Last edited:
Is it possible that correlation is not causation, at least to some degree?
That is, as the industry was moving away from linear triode output devices to higher-gain beam tetrode and pentode devices, it also happened to be moving away from directly to indirectly heated at the same time? Obviously there's exceptions for things like big transmitter triodes.
Unless it's just not possible to make a reasonable power, indirectly heated tube with gm and triode curves similar to the 2A3/300B. I guess my theory would be the differences are more about what they're optimizing for rather than a strict consequence of going indirectly heated.
Of course in practice, doesn't make any difference if you just can't get anything with curves as nice as a 300B that's indirectly heated, even if in theory someone could make one. It's not like they're going to at this point.
That is, as the industry was moving away from linear triode output devices to higher-gain beam tetrode and pentode devices, it also happened to be moving away from directly to indirectly heated at the same time? Obviously there's exceptions for things like big transmitter triodes.
Unless it's just not possible to make a reasonable power, indirectly heated tube with gm and triode curves similar to the 2A3/300B. I guess my theory would be the differences are more about what they're optimizing for rather than a strict consequence of going indirectly heated.
Of course in practice, doesn't make any difference if you just can't get anything with curves as nice as a 300B that's indirectly heated, even if in theory someone could make one. It's not like they're going to at this point.
Is the space between cathode and grid also important in IHTs?
I mean can we assume between preamp triodes like 12**7s, those which have lower gm are more linear?
I mean can we assume between preamp triodes like 12**7s, those which have lower gm are more linear?
The space between cathode and grid1 is very important for IHTs too, but it is the ratio of spacing to grid wire pitch that really matters. Low gm tubes often just use coarse grid pitch which makes them unlinear as well. The shape of the grid cross section should also match the cross section of the cathode.
For P-P or a differential stage the criteria are different. A differential stage over a CCS tail can operate well with tubes that have linear ramping gm (versus current ) since the gm of the two tubes will sum to a constant (in class A ). 12AT7 comes close for instance. hideous curves for SE, but OK for differential.
For P-P or a differential stage the criteria are different. A differential stage over a CCS tail can operate well with tubes that have linear ramping gm (versus current ) since the gm of the two tubes will sum to a constant (in class A ). 12AT7 comes close for instance. hideous curves for SE, but OK for differential.
I don't think it is that simple. If you look, for example, the EC86/PC86 is a very linear tube with pretty high gm (14mA/V).
George suggests 10% UnSet feedback, I’ve found it highly dependant on gm: EL84 and EL34 are fine with 10%, KT88 and GU50 prefers 20%, preamp tubes are fine with 1-2% only.
Main advantage of this configuration is that you can get some more Watts because of the source follower.
Any condideration without proper results by test lab on real circuit build are only personal
Walter
Years back, when I was testing out some Beam tubes in resistive Schade mode on the curve tracer, I generally posted a curve set for the internal Triode mode and then the resistive Schade curves. So as to compare apples to apples, I generally set the Fdbk R so as to get a similar curve set. This usually gave a pretty good set of Schade curves. But this may have just been serendipitous luck that the curve spacing usually came out well spaced.
I also noticed that too much resistive Schade Fdbk tended to make the HV end of the "triode" curves scrunch up closer and more vertical. This is opposite to the usual triode behavior, so it is not too surprising there would be some optimum, in between, setting.
That led me to realize that resistive Schade had a bug in it, but one that could be exploited. The R feedback is supposed to be injecting N Fdbk current proportional to output V into the grid. But the input grid V swing is corrupting the Fdbk current thru the resistor by changing the resultant V across the resistor. A lower gm tube needs more grid1 swing, so you get more corruption unless you make the resultant mu higher (higher R ) to counter that. Low gm also makes for a higher internal mu factor, so not too surprising there would be some mu correlation (or 1/gm) to get optimum "triode" curves.
When I mentioned the Fdbk corruption issue, George was onto a year long study of alternative drive schemes. He eventually came up with the cathode drive "UnSet" scheme, which separated the R Fdbk from the grid drive signal, now at the cathode. When he mentioned he had a new scheme that increased power output too (but he was still keeping the scheme secret at that point) I knew right away what he was onto. Took me 5 minutes to reconfigure the R Schade on the curve tracer, with cathode drive, to get the same results, which he acknowledged.
I had been experimenting with using a common gate, or grounded gate, MosFet stage in the R Fdbk path to turn the Fdbk current into a true current source, that would not care about grid1 V variations. George's cathode drive scheme was already on the agenda to try out, but not tried yet. Fortunately, the Tek 576 step output is capable of high current ( for SS transistors ... ) so it could easily drive the tube cathode. I just had to move one clip lead and add a grid bias resistor. And "UnSet" was running here too.
So it may be that the simple R Schade has the best potential for "triode" linearity, if you are happy with the resulting optimum mu factor.
George HAS spent a LOT of bench time on Schade Fdbk and UnSet. Lets give credit where credit is due.
What I would say is that the available tools can make for speeding things up considerably. 5 minutes on the curve tracer, versus a year of wire bending. The curve tracer was essential for finding optimized Crazy Drive quickly too. Testing g2 drive on TV tubes, g2 versions of Schade, matching tubes, paralleling tubes, tube emulation schemes....(you can make a hulking Sweep tube emulate any DHT, including little $3 1E7G or 1F5 tube models. See below: ) You DO want a curve tracer in this hobby.
1E7G:
I also noticed that too much resistive Schade Fdbk tended to make the HV end of the "triode" curves scrunch up closer and more vertical. This is opposite to the usual triode behavior, so it is not too surprising there would be some optimum, in between, setting.
That led me to realize that resistive Schade had a bug in it, but one that could be exploited. The R feedback is supposed to be injecting N Fdbk current proportional to output V into the grid. But the input grid V swing is corrupting the Fdbk current thru the resistor by changing the resultant V across the resistor. A lower gm tube needs more grid1 swing, so you get more corruption unless you make the resultant mu higher (higher R ) to counter that. Low gm also makes for a higher internal mu factor, so not too surprising there would be some mu correlation (or 1/gm) to get optimum "triode" curves.
When I mentioned the Fdbk corruption issue, George was onto a year long study of alternative drive schemes. He eventually came up with the cathode drive "UnSet" scheme, which separated the R Fdbk from the grid drive signal, now at the cathode. When he mentioned he had a new scheme that increased power output too (but he was still keeping the scheme secret at that point) I knew right away what he was onto. Took me 5 minutes to reconfigure the R Schade on the curve tracer, with cathode drive, to get the same results, which he acknowledged.
I had been experimenting with using a common gate, or grounded gate, MosFet stage in the R Fdbk path to turn the Fdbk current into a true current source, that would not care about grid1 V variations. George's cathode drive scheme was already on the agenda to try out, but not tried yet. Fortunately, the Tek 576 step output is capable of high current ( for SS transistors ... ) so it could easily drive the tube cathode. I just had to move one clip lead and add a grid bias resistor. And "UnSet" was running here too.
So it may be that the simple R Schade has the best potential for "triode" linearity, if you are happy with the resulting optimum mu factor.
George HAS spent a LOT of bench time on Schade Fdbk and UnSet. Lets give credit where credit is due.
What I would say is that the available tools can make for speeding things up considerably. 5 minutes on the curve tracer, versus a year of wire bending. The curve tracer was essential for finding optimized Crazy Drive quickly too. Testing g2 drive on TV tubes, g2 versions of Schade, matching tubes, paralleling tubes, tube emulation schemes....(you can make a hulking Sweep tube emulate any DHT, including little $3 1E7G or 1F5 tube models. See below: ) You DO want a curve tracer in this hobby.
1E7G:
Last edited:
Now I have a directory with around 40Mb of test around UL /scahde/pentode with different configuration and tested on real circuit. ( see my other thread.)
But I am not sure about the good setting of Schade feedback.
And I suspect is the worst way to get a good results for a power stage; too many variables on it and if you put a sand it will be another thing not a tube amp.
If Sofia will work I try also to get curves
But this is OT and it is my opinion
And I will read carefully your last post.
Walter
It’s a good and simple way that needs no SS hybridization, but you are new to the topic and you still need to understand key points and limits. You need a strong driver able to drive the low load seen through the Schade resistor.
I’m happy you looked at Bertola site I suggested you, as he investigated on g2 voltages and I haven’t seen many others work on that point.
@Wavebourn did a very good work on his site too, but most of it went lost few years ago.
you can help if post your test lab on real circuit.
About Bartola I haven't seen the results from lab. But maybe it is my error. In addition the circuit published have the sand so they can't called tube amp.
About strong driver I posted the circuit sfor test on other thread and I am workin on them.
Now for s.e. in near future with p-p
Walter
Hi Smoking-Amp
Do You maybe have measurements in triode mode of 1F5G (or 1F4) please?
Thanks 🙂
Sorry, I don't have those. But I see a note written in my GE tube handbook that says excellent triode for both of those. Someone else must have traced them online.
Revealed years ago:
The real question is how to ADD "magic" sound to IDHTs. I suggest using a 6" diam. PVC pipe with a microphone at one (sealed ) end (foam mounted ) and putting 1 inch holes regularly along the length (maybe 6 ft long, holes every 2 inches. ). Masking ( or Gorilla brand ) tape would be used to program which microphonics you want by plugging unwanted ones. Then the microphone gets a pot. attenuator and sums into the input of the amplifier (with a mic polarity switch ). Adjust for best sound. You might want to make the holes nearest to the mic smaller to keep the higher freqs. quieter. (masking tape partial hole blockings. )
Last edited: