re: Zintolo
Unfortunately I don't have the R Fdbk info on those 6HB6 Schade curves anymore, but I would estimate that those were using high N Fdbk. Normal pentode Ri is 24000 Ohms. Internal triode Mu is around 33 with the triode Ri around 1650 Ohms. To get down to 200 Ohms Ri would have needed a Schade Mu of 4.
Another interesting option for Schade Fdbk is to use a higher Mu Vertical Amp dual triode like the 6FJ7. The 10 Watt triode there has a Mu of 15.4. If you drop that to around Mu 4 using Schade Fdbk, the triode curves improve to look more like a 300B. Will take more drive then of course, but the 300B has a Mu of 3.9 anyway.
Unfortunately I don't have the R Fdbk info on those 6HB6 Schade curves anymore, but I would estimate that those were using high N Fdbk. Normal pentode Ri is 24000 Ohms. Internal triode Mu is around 33 with the triode Ri around 1650 Ohms. To get down to 200 Ohms Ri would have needed a Schade Mu of 4.
Another interesting option for Schade Fdbk is to use a higher Mu Vertical Amp dual triode like the 6FJ7. The 10 Watt triode there has a Mu of 15.4. If you drop that to around Mu 4 using Schade Fdbk, the triode curves improve to look more like a 300B. Will take more drive then of course, but the 300B has a Mu of 3.9 anyway.
@smoking-amp
Thanks, looking at the datasheet (http://www.nj7p.org/Tubes/PDFs/Frank/135-GE/6HB6.pdf ) they need just 8Vpp to be driven in pentode mode (100 Ohm and 40 mA), so the driver requirement is mostly based on Schade amount of feedback (30% would mean to add 60Vp to the driver). But high Schade means also high increase of power due to the fact that the p-Mosfet is taking care of the extra voltage.
So first thing is to define a % of Schade, then draw a loadline based on that.
g2 has to be fixed compared to gnd or cathode? I remember I saw a schematic drawn by Wavebourn showing g2 referenced to cathode.
Thanks, looking at the datasheet (http://www.nj7p.org/Tubes/PDFs/Frank/135-GE/6HB6.pdf ) they need just 8Vpp to be driven in pentode mode (100 Ohm and 40 mA), so the driver requirement is mostly based on Schade amount of feedback (30% would mean to add 60Vp to the driver). But high Schade means also high increase of power due to the fact that the p-Mosfet is taking care of the extra voltage.
So first thing is to define a % of Schade, then draw a loadline based on that.
g2 has to be fixed compared to gnd or cathode? I remember I saw a schematic drawn by Wavebourn showing g2 referenced to cathode.
Question:
Reading through this whole discussion and several others like it, seems like this type of feedback, when applied A to G1 of a power tube, requires a high impedance at the grid, i.e. will not work with a driver triode because of low rp and Ra. That makes sense. So we can use a MOSFET, or interstage transformer as Schade did, or various other workarounds.
So how/why does it work fine in the RH84 with an ordinary 12AT7 or similar driver?
Reading through this whole discussion and several others like it, seems like this type of feedback, when applied A to G1 of a power tube, requires a high impedance at the grid, i.e. will not work with a driver triode because of low rp and Ra. That makes sense. So we can use a MOSFET, or interstage transformer as Schade did, or various other workarounds.
So how/why does it work fine in the RH84 with an ordinary 12AT7 or similar driver?
I used a Hammond 1650R, wired for 5k:8. Resistive losses in that transformer are pretty low so that helps.
Re: RH84, the distortion measurements I've seen on the web indicated it was a pretty high distortion amp. I'm not going to try to argue with taste, but let's just say "working fine" is in the eye of the beholder. Some people love it. To each his own.
Re: RH84, the distortion measurements I've seen on the web indicated it was a pretty high distortion amp. I'm not going to try to argue with taste, but let's just say "working fine" is in the eye of the beholder. Some people love it. To each his own.
The limit of the RH and the BH is the current of the driver: not an issue with EL84 and 6V6 that requires quite low lnfb, but can become harder to drive bigger bottles (with the BH I've found a compromise with 20% shade and 23% UL using KT88).
This kind of lnfb bypasses this problem and just need a driver with more swing into a high load, so a simpler task, and you can chose the driver you prefer in the configuration you prefer.
This kind of lnfb bypasses this problem and just need a driver with more swing into a high load, so a simpler task, and you can chose the driver you prefer in the configuration you prefer.
I think there is definitely some distortion cancellation in the RH84 amp, but it is still a >1% @ 1W amp. That's too much distortion for me.
I had to apply more feedback with an ideal (solid-state) V-to-I converter to get distortion down to 1% @ 1W with a 6L6 cousin. You definitely want a non-ideal V-to-I converter for these kinds of amps (one that generates some 2nd in opposite phase of the output tube).
I had to apply more feedback with an ideal (solid-state) V-to-I converter to get distortion down to 1% @ 1W with a 6L6 cousin. You definitely want a non-ideal V-to-I converter for these kinds of amps (one that generates some 2nd in opposite phase of the output tube).
Take a simple 2 stage single ended tube amplifier:
I think that the serial two stages can cause 2nd harmonic distortion cancellation; it works fine for a resistor load.
But, when a loudspeaker load is connected to the output stage, the load is elliptical.
Elliptical loads do not have a simple 2nd harmonic distortion pattern on the plate curves.
And that means the series two stage 2nd harmonic distortion cancellation no longer works.
That is because the first stage still sees a resistive load line; but the output stage sees an elliptical load line.
Just my un-studied opinion.
I think that the serial two stages can cause 2nd harmonic distortion cancellation; it works fine for a resistor load.
But, when a loudspeaker load is connected to the output stage, the load is elliptical.
Elliptical loads do not have a simple 2nd harmonic distortion pattern on the plate curves.
And that means the series two stage 2nd harmonic distortion cancellation no longer works.
That is because the first stage still sees a resistive load line; but the output stage sees an elliptical load line.
Just my un-studied opinion.
Many triode plate curves, with reasonably selected load lines, and that are not driven from near 0V to near cutoff (actually driven much less than that), have the following harmonic distortion characteristics:
Dominant 2nd harmonic
3rd harmonic
and all the higher order harmonics that fall off very rapidly (the higher the order, the further down they are relative to the fundamental, 2nd, and 3rd).
Rates of falling distortion levels referred to the fundamental, versus a 10dB decrease of fundamental amplitude:
2nd 10dB
3rd 20dB
4th 30dB
5th 40dB
6th 50dB
7th 60dB
and so forth.
The key is to use a triode that has enough power, so that you never use it near its maximum output power.
Try using a non-negative feedback triode amp, such as an 8 Watt amp at 2 Watts, or a 20 Watt amp at 5 Watts.
Your listening mileage may vary, especially depending on your speakers.
Do you want to reduce the effect of the elliptical load lines on the tube output stage . . .
Then try designing a good Class A push pull output stage.
Dominant 2nd harmonic
3rd harmonic
and all the higher order harmonics that fall off very rapidly (the higher the order, the further down they are relative to the fundamental, 2nd, and 3rd).
Rates of falling distortion levels referred to the fundamental, versus a 10dB decrease of fundamental amplitude:
2nd 10dB
3rd 20dB
4th 30dB
5th 40dB
6th 50dB
7th 60dB
and so forth.
The key is to use a triode that has enough power, so that you never use it near its maximum output power.
Try using a non-negative feedback triode amp, such as an 8 Watt amp at 2 Watts, or a 20 Watt amp at 5 Watts.
Your listening mileage may vary, especially depending on your speakers.
Do you want to reduce the effect of the elliptical load lines on the tube output stage . . .
Then try designing a good Class A push pull output stage.
Last edited:
Hi guys, I am back after many years off, I can't remember anything about valves... however I still enjoy them everyday.
I am posting this message in this thread to re-alive the subject of Cathode Feedback. The last time I did an experiment it was with UL + CF ... and I unfortunately reversed the 16-0 taps for the negative and positive phase, it resulted in a faint super distorted sound, I was shocked after listening in UL only and I left the amp for 7 years in the basement...
I am about to reverse the CF which is around 8 db , it has already UL also, and GNF around 10 db, it is driven by Roy Mottram of Tube4hifi PCB with 6sn7gt with the transistor sink.
(just a note , I think the transformers are not damaged by the operation... I connected a 16 ohm winding of the left tube with the anode current of the right tube, and vice versa
I am posting this message in this thread to re-alive the subject of Cathode Feedback. The last time I did an experiment it was with UL + CF ... and I unfortunately reversed the 16-0 taps for the negative and positive phase, it resulted in a faint super distorted sound, I was shocked after listening in UL only and I left the amp for 7 years in the basement...
I am about to reverse the CF which is around 8 db , it has already UL also, and GNF around 10 db, it is driven by Roy Mottram of Tube4hifi PCB with 6sn7gt with the transistor sink.
(just a note , I think the transformers are not damaged by the operation... I connected a 16 ohm winding of the left tube with the anode current of the right tube, and vice versa
Hi, and especially Spreadspect rum lol A comparison of the 1650R on KT120 versus various configurations on a CFB C core transformer. (note smaller hammonds extends more to 7hz, to 40khz at 10 watts, no problem)
(note all measurements no A-weighted)
Band -3 Db at 10 watts, on the 1650R 6hz-24.5khz, on the C-core : 10hz - 33khz.
I was able to obtain 0.1% thd+noise 0NFB on CFB+UL at 16khz, 15 Watts!!! , the 1650R makes 0.5% Thd with 0NFB
THD % Values for 15 Watts, 1khz signal, 1650R UL , 1650R UL 0NFB C-core UL, C-core-UL+CBF, C-Core UL+CFB 0NFB
0.162 - 0.471 - 0.076 - 0.348 - 0.341
Following this, many would suggest try the amp with CFB and Pendote with regulated screen, etc, however I greatly prefer the amp in UL mode, it tolerates well the feedback which makes this ideal sounding, the bass is fast, full of beat, the details are captivating, the sound is not fatiguing, in CFB the sound is not as good, you lose the explosiveness, it becomes overly controlled, maybe the high are airier but everything else is lacking, the difference is huge in the listening session, and UL is VERY satisfying, you just go quiet and marvels at the sound....
The transformer handle so well the GNF and the UL that the sound is not dragging like on other tube gears, nor does it sounds warm or hot like powerful solid state amplifiers. It sound extremely lively and present.
Same tests that I concluded that KT88 beats KT120 in the 50Watts region, or that EL34 is way better than KT77, and sounds way better at low volume than KT88s. The KT120 is a more reserved tube which lets you focus on the music and also can dig up some serious power.
(note all measurements no A-weighted)
Band -3 Db at 10 watts, on the 1650R 6hz-24.5khz, on the C-core : 10hz - 33khz.
I was able to obtain 0.1% thd+noise 0NFB on CFB+UL at 16khz, 15 Watts!!! , the 1650R makes 0.5% Thd with 0NFB
THD % Values for 15 Watts, 1khz signal, 1650R UL , 1650R UL 0NFB C-core UL, C-core-UL+CBF, C-Core UL+CFB 0NFB
0.162 - 0.471 - 0.076 - 0.348 - 0.341
Following this, many would suggest try the amp with CFB and Pendote with regulated screen, etc, however I greatly prefer the amp in UL mode, it tolerates well the feedback which makes this ideal sounding, the bass is fast, full of beat, the details are captivating, the sound is not fatiguing, in CFB the sound is not as good, you lose the explosiveness, it becomes overly controlled, maybe the high are airier but everything else is lacking, the difference is huge in the listening session, and UL is VERY satisfying, you just go quiet and marvels at the sound....
The transformer handle so well the GNF and the UL that the sound is not dragging like on other tube gears, nor does it sounds warm or hot like powerful solid state amplifiers. It sound extremely lively and present.
Same tests that I concluded that KT88 beats KT120 in the 50Watts region, or that EL34 is way better than KT77, and sounds way better at low volume than KT88s. The KT120 is a more reserved tube which lets you focus on the music and also can dig up some serious power.
Last edited:
Sorry for the late thank you, I missed your post here.
Is it ok just 10dB of decrease from harmonic to harmonic? I would have expected to be one order of magnitude (20dB).
Having dominant 2nd or 3rd and then the 4th 40dB below the dominant (falling as written above) make the dominant too much prominent, whilst a more homogeneous fall would mask better dominant harmonic(s)?
Thanks,
Roberto
Good point Roberto, I can respond (the Q is not addressed to me but ...)
that I observe that waterfall only when the triode is saturated, about to go on A2 and clipping. 2nd harmonics are canceled in the output stage, that leaves out only 3rd harmonics dominance in a PP class A, that is why if you use an underrated PP amp the sound will not be 'warm' , if you use a SET underpowered, there you have a 2nd dominant harmonics and close a 3rd , this adds warmth to the sound, a PP amp don't do that unless you purposely bias it incorrectly , then at around 5 ma you saturate your poor output transformer, resulting in a 'warm' nose sound with a lack of bass... SET would tend to play no bass at all on the wrong speakers or too much of it due to 2nd increase in the 'correct speaker'
SET = coloring machine, it almost doesn't matter at that point the quality of the source, everything will sound about the same, the better the amplifier, the more contrast you get, some records will sound bad, some will put a huge smile on your
To be honest, UL+CFB ... it has its pro, like on electronic music, it just sounds powerful, no bass hanging out at all, it is very very powerful too....... I will keep investigate.
that I observe that waterfall only when the triode is saturated, about to go on A2 and clipping. 2nd harmonics are canceled in the output stage, that leaves out only 3rd harmonics dominance in a PP class A, that is why if you use an underrated PP amp the sound will not be 'warm' , if you use a SET underpowered, there you have a 2nd dominant harmonics and close a 3rd , this adds warmth to the sound, a PP amp don't do that unless you purposely bias it incorrectly , then at around 5 ma you saturate your poor output transformer, resulting in a 'warm' nose sound with a lack of bass... SET would tend to play no bass at all on the wrong speakers or too much of it due to 2nd increase in the 'correct speaker'
SET = coloring machine, it almost doesn't matter at that point the quality of the source, everything will sound about the same, the better the amplifier, the more contrast you get, some records will sound bad, some will put a huge smile on your
To be honest, UL+CFB ... it has its pro, like on electronic music, it just sounds powerful, no bass hanging out at all, it is very very powerful too....... I will keep investigate.
zintolo,
Please do not forget that I was talking about a single triode.
The triode needs to be without negative feedback.
And not a push pull state with obvious known cancellations.
I preceded the chart with:
"Many triode plate curves, with reasonably selected load lines, and that are not driven from near 0V to near cutoff (actually driven much less than that), have the following harmonic distortion characteristics:
Dominant 2nd harmonic
3rd harmonic
and all the higher order harmonics that fall off very rapidly (the higher the order, the further down they are relative to the fundamental, 2nd, and 3rd)."
My chart of the harmonic distortion amplitude falloff rate can only be reasonably accurate under reasonable conditions.
But, in order for the chart to even be close . . .
1. You need to look at a single stage that has no negative feedback: no local negative feedback, and that stage can not be inside a global negative feed back loop.
No positive feedback, local or global loop around the stage.
Two or more stages, one after the other may experience some cancellations and/or additions as the dynamic levels change, so the rule is broken in that case too.
Push Pull, Balanced stages, etc, do cancel the 2nd harmonic, so the chart does not apply there either.
2. It must be in the well behaved dynamic region of a single stage:
No grid current (A2, AB2, B2 are not well behaved dynamic regions).
No clipping, it is not a well behaved dynamic region.
The harmonic level falloff chart is related to electronic measurements.
I hope that clears it up for you.
As to Masking, that is Physiological effect. I am not a hearing expert.
Once a playback system has been decided on and set up, then,
The sound that you hear . . .
is up to your ear.
Please do not forget that I was talking about a single triode.
The triode needs to be without negative feedback.
And not a push pull state with obvious known cancellations.
I preceded the chart with:
"Many triode plate curves, with reasonably selected load lines, and that are not driven from near 0V to near cutoff (actually driven much less than that), have the following harmonic distortion characteristics:
Dominant 2nd harmonic
3rd harmonic
and all the higher order harmonics that fall off very rapidly (the higher the order, the further down they are relative to the fundamental, 2nd, and 3rd)."
My chart of the harmonic distortion amplitude falloff rate can only be reasonably accurate under reasonable conditions.
But, in order for the chart to even be close . . .
1. You need to look at a single stage that has no negative feedback: no local negative feedback, and that stage can not be inside a global negative feed back loop.
No positive feedback, local or global loop around the stage.
Two or more stages, one after the other may experience some cancellations and/or additions as the dynamic levels change, so the rule is broken in that case too.
Push Pull, Balanced stages, etc, do cancel the 2nd harmonic, so the chart does not apply there either.
2. It must be in the well behaved dynamic region of a single stage:
No grid current (A2, AB2, B2 are not well behaved dynamic regions).
No clipping, it is not a well behaved dynamic region.
The harmonic level falloff chart is related to electronic measurements.
I hope that clears it up for you.
As to Masking, that is Physiological effect. I am not a hearing expert.
Once a playback system has been decided on and set up, then,
The sound that you hear . . .
is up to your ear.
Last edited: