SEEEERIOUS DUDE
Mr. Turner is a serious dude, obviously...
Lots of calculus and a mix of overly complicated and overly simple. Anyway, as I understood, he did not do it, either! The man is a commercial source, for Christ sake!
Anyway, to him "shunt feedback" is nice for armchair diyers, not serious diyers who inhale smokes. Whatever that means.
Last but not least, his site is new, most was prepared before 2006, though
Mr. Turner is a serious dude, obviously...
Lots of calculus and a mix of overly complicated and overly simple. Anyway, as I understood, he did not do it, either! The man is a commercial source, for Christ sake!
Anyway, to him "shunt feedback" is nice for armchair diyers, not serious diyers who inhale smokes. Whatever that means.
Last but not least, his site is new, most was prepared before 2006, though
Mr. Kitic,
I am sorry I don´t get what you want to express in regard to Mr. Turner. Mr. Turner is actively sharing his knowledge with DIYers for more than ten years now (f.e, rec.audio.tubes).
So, what is your point, exactly? What do you think is wrong in the link I gave, exactly?
Thank you for clarifying,
Tom Schlangen
I am sorry I don´t get what you want to express in regard to Mr. Turner. Mr. Turner is actively sharing his knowledge with DIYers for more than ten years now (f.e, rec.audio.tubes).
So, what is your point, exactly? What do you think is wrong in the link I gave, exactly?
Thank you for clarifying,
Tom Schlangen
Re: COMMENTS
OK, you used to have a lot of solid credibility in my eyes, but with this, it has all flown out the window.
I fail to understand what is so problematic in the statement that in general 'overloaded' triode stages exhibit more high order distortion, to warant such posts. How does that impact how well RH84 amps are received by the public? A techincal argument is one thing, and sonics more often than not another. 'Try it and see' is NOT a technical argument, for it to be a technical argument, one has to provide somewhat solid explanations so people have incentive to try it at their own cost. While at it, your simulations show FFT measurements with a linear amplitude scale. Having spent hundreds uppon hundreds of hours with Circuitmaker (and other simms), this sort of diagram shows nothing useful - it should be on a dB scale.
Be that as it may, it is obvious that the choice of input tube (pentode, and especially triode) heavily influences the non-linearity of the driver, but it should be noted that in that position, the driver will exhibit 'complementary nonlinearity' with regard to the output tube. If higher harmonics can be kept low and the driver and output tube are a good match, the total THD will be lower and may still retain a very pleasing harmonic distribution.
With regards to Baby Huey, I have looked at resulting loadlines for the ECC83/12AX7. If the driver section itself was to be kept low distortion (but as i said above, one might want to exploit distortion cancelation), it seems to me the operating points fall into a region that will produce rather high THD from the driver. The resulting load line is very steep and passes through the part of the plate curves where the distances between the curves get considerably 'squashed' as plate voltage goes up.
It seems apparent the plate current should be kept quite a bit higher to push the operating point away from this region. In particular, keeping the plate voltage low (100-something volts) and plate current high (~3mA), look around the -0.5V grid potential line, and you get to a region where gm is quite linear and so is mu (draw horizontal and vertical line through operating point and look at intersections with adjecent grid potential lines).
The only problem here is that it is a very atypical operating point for an ECC83, but then so is the intended use. The unknown here is grid current at -0.5V on the grid... unfortunately there is nothing about this in the datasheet. Does anyone have some insight?
Alex Kitic said:
OK, you used to have a lot of solid credibility in my eyes, but with this, it has all flown out the window.
I fail to understand what is so problematic in the statement that in general 'overloaded' triode stages exhibit more high order distortion, to warant such posts. How does that impact how well RH84 amps are received by the public? A techincal argument is one thing, and sonics more often than not another. 'Try it and see' is NOT a technical argument, for it to be a technical argument, one has to provide somewhat solid explanations so people have incentive to try it at their own cost. While at it, your simulations show FFT measurements with a linear amplitude scale. Having spent hundreds uppon hundreds of hours with Circuitmaker (and other simms), this sort of diagram shows nothing useful - it should be on a dB scale.
Be that as it may, it is obvious that the choice of input tube (pentode, and especially triode) heavily influences the non-linearity of the driver, but it should be noted that in that position, the driver will exhibit 'complementary nonlinearity' with regard to the output tube. If higher harmonics can be kept low and the driver and output tube are a good match, the total THD will be lower and may still retain a very pleasing harmonic distribution.
With regards to Baby Huey, I have looked at resulting loadlines for the ECC83/12AX7. If the driver section itself was to be kept low distortion (but as i said above, one might want to exploit distortion cancelation), it seems to me the operating points fall into a region that will produce rather high THD from the driver. The resulting load line is very steep and passes through the part of the plate curves where the distances between the curves get considerably 'squashed' as plate voltage goes up.
It seems apparent the plate current should be kept quite a bit higher to push the operating point away from this region. In particular, keeping the plate voltage low (100-something volts) and plate current high (~3mA), look around the -0.5V grid potential line, and you get to a region where gm is quite linear and so is mu (draw horizontal and vertical line through operating point and look at intersections with adjecent grid potential lines).
The only problem here is that it is a very atypical operating point for an ECC83, but then so is the intended use. The unknown here is grid current at -0.5V on the grid... unfortunately there is nothing about this in the datasheet. Does anyone have some insight?
Re: Re: COMMENTS
Hi ilimzn,
Hmmmpf ... the usual (historical) definition of "onset of grid current" mostly is clamped either to Ig1 >= 0,3µA or to >= 0,5µA going. As I understand these 0,3µA or 0,5µA were about the lowest currents easily measurable using a plain current meter (shunted galvanometer).
According to several German spec sheets for ECC83, those 0,3µA of grid current are regarded as a limiting value when occuring at Eg1 = -1,3V already.
A most simple and practical setup/method for measuring actual grid current can be found in the Telefunken Laborbuch, 2. Ausgabe, 1955, Seite 210. Some time ago I described this method in English language over at the Geek Zone.
Regards,
Tom Schlangen
Hi ilimzn,
Hmmmpf ... the usual (historical) definition of "onset of grid current" mostly is clamped either to Ig1 >= 0,3µA or to >= 0,5µA going. As I understand these 0,3µA or 0,5µA were about the lowest currents easily measurable using a plain current meter (shunted galvanometer).
According to several German spec sheets for ECC83, those 0,3µA of grid current are regarded as a limiting value when occuring at Eg1 = -1,3V already.
A most simple and practical setup/method for measuring actual grid current can be found in the Telefunken Laborbuch, 2. Ausgabe, 1955, Seite 210. Some time ago I described this method in English language over at the Geek Zone.
Regards,
Tom Schlangen
Re: Re: COMMENTS
Yes, 2nd order distortion always cancels when using differential structure !
In doubt, just measure :
http://www.dissident-audio.com/AutoIndex/index.php?dir=PP_ECL86/&file=DA36.pdf
Yves.
ilimzn said:
Yes, 2nd order distortion always cancels when using differential structure !
In doubt, just measure :
http://www.dissident-audio.com/AutoIndex/index.php?dir=PP_ECL86/&file=DA36.pdf
Yves.
This Is Not Rocket Science
Applying parallel NFB is not that big a deal, nor is it so difficult. Yes, applying parallel feedback will increase the load the driver sees in exactly the same way that Miller Effect does. I've used parallel NFB, as 807s tend to sound really nasty without it, and the developer of the type, O. Schade, recommends it. In particular, Schade recommended feeding back 10% of the plate AC. So how does that work?
Po= 26.5W (Class AB1, from spec sheet)
V= (26.5 X 6600)^0.5= 418.21Vrms
V= 418.21 X sqrt(2)= 591.44Vp-p= 295.72Vp
Vfb= 0.1(295.72)= 29.56Vp
Since I decided to "jump over" the 807s so as to avoid the Crt and run the local NFB to the cathode follower grid drivers, that means:
Vfb`= 29.56 / 0.94= 31.45Vp
That added to the original 22.5Vp of input gives:
22.5 + 31.45= 53.95Vp at the grid of the cathode follower, since the voltage is in anti-phase. Combined with the 1.0M grid DC return, that gives:
53.95 / 1E6= 53.95uA
Since the input from the driver is still 22.5Vp, that grid resistor looks like:
22.5 / 53.95E-6= 417K
The LTP plate loads are 220K, and so the effective AC load looks like:
220K || 417K= 144K
So use that value for the AC loadline, and see what you come up with. Av= 50; THD= 4.58% (est) That THD figure may not look so swell at first, however, it is under the 5.0% THD criterion that the VT spec sheets use all the time in their recommended example circuits. Secondly, this is estimated based on the full output swing of 150Vp-p, and we're using just 45Vp-p of that, so it's actually more like 1.37%. Additionally, this graphical method tends to overestimate the THD anyway, and this is part of an LTP splitter. As a balanced circuit, it eliminates all even harmonics, so that the THD will be even lower.
Well, guess what? After building this project, it worked just like it was supposed to. Even with Sovtek 6SL7s in a circuit designed against RCA plate characteristics, I got a measured gain of 25 from the LTP. Any distortion at the grids of the 807s was so tiny that it disappeared below the noise floor while o'scoping. Compared to open loop listening, the local NFB cleaned up that pentode nastiness, and all that was required was 6db(v) of gNFB to mop up an overly aggressive sound, and to tighten up the bass.
Parallel local NFB works just fine.
6SL7 Loadline
Applying parallel NFB is not that big a deal, nor is it so difficult. Yes, applying parallel feedback will increase the load the driver sees in exactly the same way that Miller Effect does. I've used parallel NFB, as 807s tend to sound really nasty without it, and the developer of the type, O. Schade, recommends it. In particular, Schade recommended feeding back 10% of the plate AC. So how does that work?
Po= 26.5W (Class AB1, from spec sheet)
V= (26.5 X 6600)^0.5= 418.21Vrms
V= 418.21 X sqrt(2)= 591.44Vp-p= 295.72Vp
Vfb= 0.1(295.72)= 29.56Vp
Since I decided to "jump over" the 807s so as to avoid the Crt and run the local NFB to the cathode follower grid drivers, that means:
Vfb`= 29.56 / 0.94= 31.45Vp
That added to the original 22.5Vp of input gives:
22.5 + 31.45= 53.95Vp at the grid of the cathode follower, since the voltage is in anti-phase. Combined with the 1.0M grid DC return, that gives:
53.95 / 1E6= 53.95uA
Since the input from the driver is still 22.5Vp, that grid resistor looks like:
22.5 / 53.95E-6= 417K
The LTP plate loads are 220K, and so the effective AC load looks like:
220K || 417K= 144K
So use that value for the AC loadline, and see what you come up with. Av= 50; THD= 4.58% (est) That THD figure may not look so swell at first, however, it is under the 5.0% THD criterion that the VT spec sheets use all the time in their recommended example circuits. Secondly, this is estimated based on the full output swing of 150Vp-p, and we're using just 45Vp-p of that, so it's actually more like 1.37%. Additionally, this graphical method tends to overestimate the THD anyway, and this is part of an LTP splitter. As a balanced circuit, it eliminates all even harmonics, so that the THD will be even lower.
Well, guess what? After building this project, it worked just like it was supposed to. Even with Sovtek 6SL7s in a circuit designed against RCA plate characteristics, I got a measured gain of 25 from the LTP. Any distortion at the grids of the 807s was so tiny that it disappeared below the noise floor while o'scoping. Compared to open loop listening, the local NFB cleaned up that pentode nastiness, and all that was required was 6db(v) of gNFB to mop up an overly aggressive sound, and to tighten up the bass.
Parallel local NFB works just fine.
6SL7 Loadline
Tom,
The Turner site is an interesting read on shunt plate feedback, and it appears to have its technical aspects well in order. The Mu followers eliminating the loading problems on the driver. I was disappointed however by Patrick T's final So-So prognosis...., without any explanation. I see that he generally uses CFB outputs for output stage linearization, so this would seem to be the "competition" here.
Since Patrick winds custom CFB xfmrs, I have to wonder if there could be a little bias in this conclusion. So I do find myself agreeing with Alex on the possible commercial issue there. Miles has found the shunt feedback scheme to work as advertised when optimised correctly. And Ian is a fan too. So my guess is that local shunt NFB is, like any NFB, implementation dependent .
--------------------------
A little OT, but still of interest to driver stage design:
Patricks following circuit analysis, on balanced screen feedbacks to the driver, also ends in a So-So prognosis. I believe the problem there is the use of plate bootstraps for the screen DC screen bias. This mixes feedback with driver output on the screens, and compromises the gain of the pentode by loading as well. I believe screen feedbacks can be made to "fly" when well implemented. Patrick does suggest further work needed.
And Patrick's Error correction amplifier design, analyzed subsequently, has a subtle design error in it's implementation. (the calculated error term has to be inserted back in BEFORE the input signal sampling point to be considered EC, putting it in AFTER, ends up as just conventional NFDBK. ) Considering that many 100's of pages of argued details, surrounding EC, occur on the SS forum, getting EC wrong the 1st time is not that uncommon, and I do salute Patrick on bravely tackling this subject in tube design. A little more work is needed though.
And YES, measure when in doubt! ( most any cheap 16 bit sound card is a quantum leap over nothing)
Don
The Turner site is an interesting read on shunt plate feedback, and it appears to have its technical aspects well in order. The Mu followers eliminating the loading problems on the driver. I was disappointed however by Patrick T's final So-So prognosis...., without any explanation. I see that he generally uses CFB outputs for output stage linearization, so this would seem to be the "competition" here.
Since Patrick winds custom CFB xfmrs, I have to wonder if there could be a little bias in this conclusion. So I do find myself agreeing with Alex on the possible commercial issue there. Miles has found the shunt feedback scheme to work as advertised when optimised correctly. And Ian is a fan too. So my guess is that local shunt NFB is, like any NFB, implementation dependent .
--------------------------
A little OT, but still of interest to driver stage design:
Patricks following circuit analysis, on balanced screen feedbacks to the driver, also ends in a So-So prognosis. I believe the problem there is the use of plate bootstraps for the screen DC screen bias. This mixes feedback with driver output on the screens, and compromises the gain of the pentode by loading as well. I believe screen feedbacks can be made to "fly" when well implemented. Patrick does suggest further work needed.
And Patrick's Error correction amplifier design, analyzed subsequently, has a subtle design error in it's implementation. (the calculated error term has to be inserted back in BEFORE the input signal sampling point to be considered EC, putting it in AFTER, ends up as just conventional NFDBK. ) Considering that many 100's of pages of argued details, surrounding EC, occur on the SS forum, getting EC wrong the 1st time is not that uncommon, and I do salute Patrick on bravely tackling this subject in tube design. A little more work is needed though.
And YES, measure when in doubt! ( most any cheap 16 bit sound card is a quantum leap over nothing)
Don
Here is an article about a 6V6 PP amp using symmetrical shunt FB to the splitter/driver stage by Mr. Jean Hiraga published in Revue du son et du home cinema, no. 252, Mai 2001.
Regards,
Tom Schlangen
Regards,
Tom Schlangen
Hi Don,
I think everyone who thinks he knows even the slightest bit more than nothing is "biased" somehow, or else there hardly would be a point for discussion at all
BTW, myself pointing to Mr. Turners and Mr. Hiragas articles on PP/SFB amps doesn´t mean I am biased towards Mr. Turner and/or Mr. Hiraga. I just wanted to throw in some additional existing and already published stuff as fodder for those who are interested in this topic: At least Mr. Turner supplies some theory along a working circuit, and Mr. Hiraga supplies a published real world design dating back about 8 years or something.
There hardly is something really new under the sun regarding tube tech
Regards,
Tom Schlangen
I think everyone who thinks he knows even the slightest bit more than nothing is "biased" somehow, or else there hardly would be a point for discussion at all
BTW, myself pointing to Mr. Turners and Mr. Hiragas articles on PP/SFB amps doesn´t mean I am biased towards Mr. Turner and/or Mr. Hiraga. I just wanted to throw in some additional existing and already published stuff as fodder for those who are interested in this topic: At least Mr. Turner supplies some theory along a working circuit, and Mr. Hiraga supplies a published real world design dating back about 8 years or something.
There hardly is something really new under the sun regarding tube tech
Regards,
Tom Schlangen
Hi Tom,
The Hiraga 6V6 shunt P to P design would seem rather short on current for the 6SL7's to operate on. But then this is similar to the Baby Huey scheme, with the feedback resistors providing the driver operating currents. Being P-P, I would presume the circa 1% distortion figure given is mostly odd harmonic. The 6SL7's (Mu 70) would have high output impedance, especially at such low op. currents, to sum with the feedback currents. But the 470K feedbacks probably aren't feeding back much signal either. I would be concerned with Rp modulation, with the triode drivers, causing large driver distortion. But, this may be largely 2nd H, which ends up canceling in the outout stage. Still, 1% odd harmonics would not generally be considered HiFi. Not a design I would bother with.
On Patrick T's shunt feedback example, the So-So results concluded, could be due to the drivers being pentodes operating open loop, which is not so clean. This appears to be a problem with shunt P to P feedback, the driver prefers to be a pentode unless some kind of buffer is added. The CFB approach, on the other hand, achieves the same result without any restrictions on the drivers. But you need an often custom xfmr.
This is why I like the driver screen feedback idea. Triode driver operation in effect, pentode high loop gain outputs. No effect on the driver g1 input Z. Screen voltages proportionally track the plates to give better screen input Z. But does require a very sharp cutoff pentode to avoid gm2/gm1 Mu distortion at low current.
Don
The Hiraga 6V6 shunt P to P design would seem rather short on current for the 6SL7's to operate on. But then this is similar to the Baby Huey scheme, with the feedback resistors providing the driver operating currents. Being P-P, I would presume the circa 1% distortion figure given is mostly odd harmonic. The 6SL7's (Mu 70) would have high output impedance, especially at such low op. currents, to sum with the feedback currents. But the 470K feedbacks probably aren't feeding back much signal either. I would be concerned with Rp modulation, with the triode drivers, causing large driver distortion. But, this may be largely 2nd H, which ends up canceling in the outout stage. Still, 1% odd harmonics would not generally be considered HiFi. Not a design I would bother with.
On Patrick T's shunt feedback example, the So-So results concluded, could be due to the drivers being pentodes operating open loop, which is not so clean. This appears to be a problem with shunt P to P feedback, the driver prefers to be a pentode unless some kind of buffer is added. The CFB approach, on the other hand, achieves the same result without any restrictions on the drivers. But you need an often custom xfmr.
This is why I like the driver screen feedback idea. Triode driver operation in effect, pentode high loop gain outputs. No effect on the driver g1 input Z. Screen voltages proportionally track the plates to give better screen input Z. But does require a very sharp cutoff pentode to avoid gm2/gm1 Mu distortion at low current.
Don
It occurs to me that the shunt plate feedback to a triode driver configuration could be patched up, if it could be run at constant current. Then its plate Z would be constant, eliminating driver distortion (well, most), and loading on the triode.
This scheme (among others) came up earlier in another thread when Rich Walters mentioned mods to the dePalma splitter. Not everyone will be happy with SS in the circuit, but its function here is to act as a "near" high Z current source load for the triode driver below. The triode still sets the voltage output and Zout.
The active CCS is adjusted so as to maintain as near as possible a constant current thru the triode AFTER all other effects are combined (feedback, output tube g1 load). The cathode resistor could be used to sense current thru the tube (remove jumper) while one of the gain affecting resistors around the Mosfet "CCS" are adjusted for a constant current or best null condition. If the null turns out to be darn close to constant current, then maybe the cathode bypass cap can just be eliminated.
Can think of this approach as relieving the triode of linear currents, but still generating error correction currents when its plate gets tugged off linear by feedback errors. These small reactions have next to no effect on plate Zout compared to the linear current holding constant Zout.
Don
This scheme (among others) came up earlier in another thread when Rich Walters mentioned mods to the dePalma splitter. Not everyone will be happy with SS in the circuit, but its function here is to act as a "near" high Z current source load for the triode driver below. The triode still sets the voltage output and Zout.
The active CCS is adjusted so as to maintain as near as possible a constant current thru the triode AFTER all other effects are combined (feedback, output tube g1 load). The cathode resistor could be used to sense current thru the tube (remove jumper) while one of the gain affecting resistors around the Mosfet "CCS" are adjusted for a constant current or best null condition. If the null turns out to be darn close to constant current, then maybe the cathode bypass cap can just be eliminated.
Can think of this approach as relieving the triode of linear currents, but still generating error correction currents when its plate gets tugged off linear by feedback errors. These small reactions have next to no effect on plate Zout compared to the linear current holding constant Zout.
Don
Attachments
Guys,
I did quite a bit of listening to the fixed bias (latest) version of the Baby Huey on the weekend and some comparative listening against a top of the line SS Amp (a 55W version of the Hugh Dean "Maya" amp, same as the Maya but with only one pair of output transistors).
One thing the BH does really well is time accuracy - I could rave about 3D imaging benefits etc. BUT in fact this shows up as a "easy, effortless loping pace". Counter intuitively this aspect of the amp improved when I moved the shunt feedback pick offs from the output tube anodes to the ultralinear taps thus including part of the output transformer in the loop). The amp seems to "breath in time with the music". I feel this is due to the very low levels of feedback and the short loops.
On "busy" music it is perhaps a little muddy indicating that a little more feedback may help BUT I've found that that impacts the "spot on" pace and rhythm of the amp.
My future testing plans.
Before trying Sveins ideas as illustrated and modeled by ReVintage (Lars) I'm going to do one thing - Bootstrap the Source Follower (the SLCF treatment) and up its running current a bit.
Cheers,
Ian.
I did quite a bit of listening to the fixed bias (latest) version of the Baby Huey on the weekend and some comparative listening against a top of the line SS Amp (a 55W version of the Hugh Dean "Maya" amp, same as the Maya but with only one pair of output transistors).
One thing the BH does really well is time accuracy - I could rave about 3D imaging benefits etc. BUT in fact this shows up as a "easy, effortless loping pace". Counter intuitively this aspect of the amp improved when I moved the shunt feedback pick offs from the output tube anodes to the ultralinear taps thus including part of the output transformer in the loop). The amp seems to "breath in time with the music". I feel this is due to the very low levels of feedback and the short loops.
On "busy" music it is perhaps a little muddy indicating that a little more feedback may help BUT I've found that that impacts the "spot on" pace and rhythm of the amp.
My future testing plans.
Before trying Sveins ideas as illustrated and modeled by ReVintage (Lars) I'm going to do one thing - Bootstrap the Source Follower (the SLCF treatment) and up its running current a bit.
Cheers,
Ian.
Here is an all tube re-work of the triode shunt driver idea presented in post 751. The pentode would be a hi Zout tube. It's gain is adjusted (Rset) until the current thru the triode remains constant or as near as possible to constant (with shunt feedback and output tube g1 loading present). The triode then maintains a constant Zout and operates linearly under near constant current.
The triode controls the voltage output, since the pentode appears as a high Z current source. Any error feedback thru Rfdbk is seen by the triode as deviation from its required plate voltage, allowing the triode to correct by adjusting its current. The triode therefore only varies its current a tiny amount in response to detected errors, maintaining an essentially constant Zout.
Don
The triode controls the voltage output, since the pentode appears as a high Z current source. Any error feedback thru Rfdbk is seen by the triode as deviation from its required plate voltage, allowing the triode to correct by adjusting its current. The triode therefore only varies its current a tiny amount in response to detected errors, maintaining an essentially constant Zout.
Don
Attachments
LAST TIME, NO USE, OBVIOUSLY!
Dear ilimzn, person without name, I frankly could not care less about the credibility I have in your eyes.
The FFT analysis is not important in this case. The total THD is the only thing that matter, since in SE amps odd harmonics do not play an important role (they are low or statistically unimportant). It might play some role in the PP version of the amp, but yet again we all know that distortions in PP amps will be prevalently odd harmonics. One more time, it is the total value that matters, at least to make a point about something.
Please explain how could we induce enough higher order harmonics in a tube amp, be it SE or PP, just to deny that the above quote is mumbo jumbo for people who are not quite shure what are they talking about?!
I sincerely cannot believe that I am invited to explain why the white wall is actually white since everyone has already been explained that white is actually yellowish.
I do not want to participate in this endless armchair discussion, I do not have the time and do not want to let others "finish it". All those who would like to try the RH84 or any of the other amps in a PP configuration, I invite to write to my email address and I will help them as time and my obviously "limited" knowledge permit.
Best regards to all, but those who express negative opinions hidden by a pseudonym.
Čoveče, koga boli uvo šta ti misliš o meni, odnosno šta si mislio? Ko tebe uopšte zna? Imaš li ti ime?
Dear ilimzn, person without name, I frankly could not care less about the credibility I have in your eyes.
The FFT analysis is not important in this case. The total THD is the only thing that matter, since in SE amps odd harmonics do not play an important role (they are low or statistically unimportant). It might play some role in the PP version of the amp, but yet again we all know that distortions in PP amps will be prevalently odd harmonics. One more time, it is the total value that matters, at least to make a point about something.
Please explain how could we induce enough higher order harmonics in a tube amp, be it SE or PP, just to deny that the above quote is mumbo jumbo for people who are not quite shure what are they talking about?!
I sincerely cannot believe that I am invited to explain why the white wall is actually white since everyone has already been explained that white is actually yellowish.
I do not want to participate in this endless armchair discussion, I do not have the time and do not want to let others "finish it". All those who would like to try the RH84 or any of the other amps in a PP configuration, I invite to write to my email address and I will help them as time and my obviously "limited" knowledge permit.
Best regards to all, but those who express negative opinions hidden by a pseudonym.
Čoveče, koga boli uvo šta ti misliš o meni, odnosno šta si mislio? Ko tebe uopšte zna? Imaš li ti ime?
I would like to make clear that I have NOT dissed the sound of the RH84 or it's derivatives at any point in the preceding discussions that I am aware of. I am puzzled by Alex's continued paranoia on this. And I will state again that it may very well be a good sounding amplifier. I have not heard it myself, but many others do seem to like it.
But since Alex continues to put up this huge smoke screen about actual measurements, I will make some comments. It's amazing how many SET designers are absolutely allergic to FFT measurements. It's like holding up a silver cross in a graveyard, the demons head for the exits. Now I understand the usual complaints against THD numbers (they are meaningless without a spectrum), but here Alex is pushing THD as the only meaningful measurement (and from a simulation no less, with linear scaling that hides higher harmonics)!!!
When I set up my PC to do FFT measurements some time ago, I found them to be an absolute revelation in terms of understanding an amplifier's "sound" in relation to technical design aspects. This is such a low cost upgrade to a PC (free software on the Web even), I cannot understand why any amplifier designer would make do without it.
Looking at the RH84 and RH34, I see 12AT7 and 12AU7 used as the driver tubes. The 12AT7 is listed as an RF mixer tube on the GE data sheet. This is a dead giveaway that a tube has high distortion, mixers don't even work unless they are non-linear. 12AU7 probably speaks for itself. Now it may be that the 12AT7 (or 12AU7) is close to square law, like a Mosfet, due to heavy g1 "island effect", this would work well for a mixer, and would largely produce only 2nd harmonic as a triode driver with heavy loading. The large un-bypassed driver cathode resistor used also helps linearize operation.
So, without measurement data available, I would guess that the driver is producing a fair amount of 2nd harmonic with a little extra 3rd harmonic. The 2nd harmonic is anti-phase to the 2nd harmonic produced by the output tube. So some 2 H cancellation is occuring. The resultant should be an amplifier with reduced 2nd harmonic and slightly increased 3rd harmonic. Now from previous FFT experience, I would guess that this produces an amplifier with some "punch" to its sound, as long as the 3rd is not too high. This may well sound pleasing to some listeners on some types of music.
Was that so painful? Why all the Voodoo cover-up? Real measurements would of course be more illuminating. They might even tell a different story, but we will never know until someone actually makes them.
Don
But since Alex continues to put up this huge smoke screen about actual measurements, I will make some comments. It's amazing how many SET designers are absolutely allergic to FFT measurements. It's like holding up a silver cross in a graveyard, the demons head for the exits. Now I understand the usual complaints against THD numbers (they are meaningless without a spectrum), but here Alex is pushing THD as the only meaningful measurement (and from a simulation no less, with linear scaling that hides higher harmonics)!!!
When I set up my PC to do FFT measurements some time ago, I found them to be an absolute revelation in terms of understanding an amplifier's "sound" in relation to technical design aspects. This is such a low cost upgrade to a PC (free software on the Web even), I cannot understand why any amplifier designer would make do without it.
Looking at the RH84 and RH34, I see 12AT7 and 12AU7 used as the driver tubes. The 12AT7 is listed as an RF mixer tube on the GE data sheet. This is a dead giveaway that a tube has high distortion, mixers don't even work unless they are non-linear. 12AU7 probably speaks for itself. Now it may be that the 12AT7 (or 12AU7) is close to square law, like a Mosfet, due to heavy g1 "island effect", this would work well for a mixer, and would largely produce only 2nd harmonic as a triode driver with heavy loading. The large un-bypassed driver cathode resistor used also helps linearize operation.
So, without measurement data available, I would guess that the driver is producing a fair amount of 2nd harmonic with a little extra 3rd harmonic. The 2nd harmonic is anti-phase to the 2nd harmonic produced by the output tube. So some 2 H cancellation is occuring. The resultant should be an amplifier with reduced 2nd harmonic and slightly increased 3rd harmonic. Now from previous FFT experience, I would guess that this produces an amplifier with some "punch" to its sound, as long as the 3rd is not too high. This may well sound pleasing to some listeners on some types of music.
Was that so painful? Why all the Voodoo cover-up? Real measurements would of course be more illuminating. They might even tell a different story, but we will never know until someone actually makes them.
Don
CHEAP FFT ANALYSERS
I can imagine: free software seems nice, as well as your very audiophile very precise and very pricey sound card with highly precise and free of distortion analogue input, not to mention the other details related to measuring distortion with already distorting instruments.
Have you ever tried a real FFT analyser? If not, stick to total THD, because your cheap sound card will not be able to tell you anything about sound. Everyone has cheap sound cards, by the way...
1) It is only 12AT7 and not 12AU7, pay more attention next time you read someones schematics.
2) Your comments on the choice of tubes is despicable. But it shows how well you know your tubes.
You do not know what you are talking about... try building one, than use your FFT PC sound card, and publish the results, instead. Or a sim, for that matter.
You mean, previous FFT cheap sound card experience?
You read correctly the part about the punch, but the solid bass is not a consequence of slightly higher 3rd harmonics or some mumbo jumbo, but plain damping factor, increased by adopting this form of feedback.
It's amazing how many SET (we are talking here about SE pentode, if you have missed it until now, so it should be more like SEP) amplifiers (or SEP amps) you have built so far and tested on your state of the art FFT analyser.
Your comments may sound pleasing to some listeners on some types of forums.
Yes, like pain in the b...
Good on 'ya, mate! Where did you learn the language? Did you notice most of you have no names -- and mine is for real (Aleksandar, actually, but that is the same). I do not need to hide, but do you? This is not a sex and cheating related forum, for Christ sake!
I can imagine: free software seems nice, as well as your very audiophile very precise and very pricey sound card with highly precise and free of distortion analogue input, not to mention the other details related to measuring distortion with already distorting instruments.
Have you ever tried a real FFT analyser? If not, stick to total THD, because your cheap sound card will not be able to tell you anything about sound. Everyone has cheap sound cards, by the way...
1) It is only 12AT7 and not 12AU7, pay more attention next time you read someones schematics.
2) Your comments on the choice of tubes is despicable. But it shows how well you know your tubes.
You do not know what you are talking about... try building one, than use your FFT PC sound card, and publish the results, instead. Or a sim, for that matter.
You mean, previous FFT cheap sound card experience?
You read correctly the part about the punch, but the solid bass is not a consequence of slightly higher 3rd harmonics or some mumbo jumbo, but plain damping factor, increased by adopting this form of feedback.
It's amazing how many SET (we are talking here about SE pentode, if you have missed it until now, so it should be more like SEP) amplifiers (or SEP amps) you have built so far and tested on your state of the art FFT analyser.
Your comments may sound pleasing to some listeners on some types of forums.
Yes, like pain in the b...
Good on 'ya, mate! Where did you learn the language? Did you notice most of you have no names -- and mine is for real (Aleksandar, actually, but that is the same). I do not need to hide, but do you? This is not a sex and cheating related forum, for Christ sake!
OT: As RH84 is the subject discussed, with its plate to plate local NFB, this is what we can expect from it in theory:
Ca 3W into 8ohm (maybe a little more)
Zout just above 1ohm
Output at 1%THD under 2W
-1db at 20Hz
2nd and 3rd harmonic being near each other above 2W
Below 2W 2nd harmonic is clearly dominating
As Don says there is distortion cancelling taking place as THD from ECC81 will be ca 5% when total THD at output is ca 1% at 2W.
Anyway, I think it will sound quite good with high sensitivity speakers due to the absence of total NFB around the transformer and only local NFB over each stage.
Ca 3W into 8ohm (maybe a little more)
Zout just above 1ohm
Output at 1%THD under 2W
-1db at 20Hz
2nd and 3rd harmonic being near each other above 2W
Below 2W 2nd harmonic is clearly dominating
As Don says there is distortion cancelling taking place as THD from ECC81 will be ca 5% when total THD at output is ca 1% at 2W.
Anyway, I think it will sound quite good with high sensitivity speakers due to the absence of total NFB around the transformer and only local NFB over each stage.
Mr. Alex
You have some points but the aggressive style doesn't match here.
Maybe you have had some mails "beside" this thread? Otherwise I don't have a clue why you have this anger. Divorce going on or ?
"endless armchair discussion"
Yes, this is nomal at an open forum like this. I also like to act rather then just read and discuss, but you have to take it as it is around here. There is not just one "right" opinion.
Take the goodies and do leave some comments about the things you don't like, but don't take it as a contest for the next Nobel prize.
Thats my two cents, and I know I'm not alone to think and also behave in that manner around here.
This is an open forum for DIY's, the experienced and the noobs. Take out the raisin's and leave the bad manner at home please.
You have some points but the aggressive style doesn't match here.
Maybe you have had some mails "beside" this thread? Otherwise I don't have a clue why you have this anger. Divorce going on or ?
"endless armchair discussion"
Yes, this is nomal at an open forum like this. I also like to act rather then just read and discuss, but you have to take it as it is around here. There is not just one "right" opinion.
Take the goodies and do leave some comments about the things you don't like, but don't take it as a contest for the next Nobel prize.
Thats my two cents, and I know I'm not alone to think and also behave in that manner around here.
This is an open forum for DIY's, the experienced and the noobs. Take out the raisin's and leave the bad manner at home please.
Distortion is lower, and power is higher than above mentioned.
Distortion cancellation has very little or nothing to do with the results, it is all about the plate to plate feedback acting upon several factors in a way different than nfb from output to input does.
That is also the reason for improved sound: sounds very much as if in triode mode while retaining the power of the pentode.
To DIYers that is good enough as explanation. They try and are satisfied. The "tube scientists" do not try anything, have no ideas of their own but like seeming smart and "learned". That is why I oppose them wholeheartedly.
Distortion cancellation has very little or nothing to do with the results, it is all about the plate to plate feedback acting upon several factors in a way different than nfb from output to input does.
That is also the reason for improved sound: sounds very much as if in triode mode while retaining the power of the pentode.
To DIYers that is good enough as explanation. They try and are satisfied. The "tube scientists" do not try anything, have no ideas of their own but like seeming smart and "learned". That is why I oppose them wholeheartedly.