Redd 47 tube selection -- an amp topology question

Shop deals on ebay
shop deals on amazon
Advertise with us
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Shattersignal

Shattersignal

Well-known member
Joined
Jul 14, 2012
Messages
114
Location
Left Coast, USA
Hi there!  I hate to start a new topic for every question, but I guess that is the custom on Group DIY.

Is triode matching critical for the V2 stage?  I have access to a few very fine used 6DJ8-type tubes and I'm not sure what my selection criteria should be.  I've read somewhere that the two triodes work in "parallel"--which seems to suggest that both triodes are single ended.  Does that mean that matching is critical or not critical?  If both triodes are single-ended, parallel, class-A, I assume that they shouldn't need to be perfectly matched.  Am I way off-base here? 

I know that this question involves a bit of theory that is beyond my current level of understanding--I'm still trying to learn some basics about tube amp typologies and I'd love a bit of guidance.

As a broader question, and for my general education, what amplifier typologies necessitate closely matched gain stages?  Am I correct in surmising that this is usually only needed for push-pull configured stages?
 
As far as I am aware, there is no need to match the triodes in the ECC88 of a REDD 47 amplifier. In any case, there are so many possible variables to consider, that matching is not really a serious proposition. The REDD 47 design is unique in that it is arranged such that the EF86 and ECC88 stages create roughly equal but opposite amounts of distortion which therefore to  a large extent cancel each other out. Of course, this happens when when tubes of that era are used. As this is a UK design it is likely that Mullard tubes were used. If you use modern equivalents, Russian substitutes or American equivalents, you may not get the same effect.

You are correct in your assumption that tube matching is only really relevant in push pull stages.

Cheers

Ian
 
Thanks a ton, Ian.  Excellent.  I am planning on using 1960s Philips tubes in V1 and V2.  One is French and one is JAN USA.  I guess I'll see how I like it and adjust as needed.  I suspect that they are both basically similar enough that matching tubes from the same factory would amount to a pretty minor difference.

Thanks!
 
ruffrecords said:
The REDD 47 design is unique in that it is arranged such that the EF86 and ECC88 stages create roughly equal but opposite amounts of distortion which therefore to  a large extent cancel each other out. Of course, this happens when when tubes of that era are used. As this is a UK design it is likely that Mullard tubes were used. If you use modern equivalents, Russian substitutes or American equivalents, you may not get the same effect.
Click to expand...

Could you elaborate, Ian,  on this some more? I think I have seen something a little like that recently, not sure what it might be.
THD measures are reasonably new to me.

Thanks :)
 
alexc said:
ruffrecords said:
The REDD 47 design is unique in that it is arranged such that the EF86 and ECC88 stages create roughly equal but opposite amounts of distortion which therefore to  a large extent cancel each other out. Of course, this happens when when tubes of that era are used. As this is a UK design it is likely that Mullard tubes were used. If you use modern equivalents, Russian substitutes or American equivalents, you may not get the same effect.
Click to expand...

Could you elaborate, Ian,  on this some more? I think I have seen something a little like that recently, not sure what it might be.
THD measures are reasonably new to me.

Thanks :)
Click to expand...

It is in the original REDD47 design document:

https://drive.google.com/file/d/0B_n67A1hN3qtSmlMMk12Yms4U2s/view?usp=sharing

Check out the Output Stage and Distortion sections on the second page.

Cheers

Ian
 
ruffrecords said:
As far as I am aware, there is no need to match the triodes in the ECC88 of a REDD 47 amplifier. In any case, there are so many possible variables to consider, that matching is not really a serious proposition. The REDD 47 design is unique in that it is arranged such that the EF86 and ECC88 stages create roughly equal but opposite amounts of distortion which therefore to  a large extent cancel each other out. Of course, this happens when when tubes of that era are used. As this is a UK design it is likely that Mullard tubes were used. If you use modern equivalents, Russian substitutes or American equivalents, you may not get the same effect.

You are correct in your assumption that tube matching is only really relevant in push pull stages.

Cheers

Ian
Click to expand...

Years ago when this preamps were still very rare we had a few discussions about mentioned canceling effect. Don't remember anyone measuring old or new tubes, might be because there were no new EF86 types produced at that time. Just went through your "6CG7 Mu Follower Distortion" paper, where distortion was measured with wave analyser and 2nd harmonic listed.
You wrote tubes from the same manufacturer were very consistent with noticable difference between different manufacturers.
Do you maybe remember how big were differences in harmonics between different manufacturers? It would be very interesting to know if newly produced tubes had approximately the same relation between 2nd and 3rd harmonic distortion as old ones, although it is about different tube than used in Redd47. I remember Telefunken used one of Eastern Europe's company to provide them with EF86s (think Svetlana, can check better if needed), this could give at least some more clues since not all have access to Redd47 type preamp and wave analyzer.
 
My3gger said:
Just went through your "6CG7 Mu Follower Distortion" paper, where distortion was measured with wave analyser and 2nd harmonic listed.
You wrote tubes from the same manufacturer were very consistent with noticable difference between different manufacturers.
Do you maybe remember how big were differences in harmonics between different manufacturers? It would be very interesting to know if newly produced tubes had approximately the same relation between 2nd and 3rd harmonic distortion as old ones, although it is about different tube than used in Redd47. I remember Telefunken used one of Eastern Europe's company to provide them with EF86s (think Svetlana, can check better if needed), this could give at least some more clues since not all have access to Redd47 type preamp and wave analyzer.
Click to expand...

I have just checked through my original hand written results book from 2009. Unfortunately I only listed 2H because I had already concluded that, in general 3H was 20dB or more lower in all case. However, there is one note in the book that says "the tube with the poorest 2H performance has the lowest 3J reading". I have noticed this several times in the past when checking tubes. I am not certain if it is a genuine difference between tubes or the effect of approaching clipping.  If you take a triode gradually up to clipping and watch the 2H and 3H levels you will find that 2H starts to go down and 3H goes up proportionally. Thereafter all the other harmonics come in and it is difficult to see any pattern to what is going on. Now, when I made the distortion tests, it was at 20V rms at 2KHz into a 10K load. For some samples of 6CG7, that level requires an input signal not far away from grid current and grid current distortion also first manifests itself as a reduction in 2H and an increase in 3H. I later did some further tests where I varied the load. I only did this with one tube but it had the typical 6CG7 distortion at 10K load. The distortion was at a minimum with a 15K load and 10dB better than at 10K. Higher and lower loads increased the distortion.

So, my only conclusion is that, at the levels used, there could be several mechanisms beginning to affect the relationship between 2H and 3H levels. At normal levels these effects will not be present.

I just looked further (later) in the book and found the result of the original 6922 SRPP tests I did when I began work on what became the Eurochannel. An SRPP produces a lot more distortion than a mu follower. The tests were conducted with 10V rms into a 5K1 load. This lists harmonics up to 5H where present and shows some interesting differences between different manufacturers. I can scan and post it if you like.

Cheers

Ian
 
ruffrecords said:
My3gger said:
Just went through your "6CG7 Mu Follower Distortion" paper, where distortion was measured with wave analyser and 2nd harmonic listed.
You wrote tubes from the same manufacturer were very consistent with noticable difference between different manufacturers.
Do you maybe remember how big were differences in harmonics between different manufacturers? It would be very interesting to know if newly produced tubes had approximately the same relation between 2nd and 3rd harmonic distortion as old ones, although it is about different tube than used in Redd47. I remember Telefunken used one of Eastern Europe's company to provide them with EF86s (think Svetlana, can check better if needed), this could give at least some more clues since not all have access to Redd47 type preamp and wave analyzer.
Click to expand...

I have just checked through my original hand written results book from 2009. Unfortunately I only listed 2H because I had already concluded that, in general 3H was 20dB or more lower in all case. However, there is one note in the book that says "the tube with the poorest 2H performance has the lowest 3J reading". I have noticed this several times in the past when checking tubes. I am not certain if it is a genuine difference between tubes or the effect of approaching clipping.  If you take a triode gradually up to clipping and watch the 2H and 3H levels you will find that 2H starts to go down and 3H goes up proportionally. Thereafter all the other harmonics come in and it is difficult to see any pattern to what is going on. Now, when I made the distortion tests, it was at 20V rms at 2KHz into a 10K load. For some samples of 6CG7, that level requires an input signal not far away from grid current and grid current distortion also first manifests itself as a reduction in 2H and an increase in 3H. I later did some further tests where I varied the load. I only did this with one tube but it had the typical 6CG7 distortion at 10K load. The distortion was at a minimum with a 15K load and 10dB better than at 10K. Higher and lower loads increased the distortion.

So, my only conclusion is that, at the levels used, there could be several mechanisms beginning to affect the relationship between 2H and 3H levels. At normal levels these effects will not be present.

I just looked further (later) in the book and found the result of the original 6922 SRPP tests I did when I began work on what became the Eurochannel. An SRPP produces a lot more distortion than a mu follower. The tests were conducted with 10V rms into a 5K1 load. This lists harmonics up to 5H where present and shows some interesting differences between different manufacturers. I can scan and post it if you like.

Cheers

Ian
Click to expand...

Hi,

very interesting observations, i compared input and output stage of your Eurocard to Redd.47.
Reply to my older question regarding Eurochannel having only 1/10th distortion in input ECC83, compared to E88CC SRPP output. At lower signal levels input tube gives only 1% of all distortion.
Redd.47 design document suggests input tube having a lot more distortion compared to yours, it seems local nfb around E88CC doesn't lower it enough to compare to your results, global nfb only lowers 3rd harmonic. Seems like both stages with local and global nfb get distortion aproximately the same. Could it be that design documents of EMI don't give enough information to figure out how much distortion each stage has?
Russell O. Hamm's paper compares, triode, pentode and transistor harmonics when distorted. I find it very interesting because it shows higher harmonics, although it kind of only complements what we are writing here:
http://www.dwfearn.com/tubes_vs_transistors.html
Yes, please post notes about E88CC SRPP tests, i would be happy to save them with other information provided by you. It is pretty hard to get this kind of information, i'm sure it will be helpful to the others too.
 
My3gger said:
Hi,

very interesting observations, i compared input and output stage of your Eurocard to Redd.47.
Reply to my older question regarding Eurochannel having only 1/10th distortion in input ECC83, compared to E88CC SRPP output. At lower signal levels input tube gives only 1% of all distortion.
Redd.47 design document suggests input tube having a lot more distortion compared to yours, it seems local nfb around E88CC doesn't lower it enough to compare to your results, global nfb only lowers 3rd harmonic. Seems like both stages with local and global nfb get distortion aproximately the same. Could it be that design documents of EMI don't give enough information to figure out how much distortion each stage has?
Russell O. Hamm's paper compares, triode, pentode and transistor harmonics when distorted. I find it very interesting because it shows higher harmonics, although it kind of only complements what we are writing here:
http://www.dwfearn.com/tubes_vs_transistors.html
Yes, please post notes about E88CC SRPP tests, i would be happy to save them with other information provided by you. It is pretty hard to get this kind of information, i'm sure it will be helpful to the others too.
Click to expand...

There are two important difference between the REDD 47 and my Eurochannel:

1. The REDD47 uses a pentode input stage; the Eurochannel has a triode input stage. The spectrum of triode and pentode distortion is different.

2. The Eurochannel output stage has a gain of about 28dB. Therefore the output stage gain dominates the overall open loop distortion. The REDD47 output stage has local feedback that sets its gain at about 10dB. The input stage distortion therefore plays a larger part in the overall open loop distortion.

The RDD47 EF86 stage  probably has 40dB  or more gain and its output level is at least 10dB higher than that of the Eurochannel first stage. Pentodes tend to distort more than triodes so it is not too hard to see that this EF86 stage distortion could be as high that of its output stage. I wills can my old test results and post them here.

Cheers

Ian
 
ruffrecords said:
My3gger said:
Hi,

very interesting observations, i compared input and output stage of your Eurocard to Redd.47.
Reply to my older question regarding Eurochannel having only 1/10th distortion in input ECC83, compared to E88CC SRPP output. At lower signal levels input tube gives only 1% of all distortion.
Redd.47 design document suggests input tube having a lot more distortion compared to yours, it seems local nfb around E88CC doesn't lower it enough to compare to your results, global nfb only lowers 3rd harmonic. Seems like both stages with local and global nfb get distortion aproximately the same. Could it be that design documents of EMI don't give enough information to figure out how much distortion each stage has?
Russell O. Hamm's paper compares, triode, pentode and transistor harmonics when distorted. I find it very interesting because it shows higher harmonics, although it kind of only complements what we are writing here:
http://www.dwfearn.com/tubes_vs_transistors.html
Yes, please post notes about E88CC SRPP tests, i would be happy to save them with other information provided by you. It is pretty hard to get this kind of information, i'm sure it will be helpful to the others too.
Click to expand...

There are two important difference between the REDD 47 and my Eurochannel:

1. The REDD47 uses a pentode input stage; the Eurochannel has a triode input stage. The spectrum of triode and pentode distortion is different.

2. The Eurochannel output stage has a gain of about 28dB. Therefore the output stage gain dominates the overall open loop distortion. The REDD47 output stage has local feedback that sets its gain at about 10dB. The input stage distortion therefore plays a larger part in the overall open loop distortion.

The RDD47 EF86 stage  probably has 40dB  or more gain and its output level is at least 10dB higher than that of the Eurochannel first stage. Pentodes tend to distort more than triodes so it is not too hard to see that this EF86 stage distortion could be as high that of its output stage. I wills can my old test results and post them here.

Cheers

Ian
Click to expand...
ruffrecords said:
I just added a Theory folder to the DIY tab of my web site http://www.customtubeconsoles.com/diy

The only item in it right now is SRPPdistortion1.png which is a copy of the original SRPP distortion tests I did back in 2011.

Cheers

ian
Click to expand...

Thanks for posting SRPP distortion tests and comments about differences between two designs.
Theory folder on your website is very good idea, i think others would benefit from Grounding 101 paper if it was included, despite being under "Power" too. It helped me understand things like pot groundings a lot better, i don't need to experiment as much as before to get really low noise. Since it was posted most of my projects (mostly with tubes) got a few dB lower noise, they are quieter than many tube or solid state gear from respected manufacturers.

I made some calculations about Redd47 NFB; because of // triodes in output stage open loop gain seems to be around 28,5dB, so with 15dB of local NFB it comes to around 13,5dB.
If this is correct, global NFB is variable from 8-20dB, although i get best sound at up to 14dB of NFB. 34dB position never sounded as good as the other two, i might take away that one when exact copy is not needed.
SRPP distortion tests are different because JJ E88CC is included, a friend and i had problems with this one because it seems first batches had problems with anode burning out when used like real E88CC. Never tried it again,  NOS Tesla E88CC and EH6922 didn't have this problem. People seem to use 6N1P in this circuit, to me it sounds like having too much distortion, guess idea of using it came from Drip who proposed it as "replacement" which is cleary not. Your tests show more higher harmonics than E88CC/ECC88.

For practice and better understanding started calculating things about your Eurochannel, didn't post results like NFB value because it is not clear to me if it is the same as Euroboard without EQ. Is it the one with ECC83 input and SRPP E88CC output with NFB switch between stages many schematics show?
Thanks again for scan and explanations, it is of great help and makes me happy when numbers fall together so simple :)
 
I agree about JJ E88CC. I find they are not as consistent as the 6922EH. I bought some JJs recently and they were all noisy.

The Eurochannel is not easy to analyse. This is the best circuit to look at:

https://drive.google.com/file/d/0B_n67A1hN3qtelg2R21Xa0dIbVk/view?usp=sharing

One problem with negative feedback is ensuring  it is always negative. because there are frequency sensitive components in the circuit (mostly capacitors), the phase of the negative feedback can change with frequency. At some frequency the phase will become 180 degrees and the negative feedback becomes positive and overshoot or worse, oscillation, will occur.  The only way to ensure stability is to make sure that the loop gain is less one at this point and also that when the loop gain is one the phase shift is a lot less than 180 degrees.

It is not too hard to do this for one value of loop gain and hence one value of closed loop gain, which is why most of the classic tube preamps only  have one one or two close together gain settings. For a given open loop gain, for high closed loop gains there is little NFB so the loop gain is quite low and it is easy to ensure stability. As you try to reduce gain, by altering the NFB, the loop gain increases so it is harder to ensure stability.

One way to combat this is to automatically vary the open loop gain as the NFB is varied with the objective of keeping the loop gain nearly constant. This is what happens in the Eurochannel. The variable resistor that sets the NFB and hence the closed loop gain, also sets the gain of the ECC83 stage. The tracking is not perfect but the loop gain varies by only about 10dB as the closed loop gain varies from 6 to 40dB.

I have added Grounding 101 to the theory folder and also the Scroggie paper on power supply filtering. If there is anything else you think could usefully be in the Theory folder let me know.

Cheers

Ian
 

Similar threads

Dr Tony Balls
Redd 47 Type Build Troubleshooting
Replies
5
Views
2K
rock soderstrom
rock soderstrom
A
Vacuum Tube Extension Rig
2
Replies
20
Views
1K
Cqwet Dbdfte
Cqwet Dbdfte
P
v76m circuit questions
Replies
6
Views
419
ruffrecords
ruffrecords
A
Anode follower as +/- trim stage
Replies
7
Views
573
atavacron
A
Z
My journey of building a tube amp (12AX7 and EL84 SE)
2 3
Replies
54
Views
3K
zeh
Z

Latest posts

Back
Top