I suspect he will be disinterested, as the circuit you posted does the same job as his Broskie Cathode Follower, except that your circuit has a differential gain of less than half (his version has a gain of exactly half), is very unbalanced, has a much lower input impedance, and uses many more parts. It is difficult to see how it would be more useful than the BCF.
I must disagree on a few points:
"is very unbalanced"... My CMRR is every bit as good, if not better.
Either of my inputs can tolerate being grounded without drastically
changing the differential gain, only one of Broskie's can...
"differential gain of less than half (his version has a gain of exactly half)"
Both gains are less than half, by same amount, and for the same reason.
I assume we speak of DWCF only, my circuit #70 sans the LTP up front...
"It is difficult to see how it would be more useful than the BCF"
DWCF PSRR is better. As BCF omits Aikido's famous hum cancellation.
He (Broskie) would have to add a whole nother stage just for that.
I suspect Broskie is smart enough, those features would not slip past
his notice. The real question, weather my email slips past his notice.
He claims over 6000 per day, I have no problem believing that figure...
"is very unbalanced"... My CMRR is every bit as good, if not better.
Either of my inputs can tolerate being grounded without drastically
changing the differential gain, only one of Broskie's can...
"differential gain of less than half (his version has a gain of exactly half)"
Both gains are less than half, by same amount, and for the same reason.
I assume we speak of DWCF only, my circuit #70 sans the LTP up front...
"It is difficult to see how it would be more useful than the BCF"
DWCF PSRR is better. As BCF omits Aikido's famous hum cancellation.
He (Broskie) would have to add a whole nother stage just for that.
I suspect Broskie is smart enough, those features would not slip past
his notice. The real question, weather my email slips past his notice.
He claims over 6000 per day, I have no problem believing that figure...
Last edited:
You are right about the gain, not sure how I screwed that one up. Anyway, the results of a quick SIM comparison using 12AU7s and 250V supply:
DWCF:
Differential Gain = 0.48
Inv. gain = 0.5
Non-inv. gain = 0.44
CMRR = 0.08
PSRR = 0.01
BCF
Gain = 0.45
Inv. gain = 0.45
Non-inv. gain = 0.45
CMRR = 0.006
PSRR = 0.05
The PSRR is better, but this is paid for by the large top-most resistor, which reduces the headroom. If a similar resistor and smoothing cap were added to the BCF, its overall PSRR would be better than the DWCF, and would still require fewer parts.
The CMRR is considerably worse however, at -22dB versus -44dB, and this is more or less the raison d'etre of both circuits. It can be improved a few dB by LED biasing the upper triode, but it still does not compete with the BCF.
The much lower input impedance also seems like a big handicap just to obtain better PSRR.
My circuit was not drop-in optimized for 12AU7, and would require
some empirical screwin'-round several resistor values to achieve
optimum results with a different triode. Increasing Gm is good and
fine, if the resistors are adjusted to match.
It is an error correcting design, and prefers a medium or high Mu.
Lessening Mu lessens amplification of main error correction loop by
square law. As the cascode gain is aproximately Mu*Mu, and the
error of the lower grid is amplified and fed back from the top plate.
Error at the upper grid is amplified to the top plate only by upper
triode's Mu alone, and this strongly affects the damping factor of
the output node. Lower Mu triode is not better for these reasons.
Broskie has an article on optimizing the feedback balance of the
single ended WCF. I don't yet fully understand his methods, nor
certain that it translates directly to optimum operation of DWCF.
Broskie's circuit requires full isolation of the inputs to float, aka
a transformer... DWCF tolerates ground referenced inputs and
does not require a transformer to insure that both inputs float.
Consider this fact when comparing parts count, cost, and bulk.
some empirical screwin'-round several resistor values to achieve
optimum results with a different triode. Increasing Gm is good and
fine, if the resistors are adjusted to match.
It is an error correcting design, and prefers a medium or high Mu.
Lessening Mu lessens amplification of main error correction loop by
square law. As the cascode gain is aproximately Mu*Mu, and the
error of the lower grid is amplified and fed back from the top plate.
Error at the upper grid is amplified to the top plate only by upper
triode's Mu alone, and this strongly affects the damping factor of
the output node. Lower Mu triode is not better for these reasons.
Broskie has an article on optimizing the feedback balance of the
single ended WCF. I don't yet fully understand his methods, nor
certain that it translates directly to optimum operation of DWCF.
Broskie's circuit requires full isolation of the inputs to float, aka
a transformer... DWCF tolerates ground referenced inputs and
does not require a transformer to insure that both inputs float.
Consider this fact when comparing parts count, cost, and bulk.
Last edited:
It's About Time&Ultra-Linear Line Stages
Broskie also points out that his input impedances are unequal.
To make them equal he suggests shunting across both inputs.
So much for DWCF's balanced 47K + 47K = 94K input impedance
being so much terribly horribly "lower" than BCF...
I'm totally unsure how you figure BCF inverting and non-inverting
input gains to be the same? The lower SE gain is <Mu The upper
SE gain is <unity. Only with both inputs floating fully isolated and
balanced, could BCF gain be 0.45? I simply don't have that issue.
I can couple LTP plates to my differential inputs with only caps
for isolation. Try that with BCF...
Broskie also points out that his input impedances are unequal.
To make them equal he suggests shunting across both inputs.
So much for DWCF's balanced 47K + 47K = 94K input impedance
being so much terribly horribly "lower" than BCF...
I'm totally unsure how you figure BCF inverting and non-inverting
input gains to be the same? The lower SE gain is <Mu The upper
SE gain is <unity. Only with both inputs floating fully isolated and
balanced, could BCF gain be 0.45? I simply don't have that issue.
I can couple LTP plates to my differential inputs with only caps
for isolation. Try that with BCF...
Last edited:
I missed plate follower nfeedback from BCF output to lower grid.
Its possible both gains are equal? And either could ground ref?
Though input impedances are still very unequal, easily fixed...
BCF might work better than I thought. LTP up front could work.
I think the main performance difference: BCF error correction is
unity for top triode, divided by Mu for the bottom. DWCF error
correction is multiplied by .5*Mu*(Mu-1) for bottom triode, and
multiplied .5*Mu for the top triode... I think thats the maths???
I might still have to cipher on them numbers to be sure...
Its possible both gains are equal? And either could ground ref?
Though input impedances are still very unequal, easily fixed...
BCF might work better than I thought. LTP up front could work.
I think the main performance difference: BCF error correction is
unity for top triode, divided by Mu for the bottom. DWCF error
correction is multiplied by .5*Mu*(Mu-1) for bottom triode, and
multiplied .5*Mu for the top triode... I think thats the maths???
I might still have to cipher on them numbers to be sure...
Last edited:
Hi,
The WHAT?
Surely you jest? This circuit was documented as one of the 'dead end's" from the 1980 in Allen Wrights TPCB (Tube Preamp Cook Book). He presented it as "Unitiy Gain Balanced to Single-ended Converter.
So if any appellation is needed, surely it Wright Cathode Follower, if having precedence means to apply one's own name to a circuit. Of course, Allen is not so vain to name his circuits after himself.
So perhaps we just use UGBSC from now on?
Also, both the upper and lower tube of this circuit may be made "Super Linear", so we'd have the SUGBSC...
Ciao T
The WHAT?
Surely you jest? This circuit was documented as one of the 'dead end's" from the 1980 in Allen Wrights TPCB (Tube Preamp Cook Book). He presented it as "Unitiy Gain Balanced to Single-ended Converter.
So if any appellation is needed, surely it Wright Cathode Follower, if having precedence means to apply one's own name to a circuit. Of course, Allen is not so vain to name his circuits after himself.
So perhaps we just use UGBSC from now on?
Also, both the upper and lower tube of this circuit may be made "Super Linear", so we'd have the SUGBSC...
Ciao T
OK, seems I've severely misunderestimated BCF/UGBSC!
Especially when assisted by an LTP for voltage gain.
Might look like similar parts count, but only cause
I've added two grid stoppers that were missing in
drawing #70. BCF at the moment, few less parts
than DWCF for a similar result.
We'll count again after I get PSR completely nulled.
Especially when assisted by an LTP for voltage gain.
Might look like similar parts count, but only cause
I've added two grid stoppers that were missing in
drawing #70. BCF at the moment, few less parts
than DWCF for a similar result.
We'll count again after I get PSR completely nulled.
Attachments
Last edited:
In http://www.diyaudio.com/forums/tubes-valves/171306-my-dc-darling.html I describe what I put together in the end.
What is that preamp configuration called, since there is no p&p, is it SRPP?
What is that preamp configuration called, since there is no p&p, is it SRPP?
Output to next stage the bottom of SRPP resistor, is complementary loaded, plate output single ended.
Output to next stage at top end of SRPP resistor is Mu Follower, CCS load, cathode follower single ended.
Output to next stage at midpoint of SRPP resistor is anti-compliment loaded, single ended push pull.
One has to figure cathode impedance into the SRPP resistor, so sometimes the midpoint is the top.
Mu Follwers usually have some extra resistance (beyond necessary purely for bias) to insure they
behave like CCS + cathode follower and not like anti-triode.
Your extra 14.3K in series with top triode complimentary load?
Output to next stage at top end of SRPP resistor is Mu Follower, CCS load, cathode follower single ended.
Output to next stage at midpoint of SRPP resistor is anti-compliment loaded, single ended push pull.
One has to figure cathode impedance into the SRPP resistor, so sometimes the midpoint is the top.
Mu Follwers usually have some extra resistance (beyond necessary purely for bias) to insure they
behave like CCS + cathode follower and not like anti-triode.
Your extra 14.3K in series with top triode complimentary load?
Last edited:
kenpeter, that is there to drop the voltage - I need around 150V to the 1626 grid to bias it correctly, so the preamp output plate has to be at that voltage. Without additional voltage dropping the preamp output would be exactly half of B+, so around 222V. Since B+ at 445 volts is pretty much all I can muster without changing all the PS caps (expensive) and the PT (even more expensive), I have to make with what I have.
If the preamp output were 222V, the 1626 cathode would need to be around 250V, and the 1626 anode at 510 - 520 volts.
If the preamp output were 222V, the 1626 cathode would need to be around 250V, and the 1626 anode at 510 - 520 volts.
I know what the resistor was there for. Just you ask what type SRPP is it?
I'm not sure... It seems a blend of compliment and resistive plate loadings.
Output at the bottom of the SRPP resistor, you won't get any active help
from the top triode, so I don't really consider it a push-pull in that sense...
I'm not sure... It seems a blend of compliment and resistive plate loadings.
Output at the bottom of the SRPP resistor, you won't get any active help
from the top triode, so I don't really consider it a push-pull in that sense...
Last edited:
Mu-followers are (asymmetric) push-pull. The signal pull-down current comes from the bottom tube, not a resistor. This enables a higher current drive on the -ve going swing. A very important difference for a lot of the direct coupled circuits I build.
Another important difference between a stacked mu-follower and a CCS+CF is that the anode of the bottom tube "sees" a fraction of the load roughly defined by the ratio of the a-k resistor and gm of the top tube. Adding extra resistance reduces this effect but it's still there. If the top device is a triode, some of it's rp will be felt by the bottom tube as well as an "anti-triode" (as Ken says above). Put a choke in place of the resistor and you have a closer approximation to CCS+CF in the midband.
Alas, I don't have Allen's book, but thanks for letting me know! (You wouldn't be willing to copy the relevant page for me, would you? Perhaps if we ask Allen first?)
However, I take issue with the name "Unitiy Gain Balanced to Single-ended Converter." because it is not unity gain! It has a gain of one half.
Incidentally, why call it 'super linear' rather than ultralinear, as it used to be called?
e.g.,
P. Read (1960) Ultralinear Cathode Follower. RSI, September, pp979-82.
Last edited:
Can I ask your opinions?
Been following this thread with interest, because I have been building my first SE amp and am getting ready to wire the driver stage after having finished the power supplies. Because of this thread, I have been second guessing my choice of input stage. I was going with a "complementary loaded, plate output single ended" stage, to use Kenpeter's terminology. Then I was thinking about a Mu-follower stage. Or maybe a pentode with a CCS. But I would love to hear your opinions about what I should use. The amp is going to have a 6W6 output tube. (with the intent of using it to drive my compression horn tweeters) I have (2) nine pin sockets for each channel of the input stage. B+ is 300V. F+ is 6.3 volts DC. The 9 pin tubes that I have in the "StashO'Tubes" (tm) box are:
404a/5847, 6DJ8, 6922, 6N1P, EL84, 6BM8, 12AX7, 12AT7, 12AU7, 6688, 6fq7/6cg7.
So if you had this amp sitting on your workbench, what input stage would you use in it? Any and all ideas or opinions will be appreciated.
Been following this thread with interest, because I have been building my first SE amp and am getting ready to wire the driver stage after having finished the power supplies. Because of this thread, I have been second guessing my choice of input stage. I was going with a "complementary loaded, plate output single ended" stage, to use Kenpeter's terminology. Then I was thinking about a Mu-follower stage. Or maybe a pentode with a CCS. But I would love to hear your opinions about what I should use. The amp is going to have a 6W6 output tube. (with the intent of using it to drive my compression horn tweeters) I have (2) nine pin sockets for each channel of the input stage. B+ is 300V. F+ is 6.3 volts DC. The 9 pin tubes that I have in the "StashO'Tubes" (tm) box are:
404a/5847, 6DJ8, 6922, 6N1P, EL84, 6BM8, 12AX7, 12AT7, 12AU7, 6688, 6fq7/6cg7.
So if you had this amp sitting on your workbench, what input stage would you use in it? Any and all ideas or opinions will be appreciated.
Some people will be happy to call it simply an SRPP, because it looks like one (and I suppose you could argue that the Miller capacitance of the next stage serves as the load, so it will be push-pull at very high frequencies).
However, I would prefer to call it simply an actively-loaded, common-cathode gain stage, and nothing more complicated than that!
I prefer the "complementary loaded, plate output single ended" stage, or "half-mu" amplifier as I call it. It's beautiful symmetry and thermal stability appeal to my aesthetic nature, even if they're not really necessary for most audio work.
In this situation the load is not that heavy (grid leak for the power valve), so you should ask yourself whether you really need a low output impedance from the driver stage. If you do really need it, then mu-follower is the obvious choice.
Obviously the next consideration is how much gain you need. The mu-follower will give you a gain of mu, while the half-mu amp will give you a gain of half-mu (surprise surprise).
A pentode with a CCS is likely to have extraordinarily high gain! This seems an unlikely option, unless you are planning some sort of negative feedback.
Thanks, Merlinb! I wasn't planning on using NFB, so maybe the pentode is out, although in an article by Thorsten L ("Some thoughts about singleended valve Amplifiers") he says he chose a pentode input stage, which is where I got the idea. I don't think a 6W6 is that hard to drive, so maybe I don't need low output impedance. It was just an idea I was playing with. I had started out with the intention of using a CL,POSE, but then this thread came along and I started second guessing. You know how it goes. Thanks again for your thoughts.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.