Trying to learn more about phase splitters from this forum and the web (so far I have only tried simple split-load and stock Mullard long-tailed pair). It appears that even though the isodyne topology gets lots of positive remarks (can be tuned to balance very well, etc.), it is seldom used in diy circuits. One of the "cons" from the search is that "tubes (6an8 in the original E.F. Worthen article) need to be closely matched ". Are there other well known problems with this approach that makes it inferior to the others and not competitive? I like the fact that there are adjustable "knob"s for optimization. (Sorry if this topic had been discussed in length in the past, but very little turned out during my last search.)
TIA.
TIA.
Actually, I see the feature you mention, that this circuit needs to be "fine tuned", I see as a disadvantage. I perfer to design circuits that don't need fine tuning. Any circuit that needs to be fine tuned will eventually get out of tune as parts and tubes age unevenly. Must of the manufacturing industry perfers circuits that can be assembled and work "out of the box" without adjustments. A circuit that gives high performance no matter what is the best.
I've tried Worthens type phase splitter with other tube types and despite burning much midnight oil, I got worse results (thd) when compared to the simple cathodyne. I never liked the idea of matched tubes in input areas.
There is a raft of triode/pent tubes around which make good cathodyne user potential at decent prices. The beauty with the basic cathodyne is a far more reliable phase splitter solution which functions well with most med mu triode/pent combinations. Setting up the pentode screen grid for lowest thd is another hassle but whatever the design one is faced with something to adjust. If one puts a push-pull voltage driver after the cathodyne (i.e Wiliamson amps) then the impedance mismatch from "upper and lower" outputs really comes to nothing. The trick is to set a stage up with tight layout and look at the common cathode......the 2freq symmetry will show how good the balance as freq is increased upwards and I find it behaves well at a far lower THD than anyother configuration.
Even with LTP I'm not keen using this for HiFi, but for MI is the norm.
The real drawback with the Worthens is a much higher input volts and potty adjustment.
Bin it.
richj
There is a raft of triode/pent tubes around which make good cathodyne user potential at decent prices. The beauty with the basic cathodyne is a far more reliable phase splitter solution which functions well with most med mu triode/pent combinations. Setting up the pentode screen grid for lowest thd is another hassle but whatever the design one is faced with something to adjust. If one puts a push-pull voltage driver after the cathodyne (i.e Wiliamson amps) then the impedance mismatch from "upper and lower" outputs really comes to nothing. The trick is to set a stage up with tight layout and look at the common cathode......the 2freq symmetry will show how good the balance as freq is increased upwards and I find it behaves well at a far lower THD than anyother configuration.
Even with LTP I'm not keen using this for HiFi, but for MI is the norm.
The real drawback with the Worthens is a much higher input volts and potty adjustment.
Bin it.
richj
2004ex said:
I found an article about this phase splitter, and it sure does look overhyped. This thingy isn't much more than taking one voltage amp, connecting it to a cathode follower, and sending the signal back to another voltage amp for another 180deg inversion. It's still two stages in series, and like the more common "See Saw" and its variants, it'll have uneven harmonic distortion.
An LTP with a good CCS in the tail will work better, isn't so complex, nor does it require the alignment that this thingy does.
I don't see how that got a patent.
I agree with others' comments. I see the Isodyne as an unnecessarily fussy design for the job it does. It's basically an attempt to make the see-saw splitter behave well despite its inherent drawbacks. I cannot imagine that the end result would be any better than the two most popular and much simpler tube-based splitters the others have mentioned.
Addendum: I think the same criticism probably applies to the Van Scoyoc cross-coupled splitter, which wastes two triodes and also requires tuning.
All of the foregoing ignores transformer splitters, which are a different animal altogether.
Addendum: I think the same criticism probably applies to the Van Scoyoc cross-coupled splitter, which wastes two triodes and also requires tuning.
All of the foregoing ignores transformer splitters, which are a different animal altogether.
Hi all ,
The CATHODYNE or SPLIT-LOAD or CONCERTINA
PHASE SPLITTER , no matter the designation , is the
most simple , the most functional and the most reliable
tube circuit ever made .
The opinions have been divided about it , but the exce-
llent results obtained with it , is unquestionable .
Bruce Rozenblit says : “ With all of its drawbacks , the
simple phase splitter is quite useful , and I have succes-
fully implemented it in several designs. Many applications
can easy tolerate the wide variation in output impedance
between the two signals”
Morgan Jones says : “ Although the Concertina phase
splitter does not provide gain , its output balance is almost
TOTALLY determined by passive components ( anode load
and cathode load resistors ) , and valve characteristics
hardly enter the picture”
It’s very , very good , because no matter if valve ages or is
replaced , the balance ( or equilibrium ) remains the same .
Julio Rueda ( electronic engineer from Argentina and dedi-
cated only to Hi-Fi designs , in the 50’s ) says :
Note : Text was translated from Spanish to English , by me
“ The common cathodine or plate and cathode split-load
phase inverter , is an inverter self-equilibrated and is a good
example of relative theoretical imperfection against actual
practical advantages that overcome very much , anothers
designs that at first sight , seem like most perfects “
And more :
Morgan Jones says : “The phase-splitters based on the diffe-
rential pair were all able to provide overall gain , but this was
obtained at the expense of an output balance that is partially
( ... a lot , if you prefer ) dependent on the matching of “mu”
between the valves “ (... or the two halves of them )
My own experience , I have built all kind of phase-splitters ,
but the one that has the best tonal quality , and wide band-
width , undoubtedly is the cathodine. ( ... in my humble opinion ).
Ray Moth , good to see you on the thread , but I’m sorry
to disagree from you . The J.N. Van Scoyoc phase-splitter does
not need any kind of fine tuning or adjustment , because it is totally self-equilibrated due to the “cross – connections” , and has the great advantage that it cancels the 2nd harmonic , but I agree that it wastes two triodes .
Regards for all ,
Carlos
The CATHODYNE or SPLIT-LOAD or CONCERTINA
PHASE SPLITTER , no matter the designation , is the
most simple , the most functional and the most reliable
tube circuit ever made .
The opinions have been divided about it , but the exce-
llent results obtained with it , is unquestionable .
Bruce Rozenblit says : “ With all of its drawbacks , the
simple phase splitter is quite useful , and I have succes-
fully implemented it in several designs. Many applications
can easy tolerate the wide variation in output impedance
between the two signals”
Morgan Jones says : “ Although the Concertina phase
splitter does not provide gain , its output balance is almost
TOTALLY determined by passive components ( anode load
and cathode load resistors ) , and valve characteristics
hardly enter the picture”
It’s very , very good , because no matter if valve ages or is
replaced , the balance ( or equilibrium ) remains the same .
Julio Rueda ( electronic engineer from Argentina and dedi-
cated only to Hi-Fi designs , in the 50’s ) says :
Note : Text was translated from Spanish to English , by me
“ The common cathodine or plate and cathode split-load
phase inverter , is an inverter self-equilibrated and is a good
example of relative theoretical imperfection against actual
practical advantages that overcome very much , anothers
designs that at first sight , seem like most perfects “
And more :
Morgan Jones says : “The phase-splitters based on the diffe-
rential pair were all able to provide overall gain , but this was
obtained at the expense of an output balance that is partially
( ... a lot , if you prefer ) dependent on the matching of “mu”
between the valves “ (... or the two halves of them )
My own experience , I have built all kind of phase-splitters ,
but the one that has the best tonal quality , and wide band-
width , undoubtedly is the cathodine. ( ... in my humble opinion ).
Ray Moth , good to see you on the thread , but I’m sorry
to disagree from you . The J.N. Van Scoyoc phase-splitter does
not need any kind of fine tuning or adjustment , because it is totally self-equilibrated due to the “cross – connections” , and has the great advantage that it cancels the 2nd harmonic , but I agree that it wastes two triodes .
Regards for all ,
Carlos
Now, this kind of schtuffs I don't buy. I got excellent results from the differential by using Sovtek 6SL7s right out of the box. Whatever imbalance between sections, it wasn't much, and certainly no worse than what you'd find when implementing the same topology with discreet BJTs. Of course, Sovtek made these 6SL7s right since the measured performance was quite close to calculated performance, even though designed with RCA 6SL7 data sheets. The resulting LTP gave CMRRs so high as to be immeasureable: better than 80db(v). If it's a question of using bad tubes, then don't buy bad tubes.
The key to getting these results is to overcome that sandophobia and implement the tail CCS with BJTs. Cascoded BJTs give far higher output impedance, and thus LTP balance than anything else. Certainly far better than a simple tail resistor, even if you can make it bigger with a negative rail, and better than a small signal pentode. Besides, that's one less hole you need to make in the chassis.
A 6SL7 (high mu triode) in LTP phasesplitter confirms the best balance results for such configuration.
When wanting to use a (split-load)concertina with a full gain pentode preceeding for a particular design, do the stage gain maths. The pentode will heave 44dB front end gain (the split-load concertina gain = 0 ) and a following inter-driving stage will only need a gain of 7-8 to drive output tubes to 50+50V RMS. Using 20dB global NFB one arrives around the 0.5V input for full drive for the output tubes (with plenty of reserve). This was one of many standard setups for 3/4 century. Without NFB from o/p tranny sec the figures look like this .
...-> input 50mV (pent 44dB) split = 8+8V (triode 16dB) = 50+50V RMS. This is enough drive for P-P KT88 class to 100W. The trick is to select a lowish mu second stage tube which won’t increase the pentode noise in preceeding stage. I use ECL82‘s (have plenty of them) config as triodes running at 16mA per sect - despite poor looking curves. . the bandwidth is excellent for driving parallel P-P stages. The noise factor from the 1st stage pentode is the mitigating design problem. It pays to study the 7199 app notes (Neds- triode electronics etc).
The distortion from the pentode driven concertina is quite low (lower than any other configuration and depends on pentode g2 setup. The ECL82 in triode mode will give approx 1% thd at 50V RMS output drive (depending on load), but one can’t throw out everything for nothing. 20dB global NFB will drop all this down to around 0.1%. Simple „ ain’t it.??
With the figures shown without NFB, the A weighted noise figure is around ca-55dB down and with 20dB global feedback this improves to an acceptable –75dB.
Using the LTP, this itself has more gain so the input stage can be strapped as a triode or two cascaded so even more noise reduction results to -80 or -90dB levels.
Some say a noise level of –75db/w down isn’t low enough for HiFi...(WHAT RUBBISH)
With 93dB/w sens LS, there’s barely a whisper.. ..
richj.
When wanting to use a (split-load)concertina with a full gain pentode preceeding for a particular design, do the stage gain maths. The pentode will heave 44dB front end gain (the split-load concertina gain = 0 ) and a following inter-driving stage will only need a gain of 7-8 to drive output tubes to 50+50V RMS. Using 20dB global NFB one arrives around the 0.5V input for full drive for the output tubes (with plenty of reserve). This was one of many standard setups for 3/4 century. Without NFB from o/p tranny sec the figures look like this .
...-> input 50mV (pent 44dB) split = 8+8V (triode 16dB) = 50+50V RMS. This is enough drive for P-P KT88 class to 100W. The trick is to select a lowish mu second stage tube which won’t increase the pentode noise in preceeding stage. I use ECL82‘s (have plenty of them) config as triodes running at 16mA per sect - despite poor looking curves. . the bandwidth is excellent for driving parallel P-P stages. The noise factor from the 1st stage pentode is the mitigating design problem. It pays to study the 7199 app notes (Neds- triode electronics etc).
The distortion from the pentode driven concertina is quite low (lower than any other configuration and depends on pentode g2 setup. The ECL82 in triode mode will give approx 1% thd at 50V RMS output drive (depending on load), but one can’t throw out everything for nothing. 20dB global NFB will drop all this down to around 0.1%. Simple „ ain’t it.??
With the figures shown without NFB, the A weighted noise figure is around ca-55dB down and with 20dB global feedback this improves to an acceptable –75dB.
Using the LTP, this itself has more gain so the input stage can be strapped as a triode or two cascaded so even more noise reduction results to -80 or -90dB levels.
Some say a noise level of –75db/w down isn’t low enough for HiFi...(WHAT RUBBISH)
With 93dB/w sens LS, there’s barely a whisper.. ..
richj.
The Van Scoyoc x-coupled splitter is basically unbalanced. It is normally tuned using a pot between the cathode returns of the two cathode followers, with the slider going to ground.
Another drawback of this splitter is its low signal-level tolerance. The consequence of this is that the splitter must be used early in the amp, usually as the first stage. This forces the designer to use an all-balanced circuit - not necessarily a bad thing in itself, because you get very low PS noise with such a topology.
Another drawback of this splitter is its low signal-level tolerance. The consequence of this is that the splitter must be used early in the amp, usually as the first stage. This forces the designer to use an all-balanced circuit - not necessarily a bad thing in itself, because you get very low PS noise with such a topology.
Hi,
Rubbish
Auto balancing LTP:
http://dogstar.dantimax.dk/tubestuf/driver02.htm
My SS CCS equivalent attached.
The imbalance was 5V @ 60V P-P swing without common-mode FB, only 20mV @ 60V P-P swing with
Cheers!
refference said:
Rubbish
Auto balancing LTP:
http://dogstar.dantimax.dk/tubestuf/driver02.htm
My SS CCS equivalent attached.
The imbalance was 5V @ 60V P-P swing without common-mode FB, only 20mV @ 60V P-P swing with
Cheers!
Attachments
SY said:
Morgan Jones was referring to LTPs with passive tail loads. The tip-off is the mention of imbalance. VTs are naturally low gain devices. After all, the u's (highest possible theoretical voltage gain) maxes out at 100 for small signal triodes, and doesn't get above 200 even for RF "zero bias" triode finals. Even high gain small signal pentodes have g(m)'s that are pretty low: 12mA/V (12BY7A), 16mA/V (6EW6) and so forth. The g(m)'s of BJTs are limited by how much current you can force through the device before the silicon die melts. Even running at a very modest Ie= 1.0mA, the BJT has a g(m)= 38.46mA/V. How many VTs have g(m)'s that high?
With very limited device gain available, you aren't going to be getting good performance from passively loaded LTPs. You can force output balance by using unequal plate resistors as a "hack". The CMRRs remain relatively poor, and if that's all you had, you'd've been better off with a cathodyne, by far.
Hi RayMoth ,
The original design of J.N. Van Scoyoc , ( originally
Published on Radio Electronic Engineering in
1948, November ) uses one 6SN7 and one 6SL7 .
The 6SN7 at the input , acts like cathode-follower ,
coupled to the next tube’s cathodes ( 6SL7 )
that amplifies in push-pull way , with tails
cross-connected to the 6SL7’s grids .
You can use a 12AU7 ( instead 6SN7 ) and 12AX7
( instead 6SL7 ) without change in resistor’s values .
You are correct , the Scoyoc phase-splitter must be
used in the input stage . ( It works with low-level signals )
You are correct too , about the trim-pot between the two
cathodes , to fine-adjust the currents ( working point )
But I have to disagree from you , when you said that :
“this forces the designers to use an all balanced circuit ... “ .
It’s not necessary !! You can use two signals from
two different sources ( works like a mixer ) .
You can use a push-pull balanced or unbalanced signal , or
you can apply the signal at only ONE input , that you
will get at the output two identical signals with inverted
phases .
Bibliography : Eng. Julio Rueda – Audio Amplifiers Circuits
Argentina – 1957
About the Tubes 4e4’s comment , the original design
have only the two output capacitors , as usual in every
phase splitter .
Best Regards ,
Carlos
The original design of J.N. Van Scoyoc , ( originally
Published on Radio Electronic Engineering in
1948, November ) uses one 6SN7 and one 6SL7 .
The 6SN7 at the input , acts like cathode-follower ,
coupled to the next tube’s cathodes ( 6SL7 )
that amplifies in push-pull way , with tails
cross-connected to the 6SL7’s grids .
You can use a 12AU7 ( instead 6SN7 ) and 12AX7
( instead 6SL7 ) without change in resistor’s values .
You are correct , the Scoyoc phase-splitter must be
used in the input stage . ( It works with low-level signals )
You are correct too , about the trim-pot between the two
cathodes , to fine-adjust the currents ( working point )
But I have to disagree from you , when you said that :
“this forces the designers to use an all balanced circuit ... “ .
It’s not necessary !! You can use two signals from
two different sources ( works like a mixer ) .
You can use a push-pull balanced or unbalanced signal , or
you can apply the signal at only ONE input , that you
will get at the output two identical signals with inverted
phases .
Bibliography : Eng. Julio Rueda – Audio Amplifiers Circuits
Argentina – 1957
About the Tubes 4e4’s comment , the original design
have only the two output capacitors , as usual in every
phase splitter .
Best Regards ,
Carlos
Hi RayMoth ,
OK ! OK ! Now I understand your point of view . You are correct .
But don't forget , that the output level of a Scoyoc phase splitter
is relatively high , and if you are planning to build a low power
amp ( until 10 - 12 Watts ) , perhaps you won't need another drive stage .
Regards ,
Carlos
OK ! OK ! Now I understand your point of view . You are correct .
But don't forget , that the output level of a Scoyoc phase splitter
is relatively high , and if you are planning to build a low power
amp ( until 10 - 12 Watts ) , perhaps you won't need another drive stage .
Regards ,
Carlos
True, Carlos. With most tube-based splitters using a double triode, you effectively lose half the gain: concertina first stage is normally a voltage amplifer with full gain, feeding the splitter itself which has no gain; LTP splitter, with its single input, gives half the gain of a differential stage with two inputs; and the see-saw gets gain from one half but not the other.
I don't know how the gain of the Scoyoc is calculated, because I can't figure out if the second double triode (e.g. 6SL7, the one that feeds off the cathode followers, e.g. 6SN7) has current NFB or not. In this example the 6SL7 gets its bias from the cathode resistors of the 6SN7, which aren't bypassed (they can't be or there would be no signal), but does this mean that the 6SL7 gain is reduced by current NFB?
I don't know how the gain of the Scoyoc is calculated, because I can't figure out if the second double triode (e.g. 6SL7, the one that feeds off the cathode followers, e.g. 6SN7) has current NFB or not. In this example the 6SL7 gets its bias from the cathode resistors of the 6SN7, which aren't bypassed (they can't be or there would be no signal), but does this mean that the 6SL7 gain is reduced by current NFB?
dynamic mu matching
Carlos, there is a simple solution to the difference in mu between halves. Even if tightly matched, differences occur. In Electronics Engineering (from the sixties) a cunning solution was presented. In most ECC-series tubes from Philips, both halves have a separate heater for each triode.
--> By supplying each part from a variable voltage source the temperature of the cathode is varied and hence the operating point and mu. The difference in mu can easily be completely taken away with a few components.
albert
refference said:
Carlos, there is a simple solution to the difference in mu between halves. Even if tightly matched, differences occur. In Electronics Engineering (from the sixties) a cunning solution was presented. In most ECC-series tubes from Philips, both halves have a separate heater for each triode.
--> By supplying each part from a variable voltage source the temperature of the cathode is varied and hence the operating point and mu. The difference in mu can easily be completely taken away with a few components.
albert
Now, that is cunning. I daresay, in these days of microelectronic miracles, somebody could dream up a servo arrangement to control the currrent in each half of the heater, so as to correct the mu automatically as th etube ages. It could even be encapsulated within the base of the tube and be invisible to the naked eye
Afterthought: if you have a CCS in the tail of an LTP, there is no need to worry about the mu in each half being different. AC balance is assured, as long as the plate loads are identical. All you have to do is make sure the quiescent current through each triode is the same, which you can correct with resistors in the cathodes.
Afterthought: if you have a CCS in the tail of an LTP, there is no need to worry about the mu in each half being different. AC balance is assured, as long as the plate loads are identical. All you have to do is make sure the quiescent current through each triode is the same, which you can correct with resistors in the cathodes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.