Yes absolutely.
And as noted, sharing bias resistors between channels is bad engineering.
I suggest either 4 individual Self Bias resistors with 4 individual large bypass caps,
Or
Fixed Bias with 4 individual Fixed Bias pots.
For a stereo push pull amp that is fixed bias, you need to have a very highly regulated plate B+, and a very highly regulated B+ for screens.
Otherwise you are going to have to go back and forth quite a few times on the 4 pots (each one you adjust will affect the current in the other 3 tubes, because the plate and screen B+ are not regulated.
As to the importance of balancing the DC currents, take a look at the Heathkit W-5M graph. That graph is the voltage across 2 each 30 Ohm resistors in series, and 0.1V is only a 1.7 mA difference in plate currents. The quality output transformer can not take lots of DC imbalance, just look at the distortion, for 100 Hz, 50 Hz, and 20 Hz.
This distortion is there even with negative feedback, and a fairly good quality circuit.
(Graph and circuit attached)
Or
Fixed Bias with 4 individual Fixed Bias pots.
For a stereo push pull amp that is fixed bias, you need to have a very highly regulated plate B+, and a very highly regulated B+ for screens.
Otherwise you are going to have to go back and forth quite a few times on the 4 pots (each one you adjust will affect the current in the other 3 tubes, because the plate and screen B+ are not regulated.
As to the importance of balancing the DC currents, take a look at the Heathkit W-5M graph. That graph is the voltage across 2 each 30 Ohm resistors in series, and 0.1V is only a 1.7 mA difference in plate currents. The quality output transformer can not take lots of DC imbalance, just look at the distortion, for 100 Hz, 50 Hz, and 20 Hz.
This distortion is there even with negative feedback, and a fairly good quality circuit.
(Graph and circuit attached)
By the way, Single Ended Amplifiers do not have the problem of imbalanced primary currents (there is only one primary, not the 2 x 1/2 primary off push pull).
Using a pair of very good quality single ended transformers on a push pull amplifier (instead of a single push pull transformer) does 2 things:
Eliminates the Magnetic Zero Crossings on the Laminations.
Makes 2 balanced currents in the output transformer(s) of less importance.
Using a pair of very good quality single ended transformers on a push pull amplifier (instead of a single push pull transformer) does 2 things:
Eliminates the Magnetic Zero Crossings on the Laminations.
Makes 2 balanced currents in the output transformer(s) of less importance.
Er - somewhat strongly worded ....
but I must agree. There can be an economical reason only for that - and miscalculated in the instance of something as inexpensive as a resistor (even hi-power). Thus my agreement - such 'grouping-together' does not belong in quality equipment.
Not negating my above reason, but one must also consider that separate cathode resistors (I mean one each per power tube, not one per channel) do not solve the associated problem per se; they only lessen it. The rather unacceptable spread in valve characterisitcs these days has been mentioned before.
The extra effect of a further l.f. pole in the NFB mix (possible problem) has also been mentioned. In practice - yes - fixed (adjustable then) bias is a solution, but then so will be a balance control in a cathode bias arrangment - also one extra adjustment required and all that, plus retaining some self-compensating effect. For now for me the fixed bias option depends on other matters; I see it as a kind of 'horses-for-courses' thing.
The way I read it, you're off by a factor of 2. The 30R parts aren't in series - they are separate, one per cathode circuit. So 0.1V across 30R = 3.33mA imbalance. So at the extreme of the graph (+/-0.5V) the imbalance is 16.7mA for 1.75% THD @ 20Hz - falling to 0.25% THD @ 100Hz.
Doesn't seem like that big of an issue, from my perspective - at least so far as the W5M is concerned. And that holds even if the figures were double, as per your claim.
Careful here.
As far as I can notice from the graph and schematic, the voltage shown on the X-axis is the voltage seen by the indicated voltmeter in the schematic as measured across both 30 ohm resistors (in series), not only one. V = 0,5V indicates a current of 8,3 ma flowing through the 2 x 30 ohm resistors. This will be the unbalance current in the OPT, irrespective of its origin - 4,15 mA/tube, if you will.
The distortion figure given for this could be cause for concern, but then how often will a 20 Hz signal reach an amplifer? Also the graphs refer to one OPT, to me of unknown characteristics.
6A3:
Not sure I understand your reasoning regarding separate S.E. output transformers for push-pull operation. S.E. transformers carry a large d.c. at all times; I cannot see that as no problem, since immediately previously you have tried to show how even a small d.c. can mess up the works. Those OPTs combat the problem by having to be quite large, thus expensive.
That still does not render two separate transformers capable of eliminating even harmonic distortion, as happens with a single balanced transformer.
As far as I can notice from the graph and schematic, the voltage shown on the X-axis is the voltage seen by the indicated voltmeter in the schematic as measured across both 30 ohm resistors (in series), not only one. V = 0,5V indicates a current of 8,3 ma flowing through the 2 x 30 ohm resistors. This will be the unbalance current in the OPT, irrespective of its origin - 4,15 mA/tube, if you will.
The distortion figure given for this could be cause for concern, but then how often will a 20 Hz signal reach an amplifer? Also the graphs refer to one OPT, to me of unknown characteristics.
6A3:
Not sure I understand your reasoning regarding separate S.E. output transformers for push-pull operation. S.E. transformers carry a large d.c. at all times; I cannot see that as no problem, since immediately previously you have tried to show how even a small d.c. can mess up the works. Those OPTs combat the problem by having to be quite large, thus expensive.
That still does not render two separate transformers capable of eliminating even harmonic distortion, as happens with a single balanced transformer.
No, if the OPTs don't share the same core it will not operate push-pull mode.
It will only be parallel SE.
Jacques
Last edited:
It is true that even at 20 Hz, the current imbalance has to be quite large in order to approach 1.5% distortion. But I would have expected negative feedback to have corrected it.
The voltage on the W-5M is measured from the top of one 30 Ohm resistor to the top of the other 30 Ohm resistor. It is a differential measurement. If the currents are the same, there is no differential voltage.
If one resistor has 0.05V More than say 'ideal', and the other resistor has 0.05V Less than 'ideal', that = a total differential of 0.1V
0.05V/30 Ohms is 1.7mA. Which is the same as 0.1V/60 Ohms that I originally stated.
But for the w-5M, we all might find the distortion due to the current imbalance to not be very bad, because as you stated how often do we have music at 20 Hz.
So far, so good, no problem.
Now, let us consider a push pull amp that does Not use negative feedback **.
Current imbalance has the same kind of effect, except now it might be significant at 50 Hz or 100 Hz. Part of what made the W-5M look OK at 50 Hz and 100 Hz was negative feedback (but we no longer have negative feedback **).
The voltage on the W-5M is measured from the top of one 30 Ohm resistor to the top of the other 30 Ohm resistor. It is a differential measurement. If the currents are the same, there is no differential voltage.
If one resistor has 0.05V More than say 'ideal', and the other resistor has 0.05V Less than 'ideal', that = a total differential of 0.1V
0.05V/30 Ohms is 1.7mA. Which is the same as 0.1V/60 Ohms that I originally stated.
But for the w-5M, we all might find the distortion due to the current imbalance to not be very bad, because as you stated how often do we have music at 20 Hz.
So far, so good, no problem.
Now, let us consider a push pull amp that does Not use negative feedback **.
Current imbalance has the same kind of effect, except now it might be significant at 50 Hz or 100 Hz. Part of what made the W-5M look OK at 50 Hz and 100 Hz was negative feedback (but we no longer have negative feedback **).
Take the in-phase output plate of a push pull amp and connect it to the in-phase primary terminal of an SE transformer, and connect B+ to the out-of-phase primary terminal.
Take the out-of-phase output plate of a push pull amp and connect it to the out-of-phase primary terminal of an SE transformer, and connect B+ to the in-phase primary terminal.
The output plates are out of phase with respect to each other. But we connected the SE transformers, one plate to the in-phase of the primary, and the other plate to the out-of-phase primary.
We will find that the secondaries of both SE transformers are now in phase.
Now, connect the secondaries of those 2 SE transformers together: C-C; 4-4; 8-8; 16-16.
This is for all intents and purposes, a form of 'push pull' amplifier.
As long as we are in class A for both the push and pull tubes, each SE transformer has current through it at all times. Therefore, it does not have magnetic zero crossings on the laminations (the laminations always have some magnetization, and do not reverse polarity).
Also, the 2nd harmonic is cancelled to some extent, because as one plate current is increasing, the other plate current is decreasing. The signal is putting one tube in the increasing Gm range, at the same time the other tube is going into the decreasing Gm range (just like in push pull).
It also has the characteristic that a slight imbalanced current in one plate versus the other plate does not cause the kind of distortion that was referred to in the W-5M graph above.
That is because there is no common lamination (with no air gap) that needs the currents balanced.
"Diagrammatically":
Let * = In-Phase
Let Anti-* = Anti-Phase
Connect Plate* to primary*
Connect Plate Anti-* to primary Anti-*
Secondaries are now in-phase with respect to each other.
I have noted that most people do not understand the theory and application of this mode, no matter what it may be called. But it is not parallel single ended.
Take the out-of-phase output plate of a push pull amp and connect it to the out-of-phase primary terminal of an SE transformer, and connect B+ to the in-phase primary terminal.
The output plates are out of phase with respect to each other. But we connected the SE transformers, one plate to the in-phase of the primary, and the other plate to the out-of-phase primary.
We will find that the secondaries of both SE transformers are now in phase.
Now, connect the secondaries of those 2 SE transformers together: C-C; 4-4; 8-8; 16-16.
This is for all intents and purposes, a form of 'push pull' amplifier.
As long as we are in class A for both the push and pull tubes, each SE transformer has current through it at all times. Therefore, it does not have magnetic zero crossings on the laminations (the laminations always have some magnetization, and do not reverse polarity).
Also, the 2nd harmonic is cancelled to some extent, because as one plate current is increasing, the other plate current is decreasing. The signal is putting one tube in the increasing Gm range, at the same time the other tube is going into the decreasing Gm range (just like in push pull).
It also has the characteristic that a slight imbalanced current in one plate versus the other plate does not cause the kind of distortion that was referred to in the W-5M graph above.
That is because there is no common lamination (with no air gap) that needs the currents balanced.
"Diagrammatically":
Let * = In-Phase
Let Anti-* = Anti-Phase
Connect Plate* to primary*
Connect Plate Anti-* to primary Anti-*
Secondaries are now in-phase with respect to each other.
I have noted that most people do not understand the theory and application of this mode, no matter what it may be called. But it is not parallel single ended.
Unbalanced current in a push pull transformer primary causes saturation in one direction before it saturates in the other direction. That is a form of 2nd harmonic distortion.
If it saturates equally in both directions, that is 3rd harmonic distortion.
If it saturates equally in both directions, that is 3rd harmonic distortion.
This is what I understood. And when one power tube sees a positive signal on its grid the plate current increase while the other power tube sees a negative signal on its grid and decrease its plate current.
How do you connect the two secondaries together ?
The load should see the sum of the secondary currents.
How they will react on each primary hen secondaries tied together ?
Jacques
Last edited:
What harm does a zero crossing do? Zero crossing merely means an equal number of magnetic domains pointing in each direction. How is this harmful?6A3sUMMER said:
I can see no benefit in using a pair of SE OPTs for a PP amp. You just increase size, weight and cost - or sacrifice bass due to low inductance. You are also likely to increase resistive losses, as the primary needs more turns to achieve a given inductance. You reduce coupling between the two anodes, so at higher frequencies the stage has less balance.
First, some admitted Disadvantages of using 2 SE transformers to accomplish "Push Pull":
Yes, higher Primary DCR and higher Secondary DCR causes more power loss.
Yes, size and weight increase (and price increase **).
Yes, if there is not enough primary inductance, bass response drops off.
Yes, if there is high Primary DCR and high Secondary DCR, and if there is terrible leakage reactance, the Push and Pull plates will not be very well coupled.
Now, much of the above disadvantages can be reduced by using very good quality SE transformers.
(but price will increase **)
Any poor quality push pull transformer can have any or all of the bad characteristics that are listed above (with the exception of size and weight). And most quality push pull transformers have reduced those disadvantages (but price will increase **; though they admittedly are not as expensive as SE transformers).
The two most non-linear regions of transformers are:
1. Saturation
2. Zero Crossing (where all the magnetic domains first point in one direction, then when the magnetic domains have to point in the other direction.
A Push Pull transformer has magnetic domains changing directions (magnetic zero crossings).
Push tube 60 mA, Pull tube 60 mA, the opposing currents cancel, resulting in no magnetic domain direction (except if the currents are not EXACTLY equal, then they Do point in one direction).
Push tube 61 mA, Pull tube 59 mA, the magnetic domains point in one direction.
Push tube 59 mA, Pull tube 61 mA, the magnetic domains point in the Opposite direction.
Push tube 10 mA, Pull tube 110 mA, the magnetic domains point in one direction.
Push tube 110 mA, Pull tube 10 mA, the magnetic domains point in the Opposite direction.
A Single Ended transformer is different:
SE transformer, Push tube 60 mA, 61 mA, 59 mA, 10 mA, 110 mA. In Every case, the magnetic domains point in one direction.
The other SE transformer, Push tube 60 mA, 59 mA, 61 mA, 110 mA, 10 mA. In Every case, the magnetic domains point in one direction.
There are no magnetic zero crossings.
Whether the fact that there are no magnetic zero crossings counts for anything or not can be argued.
But what is not arguable is that the fact that there are no magnetic domains changing directions in this case.
Yes, higher Primary DCR and higher Secondary DCR causes more power loss.
Yes, size and weight increase (and price increase **).
Yes, if there is not enough primary inductance, bass response drops off.
Yes, if there is high Primary DCR and high Secondary DCR, and if there is terrible leakage reactance, the Push and Pull plates will not be very well coupled.
Now, much of the above disadvantages can be reduced by using very good quality SE transformers.
(but price will increase **)
Any poor quality push pull transformer can have any or all of the bad characteristics that are listed above (with the exception of size and weight). And most quality push pull transformers have reduced those disadvantages (but price will increase **; though they admittedly are not as expensive as SE transformers).
The two most non-linear regions of transformers are:
1. Saturation
2. Zero Crossing (where all the magnetic domains first point in one direction, then when the magnetic domains have to point in the other direction.
A Push Pull transformer has magnetic domains changing directions (magnetic zero crossings).
Push tube 60 mA, Pull tube 60 mA, the opposing currents cancel, resulting in no magnetic domain direction (except if the currents are not EXACTLY equal, then they Do point in one direction).
Push tube 61 mA, Pull tube 59 mA, the magnetic domains point in one direction.
Push tube 59 mA, Pull tube 61 mA, the magnetic domains point in the Opposite direction.
Push tube 10 mA, Pull tube 110 mA, the magnetic domains point in one direction.
Push tube 110 mA, Pull tube 10 mA, the magnetic domains point in the Opposite direction.
A Single Ended transformer is different:
SE transformer, Push tube 60 mA, 61 mA, 59 mA, 10 mA, 110 mA. In Every case, the magnetic domains point in one direction.
The other SE transformer, Push tube 60 mA, 59 mA, 61 mA, 110 mA, 10 mA. In Every case, the magnetic domains point in one direction.
There are no magnetic zero crossings.
Whether the fact that there are no magnetic zero crossings counts for anything or not can be argued.
But what is not arguable is that the fact that there are no magnetic domains changing directions in this case.
Hyperman75,
You got it.
Yes, since the two output tubes were in opposite phases, and since we connected them to the SE transformer primaries in opposite directions (opposite phases), the secondaries are now in phase with each other. So connect the secondaries in Parallel.
The minimum connections that are required are secondary Common to secondary Common, and for the impedance you will use, tap to tap (for example 8 Ohm tap to 8 Ohm tap).
You do not have to tie the other taps together if you will not want to use them, but it is a very good idea, since the coupling will be even tighter.
I have actually built a "push pull' amplifier using 2 SE transformers (it is not just on paper).
There were some very interesting and revealing listening sessions why that amp.
It may just have been the other parts and circuits in the amp, or it may have been the fact of the special topology of using 2 SE transformers.
You got it.
Yes, since the two output tubes were in opposite phases, and since we connected them to the SE transformer primaries in opposite directions (opposite phases), the secondaries are now in phase with each other. So connect the secondaries in Parallel.
The minimum connections that are required are secondary Common to secondary Common, and for the impedance you will use, tap to tap (for example 8 Ohm tap to 8 Ohm tap).
You do not have to tie the other taps together if you will not want to use them, but it is a very good idea, since the coupling will be even tighter.
I have actually built a "push pull' amplifier using 2 SE transformers (it is not just on paper).
There were some very interesting and revealing listening sessions why that amp.
It may just have been the other parts and circuits in the amp, or it may have been the fact of the special topology of using 2 SE transformers.
I have actually built a "push pull' amplifier using 2 SE transformers (it is not just on paper).
There were some very interesting and revealing listening sessions WITH that amp.
It may just have been the other parts and circuits in the amp, or it may have been the fact of the special topology of using 2 SE transformers.
I typed 'with' above, auto-spell checker typed 'why'.
I thought auto-spell checkers only worked on Halloween.
There were some very interesting and revealing listening sessions WITH that amp.
It may just have been the other parts and circuits in the amp, or it may have been the fact of the special topology of using 2 SE transformers.
I typed 'with' above, auto-spell checker typed 'why'.
I thought auto-spell checkers only worked on Halloween.
This discussion is way off topic for too many posts now. There is an old thread that covers pseudo PP, parallel SE amps.
Apology for that; it is wrong.
A star-delta conversion should be done on the whole cathode bias network, to find the equivalent resistor over which V is measured. Then un-balance current can be calculated.
At low signal level, even without feedback, a quality amp would have very low THD% - even a vintage Williamson without feedback connected has THD below 0.1% up to quite a few watt. I'd suggest that any 'zero crossing magnetic domain' contribution would be very insignificant in the practical mix of distortion contributors.
Last edited:
- Status
- Not open for further replies.