Hello friends,
As per subject line, a long time ago I stumbled on a page explaining how to proceed but I can't find it anymore. I'm confident someone here knows how.
Off the top of my head I'd first DC balance using DVMs and once the plates sit at same voltage with no signal, hook a scope probe between ground and each plate, AC coupled.
Feed a signal at the input, check if each output looks ok. Switch the scope in "add" mode, then adjust balance until lowest residual voltage.
Is that about it?
Some details : the circuits to be adjusted are both an input and an output stage, both class A sitting on CCSes. Scope is my old Tek cathode ray tube, and of course with respect to the HV probes will be set at 10X. All elements (mains, circuit, scope, signal generator) are all grounded at a single point.
Thanks for any insights, and have a nice week-end everyone!
- Joris
As per subject line, a long time ago I stumbled on a page explaining how to proceed but I can't find it anymore. I'm confident someone here knows how.
Off the top of my head I'd first DC balance using DVMs and once the plates sit at same voltage with no signal, hook a scope probe between ground and each plate, AC coupled.
Feed a signal at the input, check if each output looks ok. Switch the scope in "add" mode, then adjust balance until lowest residual voltage.
Is that about it?
Some details : the circuits to be adjusted are both an input and an output stage, both class A sitting on CCSes. Scope is my old Tek cathode ray tube, and of course with respect to the HV probes will be set at 10X. All elements (mains, circuit, scope, signal generator) are all grounded at a single point.
Thanks for any insights, and have a nice week-end everyone!
- Joris
Last edited:
Thanks for your reply.
Won't the sines be in anti-phase by way of the differential output?
Which TEK old scope is it? I spent 10+ years in a research lab
in the 50s & 60s. Used 530 & 540 series as my daily drivers.🙂👍
Adding or subtracting the two scope inputs to evaluate their difference has some limitations. Each channel has to remain in its linear region so you can't arbitrarily crank up each channel's gain to see a small difference between them. Gain is applied before the summation. So the usefulness of this method depends on your requirements. If you'e looking for a 100mv balance in the presence of 100V signals you might not be able to use this scheme to good effect.
An opamp adder doesn't have the same limitation because its output is proportional to the sum of all the input currents -- gain is applied after the summation.
Since the outputs of a differential amplifier are 180 degrees out of phase you also can use two resistors to passively obtain the difference. In that situation, you can set your scope's input sensitivity to whatever you need in order to evaluate the diff amp's balance.
An opamp adder doesn't have the same limitation because its output is proportional to the sum of all the input currents -- gain is applied after the summation.
Since the outputs of a differential amplifier are 180 degrees out of phase you also can use two resistors to passively obtain the difference. In that situation, you can set your scope's input sensitivity to whatever you need in order to evaluate the diff amp's balance.
One additional requirement is that the resistors used for performing the sum/difference have to be well-matched.
Doing it in that order should avoid the issue @Mark'51 just wrote about, as long as you try to get as large a picture as possible before going to add mode and don't change the vertical axis settings anymore after going to add mode.
You will have to make very sure that the probes are properly adjusted.
DC balancing will make sure both tubes draw the same current. Measuring plate voltages will NOT ive this as DC resistance is usually differing in OT transformers.
The "classical method" is to install a small resistor in each cathode and mesure the voltage across them, adjusting until they
are equal.
AC adjutment is best done with a signal, adjusting for the lest distortion
Thanks Mark for your reply, very informative.
From this and Marcel's remarks I conclude the following :
1. Don't attempt balancing at high voltage. Perhaps midway between 0 and full output swing?
2. When checking each output before balancing, adjust V/div to fill the screen with the sine to get maximum visibility.
And last, can the 10X magnification function be used for fine adjsutment during balancing?
From this and Marcel's remarks I conclude the following :
1. Don't attempt balancing at high voltage. Perhaps midway between 0 and full output swing?
2. When checking each output before balancing, adjust V/div to fill the screen with the sine to get maximum visibility.
And last, can the 10X magnification function be used for fine adjsutment during balancing?
My main one is the 475. I also have this collectible vacuum tube 310A 😛
I don't use it much now to spare the tubes, but keep it if I need to work with voltages over 500V. The construction of this unit is a work of art - The case folds open like a wallet for maintenance. Despite its obsolete technology and age this thing can still produce the sharpest trace you can imagine.
From what I undertand the 530 & 540 series were much more advanced 👍
Thanks Peter for your reply.
Now think of it, could one pass the wires going to each cathode in a current detector coil (going in opposite direction) then measure the DC error as voltage induced in the coil? A bit like a ground-fault outlet works. I think such a setup could also detect AC inbalance?
That can be arranged as the circuit is still in breadboard stage.
Now think of it, could one pass the wires going to each cathode in a current detector coil (going in opposite direction) then measure the DC error as voltage induced in the coil? A bit like a ground-fault outlet works. I think such a setup could also detect AC inbalance?
Last edited:
I'll be adjusting AC balance at each stage. If I'm not mistaken the CCS won't help balance currents if the tubes' transconductance doesn't match ?
With a current source and no grid currents, the AC components of the anode currents of a triode differential pair are automatically balanced, also when the triodes don't match. The AC current through one triode has nowhere else to go than into the other triode.
Thanks Marcel for the explanation, now I see the point of the CCS high impedance argument mentionned by nerdorama above. Boy do I have a lot to learn... Or maybe connect dots lol
That tackles the input stage case but does it applies as well to a pentode output stage connected in ultralinear?
Been experimenting on the breadboard and indeed there's not much difference between the plates from one extreme of the balance pot to the other. Is the difference due to the ultralinear connection or a slight bias difference sue to the voltage drop across the side that has all the pot resistance?
That tackles the input stage case but does it applies as well to a pentode output stage connected in ultralinear?
Been experimenting on the breadboard and indeed there's not much difference between the plates from one extreme of the balance pot to the other. Is the difference due to the ultralinear connection or a slight bias difference sue to the voltage drop across the side that has all the pot resistance?
Last edited:
With a resistor DC may be measured. With a coil DC is not observable.
AC inbalance could be detected by distortion measurment and that would apply to the whole stage.
Note that if the phase splitter is "perfect" and the tubes are matched (at working voltage) the stage will be balanced.
True. A Hall effect sensor could be used for both AC and DC but that's beyond the scope of this post.
I'm trying to get away with the need to match tubes or at least being able to compensate slight unmatches.
I think you should trim for DC rather than AC balance. The AC currents through two differential-pair triodes are equal and opposite no matter whether the triodes match, assuming a perfect tail current source, but mismatch between the triodes still causes DC offsets, unequal clipping on positive and negative excursions and reduced even-order distortion cancellation. Trimming for DC balance should solve that.
When you have two pentodes and their cathodes are connected to each other and to a current source and to nothing else, then their AC cathode currents must be equal but opposite. If the cathode current of one pentode increases with an amount ∆i, the cathode current of the other must decrease with ∆i because their sum has to stay constant (as forced by the tail current source). The way the cathode AC currents split between the screen grid and the anode could be a bit different between one pentode and the other, but I don't see how you could trim that.
However, the same applies as for the triodes: you can still have DC imbalance, unequal clipping on positive and negative excursions and reduced even-order distortion cancellation. In this case, you can also add potential output transformer saturation due to DC imbalance to the list. Trimming for equal DC currents should again help.
When you have two pentodes and their cathodes are connected to each other and to a current source and to nothing else, then their AC cathode currents must be equal but opposite. If the cathode current of one pentode increases with an amount ∆i, the cathode current of the other must decrease with ∆i because their sum has to stay constant (as forced by the tail current source). The way the cathode AC currents split between the screen grid and the anode could be a bit different between one pentode and the other, but I don't see how you could trim that.
However, the same applies as for the triodes: you can still have DC imbalance, unequal clipping on positive and negative excursions and reduced even-order distortion cancellation. In this case, you can also add potential output transformer saturation due to DC imbalance to the list. Trimming for equal DC currents should again help.