These were my words in #22. But alas, artosalo doesn't appear to believe it. Or to comprehend it
.Best regards!
I would use a DN2540 cascode sink connected to the -60V bias with a 10K resistor in series with the sink.
As im building the bias arrangement, im currently thinking of regulating this with TL783. From what i recall the topic starter has 50VAC available and there is plenty of current on this winding.
As im building the bias arrangement, im currently thinking of regulating this with TL783. From what i recall the topic starter has 50VAC available and there is plenty of current on this winding.
Even then it most probably is preferable to start with a basically very well balanced PI. You might put a trimpot between the plate resistors and the supply, if that's really a concern. For my amplifiers it never was.
Best regards!
Ofcourse.
This happens when the AC-balance is initially set to "middle", where output signals of the LTP are balanced. Then AC-balance is adjusted and simultaneously THD at the output is monitored and adjusted to minimum. Simply this means that sometimes/quite often LTP must be adjusted to small unbalance to get output stage to balance, which is most essential.
Or is it so that you have not so far noticed that with AC-balance adjusted you get better THD ?
One question ? What distortion meter/analyzer you use ?
Last edited:
What Artasalo is saying: If the output stage is unbalanced (whether through tube mismatch or unequal transformer halves) and it almost always will be, unbalancing the PI - which is just changing the signal levels - can compensate, and since output stage distortion is usually a magnitude larger than PI distortion overall distortion will improve.
Phase inverters are fairly but not totally useless on their own, so it makes sense that their contribution to performance should be the real aim of their design.
Phase inverters are fairly but not totally
useless on their own, so it makes sense that their contribution to performance should be the real aim of their design.
No mistake on my part. You need to design a better output stage, apparently.
Purposely imbalancing the LTP to make corrections in a compromised output stage is not a solution, that's a kludge. 4 days later the tubes drift; you going to constantly fiddle with AC balance? Go right ahead. Please explain how this one adjustment deals with tube differences in bias, mu, rp, etc. You may think you have resolved by placing 30V on one tube and 35V on the other, but done nothing about the parameter variation from tube to tube (which can be opposite your 30V/35V pedantry).
The output stages I apply do not require 30V on one tube of the PP and 35V on the other. They are more than happy to receive 32.5V on both, and the transformer accommodates the difference (as it is supposed to).
Like Kay Pirinha, never a problem with my amplifiers...
AC-balance is one adjustment, but it does not solve all. As important is DC-balance adjustment so that the idle current is same at both output tubes.
This concerns mainly fixed bias systems, and must be adjusted first.
These two adjustment should exist in all well designed/professional tube PP-amplifier.
Leo Fender's 82K:100K "unbalance" is intended to balance the outputs of a resistor-tail longtail. As Artosalo said:
This is not a "kludge". The amount of off-balance can be closely figured by tail resistor and Gm/Mu. When looking at a complete design, include the fact that the 82k:100k pair also has typ. 220k:220k after it.
It is curious why Leo was trying to balance (lower 2nd dist) his driver. I think electric guitar "sound" was evolving, or going through phases, rapidly. Many different sounds can be "good" in the hands of the right player.
My kludge was very specifically stated in reference to a quality CCS in the tail, and one attempting to mess with the stage performance by forcing unequal plate resistors. This most certainly IS a kludge. Doesn't matter if you are trying to address the output stage imbalance...
What is fine and acceptable is the original Mullard 5-20 design with a resistor in the tail and unequal plate resistors. Not ideal, but not a kludge.
We need to understand what is good and what is not good advice. Morgan Jones exhausts the LTP pretty well. CCS wants matched plate loads.
What is fine and acceptable is the original Mullard 5-20 design with a resistor in the tail and unequal plate resistors. Not ideal, but not a kludge.
We need to understand what is good and what is not good advice. Morgan Jones exhausts the LTP pretty well. CCS wants matched plate loads.
actually what we want is simply a way to independently adjust the level of the two phases feeding the output stage. unequal plate resistors is one way and having an AC balance control is the same thing. Of course it is the output stage that needs adjusting but we cannot adjust the anode loads of the output stage - unless of course someone hack a variac and then grafts it onto the OPT......
Artosolo - I get what you are saying. It's all well and good having a perfectly balanced phase inverter, but as you say, even with "matched" tubes gm will be different. If you run REW or similar with real time distortion measurement (and I have), it's easy to null the second harmonic by tweaking the gain of one half of the phase inverter (even one using a CCS as a tail). The optimum setting for low levels might not be the same as for high levels, but usually a compromise can be reached. The alternative is to have very well matched tubes, or not measure it and don't worry about it Of course some second harmonic "enhancement" might be the desired outcome.
Of course some second harmonic "enhancement" might be the desired outcome.
Not so good idea.If you change the screen level you must re-adjust the DC-balance and level.Exept with automatic bias control.Individually adjustable Vg2 to fine-tune each output tube's gm?
Balancing the AC with different anode resistors is ok with a CCS at the cathodes.
Not with a cathode resistance, there the different anode resistors are also to compensate for the error from the fact that one side is a common cathode stage an the other a common grid stage, both having a different non constant gain (signal level dependent).
Mona
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.