Hello,
I'm wondering how a differential stage -like a LTP for example- handles noise. What theory predicts about CMRR assuming ideal implementation. PSU ripple should be perceived as common mode and rejected, or not? But what about heaters injected noise or EMI from transformers? What is common mode and what differential regarding noise sources in a case like this?
I'm wondering how a differential stage -like a LTP for example- handles noise. What theory predicts about CMRR assuming ideal implementation. PSU ripple should be perceived as common mode and rejected, or not? But what about heaters injected noise or EMI from transformers? What is common mode and what differential regarding noise sources in a case like this?
It depends a lot on what you do with the output of the differential stage. If everything is differential, PSU ripple and interference from the heaters should be suppressed to some extent. If you take the output signal single-endedly from one of the two anodes and bias it with a resistor to the supply, supply ripple will just pass right through; you get some voltage division between the resistor that biases the anode and twice the internal resistance of a valve, but that's it. You won't get much suppression from anything injected by the heaters either.
Thanks Marcel! I understand that CMRR happens at the input of the next stage. Assuming a fully differential push pull amp, drivers' input should cancel VAS noise, output stage should cancel drivers' noise and OPT should cancel output stage noise, isn't it? Is it possible psu or heaters noise to sneak in to a scheme like this?
When you have a completely differential amplifier, you still have mismatch between the components limiting the suppression of any common-mode interference. The worst possible resistor mismatch can be calculated from the resistor tolerances, but data about valve mismatch is typically nowhere to be found.
Regarding noise, as in circuit noise (thermal, 1/f, shot, partition and so on), the noise-generating processes in one half of the amplifier generate their noise completely independently from those in the other half. The standard deviations and probability distributions should match to some extent between the halves, but the momentary voltages and currents do not. Therefore, it doesn't cancel.
Regarding noise, as in circuit noise (thermal, 1/f, shot, partition and so on), the noise-generating processes in one half of the amplifier generate their noise completely independently from those in the other half. The standard deviations and probability distributions should match to some extent between the halves, but the momentary voltages and currents do not. Therefore, it doesn't cancel.
Agreed, it's unreasonable to expect fully matched pairs in real life so CMRR will be compromised. I just wanted to know how does a tube pair faces each source of noise. So far I'm gathering that anode and heater psu ripple appear as common mode signal. What about air induced EMI? From experience I know that balanced input, by contrast to SE, is not sensitive to noise picking if it's left floating instead of shorted to ground. So grids should be rather immune to EMI(?) What about anodes and cathodes? Theoretically I mean.
Theoretically you could get some cancellation if the source of interference is placed at the same distance and direction from both of the valves, so they both receive the same interference. That seems rather unlikely, though.