I agree with Sacpul that the bass has something to do with the earth connection. Please check with it.
Maybe replace the 12AX7 with a 12DW7, and make a few small adjustments on the plate and cathode resistors of the second half so you can get some decent current through the splitter? The EL84 is a rather easy final to drive, but doesn't a little extra current help keep slew limiting at bay?
You could improve the sound by disconnecting the rectifier feeding the magic eye envelope detector. At present any isolated negative-going transient will be clipped as it charges the 470nF capacitor. This could then cause problems within the feedback loop, as the amplifier temporarily loses loop gain. Why would someone want to degrade the sound of a basically OK ampliifer by adding pretty lights? I would do this before adding 'boutique' capacitors.
The amplifier can probably drive a permanently connected 470nF, although it would increase HF distortion. Much harder to drive a capacitor which is only connected on signal peaks, as it takes a transient charging current.
The amplifier can probably drive a permanently connected 470nF, although it would increase HF distortion. Much harder to drive a capacitor which is only connected on signal peaks, as it takes a transient charging current.
I have the smaller Ming-Da EL84 Push-Pull (the Mini). The schematic is exactly the same, except the circuitry that drives the "magic Eye". Ha! Seems like I had a good deal!
The bridge rectifier waw on the air in mine amplifier, too, just soldered on the capacitor's pins. I replaced it with 4 good quality diodes, still "airborne". The next thing to do, which might improve the bass and is generally a good idea in PP amps, is to give each EL84 its own cathode resistor and bypass capacitor. Resistors will have to be twice the ohmage. And definitely bypass the electrolytic cathode capacitors with plastic caps!
The bridge rectifier waw on the air in mine amplifier, too, just soldered on the capacitor's pins. I replaced it with 4 good quality diodes, still "airborne". The next thing to do, which might improve the bass and is generally a good idea in PP amps, is to give each EL84 its own cathode resistor and bypass capacitor. Resistors will have to be twice the ohmage. And definitely bypass the electrolytic cathode capacitors with plastic caps!
Giving each EL84 its own cathode resistor and bypass is a mixed blessing. It improves P-P balance and means that the output valves don't need to be so well matched. On the other hand, it places an extra burden on the bypass capcitor as the audio signal current is no longer partially cancelled so distortion could rise. Try it and see; it might work, it might not.
I forget to say that separating the cathodes of the EL84 could also reduce the bass, unless a much larger bypass capacitor is used. This amplifier only has a low level of feedback*, so the output impedance will be high. This gives soft flabby bass, which some people seem to prefer. Too low a capacitance at the EL84 cathode will make this worse by raising the EL84 anode impedance at low frequencies - it doesn't matter too much while the cathodes are commoned.
* assuming the circuit diagram is correct - should the feedback resistor be 1K rather than 10K?
* assuming the circuit diagram is correct - should the feedback resistor be 1K rather than 10K?
Are we sure we want to "partially cancel" the audio signal? I don't get why we would like to cancel any audio. Any audio that is canceled will detract from what we listen! If you hear more distortion it is actually more of the signal that is already there, this means there are problems elsewhere in your circuit.
And you are right, I forgot to mention that bypass capacitors should be double the μF.
And you are right, I forgot to mention that bypass capacitors should be double the μF.
At the output cathodes there is a signal current from both valves. These are antiphase so they cancel. This is good! If the phase splitter and output stages were perfectly balanced then there would be no net signal current. Then the cathode resistor would not need any bypass.
Two things upset this. First, the balance will not be perfect. Second, the output valves will generate significant second-harmonic currents which will be in phase at the cathode so we need a bypass for them. At the anodes the second-harmonics cancel because it is P-P. However, any second-harmonic at the cathodes will mix with the signal to produce third-harmonic which will not cancel at the anodes.
All this means that separating the cathodes needs a much bigger bypass capacitor - much more than twice the value.
Two things upset this. First, the balance will not be perfect. Second, the output valves will generate significant second-harmonic currents which will be in phase at the cathode so we need a bypass for them. At the anodes the second-harmonics cancel because it is P-P. However, any second-harmonic at the cathodes will mix with the signal to produce third-harmonic which will not cancel at the anodes.
All this means that separating the cathodes needs a much bigger bypass capacitor - much more than twice the value.
An excellent point. Seems like 47n, but still. I would also remove the tubes from the sockets in order to reduce the heaters consumption. One can always plug them back in during the Chinese New Year 🙂
Yes, you are right - I misread the value. Not quite so bad as 470n, but still unnecessary. Was this amp designed to impress eyes, rather than ears?
I have been taking another look at the schematic. All my remarks are based on that being accurate. First thing is that there is very little feedback (about 2dB). The open loop gain is about 60, closed-loop gain about 50 (which matches the claimed spec of 12W out for 200mV in). This is not enough feedback to be worth doing.
This means that the frequency response, distortion etc. are largely set by the amp circuitry rather than the feedback. At both LF and HF extremes this means the output transformer, so this had better be very good!
At LF the only limit before the transformer is the coupling capacitors. I calculate an LF limit of 1Hz! This is far too low. It means that the output has to handle all bass in the input signal, which is likely to lead to distortion in the transformer. Better to limit the bass by reducing the couplers or adding a capacitor at the top of the volume control. Then you will have less bass, but better quality bass.
At HF the main limit before the transformer is the lead-lag network in the anode of the first stage. This gives an HF shelf above 60kHz. With such low feedback I wonder why this is necessary. Does the transformer have a nasty ultrasonic resonance which boosts the forward gain and hence the feedback? If so, the network is needed. If not, it seems superfluous.
It is always interesting to see other people's designs.
This means that the frequency response, distortion etc. are largely set by the amp circuitry rather than the feedback. At both LF and HF extremes this means the output transformer, so this had better be very good!
At LF the only limit before the transformer is the coupling capacitors. I calculate an LF limit of 1Hz! This is far too low. It means that the output has to handle all bass in the input signal, which is likely to lead to distortion in the transformer. Better to limit the bass by reducing the couplers or adding a capacitor at the top of the volume control. Then you will have less bass, but better quality bass.
At HF the main limit before the transformer is the lead-lag network in the anode of the first stage. This gives an HF shelf above 60kHz. With such low feedback I wonder why this is necessary. Does the transformer have a nasty ultrasonic resonance which boosts the forward gain and hence the feedback? If so, the network is needed. If not, it seems superfluous.
It is always interesting to see other people's designs.
- Status
- Not open for further replies.