Me again, Tobruk,
I don't want to lead you astray, but there was comment about the frequency response of the LM317 elsewhere. You must remember that such a device would have to be a CCS over at least the whole audio band. It is not clear that all power chips could do that. But I regret not recalling where I read that. Perhaps others could help.
Then to repeat, I would suggest as I did on another one of your threads that you ask a moderator to group the lot (if I am correct that it is about the same project) for your benefit.
Regards.
I don't want to lead you astray, but there was comment about the frequency response of the LM317 elsewhere. You must remember that such a device would have to be a CCS over at least the whole audio band. It is not clear that all power chips could do that. But I regret not recalling where I read that. Perhaps others could help.
Then to repeat, I would suggest as I did on another one of your threads that you ask a moderator to group the lot (if I am correct that it is about the same project) for your benefit.
Regards.
The DIYparadise EL84 amp drives the output push-pull pair from one side only. So the output stage MUST behave as a differential amplifier to work properly. To cause diff-amp behavior, there has to be a large tail resistor at the EL84 cathodes. The DIYparadise schematic shows just a 135 ohm resistor, which in my view, is inadequate. He later suggests using a CCS. And, in this case, you would not want to bypass the CCS with a cap, because one EL84 cathode must be able to "talk" to the other cathode for push-pull behavior to occur, when only one side is driven.
As Johan notes, an LM317 used as a CCS will fall down at higher frequencies. AudioXpress magazine recently published a very interesting two-part article by Walt Jung (April and May 2007) about CCS circuits and measurements. In effect, the raw LM317 has a roughly 400K resistance in shunt with a 350pF capacitance, and both components will be non-linear to a degree. A CCS in the cathode side of a tube circuit can be less critical than in the plate side because the impedances are much lower, but I would prefer to see the LM317 cascoded with a BJT or FET to reduce the effective capacitance (the resistance will go higher too in cascode, but 400K at low frequencies was already plenty good enough for this situation). Jung shows a circuit with an LM317 cascoded by a DN2540, giving outstanding performance, close to measurement floor across the entire frequency band.
As Johan notes, an LM317 used as a CCS will fall down at higher frequencies. AudioXpress magazine recently published a very interesting two-part article by Walt Jung (April and May 2007) about CCS circuits and measurements. In effect, the raw LM317 has a roughly 400K resistance in shunt with a 350pF capacitance, and both components will be non-linear to a degree. A CCS in the cathode side of a tube circuit can be less critical than in the plate side because the impedances are much lower, but I would prefer to see the LM317 cascoded with a BJT or FET to reduce the effective capacitance (the resistance will go higher too in cascode, but 400K at low frequencies was already plenty good enough for this situation). Jung shows a circuit with an LM317 cascoded by a DN2540, giving outstanding performance, close to measurement floor across the entire frequency band.
Brian,
Your suggestion about caps between the cathodes, and then the earth reference lifted with a 1meg resistor seems appropriate here. Worked for me with the LM317 and avoids the need for bias adjustment.
With this circuit I always wondered what was the mechanism by which the second valve was driven - now I know. I wouldn't dismiss it as a design as a differential output stage has outstanding performance.
Shoog
Your suggestion about caps between the cathodes, and then the earth reference lifted with a 1meg resistor seems appropriate here. Worked for me with the LM317 and avoids the need for bias adjustment.
With this circuit I always wondered what was the mechanism by which the second valve was driven - now I know. I wouldn't dismiss it as a design as a differential output stage has outstanding performance.
Shoog
Shoog,
I guess you're referring to the the two-CCS circuit we discussed in another thread. Yes, it should work here too, keeping each tube separately DC biased, while at the same time connecting the cathodes with caps for diff-amp behavior.
For a very simple design, I agree that this one-sided-drive diff-amp output stage amp should not be dismissed. But the CCS(s) has to be stout.
I guess you're referring to the the two-CCS circuit we discussed in another thread. Yes, it should work here too, keeping each tube separately DC biased, while at the same time connecting the cathodes with caps for diff-amp behavior.
For a very simple design, I agree that this one-sided-drive diff-amp output stage amp should not be dismissed. But the CCS(s) has to be stout.
Can anybody suggest a circuit for two pairs of output tubes in push pull? My idea would be to use a single pair of coupling caps and two CCS - one under each pair of tubes. Others might disagree. this is quite an important subject because pairs of output tubes alow us to use a lot of the smaller good sounding triodes like 12b4, 6s4, 1626, 6CK4, 6AH4 etc.
zobsky said:
You probably could, but the 10M45 is a 450 volt part, and is overkill here. In the cathode circuit, we only have to contend with a few volts and that gives us more options. Surprisingly, in Walt Jung's recent CCS tests in AudioXpress, the LM317 and 10M45 were comparable (at low voltages). In fact, the LM317 was slightly better at frequencies below 20KHz, and the 10M45 was slightly better above 20KHz, but the differences were slight (and probably would vary unit-to-unit). In any case, an LM317 plus DN2540 cascode is far superior to either the LM317 or 10M45 alone.
Yvesm said:
Yvesm,
I think we actually agree here. Tobruk appeared to be asking about the same project, but making each question a separate thread. His choice, but exactly as you say. If we had all the questions under one heading, various bits of advice would appear more consolidated than a number of replies not taking the whole into account.
Regards.
CCSs in LTPs
On a practical note one need also take into account that the required internal impedance of a CCS will depend on the application. I mentioned elsewhere that for a typical ECC81 ltp phase inverter, the impedance at the cathodes is of the order of 1K5. I see diminishing returns to use a super CCS there with a Z of say several megohm. I would believe the effective combined cathode impedance in PP EL84s would be considerably lower. Some CCSs can give more problems with temperature dependance than their extra high Z will solve.
On a practical note one need also take into account that the required internal impedance of a CCS will depend on the application. I mentioned elsewhere that for a typical ECC81 ltp phase inverter, the impedance at the cathodes is of the order of 1K5. I see diminishing returns to use a super CCS there with a Z of say several megohm. I would believe the effective combined cathode impedance in PP EL84s would be considerably lower. Some CCSs can give more problems with temperature dependance than their extra high Z will solve.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.