Carl,
Nice circuit, beautifully drafted.
What is the consensus here about 0.1R emitter resistors? If the outputs are not carefully matched this will lead to substantial differences in current sharing. I found 0.22R worked much better.
Jeremy, what are your thoughts?
I spent about ten hours in simulation on this circuit, keeping the same topology so the same pcb could be used, and was unable to come up with anything which would enhance the sonics. I don't see this as a disadvantage, perhaps it's vindication this is an OK design. As a designer, I would naturally have done things differently, but working with the existing topology it was difficult to improve upon. I do feel the pcb design is ordinary, however.
Hugh
Nice circuit, beautifully drafted.
What is the consensus here about 0.1R emitter resistors? If the outputs are not carefully matched this will lead to substantial differences in current sharing. I found 0.22R worked much better.
Jeremy, what are your thoughts?
I spent about ten hours in simulation on this circuit, keeping the same topology so the same pcb could be used, and was unable to come up with anything which would enhance the sonics. I don't see this as a disadvantage, perhaps it's vindication this is an OK design. As a designer, I would naturally have done things differently, but working with the existing topology it was difficult to improve upon. I do feel the pcb design is ordinary, however.
Hugh
ULD Extreme
"What is the consensus here about 0.1R emitter resistors? If the outputs are not carefully matched this will lead to substantial differences in current sharing. I found 0.22R worked much better."
Hugh
Although these devices are more closely specced than many previous output devices, I think that it could be an advantage to achieve the better current sharing as you suggested.
Years ago I was occasionally annoyed by the slight distortion l noticed from one channel of my AEM6000 Mosfet Power Amplifier. I eventually traced the problem to uneven current sharing with the output FETs in this channel. Neither device was faulty.
Alex
"What is the consensus here about 0.1R emitter resistors? If the outputs are not carefully matched this will lead to substantial differences in current sharing. I found 0.22R worked much better."
Hugh
Although these devices are more closely specced than many previous output devices, I think that it could be an advantage to achieve the better current sharing as you suggested.
Years ago I was occasionally annoyed by the slight distortion l noticed from one channel of my AEM6000 Mosfet Power Amplifier. I eventually traced the problem to uneven current sharing with the output FETs in this channel. Neither device was faulty.
Alex
Although I can't add anything from a technical point of view, I can say that from a practical point of view I've switched between 0R1, 0R22, and 0R33 a few times now, and found the 0R22 much easier to work with, with no audible difference. I didn't find 0R33 an improvement in robustness - stability and matching were readily achieved with 0R22's. This does however assume the transistors are reasonably matched and heatsink equally bonded across output devices, otherwise you'll get the problem of one device getting all the load.
ULD Extreme
Carl
Personally, I would prefer to see the Zobel network reworked to
achieve a higher frequency rolloff point, perhaps with the capacitor reduced to 100nF,and the inductance more in the vicinity of 4.7uH. Vhfman has also reported that an inductor not wound on a former, but with no overlapping of the turns has been reported to result in improved SQ.
Perhaps it could be mounted vertically . Any comments regarding this area ? I do not expect our guru from Victoria to agree with me on this issue , however.
Alex
Carl
Personally, I would prefer to see the Zobel network reworked to
achieve a higher frequency rolloff point, perhaps with the capacitor reduced to 100nF,and the inductance more in the vicinity of 4.7uH. Vhfman has also reported that an inductor not wound on a former, but with no overlapping of the turns has been reported to result in improved SQ.
Perhaps it could be mounted vertically . Any comments regarding this area ? I do not expect our guru from Victoria to agree with me on this issue , however.
Alex
Alex,
I'm certainly no guru. I'm trained in the school of hard knocks, I simply know what works and what sounds good. Topologically this amp is extremely conventional. I won't argue on the 'Zobel' (it lacks a 10R resistor), but it is important to stability. This cap should have a 10R resistor under it, and it should precede the output inductor, not follow it!
I did find this amp very unstable with more than 15nF across the load. This makes it unusable with electrostatics, but then, that's nothing unusual, many SS amps are unsuitable with reactive loads.
Cheers,
Hugh
I'm certainly no guru. I'm trained in the school of hard knocks, I simply know what works and what sounds good. Topologically this amp is extremely conventional. I won't argue on the 'Zobel' (it lacks a 10R resistor), but it is important to stability. This cap should have a 10R resistor under it, and it should precede the output inductor, not follow it!
I did find this amp very unstable with more than 15nF across the load. This makes it unusable with electrostatics, but then, that's nothing unusual, many SS amps are unsuitable with reactive loads.
Cheers,
Hugh
Attached here is yet another revision. This one has the emitter resistors moved from 0R1 to 0R22 ohms.
What about the 6K8 1W resistor that is hanging off the BC556 collector and the 2 100R resistors in the upper left corner? I have in my notes that someone had said that it was not needed and should be replaced with a jumper.
Any thoughts on this? A concensus??
What about the 6K8 1W resistor that is hanging off the BC556 collector and the 2 100R resistors in the upper left corner? I have in my notes that someone had said that it was not needed and should be replaced with a jumper.
Any thoughts on this? A concensus??
ULD Extreme
Carl
It has been suggested by Douglas Self and others, that this resistor be removed. The main reason that was quoted appears to be due to problems when starting up with a reduced voltage from a Variac or similar.
I completely disagree with this suggestion, as this resistor reduces the dissipation through Q5, thus reducing current source variations as it heats up. I found with my Class A amplifier that this helped reduce temperature related drift of the front end during warm up, which also helps keep the differential pair balanced over a wider temperature range.
Alex
Carl
It has been suggested by Douglas Self and others, that this resistor be removed. The main reason that was quoted appears to be due to problems when starting up with a reduced voltage from a Variac or similar.
I completely disagree with this suggestion, as this resistor reduces the dissipation through Q5, thus reducing current source variations as it heats up. I found with my Class A amplifier that this helped reduce temperature related drift of the front end during warm up, which also helps keep the differential pair balanced over a wider temperature range.
Alex
Carl,
LTP stage current is a tad over 6mA. If we bridge this 6k8 resistor, then close to full rail voltage will be dropped across Q3, dissipating 330mW in this device. It's rated to 500mW, so this is far too much; as a general rule, TO92s should not be taken over 200mW for reliability.
In addition, this resistor prevents the LTP turning on until rail voltage reaches 43 volts, and this prevents the offset control mechanism and the bias system from activating, so that no current flows through the outputs and they remain off.
IMHO the 6k8 resistor should be left in.
Cheers,
Hugh
LTP stage current is a tad over 6mA. If we bridge this 6k8 resistor, then close to full rail voltage will be dropped across Q3, dissipating 330mW in this device. It's rated to 500mW, so this is far too much; as a general rule, TO92s should not be taken over 200mW for reliability.
In addition, this resistor prevents the LTP turning on until rail voltage reaches 43 volts, and this prevents the offset control mechanism and the bias system from activating, so that no current flows through the outputs and they remain off.
IMHO the 6k8 resistor should be left in.
Cheers,
Hugh
Carl,
If you go three pairs, you have the option of using 5 diodes in a string, which Mihai Rauta (Roender designer) feels is optimum in light of the 1.7mV/C v. 2.1mV/C tempco discrepancy in diodes/outputs.
However, to achieve the correct bias voltage, the current through this string should be around 2.5mA, but the fact is that the VAS (Q9 - BF469) is passing just over 12mA. I feel that four diodes should be used, as in the present design, with a pot. This makes the pcb pretty much OK for reuse, too.
Further, by increasing the emitter resistors from 0.1R to 0.22R we reduce the likelihood of thermal runaway, and thus bias control will be more stable in all thermal conditions.
This is a good compromise. The best option would be a Vbe multiplier carefully tailored for the 2.1mV/C tempco of the outputs, but this negates the benefits of using these new output devices.
Hi John,
To control switch on thumps, I generally try to get the LTP operational as soon as possible, and use a TO-126 as CCS LTP. I believe you take the same approach?
Cheers,
Hugh
If you go three pairs, you have the option of using 5 diodes in a string, which Mihai Rauta (Roender designer) feels is optimum in light of the 1.7mV/C v. 2.1mV/C tempco discrepancy in diodes/outputs.
However, to achieve the correct bias voltage, the current through this string should be around 2.5mA, but the fact is that the VAS (Q9 - BF469) is passing just over 12mA. I feel that four diodes should be used, as in the present design, with a pot. This makes the pcb pretty much OK for reuse, too.
Further, by increasing the emitter resistors from 0.1R to 0.22R we reduce the likelihood of thermal runaway, and thus bias control will be more stable in all thermal conditions.
This is a good compromise. The best option would be a Vbe multiplier carefully tailored for the 2.1mV/C tempco of the outputs, but this negates the benefits of using these new output devices.
Hi John,
To control switch on thumps, I generally try to get the LTP operational as soon as possible, and use a TO-126 as CCS LTP. I believe you take the same approach?
Cheers,
Hugh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.