Are the decoupling caps for the local regulators close to the regulators?
Also note that the pinouts for the positive and negative regulators are different. Pin 2 of the package is the lowest voltage on the regulator die. Thus, for the positive regulator pin 2 is GND, whereas for the negative regulator it's -Vin.
Tom
Also note that the pinouts for the positive and negative regulators are different. Pin 2 of the package is the lowest voltage on the regulator die. Thus, for the positive regulator pin 2 is GND, whereas for the negative regulator it's -Vin.
Tom
Well, after some debugging I found that removing diode across pins 2 and 3 of the 79L15 did the trick. I think I put it wrong way and let the ps -v go directly to the 411 pin 4.
At the moment I get the required voltage without the diode, so I think I´ll leave it this way.
Tom, the decoupling caps are close to the regulators, so no problems here.
Andrew, exactly, I got around 47v between pins before solving the problem, but not sure if they are toasted. I´m measuring offset between pins 9 and 10 and I got like 0,2-0,3mv with and without sound source playing. You think the 411s are working or not.
Whats the best procedure to measure offset?
Thanks again for your help.
At the moment I get the required voltage without the diode, so I think I´ll leave it this way.
Tom, the decoupling caps are close to the regulators, so no problems here.
Andrew, exactly, I got around 47v between pins before solving the problem, but not sure if they are toasted. I´m measuring offset between pins 9 and 10 and I got like 0,2-0,3mv with and without sound source playing. You think the 411s are working or not.
Whats the best procedure to measure offset?
Thanks again for your help.
Last edited:
Some updates.
Tested the amp with no 411s and the offset goes to 12mv and 20mv on the other channel.
Put back the 411s and Im reading 0,3mv and 0,8mv. Turn the volume pot and the offset goes up, and then drops again to those values, so I think it´s working.
Had to change some resistors to set a gain of 34db since my sound source is too low on voltage.
Im stunned with the sound, and even after increasing the gain, I can´t hear any distortion or quality drop.
Andrew, Tom; not only in this thread, but all your posts in that forum helped me a lot.
Thanks!!
Tested the amp with no 411s and the offset goes to 12mv and 20mv on the other channel.
Put back the 411s and Im reading 0,3mv and 0,8mv. Turn the volume pot and the offset goes up, and then drops again to those values, so I think it´s working.
Had to change some resistors to set a gain of 34db since my sound source is too low on voltage.
Im stunned with the sound, and even after increasing the gain, I can´t hear any distortion or quality drop.
Andrew, Tom; not only in this thread, but all your posts in that forum helped me a lot.
Thanks!!
measure the opamp output voltage on both channels.
output offset is measured with no input signal and the input loaded with a zero ohm dummy plug. The output is disconnected from the speaker/load.
Then it is worth checking with the input connected with a cable but an open end and then with a Source connected first with volume turned down to zero and then at ~ -6dB ref maximum (no speaker or it might be too loud). If the output is near clipping when vol is set to -6dB, then you have far too much gain in the amplifier.
+34dB gain for a 20Vac output (50W into 8ohms) needs only 400mVac as Maximum input. That's a very sensitive setting !
output offset is measured with no input signal and the input loaded with a zero ohm dummy plug. The output is disconnected from the speaker/load.
Then it is worth checking with the input connected with a cable but an open end and then with a Source connected first with volume turned down to zero and then at ~ -6dB ref maximum (no speaker or it might be too loud). If the output is near clipping when vol is set to -6dB, then you have far too much gain in the amplifier.
+34dB gain for a 20Vac output (50W into 8ohms) needs only 400mVac as Maximum input. That's a very sensitive setting !
Last edited:
If the inherent output offset were zero, then the opamp output would be near zero mV ref audio ground.
But we know the servo is applying some correction (injecting current) to bring the output offset from tens of mV to <1mV
That implies some opamp output voltage.
If the output voltage is over half of opamp rail voltage, then you should be looking at changing the injection resistor value.
If it's already near rail voltage, then you have a serious problem and need to re-design the servo and/or the power amplifier.
But your correction is very small so I would expect the opamp output voltage to be less than 1000mV ref audio ground.
In which case you could be looking at increasing the injection resistor value.
The injection resistor bypasses the normal feedback resistor. Any noise, or distortion, or remaining audio signal that passes from the opamp through the injecting resistor is applied to the -IN input and gets multiplied by the amplifier gain.
A larger injection resistor reduces the unwanted signal relative to the normal feedback signal.
A too large injection resistor makes the opamp work harder and in the extreme forces the opamp output towards the supply rail, after which it cannot provide an further correction.
BTW,
A DC coupled amplifier needs speaker protection. The DC servo simply minimises the output offset when everything is working correctly.
You NEED speaker protection for the rare occasion when something is not working correctly.
eg.
window comparator monitoring the opamp output voltage and turning ON a warning beacon/klaxon when opamp output voltage exceeds 50% of rail voltage.
a second window comparator monitoring opamp voltage and shuts down the amplifier when output exceeds 90% of rail voltage OR output offset detection that isolates speaker when a threshold is exceeded.
Last edited:
+ve voltage sends current into the -IN node.
-ve voltage sinks current from the -IN node.
You should find that one channel has a +ve offset and the other a -ve offset when the servo is disabled.
The voltage value and the resistor values give you an indication of the nA, or uA that is being sourced sunk to correct the offset.
Those very low output values tell me you should be quadrupling the injection resistor value. Maybe worth trying ten times value opn the lower channel, but don't make them different (it's just a stack up of different tolerances).
Did you see my edit above?
-ve voltage sinks current from the -IN node.
You should find that one channel has a +ve offset and the other a -ve offset when the servo is disabled.
The voltage value and the resistor values give you an indication of the nA, or uA that is being sourced sunk to correct the offset.
Those very low output values tell me you should be quadrupling the injection resistor value. Maybe worth trying ten times value opn the lower channel, but don't make them different (it's just a stack up of different tolerances).
Did you see my edit above?
Last edited:
Yeah, I know, but my sound source outputs only 250mV. Just will be careful when connecting other sources.
Last edited:
Yes! Exactly. The one injecting negative voltage had negative offset without the opamp plugged. didn´t realise it was negative before!
The injection resistor is the one between Rf and opamp pin 6? or the one between Rf and opamp pin 3.
250mVac is tiny for a maximum output level.
Could that 250mVac be the average level with normal music/audio files?
250mVac playing 440hz square wave at full volume. with normal music I think it can even drop.
post1:
R8 is the lower leg NFB
R3 is the upper leg NFB
Gain = 1+ R3/R8 = 50.1times
R4 is the injection resistor.
The injection voltage is divided down by the ratio of R4:R8
i.e. injection voltage applied to -IN is Vout * R8/(R8+R4)
Injection current is Vout / (R4+R8)
Have you gone back and studied that old servo Thread?
You appear to not understand what you are building.
To design a new, or modified circuit you need to understand it's operation.
If not, then the best you can achieve is to copy exactly an existing known and proved working design. Surely you are above that !
R8 is the lower leg NFB
R3 is the upper leg NFB
Gain = 1+ R3/R8 = 50.1times
R4 is the injection resistor.
The injection voltage is divided down by the ratio of R4:R8
i.e. injection voltage applied to -IN is Vout * R8/(R8+R4)
Injection current is Vout / (R4+R8)
Have you gone back and studied that old servo Thread?
You appear to not understand what you are building.
To design a new, or modified circuit you need to understand it's operation.
If not, then the best you can achieve is to copy exactly an existing known and proved working design. Surely you are above that !
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.