Those two op amps are bipolar input, and higher speed, the 8066, fet input op amp just didn’t like that circuit apparently, and may be oscillating it appears. Have you measured for dc offset at the outputs?
Looking at the feedback arrangement, I would stick with bipolar op amps for this one.
I always try to connect and disconnect my headphones with the power off to prevent that, but my amp also has a relay to delay the connection of the load when turning on, and immediately disconnect the load when shutting off.
Looking at the feedback arrangement, I would stick with bipolar op amps for this one.
I always try to connect and disconnect my headphones with the power off to prevent that, but my amp also has a relay to delay the connection of the load when turning on, and immediately disconnect the load when shutting off.
yeah, you are right. rail voltages are +/- 9v so no risk blowing it but the only reason I tried it was because a few people said it sound better than LM4562 in their amps, I suppose audio forums arent the best place to get electronics recommendations
i added a bunch 470n, huge 10u/600 polyprops into my tube amp, it was very noticeable. Even so the 1000u/450 wet cap.
but as i said, tube amps have low psrr..so every effort counts.
chargeable batteries are low impedance supply. i believe that is the reason why some prefer batteries.
Im thinking about adding voltage regulators to this amp and I noticed the RC circuits on the Op amp rails. Would these have any use or even negatively effect the circuit if there were voltage regulators?
Also if the purpose of R5/105 is to increase output impedance, what effects does that have? the headphones are 45 ohm if that means anything
Also if the purpose of R5/105 is to increase output impedance, what effects does that have? the headphones are 45 ohm if that means anything
Last edited:
It does not. The NFB resistors R3 R103 tap after the 22r resistors. Output impedance is essentially zero. R5 R105 prevent "infinite" current flowing through the transistors to a short.
_I_don't see any "sloppy" here. Very conventional design with ample values. The change from 4700u to 8000u should not be "wow". If your ears are so discerning, I would look at C2 as directly affecting the low end; a crap cap here will sound bad, so a super-duper cap may please you more.
Imho the output stage has a low PSRR at the neg. rail, causing the opamp to constantly "fight" against it as it gets injected directly into the nfb. As each chanell idles at 120mA-ish each, there sure must be some ripple in the rails (should sim that but ...). So any upping the psu caps would in consequence be an improvement. Perhaps just upping the neg. reservoir cap would be an improvement?
Would it be possible/advisable to add another op amp to each channel to increase the current output? the NJM4556 is a high current op amp and its drives the headphones noticeably better at higher volumes, it doesnt sound very good so im looking for a way to increase current of my preferable op amps such as LM4562 and AD823.
I am not sure why any reasonably powerful opamp would be running out of current in this circuit - I mean, the BD139 is running at 120 mA, 240 mA worst-case, that's pretty much where it has highest beta (>100), well before beta droop sets in.
That said, the circuit has its share of issues AFAICS:
1. No compensation cap between opamp output and -in. (10-100 pF NP0 or styrene, typ.)
2. Feedback taken after 22 ohm output series resistor. While it eliminates its effect on output impedance, it makes the circuit more prone to oscillation as well. I would rather remove the 22R, add anywhere from 1 to 10 ohms of output series resistance outside the loop (depending on what you can afford) and include series resistors in the collector leads if need be.
3. It is noisier than it would have to be, due to the feedback network being a tad high in impedance. There would also be a bit of DC offset due to DC resistances being mismatched - several 5532 types spec a maximum 500 nA of input bias current, which could get output DC offset to about 40 mV. Likewise, AC impedance is also mismatched, depending on what kind of source or volume pot is attached.
4. Distortion performance into mid-high impedance loads could be improved a fair bit by replacing the 100R/5W emitter resistors by basic current sources (sinks) - nothing too fancy required.
5. It is uncommon for a headphone amp not to have regulated supplies and rely on feedback to sort things out. Class A jobs in particular usually feature them due to their current demands creating greater amounts of ripple. At the very least, I would include some capacitance multipliers to clean up the unregulated supplies - make sure you get maybe 2-3 V of drop over them to work with (if bipolar, MOSFETs may require more), add feedback caps to slow the transistors down if needed for stability, and consider going with a higher-voltage transformer in order to maintain decently high supplies (maybe 2x 12 V).
At least some attention should be paid to ground routing, even though the circuit doesn't strike me as super-critical either. It is advised to combine headphone jack and C2 ground return before going back to the power supply, maybe add input and R1 ground as well. Power ground from the filter capacitors should be returned separately for good measure, though in this circuit there really isn't much that could happen.
That said, the circuit has its share of issues AFAICS:
1. No compensation cap between opamp output and -in. (10-100 pF NP0 or styrene, typ.)
2. Feedback taken after 22 ohm output series resistor. While it eliminates its effect on output impedance, it makes the circuit more prone to oscillation as well. I would rather remove the 22R, add anywhere from 1 to 10 ohms of output series resistance outside the loop (depending on what you can afford) and include series resistors in the collector leads if need be.
3. It is noisier than it would have to be, due to the feedback network being a tad high in impedance. There would also be a bit of DC offset due to DC resistances being mismatched - several 5532 types spec a maximum 500 nA of input bias current, which could get output DC offset to about 40 mV. Likewise, AC impedance is also mismatched, depending on what kind of source or volume pot is attached.
4. Distortion performance into mid-high impedance loads could be improved a fair bit by replacing the 100R/5W emitter resistors by basic current sources (sinks) - nothing too fancy required.
5. It is uncommon for a headphone amp not to have regulated supplies and rely on feedback to sort things out. Class A jobs in particular usually feature them due to their current demands creating greater amounts of ripple. At the very least, I would include some capacitance multipliers to clean up the unregulated supplies - make sure you get maybe 2-3 V of drop over them to work with (if bipolar, MOSFETs may require more), add feedback caps to slow the transistors down if needed for stability, and consider going with a higher-voltage transformer in order to maintain decently high supplies (maybe 2x 12 V).
At least some attention should be paid to ground routing, even though the circuit doesn't strike me as super-critical either. It is advised to combine headphone jack and C2 ground return before going back to the power supply, maybe add input and R1 ground as well. Power ground from the filter capacitors should be returned separately for good measure, though in this circuit there really isn't much that could happen.
I will try as you suggested and add 1 ohm resistors after feedback instead, When trying to use the OPA1612 there was intermittent hissing and noises. also when using non-JFET op amp in parallel there is similar problems. If the OPA1612 can work in parallel with this configuration then that will be perfect.
The headphones are low sensitivity planar magnetics and the LM4562 is 50mA, in parallel it makes a big difference especially at higher volumes.
these resistors are in the signal path, I hope after reducing impedance 20 times it could sound a microscopic bit better. removing the input attenuator and using digital volume control it sounded much cleaner, though that potentiometer was 5k ohm
The headphones are low sensitivity planar magnetics and the LM4562 is 50mA, in parallel it makes a big difference especially at higher volumes.
these resistors are in the signal path, I hope after reducing impedance 20 times it could sound a microscopic bit better. removing the input attenuator and using digital volume control it sounded much cleaner, though that potentiometer was 5k ohm
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.