Ebay amp..

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I bought this headphone kit from ebay a number of years back,what I´m thinking of is it possible to change the 2 elyts after the opampout with some smaller plastic caps?Any other changes?

Ebay%20headphone%20amp_zpsqare3d8k.jpg
 
That looks like a fairly sensible design overall. The opamp is an NE5534, I guess? There is no output series resistor? Stability may benefit from a lttle Zobel network (33R/10n) to ground.

You could actually replace these two 47µs with a single 100µ over both diodes, which works pretty much the same. You wouldn't necessarily think so, but simulation shows otherwise.

I can see potential for improvement in that right now it's clearly a Class AB output stage. Please measure the voltage drop across one of these 1 ohm emitter resistors, that should give us a decent idea of quiescent current.

A picture of the actual board would also be very useful, as there are things that you cannot judge from a schematic alone, like expected thermal stability.
 
Try put there a 0,47u or more. If you wire opamp directly to diodes, then caps make no sense, and will affect the bias of end stage. You want just to feed AC to end stage, it should bias itself correctly. (or put to one side a resitive trimmer.)

Elyt caps suck big time! They are ok in bass, but for treble- foils are required. I confirmed this with expensive lc bridge.
Elyts at trebles are more behaving like chokes..:eek: (value starts drop to half on mid freqs..)
Got big improvement with bunch of polyprop wimas in my setup! (amp+supply):D
 

Attachments

  • Ebay headphone amp_zpsqare3d8k.jpg
    Ebay headphone amp_zpsqare3d8k.jpg
    72.9 KB · Views: 186
Try put there a 0,47u or more. If you wire opamp directly to diodes, then caps make no sense, and will affect the bias of end stage. You want just to feed AC to end stage, it should bias itself correctly. (or put to one side a resitive trimmer.)
???

The whole point of using caps there is "affecting bias" - more precisely, keeping output transistor base current from modulating diode current and hence bias level and output voltage. As any circuit simulator will show, this effect generates copious amounts of even-order distortion when driving loads - current from the opamp has to travel "backwards" through the upper diode (reducing its current, potentially to zero) to the npn output for positive peaks but "forward" through the lower one to the pnp for negative peaks. So positive peaks are dragged down due to increased source impedance while negative peaks even get a bit bigger. The same circuit simulator will also show that using caps will eliminate this effect at higher frequencies, and a single cap across both diodes does so only marginally less splendidly (one diode is always forward-biased and conducting). It still works well if said caps have an ohm worth of ESR.

All of this does not affect thermal tracking, as the capacitance needed to do so would be impractically large.
Elyt caps suck big time! They are ok in bass, but for treble- foils are required. I confirmed this with expensive lc bridge.
Elyts at trebles are more behaving like chokes..:eek: (value starts drop to half on mid freqs..)
I can only assume that your 'lytics must have been old axials in the 1000s of µF, and that you may have been misinterpreting the measurement results or had big parasitic inductance in your test setup. Of course they'll have some series inductance (and ESR), but even bigger values tend not to become clearly inductive until >100 kHz, and typical 47 µF ones as suggested here ought to be good to about 1 MHz, +/-. Surely that'll do for a Class AB audio amplifier?

I don't doubt that some local supply bypassing can help performance, see e.g. the fun to be had with the LM3886 in that regard. I just don't think it'll be worth it in this spot.
 
???

I can only assume that your 'lytics must have been old axials in the 1000s of µF, and that you may have been misinterpreting the measurement results or had big parasitic inductance in your test setup. Of course they'll have some series inductance (and ESR), but even bigger values tend not to become clearly inductive until >100 kHz, and typical 47 µF ones as suggested here ought to be good to about 1 MHz, +/-. Surely that'll do for a Class AB audio amplifier?

I don't doubt that some local supply bypassing can help performance, see e.g. the fun to be had with the LM3886 in that regard. I just don't think it'll be worth it in this spot.
With this new schema current will flow always, distortion must go down.


Bunch of <3 years old caps, new caps also measured, from 10u up to 3300u. Big were worst.
That value on can is valid often max to <150-300Hz, then it goes down quite fast.
This bridge does go seamlessly up to 200kHz
At 10Khz they were like few xµF or nF (220u/25); tantal-minimum improvement. (calculated by percentual loss of capacity by freq.)
i can guarantee, none of the elyts reached 20-30kHz. Foil caps were rock solid. Of course, Kelvin coax probes were calibrated.
(ideal powersupply is hard, but how can it be if caps are behaving like chokes, supply resistance goes up)

This is picture of LM3886 PSRR (opamps are same), what i believe bad semiconductor sound or instabilities are because of soft supply on HF. It´s up to you.
 

Attachments

  • Screen Shot 11-15-15 at 06.52 PM.PNG
    Screen Shot 11-15-15 at 06.52 PM.PNG
    37.7 KB · Views: 161
  • Screen Shot 11-15-15 at 07.16 PM.PNG
    Screen Shot 11-15-15 at 07.16 PM.PNG
    16.7 KB · Views: 155
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.