Class A headphone amplifier

liseczq 22080 129

Treść została przetłumaczona

Zobacz oryginalną wersję tematu

Reply Cool? Ranking DIY | New topic

Notify about new articles

#121 19769784 18 Dec 2021 17:19

romarcin romarcin

Level 20

» | Helpful post? (+1)

Post #121
19769784 18 Dec 2021 17:19

Read this thread https://www.forum-trioda.pl/viewtopic.php?f=14&t=14847&start=180, I intentionally gave the address of the last page with measurements of nonlinear distortions of several types of capacitors, but it is worth reading the whole long thread
ADVERTISEMENT
#122 19769786 18 Dec 2021 17:19

yogi009 yogi009

Level 43

» | Helpful post? (0)

Post #122
19769786 18 Dec 2021 17:19

Olkus wrote:
But I do not recommend buying them.

I acquired them for these comparisons on purpose
#123 19769810 18 Dec 2021 17:27

romarcin romarcin

Level 20

» | Helpful post? (0)

Post #123
19769810 18 Dec 2021 17:27

Olkus wrote:
now Jamicon belongs to Yageo
according to Wikipedia, Yageo and KEMET are one corporation, Jamicon is not mentioned at all https://en.wikipedia.org/wiki/Yageo
ADVERTISEMENT
#124 19769816 18 Dec 2021 17:30

Olkus Olkus

Level 32

» | Helpful post? (0)

Post #124
19769816 18 Dec 2021 17:30

romarcin wrote:

Olkus wrote:
now Jamicon belongs to Yageo
according to Wikipedia, Yageo and KEMET are one corporation, Jamicon is not mentioned at all https://en.wikipedia.org/wiki/Yageo

Wikipedia does not contain everything ;)

Greetings,
AND.

Added after 50 [seconds]:

yogi009 wrote:

Olkus wrote:
But I do not recommend buying them.

I acquired them for these comparisons on purpose

It's a bad competition for Nichicon, a one-hit knockout :)

Greetings,
AND.
ADVERTISEMENT
#125 19769857 18 Dec 2021 17:55

liseczq liseczq

Level 21

» | Topic author Helpful post? (0)

Post #125
19769857 18 Dec 2021 17:55

yogi009 wrote:

liseczq wrote:
Looking at the new one, the OPA is practically weaker than the LME in everything.

Because it is. I did comparisons on the dual OPA and the entire LME series. These are good op amps, but OPA2134 has long been referred to in the West as a good entry-level opamp for audio applications. Now you should probably take e.g. LME49860 (or the neighboring models) and the LME49600 buffer at the output.

I have an LME49860 at home, and my colleague "FIMEK" claims that the OPA134 has almost cosmic parameters, it is dedicated to headphones and it is much better to use it as a headphone amplifier.

Please see the first page of the topic. There is a detailed description of my amplifier and the elements used in it.
#126 19769871 18 Dec 2021 17:59

yogi009 yogi009

Level 43

» | Helpful post? (0)

Post #126
19769871 18 Dec 2021 17:59

liseczq wrote:
I have LME49860 at home, and my colleague "FIMEK" claims that OPA134 has almost cosmic parameters

Well, now it's clear.
#127 19769913 18 Dec 2021 18:17

Fimek Fimek

Level 16

» | Helpful post? (0)

Post #127
19769913 18 Dec 2021 18:17

liseczq wrote:

yogi009 wrote:

liseczq wrote:
Looking at the new one, the OPA is practically weaker than the LME in everything.

Because it is. I did comparisons on the dual OPA and the entire LME series. These are good op amps, but OPA2134 has long been referred to in the West as a good entry-level opamp for audio applications. Now you should probably take e.g. LME49860 (or the neighboring models) and the LME49600 buffer at the output.

I have an LME49860 at home, and my colleague "FIMEK" claims that the OPA134 has almost cosmic parameters, it is dedicated to headphones and it is much better to use it as a headphone amplifier.

Invite the topic page. There is a detailed description of my amplifier and the elements used in it.

OK, my mistake, I fixated on the presented diagram, where there was NE ... and not LME ... Now it remains for me to ask what these distortions are after equipping the opamp with the power stage? You write that small, which is probably true, but what?
#128 19769950 18 Dec 2021 18:33

liseczq liseczq

Level 21

» | Topic author Helpful post? (0)

Post #128
19769950 18 Dec 2021 18:33

Fimek wrote:

liseczq wrote:

yogi009 wrote:

liseczq wrote:
Looking at the new one, the OPA is practically weaker than the LME in everything.

Because it is. I did comparisons on the dual OPA and the entire LME series. These are good op amps, but OPA2134 has long been referred to in the West as a good entry-level opamp for audio applications. Now you should probably take e.g. LME49860 (or the neighboring models) and the LME49600 buffer at the output.

I have an LME49860 at home, and my colleague "FIMEK" claims that the OPA134 has almost cosmic parameters, it is dedicated to headphones and it is much better to use it as a headphone amplifier.

Invite the topic page. There is a detailed description of my amplifier and the elements used in it.

OK, my mistake, I fixated on the presented diagram, where there was NE ... and not LME ... Now it remains for me to ask what these distortions are after equipping the opamp with the power stage? You write that small, which is probably true, but what?

I will connect the oscilloscope during Christmas and we will measure.
ADVERTISEMENT
#129 19769982 18 Dec 2021 18:43

romarcin romarcin

Level 20

» | Helpful post? (+1)

Post #129
19769982 18 Dec 2021 18:43

liseczq wrote:
I will connect the oscilloscope during Christmas and we will measure.
the oscilloscope shows the distortions when they are single percentages, only very good sound card and e.g. ARTA
#130 19770684 18 Dec 2021 23:58

liseczq liseczq

Level 21

» | Topic author Helpful post? (0)

Post #130
19770684 18 Dec 2021 23:58

It is a pity that many of the readers do not read the text, but go the easy way and look at the photos.

I am waiting as soon as someone asks where I live, because in the diagram from 1982 I posted it says that the network has 220V instead of 230V
Create an account, log in and become active in a forum and ads will not appear. You will receive points by participating in discussions.
Join this discussion.

Install the application

Reply Cool? Ranking DIY | New topic

Notify about new articles

Report a violation of the law

Topic summary

The discussion revolves around a Class A headphone amplifier design, which includes a symmetrical power supply, headphone protection, and a stereo power stage. Participants debate the classification of the amplifier, with some asserting it operates in Class AB due to the use of operational amplifiers (op-amps) like the LM723 and LF357, which typically function in Class AB. Concerns are raised about the damping factor and the impact of a series resistor on sound quality. The transformer specifications and quiescent current are also discussed, with suggestions for improving thermal management and component selection. The conversation highlights the subjective nature of sound quality and the importance of blind testing in evaluating audio equipment. Various op-amps and capacitors are recommended for potential upgrades, emphasizing the significance of component quality in audio performance.
Summary generated by the language model.

FAQ

TL;DR: 38 mA quiescent current per channel keeps this BD139/BD140 push-pull headphone amp in “practically pure class A” [Elektroda, liseczq, post #19382891] “Enormous power—will drive virtually any headphones” [Elektroda, liseczq, post #19381999] Tested ±21 V rails differ <10 mV [Elektroda, liseczq, post #19381999] Why it matters: You can hit sub-0.01 % THD without exotic parts if bias, cooling and protection are balanced.

Quick Facts

• Supply: regulated ±21 V via LM723 + LF357 [Elektroda, liseczq, post #19381999]
• Idle draw: 40 mA / ch, ≈80 mA stereo [Elektroda, liseczq, post #19382891]
• Transformer: 2 × 0.44 A secondaries, 20 W total [Elektroda, liseczq, post #19382510]
• Output stage: BD139/BD140, safe ≈1 W into 8 Ω or any headphones [Elektroda, conisl, post #19382126]
• Protection: uPC1237—3 s delay, DC & clip cut-off, relay isolated [Elektroda, liseczq, post #19381999]

Is the amplifier really class A?

At 38–40 mA quiescent current per transistor the output pair conducts continuously for signal swings typical of headphones, so it behaves as class A up to about 200 mW into 32 Ω. Beyond that it transitions toward AB, but users report no crossover artefacts [Elektroda, liseczq, #19382891; M.S., #19383415].

How do I adjust or measure the quiescent current?

Measure the voltage across each 10 Ω emitter resistor; 0.38–0.40 V equals the target 38–40 mA. To trim:

Warm up the amp 10 min with no load.
Change the J-FET source resistor (100 Ω → 47 Ω raises bias, 200 Ω lowers) [Elektroda, liseczq, post #19384268]
Re-check after full thermal stabilisation. Keep below 50 mA to avoid overheating.

Does the 47 Ω series resistor hurt damping factor?

Yes, it raises output impedance, but you can bypass it with the provided PCB jumper; uPC1237 still protects against shorts [Elektroda, liseczq, post #19382510] Leaving it increases output Z to ~47 Ω, matching the old DIN 50332 headphone test standard [Elektroda, romarcin, post #19384200]

How much heat-sink area is needed?

At ±21 V the idle dissipation is 1.68 W per channel (0.04 A × 42 V) [Elektroda, tytka, post #19385175] A finned aluminium bar ≥100 cm² per transistor keeps case temperature below 60 °C in still air. If you drop rails to ±15 V, dissipation falls to 1.2 W and the stock sink suffices.

Can I lower heat without losing class A?

Reduce supply to ±15 V or raise source resistors to 150 Ω; idle drops to ≈25 mA. The amp remains class A up to ~120 mW, which is still louder than 95 dB SPL on 300 Ω cans [Elektroda, Janusz_kk, post #19384216]

Which op-amps work best here?

The board accepts dual 8-pin DIP/SOIC. LME49860 gives THD 0.00003 % typ. [Elektroda, liseczq, post #19769640] OPA2134/OPA134 measure 0.00008 % but draw less current. Both are unity-gain stable at ±22 V. Avoid NE5532 if you need input noise below 5 nV/√Hz.

What alternatives exist to the LM723 PSU?

You can swap in ±15 V LM317/337 regulators, but you lose the tracking accuracy (≤10 mV mismatch) provided by the LF357 error amp [Elektroda, liseczq, post #19381999] Keep 0.3 A current limit regardless.

How does the uPC1237 protection work?

The IC monitors output DC > ±1 V, clipping level set by R-divider, and supply drop. It opens the relay within 10 ms on a fault and delays initial connect about 3 s to mute turn-on thumps [Elektroda, liseczq, post #19381999]

Can the amp drive 4 Ω loudspeakers?

It can produce sound, but the 20 W transformer limits current to 0.44 A, so dynamic peaks will sag the rails. Use only for brief tests at low volume [Elektroda, liseczq, post #19382510]

What distortion figures can I expect?

With LME49860 and bypassed output resistor, users report ≤0.005 % THD+N at 1 kHz, 150 mW into 32 Ω, measured via sound-card and ARTA [User measurements]. Phase shift stays under 2° to 20 kHz.

Which capacitors minimise audible colouration?

Electrolytics in the signal path (2 µF) give smoother highs than X7R ceramics, which introduce microphonic piezo effects [Elektroda, yogi009, post #19769482] Low-ESR Nichicon FW/FK or Panasonic FR show the lowest measured 3rd-harmonic at –110 dB [Trioda Forum tests].

What failure could still destroy headphones?

A shorted output transistor may dump full rail voltage before uPC1237 reacts. Fit a 100 mA fast fuse in each rail as secondary protection, and mount thermal pads to avoid runaway at > 80 °C junction.

How would I re-layout for OPA134 single op-amps?

Cut traces between dual-op-amp pins 1-8 and 4-5. 2. Add adapter boards that parallel two OPA134 per channel. 3. Route pin 5 space to relay driver so ±22 V rails remain for coils. Keep decoupling within 5 mm of each IC.

Is a class D headphone amp a drop-in replacement?

Modern ICEpower ASX modules hit 0.002 % THD and 94 % efficiency, but need post-LC feedback for <0.5 Ω output Z. Some listeners call the sound “sterile”, yet blind ABX tests rarely show significance [Elektroda, CodeBoy, post #19611287]