Can somone give a some insight into the pros and cons of the negative feedback used on these two very similar amps? Or perhaps some other critism or explanation about why these designs are good or bad?
Lehmann black cube - No Global negative feedback
Altronics K 5503 studio series . - Global negative feedback
I own the original lehmann and have recently built the very similar headphone amp from altronics. I would also like any suggestions about any improvements I can make to either design?
Lehmann black cube - No Global negative feedback
An externally hosted image should be here but it was not working when we last tested it.
Altronics K 5503 studio series . - Global negative feedback
An externally hosted image should be here but it was not working when we last tested it.
I own the original lehmann and have recently built the very similar headphone amp from altronics. I would also like any suggestions about any improvements I can make to either design?
The first one does have DC offset at the output, depending on the matching or not of the output stage devices. You may or may not want that.
In the second, this is, to a large extend, nulled out by the feedback.
In general, the second will have objectively lower distortion and better linearity. You may or may not want that ;-)
Personal preference for reproduction is just that, personal preference and hard from someone else to judge for you. Although, of course, many here will give their own preference and assume yours is the same ;-)
Since you own both, what do you think? Do you really need someone else to tell you how well or bad your own stuff sounds??
Jan
In the second, this is, to a large extend, nulled out by the feedback.
In general, the second will have objectively lower distortion and better linearity. You may or may not want that ;-)
Personal preference for reproduction is just that, personal preference and hard from someone else to judge for you. Although, of course, many here will give their own preference and assume yours is the same ;-)
Since you own both, what do you think? Do you really need someone else to tell you how well or bad your own stuff sounds??
Jan
To my ears they sound about the same.... More trying to better my understanding.
Another question.
Why is there no protection circuits on these and similar headphone amps?
Is this because the transistors are operating in a safe region and are considered too reliable and unlikely to fail?
Does the global negative feed back offer any protection in case of a failure?
Also how important is power supply quality with these designs, which of the two has better PSRR?
Another question.
Why is there no protection circuits on these and similar headphone amps?
Is this because the transistors are operating in a safe region and are considered too reliable and unlikely to fail?
Does the global negative feed back offer any protection in case of a failure?
Also how important is power supply quality with these designs, which of the two has better PSRR?
These designs aren't actually as similar as you may think - the Lehmann employs a two-stage buffer (a so-called diamond buffer), while the Altronics actually employs a single-stage buffer, with the two extra transistors being used as current source/sink for the bias diodes D5+D6 (D11+D12) at ~11.5 mA.
The Lehmann also tends to be reasonably far into Class A territory (about 75 mA of output stage quiescent current if memory serves), while the Altronics probably runs somewhere around 10-15 mA through its output stage, closer to Class AB. (The voltage drop across one of the 4.7 ohm emitter resistors will tell you the exact value.) That's also why it has this diode contraption (D1-D4), though I'm not exactly sure what this is supposed to achieve - you'd think the extra current from the opamp wouldn't make that much of a difference when the output stage runs out of breath. Some Class B (actually more like C) designs use a resistor there to "fill in" the dead spot in output stage response, which still makes some sense at least.
The diamond buffer circuit is competent enough to give decent results driving headphones on its own, but on the downside I haven't had very much luck running feedback across it in simulation. Due to the added pole, I had to limit the bandwidth so much that distortion ended up being substantially worse than for a single-stage buffer. I would not be surprised if negative feedback ended up being left out in the Lehmann because they had trouble getting the thing stable.
I can see a few possible tweaks on the Altronics:
1. Connect a 100 µF (or larger) electrolytic capacitor across D5+D6 (D11+D12), i.e. between the current source/sink collectors. This keeps bias diode current from ever approaching zero dynamically and results in dynamic buffer stage base current being limited by the opamp rather than current source/sink current. This design uses a healthy amount of bias current to begin with, but I'd assume a substantial reduction of even-order distortion into lower-impedance loads would still be observed.
2. Assuming the BD139/140 heatsinks (+ power supply + ventilation) approach those of the Lehmann, you can increase output stage bias current by unsoldering + lifting the opamp-facing end of D5/D6/D11/D12 and adding a Schottky diode (maybe 1N5817/18/19 or similar) in series with each 1N4148. It's not exactly a very flexible method of idle current adjustment but it does work.
3. You could apply class A biasing to the opamp output stage by unbalancing current sourcing + sinking. Increase the 100R resistor at Q1 to 150 ohms for ca. 4 mA of delta.
4. You can try a more beefy opamp for better driving of low-impedance loads, like the trusty NJM4556A (presumably the AD variety in DIP).
5. While not including a volume pot in this design at all will keep impedance imbalance at bay, this aspect may very well prove decidedly impractical. Plenty of source / headphone combinations will require overall gains of far less than 6 dB to keep the noise floor at bay. I would recommend a log pot no higher than 10k.
Input EMI filtering is not ideal on both. 100R + 47p on the Altronics may not do all that much, but whoever designed this Lehmann clone with a 10k resistor after the pot wiper decidedly overdid it (I'd be OK with a 1k, but 10k would make it the dominant input noise source).
The Lehmann also tends to be reasonably far into Class A territory (about 75 mA of output stage quiescent current if memory serves), while the Altronics probably runs somewhere around 10-15 mA through its output stage, closer to Class AB. (The voltage drop across one of the 4.7 ohm emitter resistors will tell you the exact value.) That's also why it has this diode contraption (D1-D4), though I'm not exactly sure what this is supposed to achieve - you'd think the extra current from the opamp wouldn't make that much of a difference when the output stage runs out of breath. Some Class B (actually more like C) designs use a resistor there to "fill in" the dead spot in output stage response, which still makes some sense at least.
The diamond buffer circuit is competent enough to give decent results driving headphones on its own, but on the downside I haven't had very much luck running feedback across it in simulation. Due to the added pole, I had to limit the bandwidth so much that distortion ended up being substantially worse than for a single-stage buffer. I would not be surprised if negative feedback ended up being left out in the Lehmann because they had trouble getting the thing stable.
This, plus protection circuitry adds a not-insubstantial amount of complexity and additional failure points of its own (e.g. relay contacts). DIYers will usually shy away from complexity and prefer to overbuild instead.
Depends on what said failure is, really. If an output transistor decides to go belly up, you're SOL, but there are some it can catch.
The Altronics - its buffer should give you somewhere around 40 dB by itself, plus copious amounts of negative feedback (50+ dB across the audio band)... the opamp or PCB layout might actually be the bottleneck. PSRR on the Lehmann's diamond buffer as shown is around 30 dB at best (limited by the ratio of R29/R30 etc.).
I can see a few possible tweaks on the Altronics:
1. Connect a 100 µF (or larger) electrolytic capacitor across D5+D6 (D11+D12), i.e. between the current source/sink collectors. This keeps bias diode current from ever approaching zero dynamically and results in dynamic buffer stage base current being limited by the opamp rather than current source/sink current. This design uses a healthy amount of bias current to begin with, but I'd assume a substantial reduction of even-order distortion into lower-impedance loads would still be observed.
2. Assuming the BD139/140 heatsinks (+ power supply + ventilation) approach those of the Lehmann, you can increase output stage bias current by unsoldering + lifting the opamp-facing end of D5/D6/D11/D12 and adding a Schottky diode (maybe 1N5817/18/19 or similar) in series with each 1N4148. It's not exactly a very flexible method of idle current adjustment but it does work.
3. You could apply class A biasing to the opamp output stage by unbalancing current sourcing + sinking. Increase the 100R resistor at Q1 to 150 ohms for ca. 4 mA of delta.
4. You can try a more beefy opamp for better driving of low-impedance loads, like the trusty NJM4556A (presumably the AD variety in DIP).
5. While not including a volume pot in this design at all will keep impedance imbalance at bay, this aspect may very well prove decidedly impractical. Plenty of source / headphone combinations will require overall gains of far less than 6 dB to keep the noise floor at bay. I would recommend a log pot no higher than 10k.
Input EMI filtering is not ideal on both. 100R + 47p on the Altronics may not do all that much, but whoever designed this Lehmann clone with a 10k resistor after the pot wiper decidedly overdid it (I'd be OK with a 1k, but 10k would make it the dominant input noise source).
Last edited:
Aside from the dozen-mV DC offset of the one without overall NFB... the distortion of the output stage is not corrected. Yes, the diamond buffer can be very low distortion, but not zero. This may be a "flavor choice"; some buyers/designers do not want perfect sound.
The 5503 is an AB stage without a trimmer. It is liable to crossover distortion. The Designer may have picked bias diodes and output devices for a happy idle current. Still, this is the kind of stage we "always" want NFB around to avoid hoarseness in soft passages.
Protection: The 5503 has those extra diodes. With emitter resistors, max current is about 0.133 Amps. And +/-15V supplies. The worst case for the transistor is 14V 133mA or 1.9 Watts, half the time, so 1 Watt. These BD parts are just big enough to throw-off a Watt for hours. (There is an 8 Ohm option which risks damage in extreme abuse, unless the BDs are heat-sunk.)
The Black is limited by the 1.5k pull-up resistors. Assuming BD hFE is like 200, the large-signal output impedance is around 8 Ohms and peak current on the order of 2 Amps. Yeah, unless there is a monster heatsink _I_ could kill it.
The saving graces is that headphone cables are less-likely to be run under carpet, and less-likely to ever be run FULL output like a loudspeaker amp at a party.
The 5503 is an AB stage without a trimmer. It is liable to crossover distortion. The Designer may have picked bias diodes and output devices for a happy idle current. Still, this is the kind of stage we "always" want NFB around to avoid hoarseness in soft passages.
Protection: The 5503 has those extra diodes. With emitter resistors, max current is about 0.133 Amps. And +/-15V supplies. The worst case for the transistor is 14V 133mA or 1.9 Watts, half the time, so 1 Watt. These BD parts are just big enough to throw-off a Watt for hours. (There is an 8 Ohm option which risks damage in extreme abuse, unless the BDs are heat-sunk.)
The Black is limited by the 1.5k pull-up resistors. Assuming BD hFE is like 200, the large-signal output impedance is around 8 Ohms and peak current on the order of 2 Amps. Yeah, unless there is a monster heatsink _I_ could kill it.
The saving graces is that headphone cables are less-likely to be run under carpet, and less-likely to ever be run FULL output like a loudspeaker amp at a party.
It is the simplest form of current limiting. (Yes, when it goes into limit the opamp's 0.030A adds to the BDs' 0.133A but that's incidental.)
Attachments
> only 133 mA?
I have *never* used 150mA to drive headphones on a head.
At 32 Ohms this is 280mW. This is 24dB above 1 milliWatt. Since most headphones make well over 90dB SPL@1mW, this is over 115dB SPL. I've worked enough around 115dBSPL systems for pay to know I'd never want that much for pleasure. (Unlike long-range loudspeakers, you can't walk-away from headphones for useful volume drop.)
I have *never* used 150mA to drive headphones on a head.
At 32 Ohms this is 280mW. This is 24dB above 1 milliWatt. Since most headphones make well over 90dB SPL@1mW, this is over 115dB SPL. I've worked enough around 115dBSPL systems for pay to know I'd never want that much for pleasure. (Unlike long-range loudspeakers, you can't walk-away from headphones for useful volume drop.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.