> I do have the Hot Tuna Burgers album.
Bah. You had to BE THERE when they had their BIG GYM SYSTEM on tour mid-1970s.
I must admit..... I stopped 1/4 mile from the gym and realized I didn't want to go inside.
I (my company) actually did their sound, much later, in a much quieter period, in a much smaller room.
> put the trim pot, in Vb-e multipliers, in the b-e leg with a series limit resistor
Good point. But I already said "reckless" so I better stay the course. So he blows-up two bucks of transistors. He doesn't have short-protection either, which is reckless. He can do the frills on the final prototype.
> a post-1646 Vbe multiplier
No, put it before the 1646. That's your great discovery: a DC voltage injected here gets doubled and flipped to both lo-Z outputs.
LED is cool too.
> some discussion elsewhere of a 27 ohm source impedance being desirable ..., but it isn't clear (to me) what the intent of that circuit is.
I was aiming at a "universal maximum power" headphone amp for high-level monitoring. On another forum I plotted the impedance, sensitivity, and power rating of many-many headphones. The vast majority can be driven to very high level and near max power with a source which is 7Vrms and 27 ohms. The hi-Z phones get 6V, the lo-Z phones get 3V. In older days, the low-Z phones typically had higher power ratings and lower power sensitivity, and (for no apparent logical reason) they almost-all converged toward 7V 27 ohm source.
The 27 ohms could be "real" or "virtual". You could have a zero-Z output, but hide 27 ohms worth of sag in the power supply. (Though if there's significant storage after the sagger, you could pop a weak phone before it sagged.)
> use a reasonable build-out resistor
My study put a number on "reasonable". Or semi-reasonable: 7V 27R is WAY loud.
Here's a scatter-plot:
This derives the 7.3V 29R line and some other popular plans:
Walkman/iPod devices must lay near the green line.
Note that the 2-AA, IEC, and (30mA) opamp lines won't go near 120dB SPL (nor should we usually want to go anywhere near there). The 32W speaker amp with 100R buildout is able to overpower most phones (indeed I've killed a few by leaving them plugged-in to such a source).
> If the driver in the headphone exhibits a non-flat load that will interact with any source impedance causing a response error
My tests on a few hi- and lo-Z phones say that many hi-Z phones have impedance bumps and most lo-Z ones don't, or just a mild rise at 20KHz. Some very casual abuse says that "damping" per se is not important (these are not room speakers) except that when you drive the coils out of the gap (easy to do on cheapies) damping does reduce the racket a hair (it obvious can't have any effect once the coil actually gets away from the gap). The flat-Z phones should hardly care what the source Z is. Even when there is a 20KHz rise, the "error" from low or high drive Z is less than the overall error of a pennyworth of mylar. The 300 ohm jobs that rise to 600 ohms at 400Hz should be driven from a specified impedance to give the specified response... but it is rarely specified. There is an IEC test-spec which mentions a 120 ohm resistor, but I never saw a headphone sales sheet which suggested any such thing. My tin-ear impression is that the AKG 600 are "flatter" with source of 100 ohms or less, but maybe that's my bad taste.
Even 5 years ago, most headphones were re-badged 1980s designs. But the iPod has stirred the low-Z pot. The buds I got free with my iPod appear to be more sensitive than some high-price closed-ear traditional phones wound to a similar Z. The economics of pocket batteries are harsh, and Apple can afford to develop thrifty drivers to make the most of that wafer-thin battery. (My Nano is smaller than the clicker for my Honda, and holds a LOT more tunes than the 5-CD radio in the dash.) So my 7V 27R guide may be going out of date, in the direction of "excessive", at least for iPod-tradition phones.
Oh... for portable battery use, you don't design for "universal impedance". In lo-Z loads, anything over a few volts is pure waste; in hi-Z work anything over a few mA is a waste. My aim was wall-power, where "waste" was using a 30VA transformer to get under a Watt of total audio. That's the fun of DIY home headphone amps: grotesque waste is trivial cost. I could not afford a 6,000VA transformer for my 2*100W loudspeaker amp, but at headphone levels it was junkbox dregs. (One of mine uses a power wart stolen from a burglar alarm system.)
> THAT1646 at these low voltage levels is it's -101 dBu typical noise.
Well, at gain of 2?
I expected to have to tap low line levels and judge system hiss/buzz levels. A previous hack had been inadequate at gain of 5 and had to be torqued to gain of 10, still not generous. I find that gain of 50 is "enough". Far too much 99% of the time, and gain near 2 is often comfortable as a confidence monitor. But I might also be tapping quite high-level sources. To get almost 40dB range of gain without noise or overload compromises (and with fairly high-hiss BiFet devices) I felt I had to use an active variable-gain stage rather than a simple loss-pot.
> I've always been happy with using one of my preamp cards as headphone amps
Of course. Headphone tradition IS telephone tradition, and a pro Line Output is a telco line driver. I believe many studio phones evolved to work with the 150 ohm tap of a +18dBm Line Amp. And while a single telephone earpiece is 200-500 ohms, the telco also had "bridging" phones of 2K impedance and hotter magnet so an operator could tap a line with little drop; an old Sennheiser was made in 2K impedance and could almost be dropped across an active broadcast line without being noticed.
It is the 3V Walkman and iPod which changed the world. Now it is about power at 1V signal.
Bah. You had to BE THERE when they had their BIG GYM SYSTEM on tour mid-1970s.
I must admit..... I stopped 1/4 mile from the gym and realized I didn't want to go inside.
I (my company) actually did their sound, much later, in a much quieter period, in a much smaller room.
> put the trim pot, in Vb-e multipliers, in the b-e leg with a series limit resistor
Good point. But I already said "reckless" so I better stay the course. So he blows-up two bucks of transistors. He doesn't have short-protection either, which is reckless. He can do the frills on the final prototype.
> a post-1646 Vbe multiplier
No, put it before the 1646. That's your great discovery: a DC voltage injected here gets doubled and flipped to both lo-Z outputs.
LED is cool too.
> some discussion elsewhere of a 27 ohm source impedance being desirable ..., but it isn't clear (to me) what the intent of that circuit is.
I was aiming at a "universal maximum power" headphone amp for high-level monitoring. On another forum I plotted the impedance, sensitivity, and power rating of many-many headphones. The vast majority can be driven to very high level and near max power with a source which is 7Vrms and 27 ohms. The hi-Z phones get 6V, the lo-Z phones get 3V. In older days, the low-Z phones typically had higher power ratings and lower power sensitivity, and (for no apparent logical reason) they almost-all converged toward 7V 27 ohm source.
The 27 ohms could be "real" or "virtual". You could have a zero-Z output, but hide 27 ohms worth of sag in the power supply. (Though if there's significant storage after the sagger, you could pop a weak phone before it sagged.)
> use a reasonable build-out resistor
My study put a number on "reasonable". Or semi-reasonable: 7V 27R is WAY loud.
Here's a scatter-plot:
This derives the 7.3V 29R line and some other popular plans:
Walkman/iPod devices must lay near the green line.
Note that the 2-AA, IEC, and (30mA) opamp lines won't go near 120dB SPL (nor should we usually want to go anywhere near there). The 32W speaker amp with 100R buildout is able to overpower most phones (indeed I've killed a few by leaving them plugged-in to such a source).
> If the driver in the headphone exhibits a non-flat load that will interact with any source impedance causing a response error
My tests on a few hi- and lo-Z phones say that many hi-Z phones have impedance bumps and most lo-Z ones don't, or just a mild rise at 20KHz. Some very casual abuse says that "damping" per se is not important (these are not room speakers) except that when you drive the coils out of the gap (easy to do on cheapies) damping does reduce the racket a hair (it obvious can't have any effect once the coil actually gets away from the gap). The flat-Z phones should hardly care what the source Z is. Even when there is a 20KHz rise, the "error" from low or high drive Z is less than the overall error of a pennyworth of mylar. The 300 ohm jobs that rise to 600 ohms at 400Hz should be driven from a specified impedance to give the specified response... but it is rarely specified. There is an IEC test-spec which mentions a 120 ohm resistor, but I never saw a headphone sales sheet which suggested any such thing. My tin-ear impression is that the AKG 600 are "flatter" with source of 100 ohms or less, but maybe that's my bad taste.
Even 5 years ago, most headphones were re-badged 1980s designs. But the iPod has stirred the low-Z pot. The buds I got free with my iPod appear to be more sensitive than some high-price closed-ear traditional phones wound to a similar Z. The economics of pocket batteries are harsh, and Apple can afford to develop thrifty drivers to make the most of that wafer-thin battery. (My Nano is smaller than the clicker for my Honda, and holds a LOT more tunes than the 5-CD radio in the dash.) So my 7V 27R guide may be going out of date, in the direction of "excessive", at least for iPod-tradition phones.
Oh... for portable battery use, you don't design for "universal impedance". In lo-Z loads, anything over a few volts is pure waste; in hi-Z work anything over a few mA is a waste. My aim was wall-power, where "waste" was using a 30VA transformer to get under a Watt of total audio. That's the fun of DIY home headphone amps: grotesque waste is trivial cost. I could not afford a 6,000VA transformer for my 2*100W loudspeaker amp, but at headphone levels it was junkbox dregs. (One of mine uses a power wart stolen from a burglar alarm system.)
> THAT1646 at these low voltage levels is it's -101 dBu typical noise.
Well, at gain of 2?
I expected to have to tap low line levels and judge system hiss/buzz levels. A previous hack had been inadequate at gain of 5 and had to be torqued to gain of 10, still not generous. I find that gain of 50 is "enough". Far too much 99% of the time, and gain near 2 is often comfortable as a confidence monitor. But I might also be tapping quite high-level sources. To get almost 40dB range of gain without noise or overload compromises (and with fairly high-hiss BiFet devices) I felt I had to use an active variable-gain stage rather than a simple loss-pot.
> I've always been happy with using one of my preamp cards as headphone amps
Of course. Headphone tradition IS telephone tradition, and a pro Line Output is a telco line driver. I believe many studio phones evolved to work with the 150 ohm tap of a +18dBm Line Amp. And while a single telephone earpiece is 200-500 ohms, the telco also had "bridging" phones of 2K impedance and hotter magnet so an operator could tap a line with little drop; an old Sennheiser was made in 2K impedance and could almost be dropped across an active broadcast line without being noticed.
It is the 3V Walkman and iPod which changed the world. Now it is about power at 1V signal.
I took consolation in knowing that more sound quality is controlled by how you use the parts, than what you pay for them, and being a decent sized manufacturer means we could get a lot of part for our money. I recall using something like 1 million small signal diodes a month, so when you call the vendors listen. 
