I want to build a headphone amp that has excellent technical specs, but also has few "psychoblunders". I.e, even if they have no real effect on the sound, I don't want any electrolytics in the signal path. It must be class A. No opamps, even if they have no faults. Actually, the new National parts are still tempting me. It probably won't have silver wire and Teflon caps, but I won't rule it out. I also won't rule out tubes, as long as the design is OTL. I've never been comfortable with transformers. I'm wary of Cdom compensation, having seen bad results with it in the past. I want just the right amount of negative feedback- enough to do the job, but not enough such that the amp could be labeled "high negative feedback".
What would you build? SS or tube? Bipolar or MOSFET? Massively paralleled devices (I have bags of TO-5 devices and bigger bags of TO-90 devices)? I have a lot of small signal tubes, but few power tubes. How complex a front end? Regulated or unregulated power supply. What do you feel strongly about?
Note that my headphones are ok dynamics, no fancy electrostatics. DCR is 47 ohms, 0.17 mH @ 1kHz. I don't listen terribly loud, probably quiet by most peoples standards.
What would you build? SS or tube? Bipolar or MOSFET? Massively paralleled devices (I have bags of TO-5 devices and bigger bags of TO-90 devices)? I have a lot of small signal tubes, but few power tubes. How complex a front end? Regulated or unregulated power supply. What do you feel strongly about?
Note that my headphones are ok dynamics, no fancy electrostatics. DCR is 47 ohms, 0.17 mH @ 1kHz. I don't listen terribly loud, probably quiet by most peoples standards.
What about the Cavalli-KumisaIII ?
Class A, Jfet input, Sziklai-output, fully complementary, all discrete, dc-coupled.
What more does one need ;-))
Haven't build mine yet, so unfortunately I cannot say anything about the sound.
All the best, Hannes
PS: forgot: it uses an opamp for the servo. But which servo doesn't? And without servo you have a ac-coupling cap at input.
Class A, Jfet input, Sziklai-output, fully complementary, all discrete, dc-coupled.
What more does one need ;-))
Haven't build mine yet, so unfortunately I cannot say anything about the sound.
All the best, Hannes
PS: forgot: it uses an opamp for the servo. But which servo doesn't? And without servo you have a ac-coupling cap at input.
http://headwize.com/projects/showfile.php?file=gilmore3_prj.htm
The designer's requirements list was similar to yours.
It sounds very good, btw.
The designer's requirements list was similar to yours.
It sounds very good, btw.
00940 said:
The design seems very good, as long as you can find the input dual FETs. The other design uses single FETs which seem to be available in the US.
The Gilmore's power supply design is better, even if the 2200uF capacitors are not quite recommended at the 3X7 regulators output. But a PS with those chips will sound much better than 7X15 types.
I found that one capacitor in a signal path could be better than many more active devices required to get rid of it. That was one of the reasons why I made this headphone amp few years ago:
http://www.ant-audio.co.uk/Theory/ANT Amber Headphone Amplifier.pdf
It is SE class A and the output FET could be changed to suit low- Ohm loads.
Here are the measurement results for the original version (BSP129 output):
http://www.ant-audio.co.uk/Data/Amber_0015.htm
And that for a version with FDN439N on the output.
http://www.ant-audio.co.uk/Data/Ruby02.htm
THD measurements done at 500 mV RMS output (-8dB on the graphs) - it is very loud on most headphones.
It is a simple circuit that gives a very good performance thought it requires a careful selection and matching of transistors. To my ear it sounds better than most (but I would say that, woudn't I?
)
Alex
http://www.ant-audio.co.uk/Theory/ANT Amber Headphone Amplifier.pdf
It is SE class A and the output FET could be changed to suit low- Ohm loads.
Here are the measurement results for the original version (BSP129 output):
http://www.ant-audio.co.uk/Data/Amber_0015.htm
And that for a version with FDN439N on the output.
http://www.ant-audio.co.uk/Data/Ruby02.htm
THD measurements done at 500 mV RMS output (-8dB on the graphs) - it is very loud on most headphones.
It is a simple circuit that gives a very good performance thought it requires a careful selection and matching of transistors. To my ear it sounds better than most (but I would say that, woudn't I?
)Alex
x-pro said:
Some will also say that the best cap is no cap at all, and for that you have to use more active devices. At least to make it symetric.
You could try that same design of yours bridged in mono, with no output caps, and see how it sounds as compared with the caps. That should be much better than any measurements you can provide.
PMA said:
I think you would be wrong to draw that conclusion from this particular circuit. Headphone amplifier requirements are very different to what you need from a power amp to drive loudspeakers. I prefer low distortion and a very clean (but musical) sound. In this case most of the time distortion would be much lower than as measured - it is SE class A after all. At 0.5 V on the headphones your ears most likely would distort much more! All my power amp designs were AB with high NFB and low THD and I am happy with these too
.carlmart said:
But not necessarily a better sound. I could easily improve measurements just by changing the load resistor in the output follower to a current source - and I 've tried it. However the sound was not as good - even at the levels where THD would be low anyway.
Alex
...for what it's worth, here's "my" headphone amp:
(not much of mine, as you'll see
)
I just use a "downsized" JLH.
All about the JLH can be found at
http://www.tcaas.btinternet.co.uk/
and then the idea to "downsize" I picked from the ESP pages:
http://sound.westhost.com/project70.htm
He proposes what is there called the "Headphone DoZ" amplifier (never found this name very polite...) (and DoZ is actually a variant of a JLH).
The JLH I made is the "no-caps" version, I think it's identified as "JLH 2003"; you'll find everything in the excellent site by Geoff Moss.
I used to plug the headphones on the "full JLH" for "test and debugging" purposes (most of it happens around midnight...)
of course through a "level reducer" (also inspired from ESP, basically some resistors - see project nr. 100).
Then I would just do all evening listening on headphones.
But all that heat was of course unnecessary, so I etched two more JLH cards, pulled some TIP35s from some old something, and there it is, a "low-power" JLH.
Just in case, I kept those resistors adapter. The level is just fine; voltage is +/- 18V and biased at 300mA.
Still massive heatsinks, but they were already attached behind the TIP35s
_
(not much of mine, as you'll see
)I just use a "downsized" JLH.
All about the JLH can be found at
http://www.tcaas.btinternet.co.uk/
and then the idea to "downsize" I picked from the ESP pages:
http://sound.westhost.com/project70.htm
He proposes what is there called the "Headphone DoZ" amplifier (never found this name very polite...) (and DoZ is actually a variant of a JLH).
The JLH I made is the "no-caps" version, I think it's identified as "JLH 2003"; you'll find everything in the excellent site by Geoff Moss.
I used to plug the headphones on the "full JLH" for "test and debugging" purposes (most of it happens around midnight...)
of course through a "level reducer" (also inspired from ESP, basically some resistors - see project nr. 100).
Then I would just do all evening listening on headphones.
But all that heat was of course unnecessary, so I etched two more JLH cards, pulled some TIP35s from some old something, and there it is, a "low-power" JLH.
Just in case, I kept those resistors adapter. The level is just fine; voltage is +/- 18V and biased at 300mA.
Still massive heatsinks, but they were already attached behind the TIP35s
_
...JLH does have a cap...
... oops, I don't know if the presence of the input coupling cap will disturb you...
sorry, forgot that.
(anyway, "all" my CD players have output caps as well... are you really for "strictly no cap all along the path", or just in the amp?)
(actually I'll have to remove some of those unnecessary ones...
)
_
... oops, I don't know if the presence of the input coupling cap will disturb you...
sorry, forgot that.
(anyway, "all" my CD players have output caps as well... are you really for "strictly no cap all along the path", or just in the amp?)
(actually I'll have to remove some of those unnecessary ones...
)_
stoolpigeon said:
Obviously closer to the reference - in a way that I can hear more detail and more music from the recording. I don't like distortion, however there are much more to a good reference sound than just lack of distortion - and those are very low anyway for a normal listening level here. Don't forget - it is not just the amplifier distortion that matters - headphones do produce much more distortion at these levels and is is the interaction between the amp and the headphones that counts!
Cheers
Alex
P.S. - and I am certainly one of "no capacitors in the signal path" people - all my designs for CD players and integrated/pre/power amps for Creek did not have "any" caps in the path of the signal. In this particular case of headphone amplification however I found that opposite works better. Mind you - in my design there is only one output cap - the VAS works from DC.
Many thanks- keep 'em coming! The Cavalli circuit is very appealing. Nothing jumps out at me saying, "gee, I'm not sure I'd do it that way." As I look at these circuits I'm finding that I have some strongly held opinions on topologies, not necessarily supported by anything concrete. Zero global feedback, for example, scares me. Probably a good reason to try it. IMO, knowing what's in the box can have an effect on the perceived sound, thus my comment about "psychoblunders", stuff that can affect perception, regardless of it's actual contribution to the signal. The simple coupled designs are reminiscent of some of the circuits from the early RCA transistor book. I built one of those as a headphone amp back in school around 1973, and it wasn't bad, even with several coupling caps. I've also done a class A design with an opamp and a couple MOSFETs, that I thought was excellent, but full discrete appeals to me more right now (ok, I have to use up some of my surplus discrete parts). One thing I no longer have is a preference for is symmetry, in fact even when a circuit looks symmetric, electrically it rarely is. I'm also willing to tolerate a reasonable value (film) input cap, but not an electrolytic on the output. Though I think it can be done with little or no real penalty, I don't wanna! Another requirement I didn't mention is noise level, though I most of these circuits will have a problem with it. The winning project will probably cherry pick various ideas from many circuits.
Conrad Hoffman said:
Hi Conrad,
my circuit is DC-capable, and you can try it without an output cap and with a split supply, separating the "ground" for the input and the output. It you make it "for your own ears only" than few volts DC on the headphones "ground" would not be a problem. The interesting feature of my circuit is that it lends itself very nicely for this kind of approach, as the VAS load (R15) is referenced to the second supply anyway (TL1431) and DC offset between that and the output could be made near 0 with a reasonably low drift. I may try it myself
.Cheers
Alex
Have you looked at the JE-990, or perhaps a derivative:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=107404
It's a fine design as it is and there's an AES article covering the design. Might want to consider an input cascode, perhaps JFETs, and higher output bias. Here's a power amp design that I'm working on that shows many of the input stage mods that I believe are worth taking a look at:
http://baselaudiolabs.googlepages.com/PTLB_POW1.JPG
Pete B.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=107404
It's a fine design as it is and there's an AES article covering the design. Might want to consider an input cascode, perhaps JFETs, and higher output bias. Here's a power amp design that I'm working on that shows many of the input stage mods that I believe are worth taking a look at:
http://baselaudiolabs.googlepages.com/PTLB_POW1.JPG
Pete B.
PB2 said:
Interesting Class-B design. Two things I couldn't quite understand the benefits of:
R22 feedback
D7 and D8 clamping diodes
On the diodes I could understand putting them in parallel with C8, but not at the input. I have never seen anyone using it at the input. I wonder if it may not introduce nonlinear distortions.
Have you already built this design?
I notice the thread on the JE-990 discusses the LM394. That's a very interesting device, and I built the RIAA circuit shown in the data sheet- the one that shuts off the inputs of an LM318 and uses the LM394, which contains a large number of matched transistors combined for low noise. I don't know if they're still made, but I have a few left. The preamp worked well, and my dad has it now. The amp design, however, isn't really what I'm after. At the moment I'm leaning towards Alex's no-global-feedback design, mostly because I consider it a walk on the wild side compared to the high feedback designs I usually favor. I also have most of the parts sans a couple of the semis. Odd, no one's suggested any tube circuits yet!
Headphones tubes amp which are both OTL and also without an output cap are not common you know High voltage is a tad scary to have without protection.
For an impedance of 47ohms, you need at least a 220uF output cap. It's a high value for film cap. It's possible to find though. If you want to go that way, many designs exist, just have a look in the library of the headwize website for exemple. The improved Morgan Jones maybe ? http://headwize.com/projects/showfile.php?file=cmoy5_prj.htm
But hey, if you're a bit crazy, there's Kevin Gilmore "madness" coming at your rescue. This : http://www.head-fi.org/forums/showthread.php?t=111554 is an OTL, OCL tube amp.
A last option could be an hybrid. I was always interested in that one : http://www.head-fi.org/forums/showthread.php?t=80748
High voltage is a tad scary to have without protection. For an impedance of 47ohms, you need at least a 220uF output cap. It's a high value for film cap. It's possible to find though. If you want to go that way, many designs exist, just have a look in the library of the headwize website for exemple. The improved Morgan Jones maybe ? http://headwize.com/projects/showfile.php?file=cmoy5_prj.htm
But hey, if you're a bit crazy, there's Kevin Gilmore "madness" coming at your rescue. This : http://www.head-fi.org/forums/showthread.php?t=111554 is an OTL, OCL tube amp.
A last option could be an hybrid. I was always interested in that one : http://www.head-fi.org/forums/showthread.php?t=80748
Swallow your psychoblunder pride:
Separate your 'phones into independent left and right (dual mono) channels. Then: stepped attenuator into Common-cathode amp using an EF86 strapped as a Triode, AC-coupled to a FET single-ended Class-A CCS-loaded Source follower. Careful with those supplies - shunt reg for the tube, series reg for the follower.
Go on, you know you want to.
Separate your 'phones into independent left and right (dual mono) channels. Then: stepped attenuator into Common-cathode amp using an EF86 strapped as a Triode, AC-coupled to a FET single-ended Class-A CCS-loaded Source follower. Careful with those supplies - shunt reg for the tube, series reg for the follower.
Go on, you know you want to.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.