MOSFET follower headphone amplifier

[GRollins]

If a picture's worth a thousand words, surely a schematic is worth at least five hundred.
I tried recently to describe a headphone amplifier, using only words. While it is possible, in principle, to unambiguously define all the parts of a circuit and how they fit together, it's pretty cumbersome. In an attempt to clear some of the fog, I dredged up the schematic.
It's a trivial circuit, something you can build using only junk box parts, but it performs very well, indeed.

Grey

[sek]

It is easy to, e.g., draw a 1000uF bipolar cap (and claim to have it in the junk box, for that matter).

*duck* I like it!

[EUVL]

Grey,

Shame on you !!
You should have posted a schematic with Loveltech's !!!




Patrick

[PMA]

Hi Grey,

what do you think about current source instead of source resistor, like here:

http://www.pha.inecnet.cz/macura/follower_e.html

and here:

http://sound.westhost.com/project83.htm

[GRollins]

sek,
Guilty as charged. I always draw caps with both sides as straight lines; the ones with one or both lines curved look funny to me. It's an esthetic thing, nothing more complicated than that. Yes, I could put a + on the silly thing, but I'm lazy.
It's how you know I'm not an anal-retentive EE.
For the record, the MOSFET shown is a (now discontinued) Motorola part very much like the IRF610. It's an enhancement mode device, so the Source is more negative than the Gate. Given that the Gate is at ground potential, that puts the Source negative relative to ground. Attach the DC blocking cap accordingly.
And what do you mean you don't have 1000uF film caps in your junk box? You need an upgraded junk box, man!
EUVL,
I'll be doing a Lovoltech power JFET headphone circuit, by and by. Note the date down in the corner of this one. I'm just posting something I threw together after treating myself to a new set of headphones a few years ago. I got frustrated trying to describe the circuit using only words a couple of days ago and have been trying to remember to post the schematic ever since.
PMA,
Now we're entering deeper waters...
Take any or all of the following thoughts and throw them in a pot. Stir over medium heat. Serves one.
--I have no problem with the idea of using a current source instead of a resistor to bias the circuit.
--A Darlington? Why not use another MOSFET so the characteristics match? (This may be answered in the project's writeup, but I'm short on time. I'll try to read it later tonight.)
--Given that the intended load is a pair of headphones, there's not much energy storage to be dealt with. The CCS doesn't really have to absorb a lot of backlash. Also, 99.999% of the headphones on the market are one way, hence no crossover. Again, a simple load, very easy to drive. There's no mass to speak of, so there's not much need for a stratospherically high damping factor, either. In short, simple is sufficient, and quite possibly better, which leads us to...
--It adds complexity. Whether this is good or not depends on who you're talking to. In this case, I've gotten rid of the input cap and all the biasing stuff, partially due to using a bipolar power supply.
--I have run into cases where a simple resistor actually sounded better than a current source. Why? Ugh. Ask me on the second Tuesday of next week and I may have an answer for you. My suspicion is that it's related to high frequency compliance difficulties, but I've never been able to think of a good way to test my hypothesis. It's offered only as a basis for discussion and/or ridicule. Given the relatively modest voltage swings required in this case, I honestly doubt that it's really much of a factor. Don't misquote me, as others have. I'm not saying that resistors always sound better, just sometimes. Have I built two versions of the circuit, one with a resistor, one with a CCS? No. It's possible that this is a case where a CCS might sound better. Or maybe the two would be indistinguishable. Who knows? As it sounds now, the thing sounds damned near perfect. I'm reluctant to mess with it.
In the DIY progress report thread, where I originally attempted to describe this circuit, paulb compared this circuit to the Szekeres headphone amp. There are similarities, but there are differences, too. Again, my circuit uses a bipolar power supply, which allows me to dispense with the input cap and the biasing circuit. I use caps sometimes if they will allow me to delete a bunch of other stuff. Do caps sound perfect? No, of course not. On the other hand, their imperfections are relatively predictable compared to gain devices. Drive them with enough current and they can sound pretty good (ask any tube guy). That said, if I can remove a cap from the circuit simply by putting in another rail...to me it's a no-brainer. Getting rid of the bias circuit just sort of falls in your lap when you add the negative rail. You've got the opportunity--might as well take it.
Two more quick notes about the circuit.
1) If you want more bias, decrease the value of R4. Have good heatsinks handy, as the circuit runs pretty warm. Conversely, if you want to decrease the bias or heat, increase R4. The bias will be:
I= (Rail voltage-Vgs)/R4
Multiply that current by the rail voltage to see how much heat you'll be needing to get rid of. Warning: It's more than you think.
2) You are in no way required to build the power supply as drawn. That was just some junk box stuff I wanted to use up. The transformers are literally two separate transformers. They were too puny to do much else with. Yes, I could have tied the secondaries together and made a 24-0-24VAC out of it, but I have a bunch of bridge rectifiers to use up and this seemed like a good chance to do so.

Grey

[moe29]

Grey,

I'll just say thanks for posting the schematic

I wanted to try something else besides an opamp-based amp.

[GRollins]

Opamp headphone circuits? Been there, done that. Gave the T-shirt to a homeless fellow living under the Gervais St. bridge.
Besides, it's cold down in the Dungeon. Anything that produces heat is welcome. I suppose it goes without saying that this circuit is pure, red-blooded class A. And, yes, it is capable of driving a speaker, providing your volume requirements are modest.
Or you can start increasing the bias, perhaps even paralleling output devices. After all, it's capable of something like 45V, peak-to-peak, albeit with asymmetrical clipping; call it 40Vp-p symmetrical, or about 14Vrms. That's nearly 25W/8 Ohms if you provide sufficient current.
Remember that this circuit can only swing the voltage that you put into it (actually a whisker less). In my case, I don't need much. First off, I don't listen as loud as I used to. Second, the various sources I use all have sufficient output to drive the thing directly. Finally, even if I were to want more voltage swing, I have a tube preamp, and if there's one thing tubes can do, it's voltage...though I can't remember ever needing it.

Grey

[adason]

i built this one quite a few years ago, it worked fine, but in comparison to tube headphone amp it was nothing special
http://headwize.com/projects/showfil...szeke1_prj.htm

[adason]

here is improved version, i did not try it
http://headwize.com/projects/showfil...murdey_prj.htm


but i modified the circuit to eliminate the output cap by adjustable power supply

[paulb]

Originally posted by GRollins
...That said, if I can remove a cap from the circuit simply by putting in another rail...to me it's a no-brainer. Getting rid of the bias circuit just sort of falls in your lap when you add the negative rail. You've got the opportunity--might as well take it. ...

Hi Grey, and thanks for posting the schematic.

I built a Sezerkes a while back, but needed some gain so added (gasp) an op-amp in front of it. Oh, and I used a P-channel FET. Or maybe it was one of each, N- and P- so I could draw equal currents from a dual rail supply. In any case, I removed it from the preamp...it's been dismantled.

Looking forward to the Lovoltech version...I have high impedance phones so need a bit of gain.

[EUVL]

OK, I stick my neck out and fire the first shot.




Patrick

[rjm]

The circuit posted by Grey is just a Szekeres amp with a honkin' big unregulated split power supply. Nothin' wrong with that. The split supply allows you to get rid of the gate bias network and input coupling cap, and offers better PSRR.

What I found (following up Greg's suggestion) was that if you use three channels rather than two you can get rid of the output coupling capacitors as well, which I feel are the sonic Achilles heel of the circuit. For the details see my Szekeres VE page.

There is also a thread over at headwize.

/R

[steenoe]

OK, I stick my neck out and fire the first shot.

Thanks for that, Patrick Looks nice. Did you build or simulate the circuit? Good idea feeding the zener with a ccs. Might be just the thing for the shunt-reg supply I am working on (again)

Steen

BTW Maybe you should start a new thread? Its often difficult to find things burried in other threads.

[mullardel34]

Grey,

Given your extensive background with the XA project (Nelson Pass-based SuperSymmetry/Aleph) and other PassLabs-based efforts, I thought that I'd note that Marcello Pellerano posted the following Aleph-based headphone-amp design a few years ago on the HeadWize project page:

http://headwize.com/projects/showfil...lerano_prj.htm

Like the project author, Marcello, I also enjoy the use of the low-impedance (~32 ohm) Grado headphones, so the added control provided by the Aleph current-source really brings these cans to life. The purity and the sense of ease in the presentation of the music delivered by the Aleph headphone amplifier is truly amazing. Even my entry-level in-the-ear Shure headphones deliver an astounding musical experience when driven by this amplifier.

I built this headphone-amp with just a few modifications to the published circuit. The primary modification that I made to the presented design was the use a dual-rail power-supply instead of the single-ended power-supply of the posted project; this design change eliminates the need for the output coupling capacitors. I used a 12-0-12VAC dual-secondary Plitron power transformer, resulting in +/-16VDC power-supply rails. Of course, as with any Aleph-based DC-coupled output stage, all of the output-stage components need to be suitably matched to avoid any DC offset issues...

The Aleph-based headphone amplifier avoids one of the major deficiencies of most high-end headphone-amp designs; over-damping of the output stage. When driven with an excessively low output impedance, most high-quality headphones present a very 'dry' and 'sterile' sonic character. However, the same headphones can really sing when driven with a 'critically-damped' output stage. The observed sonic characteristic of matching the output impedance to the load is quite similar to the very nice effect of driving a ribbon tweeter with a current-source amplifier. My personal preference is a damping-factor somewhere in the range of 4-15. A low damping-factor is really shines with a benign load impedance, while the higher damping-factor is potentially necessary to control a more wide-ranging load impedance. The Aleph-based headphone amplifier allows for the simple reconfiguration of the standing Class-A bias current, which sets the effective output impedance of the output stage. Just be certain to provide adequate heatsinking on the MOSFET's; with 300ma bias (and a 12-0-12VAC dual-secondary power transformer), the individual output devices will be handling about 5-6 watts of heat.

Be certain to use a high-quality input-coupling capacitor in order to avoid degrading the intrinsic performance of the Aleph-based headphone amplifier. The Solen FastCap capacitors are very neutral at a low cost, but the Mundorf capacitors offer a very audible sonic improvement (even the 'entry-level' Mundorf capacitors, which are anything but 'entry-level' from a cost standpoint).

BTW, I also took a pass on the EQ circuit included in the project by the author. I prefer to keep the signal-path as simple and uncorrupted as possible.

[steenoe]

Hi Mullard. Marcello posted about his project here:
ZEN-like headphones amp Username "tortello".
Many of us, use that headphone amp, since digi01 made several GB's on pcb's for it (his own layout, also without the Equalizer) Search for Mini Zen, or something like that.
Thanks for posting your interesting thoughts. Component choice is pretty important in a simple circuit like this one, indeed.

Steen

[rjm]

When driven with an excessively low output impedance, most high-quality headphones present a very 'dry' and 'sterile' sonic character.

If by "excessively low output impedance" you mean opamps, I'm temped to agree, though this is the first time I've seen it attributed to damping factor. Surely if it was that simple, inserting a resistance in series with headphone would solve the issue? Although I haven't tried it, from what I've read this doesn't work.

Be certain to use a high-quality input-coupling capacitor in order to avoid degrading the intrinsic performance of the Aleph-based headphone amplifier

Certainly well-intentioned advice, but can you tell me why a 4.7u film coupling cap on the input is 'degrading the intrinsic performance' when you have two 470uF electrolytics for the output coupling capacitance?

/R

[EUVL]

I cannot quite understand why people would still want to build capacitor coupled circuits when they can in many cases get round it quite easily.

Patrick

[mullardel34]

Originally posted by rjm


If by "excessively low output impedance" you mean opamps, I'm temped to agree, though this is the first time I've seen it attributed to damping factor. Surely if it was that simple, inserting a resistance in series with headphone would solve the issue? Although I haven't tried it, from what I've read this doesn't work.
/R

I've noted an audible correlation between damping-factor and audio quality. In my experience, whether we're discussing headphones or loudspeakers, damping factors greater than about '15' result in the suppression of a number of desirable sonic characteristics. One of great advantages of an Aleph output stage is its ability to dynamically maintain a balance between output impedance and the exertion of control of the load; a static series resistor does not duplicate the adaptive behavior of the Aleph design. As this is a DIY forum, I would suggest that you conduct your own experiments along these lines.

Originally posted by rjm

Certainly well-intentioned advice, but can you tell me why a 4.7u film coupling cap on the input is 'degrading the intrinsic performance' when you have two 470uF electrolytics for the output coupling capacitance?

/R

Please re-read my post; I used a dual-rail power-supply so that I eliminated the output-coupling capacitors in my implementation of the Zen headphone-amplifier.

[mullardel34]

Originally posted by steenoe
Hi Mullard. Marcello posted about his project here:
ZEN-like headphones amp Username "tortello".
Many of us, use that headphone amp, since digi01 made several GB's on pcb's for it (his own layout, also without the Equalizer) Search for Mini Zen, or something like that.
Thanks for posting your interesting thoughts. Component choice is pretty important in a simple circuit like this one, indeed.

Steen

Steen,

Thanks for the welcome and the information. I'm glad to hear that others enjoy Marcello's Aleph-based headphone-amp design. Now, if I could only find a design for a point-singularity power-supply so that I run this Class-A circuit as a mobile headphone-amp without the need to carry a car-battery...

[steenoe]

Hi again Mullard, you are welcome Just for the record; I have to be a little bit against your saying that Marcello's amp is Aleph based. It is not
It is Zen based, with an Aleph based CCS The input mosfet pair in the Aleph design is the main difference from a Zen Which doesnt have a thing like that.

Steen

[mullardel34]

Originally posted by steenoe
Hi again Mullard, you are welcome Just for the record; I have to be a little bit against your saying that Marcello's amp is Aleph based. It is not
It is Zen based, with an Aleph based CCS The input mosfet pair in the Aleph design is the main difference from a Zen Which doesnt have a thing like that.

Steen

Steen,

Good points. For me, if it has an Aleph current-source providing Class-A bias to a common-source MOSFET output device, I tend to regard it as an Aleph design.

The original Aleph amplifiers and the follow-on XA amplifiers simply insert alternative front-end circuits on a common output-stage design, while the 'Zen' designs inject the input signal directly into a 'naked' output stage. So, I can see where confusion could result from indiscriminate use of the 'Aleph' and 'Zen' labels. I'll be more circumspect in my use of these terms in the future.

A future design project for me is an XA-based headphone-amplifier. I've already modified a set of Grado's and a set of Sennheiser's for balanced-drive operations (separating the usual common headphone 'ground' into separte returns for each 'can', then soldering-on the XLR connectors), along with a perf-board XA headphone-amp prototype to drive them (which throws off about 25-30 watts of heat, just to drive headphones...). When I get some 'spare time' (in about a couple of months), I'm going to replace the high-bias power MOSFET-based differential input stage with a paralleled 6N1P-based vacuum-tube differential input stage; the sound should be glorious. With the high-current needs of the 6N1P heaters, total power dissapation will be somewhere between 80-100 watts. I just love the smell of burnin' silicon grease in the mornin'...

[steenoe]

A future design project for me is an XA-based headphone-amplifier.

That would be great Please dont hesitate to implement a J-fet input Which should be easy enough, considering the railvoltages needed for headphone amp?

Steen

BTW Better get some of those J-fets that papa fancies

2SK370 and 2SJ108, readily available from Toshiba.

This is cut from another thread

[paulb]

Adding an extra power supply rail to get rid of a coupling cap adds another cap anyway, that being the output cap of the supply voltage. If you trace the AC path it goes through the power supply capacitors.

[steenoe]

paulb, just what I was thinking Having been listening to the "tortello" amp through my headphones for a few years, I cant claim to miss/need anything! I slightly tweeked it for higher impedance phones, as far as I recall. Mine are Beyerdynamics 880's. But whatever, I am open to new ideas, so if anyone posts a nice circuit (especially with LU's ) I am all set to go!

Steen

BTW I still think that if anyone wants to start a J-fet design, it should have its own thread

[azira]

Originally posted by GRollins
If a picture's worth a thousand words, surely a schematic is worth at least five hundred.
I tried recently to describe a headphone amplifier, using only words. While it is possible, in principle, to unambiguously define all the parts of a circuit and how they fit together, it's pretty cumbersome. In an attempt to clear some of the fog, I dredged up the schematic.
It's a trivial circuit, something you can build using only junk box parts, but it performs very well, indeed.

Grey

Grey,
I just got some HD555s and I was just about to build a circuit much like this. Except for thinking about taking a page from the zen-lite book and use a bulb for R4...

Question: You direct-coupled the input, so are you assuming no dc offset coming in the input? no big deal. But I ask because you bias the gate to GND so Vsource sits around -4V. This sets the polarity of the 1000uF cap with + at the output. But if you ever drove more than 4V peak of signal (it can easily happen on higher impedance cans), wouldn't this drive the cap against it's polarity? What are your thoughts on this? I don't really know how electrolytics response in reverse polarity. I've heard you can run them back to back as well for this type of situation.

You could single-end the power supply and add an input cap to the mix but that's another cap in the signal path.

--
Danny