Ludo's Headphone Amp

AVForums

Help Support AVForums:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ludo

AVForums Super Veteran
*
Joined
Oct 14, 2008
Messages
1,729
Reaction score
20
Location
Pretoria
So many threads...

OK, so you can know what you're up against, here's the big pic, and the long story you'd expect by now. Scroll down to PRICE if that's all you wanted to know.

click for latest pic, v.0.9 at the moment.

I'll try to keep it interesting without mentioning Irishmen and bars.

The opamp supplies voltage gain, the discrete bits do the heavy lifting and feedback comes from the output stage. The output stage is biased by Q1 & Q2, feeding "identical" currents into R7 & R8 via Q4 & Q5 connected as diodes) then meeting at the output node of the opamp, and cancelling each other. As the output node of the opamp moves up and down it pushes the "outer ends" of the pair of biasing diodes up and down, as the voltage over those diodes stays the same (because the current through them remains the same due to the current sources.) So the bases of the output transistors get pushed up and down, pushing the output to the headphones up and down. Do you spot the holes in this argument? The real question is whether this is actually a diamond buffer, or not. The mind boggles.

The idea is that the opamp output only supplies enough current to drive the output stage. So somewhere between 50-200 times less than the headphones require, due to the output stage's current gain. It puts the opamp on a cruise liner sipping cocktails, in stead of down in the hole digging dirt. This keeps the opamps voltage rails quiet, so the sea is even calm. The output stage is like sunshine, as it runs very hot and never switches off. Well, unless the cans are very esoteric, it won't switch while they're are on your ears, promise.

But the same warmth can be supplied to the opamp output stage. Another current source (Q3) pulls a constant little something from the opamp output to keep it active in class A. Think of it as a continuous and creative tax write-off that makes the holidaymaker richer and richer over time. The blue stuff on the pic is there to cancel the bias current that comes out of the input pins of the opamp. 2 green LEDs make low noise voltage references that then get filtered silly. The filtered result is applied to the input pin via a suitably large resistance. This counteracts inflation in terms of offset voltage. Inflation ruins really long holidays and offset
voltage ruins the natural and comfortable seating arrangement of headphone membranes.

The pair of 1N4148 diodes from the + input pin to the supply rails are for "sailing around the horn of Africa." There will be fright and panic if an uninvited marauding voltage hops aboard, but these diodes keep excessively violent intrusions from wrecking the reception area in the opamp. Tiny things, but electrically like the heavies around the prez.

"But what does it sound like?" I hear you ask. Like nothing in particular, but loudly so, as far as I can tell. Attempts to measure the distortion have been pretty unsuccesful, but that says more about the attempts than about the distortion. I shall try again some time, and try harder.

"Why do you call it V0.9?" Because I don't believe that it is finished.

One day I have to make up my mind about the final and proper value for R14 & C12. It appears they can be left out though. The resistor values for feedback loop and input filter and pot need to be balanced properly. It should make a difference but depends on the setting of the volume knob. Pic is wrong, maar wat is wysheid? I like 1V p-p on my cans for Emil Gilels. 5V p-p on something like "The Ocean" (LedZep) but My CDP no longer has the usual output level. Too much hacking. No telling what the sensitivity of your headphones is, the gain in your phono pre, or how you like it. Easy to change though.

One day we will know if some funky opamp will do better, and why. But the possibilities of finding a prettier holidaymaker for accompaniment opens a can of worms best left to the young at heart. High-maintenance cuties and high-risk investments can bring you a lot more trouble than you care for. If you like trouble and you have to scratch, play with the caps. The ones on the board have a huge diameter, can be oversised to your contentment and be any colour. We don't talk about cables and such here, that and blue LEDs is private stuff. There is space for moderate sized Schottky diodes.

PRICE

R600 + postage buys a finished stereo board. Why so bloody expensive? That's about parts cost x 2 so a kit of parts will be the same (see ***) There is no silkscreen on PCBs from the kitchen, and a lot of the parts have to be scaled to work right. I'd rather make the thing and test than start a support service for folk who just get 15V on the output and broken headphones. It's DIY so there is no guarantee of suitability stated or implied for any purpose whatsoever etc but one obviously doesn't want to be a d**s about it. I solder with SN62Pb36Ag2 so a small quantity will be included for you to do mods. Also the little crimp connectors for input wires, which I always solder, never crimp. 1m thin cheap coax that solders nicely. Common metal film resistors where it matters, multiturn cermet offset trimmer, MKP10 input cap, Epcos 105deg PSU caps and a sprinkling of polyester cheapies. No SMT on this one but a lot of opportunity for it under the PCB. Opamp in a nice socket to tempt you not to heed my warnings.

You'll have to add:

*One or 2 transformers of 15-0-15V. Bigger is better but a single 30VA works well enough at about R100
*RCAs about R25 at Yebo/Electronics123, with bling.
*A dual potentiometer (10k logarithmic stereo type) with a knob R15(?) each at E123 but the knobs are truly ugly
*Mains socket and a fuse R15
*A box with say 20x8cm aluminium plate for a heatsink (thicker and more is better but ~1.5mm is fine) You get to use the heatsink compound that should have been on your CPU...
*Some screws and stand-offs and nuts and washers. There are 2 holes in the heatsink tab to connect to the big aluminium plate with. The tab is 12.7mm thick. R15(?)
*Different wires that cost more than the board. The sky is the limit eh?


I'd offer one for the esteemed membership to test-drive. But that would mean I have to make a box. That might take a year. I have only one now, and lots of horrid prototypes. Let me know if I must make you a fresh one, or if you want to make a sturdy box for one that travels and can get scratched/stolen by the PO without hard feelings. I'll leave it for a week or so before I start. If everybody wants one I'll probably have to go below deck till it blows over. If nobody wants I'm a free man looking for a deck chair. The pcb layout will be coming up so you can do your own thing with PressnPeel if you like.


(***)
If you are a sensible person, you'd know that one can sell a finished stereo gain-clone board similar to this for R350. But that will drive speakers. So if you're on a really tight budget this is madness. Not as mad as some other options though. If you've read all this you're not all that sensible anyway.

 
Top