Good DIY headphone amp?

[PRR]

> amp for my 32 ohm phones

A direction. Simple and effective:

The current source I2 can be replaced with about 50Ω. For higher-Z phones it can be 100Ω. For Sennheiser 600, use 200Ω and raise the rails to +/-18V, to +/-22V if your opamp can stand it.

This resistor and the transistor will dissipate about 3 Watts each, All The Time, so you need a 10W resistor, a fairly beefy heatsink, and ample space and ventilation.

The opamp can be about any good audio chip. 5532 is an obvious choice, TL072 will work fine. Newer sexier chips are around. I suggest you don't use something with very high gain-bandwidth or the emitter follower will become a radio transmitter, which won't leave much control of the audio.

NO output cap is needed. No input cap is needed if you are SURE all your sources have no DC leakage. In general you want a 0.5uFd film cap at the input just in case. Rin can be a 100K pot for volume control: one end to input, other end to input-ground, wiper to opamp.

Click on the image to view it at its original size
This is way-cool. THANKS, Ethan!!!!

 

[NewYorkDave]

That looks great, PRR. I was wondering, could performance be improved by using a second TIP41 configured as a current sink--or does that defeat the whole goal of simplicity?

 

[gyraf]

[quote author="buttachunk"] how much voltage gain ?
[/quote]

I haven't measured - but it's plenty sensitive to be driven from a standard consumer cd-player.

Jakob E.

 

[Gus]

PRR.

Question about crossover. Looking at that circuit it looks like the crossover part of the opamp output is moved away from center. It look like that to me because of the Base to Emitter drop of the TIP 41.

Headphones not needing as much drive makes me think this could be an added feature because the crossover will not happen at low drive levels.

here is the question would an added diode from emitter to the output node make sure the opamp is never in the crossover section, because the output might never be more than under 1 volt drive?

 

[LazyOne]

isn't there a way to do the design of jakob with ecc82 tubes i have a whole bunch of them.
first i wanted to make the mic pre butmy problem is transformers that cost enough money. (i am a student) .
and a good headphone amp is also something i could use.
it is probably a dumb question but if i could use the ecc82 i have also some transformers foor mains to 15 volt etc..... so i could use the trick and some resitors and caps an have a nice headphone amp for not too much money.
(srry for my bad english)

 

[PRR]

> could performance be improved by using a second TIP41 configured as a current sink

Not enough to notice, as long as you have power and voltage to spare. You can go either way.

> it looks like the crossover part of the opamp output is moved away from center.

Crossover is NOT at zero-voltage, it is at zero current. So the 0.6V shift is quite irrelevant. Yes, it does avoid crossover because the NPN base current is always one direction (unless you hit overload).

> would an added diode from emitter to the output node make sure the opamp is never in the crossover section

See above.

> the output might never be more than under 1 volt drive?

If you want more than what a 2-AA WalkMan gives in 32Ω, you are going to have more than a volt of signal. Critical listening ine some cans needs several volts of signal.

> a way to do the design of jakob with ecc82 tubes i have a whole bunch of them

ECC82/12AU7 will work but with slightly less gain and output.

> my problem is transformers that cost enough money. (i am a student)

Don't do tubes unless you have spare money. A perfectly excellent headphone amp can be built with one chip, $20 of parts. The best deal is one of the small kits sold for stereo speakers. One way to go is Velleman-kit K4003, http://www.velleman.be/common/product.Aspx?lan=1&id=9184 $29, except use a 12VCT or 18VCT 10VA transformer instead of that monster 24VCT 50VA transformer they use for speakers. http://www.velleman.be/Downloads/0/Manual_K4003.pdf Or look at two of K4001 http://www.velleman.be/common/product.Aspx?lan=1&id=9181 mono amp, $14. http://www.velleman.be/Downloads/0/Manual_K4001.pdf

 

[Kev]

Do go and have a browse around the Velleman kits as there looks like some fun in there.

for those that can't be bothered
4001 is based on the TDA2003 IC, capable of delivering 4Wrms at 4ohms
4003 TDA1521 IC, with a maximum supply capability of 2 x 15Wrms (4ohm) of 2 x 10Wrms (8ohm)

I am currently trying out the TDA1905 (5W and single ended) as Jaycar have heaps of these things on special at around $4 the moment.
It is a 16DIP with one side being the ground and heatsink ... use a large'ish area of copper on the PCB as a heatsink.

I hope it will make for a good workhorse unit for H/P's and very small speakers. I often dump a small speaker in the room so you can get the attention of people without their cans on while setting up .

I'll let you know how I go.

 

[LazyOne]

i have already build the 4001 and the 4003 (the 4005 also but the supply is not so good from velleman) but i'm doing also a project for school about tubes thats why i want to build something with it. i have some ecc82's, some sockets and some supply transformers that i can use so the only part is that i need good resistors en caps but i have that also. (getting some free parts from friends en family is usefull sometimes).
so i have done some experience on electronics and soldering
thats why i wanted tubes this times.
but with a little les gain and output i'll be content so this holiday i will give it a try with ecc82

 

[PRR]

> i'm doing also a project for school about tubes

OK, school is a valid reason to mess with tubes.

I shouldn't do your homework for you, but I was curious how the performance is with mild 12AU7 instead of hot 6DJ8 (or similar ECC types).

It works just fine, no resistor changes needed. (I was afraid that bias for a 6DJ8 would not be best for a 12AU7, but medium-Mu triodes are remarkably un-fussy.

Total supply current about 22mA. Output stage current 6.5mA each, 20mA total.

Input sensitivity is around 250mV. Plenty for hi-fi sources, even too much: you need a volume pot in front, either in another box or built into the headphone amp.

The low output stage current means this amp is NOT good for low-Z loads. At 100Ω load I get almost 1VRMS output, 10 milliWatts, at 5% distortion. The distortion is mostly 2nd harmonic so this will be very listenable, not harsh. But at 11 or 12mW it is going to clip pretty nasty. Jakob says he likes it with 300Ω loads, and that makes sense to me. At 300Ω I get 1.4VRMS, 6mW, at only 1.3%THD, 1.29% of that as 2nd harmonic. Output impedance is around 150Ω.

To get more power or lower-Z loads or lower THD, you can parallel more output tubes and reduce the output cathode resistors in proportion. For the same load and THD, power increases as the square as the number of tubes (6 output triodes gives around 40mW). However the grid resistor should be less than 680K or 1Meg per tube, so 6 triodes need a lower grid resistor, 100K or 150K. Change 100 to 50, change 3K to 1.5K.

By the way: on my simulation the output grids and input plate self-bias to the same voltage (around 60V). It may be possible to omit the 220nFd, 220K, and 100Ω, and DC-couple the two stages. That also allows up to maybe a dozen parallel triode sections for output. That puts you up around 100mW for 100-300Ω loads, which is a LOT. In stereo, use 12 twin-triodes, one for the two input stages and all the rest as the two output stages. Cathode resistor about 750Ω 20W, power supply current about 170mA total. Heater demand with 12AU7 is 6.3V at almost 2 Amps.

OR: if you use the 12V heaters all wired in series you need 151V of heater supply, very close to the 160V nominal B+. You could combine the plate and heater power demands in one supply. But that may not be best. The plate needs 170mA and very clean DC. The heaters need 150mA (for 12AU7) and can be quite dirty DC. Feeding them both from one supply means building a really clean 320mA supply instead of just a really clean 170mA supply. But: the output stage has some power supply rejection, the input stage has only a little. Only the 5mA going to the inputs needs to be super clean. So you could build a 320mA fairly-clean supply, and add a 5mA R-C filter to deliver 5mA of super-clean DC to the input stages.

 

[gyraf]

By the way: on my simulation the output grids and input plate self-bias to the same voltage (around 60V). It may be possible to omit the 220nFd, 220K, and 100Ω, and DC-couple the two stages. That also allows up to maybe a dozen parallel triode sections for output. That puts you up around 100mW for 100-300Ω loads, which is a LOT.

Hey.. That's a good idea. I'll try that..

Thanks, PRR

Jakob E.

 

[CJ]

OT: Here's an amp for driving speakers off your computer.
Might work for low imp bones also.

Runs a litle hot:

 

[jrmintz]

If I were to use a pair of API 325s for the headphone amp, how would I connect the outputs of the two transformers to the TRS heaphone output connector?

Schemo HERE

Thanks!!

 

[NewYorkDave]

For 300-ohm phones:

Jumper pins 11 and 12 on each amp.

Connect pin 10 of left amp to tip via a 22 ohm, 2-watt resistor.

Connect pin 10 of right amp to ring via a 22 ohm, 2-watt resistor.

Connect pin 13 of both amps to sleeve.

Leave pins 14 and 15 unconnected on both amps.

 

[Gus]

Chris nice thermo cam picture!

 

[jrmintz]

Thanks, Dave. Should the sleeve be isolated from the chassis?

 

[Rob Flinn]

Jakob

I`m looking at the diagram for your valve HP amp. It looks like a very nice circuit.

Question:

The input Z is the 10K pot ?

If I wanted to put say 4 of them in a box for a headphone amp & paralleled the inputs it looks like the combined input Z would be rather on the low side. Since valves are high Z devices would they be that confortable with a 2k5 input Z? Is there a good way round this ?

 

[NewYorkDave]

You could probably get away with bumping up the value of the input level pots to as high as 100K before Miller effect started biting you in the ass. If you're running four amps with their inputs in parallel, try 50K input pots. The total load impedance seen by the input would then be 12.5K. Use audio taper, natch.

 

[Rob Flinn]

Thanks Dave

 

[Rob Flinn]

One other thing:

How low impedance can the heaphones be for Jakobs circuit?

Obviously Cathode followers are very low Z output, but in parallel this must be even lower still. If for example you plugged som *ohm headphones in would you compromise frequency response or power output ?

 

[NewYorkDave]

Having not built and tested the circuit, my best guess from looking at the schematic is that it would have a hard time driving low-Z phones without bad clipping on the lower half of the waveform. That's because a single-ended cathode follower cannot sink current, it can only source. For high output into lower impedance phones, it would be better to use a push-pull output such as a White cathode follower.

When working hard, the circuit is likely generating some second-harmonic distortion that singers may very well find pleasing in their cans. This may or may not be want you want, depending on whether you prefer an "enhanced" sound or seek absolute fidelity.