advice on mic > preamp > headphone amp > headphones

[aeiah]

Hello all,

I'm trying to build a device that will do mic > preamp > headphone amp > headphones. I thought id put this thread in the headphone section since i think the mic > preamp side of things is ok.

The idea is that me and my girlfriend can play on our two xbox 360s side-by-side with headphones on, and hear eachother's voices (and our own) as well as our own game audio. I plan to mix things together passively as the last stage before the headphones.

Our headphones are Sennheiser HD 215 - 32ohm, 112dB SPL

My ability to solder and follow diagrams and instructions is reasonable, but my audio knowledge is poor This is where my experimentation has lead me so far:

I bought a preamp kit from maplin with an input impedance of 1kohm and an output gain of 100. I added some power to an electret mic and this seems to work well, but it's too quiet to use just on its own (being line level or perhaps lower).

I then got a headphone amp kit from ebay (OPA2604 x 2 Headphone Amp kits CMOY). It seems to have a gain of 1, at least with tests using an ipod, meaning its fairly useless in its current state. I understand I can change some resistor values to increase the gain, although i'm having a little bit of trouble understanding the circuit board since the thing is so damn small. At least the SMD resistors are easy to unsolder.

I grabbed a 1W amp kit from maplin, which offers a 50 gain output. This hisses horribly, but does provide the amplification needed for the mic. (mic > preamp > 1w amp > headphones). It has a fixed output but variable input, so by setting it to zero i could confirm that it was the amp making the hiss and not the input.


Should i build a new amp circuit, mod my chinese headphone amp to a gain of 50, construct a low pass filter to mitigate the 1w amp hiss or scrap the whole thing and build everything from the ground up? Really, i think i just need pointing in the right direction with regards to a 50x gain low noise headphone amp, if such a thing is possible.

Many thanks

[aeiah]

oops, meant to link to the 1w 50 gain amp:
1W Amplifier Kit : TR Kits : Maplin

[sgrossklass]

LM386... yuck. If you want cheap and cheerful, look out for something with a TDA2822M and set a 20 dB gain externally.

Your gain problem might just evaporate the moment you establish a power supply for the electret capsules. The preamp board seems to be lacking one, and the FETs don't work terribly well like that. From the V+ supply, run a 1k..2k2 resistor to the positive terminal of a 10µ..47µ (16+ V) filter cap to ground, and then go from the filtered voltage to the LIN and RIN pins via 4k7 or so each (depending on operating voltage - I'd pick 2k2 for 3 V, 4k7 to 6k8 should be fine for 9 V, and so on).

What kind of moron sells an OPA2604 in a Cmoy? It's definitely not a good performer at low voltages and prone to oscillation to boot. Any ordinary 5532 will do better, let alone a 4556. Ah, so they're using two in parallel, with 47 ohm combining Rs... still, I'm not too sure about low impedance load driving. Anyway, some good-quality shots of the top and bottom of the PCB would be helpful.

[aeiah]

Thanks sgrossklass for your assistance so far. The datasheet you linked to looks interesting, and i noticed they have a kit. If I can't get this working with what I have, do you think that amp would do the job? This is just for chatting through mics - im much more interested in introducing as little noise as possible than i am in audio quality per se. it doesn't have to be silent, as there'll be plenty of other sounds going on, but you get the idea.


I'm a little unclear on your description for powering the electret mic. I've been doing something akin to this, but do you mean something different (except for the component values)? The cap doesn't seem to make any difference for me. The preamp kit pretty much takes the signal straight into some caps, so are they needed (and does this mean I'm powering my mic properly, and should move on to the other parts)? There's a PDF with a circuit diagram of the preamp here. No matter what I do, I get about the same result by just putting a 10k resistor from V+ to the electret's positive.


Here's the OPA2604 'cmoy' pics.

http://aeiah.ninth.su/audio/OPA2604_top.jpg
http://aeiah.ninth.su/audio/OPA2604_bottom.jpg

The caps are 100µ 50v

Is it worth changing the resistors to increase the gain, or is this circuit fairly unsuitable for that sort of thing?

Thanks

[sgrossklass]

Originally Posted by aeiah

No matter what I do, I get about the same result by just putting a 10k resistor from V+ to the electret's positive.

That should give you power at least. Power supply rejection of a common source amplifier isn't too grand though, hence I was suggesting to RC filter the supply voltage.

Originally Posted by aeiah

Here's the OPA2604 'cmoy' pics.

http://aeiah.ninth.su/audio/OPA2604_top.jpg
http://aeiah.ninth.su/audio/OPA2604_bottom.jpg

The flash reflections make it a little hard to see, but it seems we're looking at an "Apheared-47" type circuit with a voltage gain of 1+ R4/R5 ~= 3 or 10 dB. R3 presumably goes to the volume pot wiper. No bypass caps close to the opamps, pretty brave for a known-finicky OP type. Any indications of the electrolytic caps' manufacturer?

There are two problems with your wiring:
1. You NEED a bipolar power supply - V+, V- AND ground. With no low-impedance ground return, it's no surprise you don't get much in terms of output level. I'm surprised it works at all.
2. We're talking low-level audio signals here. Using shielded signal cables would be advisable.

Originally Posted by aeiah

The datasheet you linked to looks interesting, and i noticed they have a kit. If I can't get this working with what I have, do you think that amp would do the job? This is just for chatting through mics - im much more interested in introducing as little noise as possible than i am in audio quality per se. it doesn't have to be silent, as there'll be plenty of other sounds going on, but you get the idea.

At stock gain (40 dB), a TDA2822M is a very hissy affair. At 20 dB, it should be OK. (Use metal film resistors for gain setting Rs R1..R4.) This circuit only needs V+ and ground.

[sgrossklass]

By 'bipolar' power supply, I mean split power supply. Engrish bad when tired...

[aeiah]

can't believe i made such a fundamental mistake with the power supply. would this be an acceptable way of rigging the power, or should i actually build a circuit for it?

I'll get back to you about the cap manufacturer when I'm not at work.

Whilst the PCB does look like it says 4.7K, the SMDs are 47K. did they give me the wrong ones? I looked the numbers up here (470, 1002, 473 are the 3 types of resistors on the circuit).

I suppose I'll have a clearer picture of the gain once I wire the thing up properly. I'd have probably spent the next few months in complete ignorance of dual/split power supplies if it wasn't for you spotting my glaring error

[sgrossklass]

Originally Posted by aeiah

can't believe i made such a fundamental mistake with the power supply. would this be an acceptable way of rigging the power, or should i actually build a circuit for it?

If you have two floating supplies, e.g. 9 V blocks, you can do it like that. Supplies sharing a common ground are, obviously, not stackable.

For mains-powered operation, you'd typically use a split-secondary transformer. The centre tap goes to ground, and a single bridge rectifier will do for both rails. If all you can find is a transformer or wall-wart with a single AC output, you can still use an oldfashioned half-wave rectifier - see the O2 amplifier for an example. The corresponding design article also mentions the tradeoffs involved.

Originally Posted by aeiah

Whilst the PCB does look like it says 4.7K, the SMDs are 47K. did they give me the wrong ones? I looked the numbers up here (470, 1002, 473 are the 3 types of resistors on the circuit).

Indeed, 473 would be 47k, resulting in a gain of a whopping 1.2 or 1.5 dB... which is promptly lost in the 23.5R/headphones voltage divider at the output. Hence, approximately unity gain.

I'm still pondering under what kind of circumstances the circuit may be working even without a power supply ground. Apparently that was provided by the source, and the supply voltages (isolated from the source) shifted in order to accomodate the output. Pure luck. I'd want something that can be counted on working somewhat more predictably...

[aeiah]

I'll see what powering it properly does. I'll use 2x9v, since eventually this thing will just sit on the sofa between me and my girlfriend, and id rather not trail a power lead.

i doubt a gain of 1.2 dB would give me any real difference from what I'm experiencing with it now. Is there anything stopping me changing the 47k resistors for, say, 1k, to give 1+10/1 = 11 dB gain? I know a lot of opamps have this kind of gain (although perhaps with 10k and 100k) but i dont want to trash something that might be useful somehow.

Thanks for your help

[sgrossklass]

Originally Posted by aeiah

i doubt a gain of 1.2 dB would give me any real difference from what I'm experiencing with it now. Is there anything stopping me changing the 47k resistors for, say, 1k, to give 1+10/1 = 11 dB gain?

No, that should work fine. A gain of 11 (or ~21 dB) would normally be considered fairly high for reasonably efficient 32 ohm 'phones, but in your case it may be what's needed.

Optimum gain always is a compromise for which source levels, headphone sensitivity and maximum output levels need to be considered. Chosen too low, things may not be loud enough. Chosen too high, the volume pot may get into a region with bad channel tracking, hiss may become audible and distortion will suffer.

Incidentally, it seems OPA2604 and TDA2822M have about the same level of voltage noise. With your headphones, resulting hiss at ~20 dB gain should be around the threshold of audibility. Your mic setup is likely to contribute more.

I just remembered you have to drive two headphones. Two 32 ohm cans in parallel is likely to be pushing it for this amp, I wouldn't recommend that. Better get a second one, or alternatively mod it (I'd suggest NJM4556AMs and swapping the 47 ohms for 10 ohms; the remaining 4k7s can also be swapped for 1k for slightly lower noise - oh wait, aren't those still 47k, too?! Make that 'a good bit lower noise' then). When getting new resistors, preferably use the "thin-film" variety instead of "thick-film" ones.

Hmm, 4556AM's DMP-8 package is a bit larger than SO-8... should still fit though.