Headphone Distribution amp question

Hey, quick headphone distribution amp question.

So i'm thinking about a 5 channel (5 in, 5 out) headphone distribution amp. Simple ne5534 non-inverting amp, similar to gyraf's design. Each channel will be able to select which input to use.

I just want to confirm or deny what i'm thinking. I was considering whether or not the inputs will need to be buffered. Worst case (when all channels select the same input) input impedance will be 20k (as per the datasheet, the ne5534 has a typical 100k input resistance, paralleled 5 times gives 20k). I am thinking this will be plenty high enough so i shouldn't need to buffer the inputs. Correct?

Thanks!

> as per the datasheet, the ne5534 has a typical 100k input resistance

Where does it say that??

OIC: 30K-100K ri BUT this is open loop.

Close some NFB around it, the inputs act like zero or infinity impedance.

BUT you have to supply the bias current with low offset voltage. Indeed you often like a 47K resistor on a 5532 input. (Use the 5532; compensating a 5534 is a bit tricky at your level, and the 5532 offers excellent performance for a hedfone box.)

AND you will want individual level controls, pots. These can be 100K or 10K, with or without some 22K-100K fixed resistor, with/without a DC block cap: these plans all work but with different issues.

Say you end up with five 10K loads. Would I drive that from one generic output? Yes, IF it was not my main program output, if loading or shorting it did not kill the show/take.

Steal (research) from those who have gone before you. Here's a RANE hedfone box similar to your goal.

http://www.rane.com/pdf/old/hc4aman.pdf
http://www.rane.com/pdf/old/hc4asch.pdf

more: http://www.rane.com/oldman.html#gpm1_7

Aside from squandering another buck to buffer, they get a balanced input while still using simple unbalanced channel level pots.

Their 4580 is not radically different from a 5532, you can drop-in 5532 with no changes. (Well, it is quite a different design internally, and the 5532 is better for some things, but moot for what you are doing.)

I always thought the 5532/5534 family was designed to drive a minimum load impedance of 600 ohms.
Wouldn't it be a problem to drive low-Z phones to an adequate level? (Even with 2 amps in parallel.)
My headphone amps use a PNP/NPN pair after the opamp, to add some more 'power'.

Ooooh okay, wow. So, looking at the linked schem, they DO indeed use a buffer. And this is in order to supply the gain stages with appropriate bias current (not so much an impedance issue), right? Am i understanding this correctly?

Indeed you often like a 47K resistor on a 5532 input.

Click to expand...

Not sure what you're referring to here... are you talking about matching the bias currents of the +/- inputs? Doesn't that depend on your feedback network?

Say you end up with five 10K loads. Would I drive that from one generic output? Yes, IF it was not my main program output, if loading or shorting it did not kill the show/take.

Click to expand...

If i understand what you're driving at correctly, wouldn't an input buffer also take care of this issue.

Thanks again for the advice, links and patience hah!

I always thought the 5532/5534 family was designed to drive a minimum load impedance of 600 ohms.
Wouldn't it be a problem to drive low-Z phones to an adequate level? (Even with 2 amps in parallel.)
My headphone amps use a PNP/NPN pair after the opamp, to add some more 'power'.

Click to expand...

Yeah, i'm just talkin about the gain stage... and issues with having multiple amps drawing from one input. I'm planning on using an output stage for each amp too, don't worry.

RuudNL said:

I always thought the 5532/5534 family was designed to drive a minimum load impedance of 600 ohms.

Click to expand...

It works fine with 100 ohm load. Distrotion increases but not much.

RuudNL said:

Wouldn't it be a problem to drive low-Z phones to an adequate level?

Click to expand...

Use series resistance. There is more than enough voltage.

> problem to drive low-Z phones to an adequate level? (Even with 2 amps in parallel.)

gg12387 specifically mentioned 553x chips.

5532 is typical 38mA into a short. This is almost 1.2V or 20mW in 32 ohms. That's ample for much living-room listening, may be shy in the studio. Two 5532 can put typical 76mA, 2.4V peak, 90mW in 32 ohms, which is generally ample even for high-level listening.

As audiox says: if the load is so small the amp is strained, pad it out. AFAICT, lo-Z headphones generally don't need damping, and this is just performer-cans.

My over-kill headphone amp, intended for use too-near loud orchestra, does 110mA peak, and I do not think I have even been near clipping. I do know I would NOT want to hear more than a few seconds of it at clipping; I want to keep my hearing. And it is not a music-monitor: it is a technician tool. If I hear clipping, I know it aint my HP amp, I need to check my main audio path. For tracking, a touch of fizz in the performer cans does no real harm.

On a purely numeric analysis: 7V rms behind 29 ohms will put 99% of headphones very near their Rated Power. However many good headphones will go MUCH louder than is safe for your hearing.

Build a one-opamp amp. Try it. It will be a bit lame for hi-level in lo-Z cans. Double-up, it will go twice as loud, and that is often ample. Yes, loud tracking may need more, and it becomes cheaper to get there with a couple booster transistors. But start modest.

please forgive the pimping of parts... but I really like TPA6120A2 for this kind of thing.
It's a beast that'll drive virtually any headphones well - from little buds through to big ol' 600Ohm DT100's.

There may be alternatives out there from other manufacturers, but I don't know their product line as well  ;D

/R

Looking at the rane schem given above, could somebody please explain R18/R19... I don't know what those are for? Thanks much!

guitarguy12387 said:

Looking at the rane schem given above, could somebody please explain R18/R19... I don't know what those are for? Thanks much!

Click to expand...

R18/19 are so the mono switch doesn't short the two opamps to each other.

2x4560 will put out some decent current. I designed a HP amp back at Peavey that used transistor current buffers on fast opamps and drove the sleeve out of phase to get 2x the voltage swing (4x power). The common sleeve signal meant stereo and mono headroom were different, but the output would melt the wax in your ears, and the transistor buffers could drive small speakers. I actually current limited the charging circuit in the power supply to prevent opening the thermal fuse in the 1 amp wall wart when the output capability was abused (read driving speakers). 

JR

Ooooh okay thank you. So can they be safely omitted if i don't use the mono switch?

You could even eliminate the opamps immediately following and buffering that mono switch circuit.

JR

Hmm now i'm a little confused. Are you saying eliminate z1B and z6B? because (in reference to your other post) Z1A and Z6A will supply enough current to the amp stages? Am i getting this correct?

The only reason for Z1B and Z6B is to buffer the relatively high output impedance of the stereo/mono switch circuit. it looks like there are some other spare parts in that circuit but I don't want to confuse you further.

If you don't use a mono switch, you don't need the resistors or following buffer opamps, just connect the outputs of Z1A and Z6A to C3 and C20 respectively.

My comment about opamp drive capability was about them directly driving headphones, but they will certainly drive a couple K worth of pots.

JR

Very interesting... Thanks alot for your incite and patience!