Poor Person's Headphone System

[chris p]

I really like the idea of people controlling their own headphone mix (AVIOM, Hearback, etc), and have been sketching out some ideas for a poor DIY version.

The basic idea is to use a be!@#$%^& ADA8000 as a base units, taking an ADAT 8 channel signal that will be 4 stereo submixes. These 4 stereo channels will then be sent over balanced lines to each satellite (the current design uses 2 CAT5 cables to achieve this, but a multicore will also work, or course). The satellite then consists of balanced line receivers (x8), stereo volume controls (x4) for each submix, a stereo virtual earth mixer and a headphone driver with volume control. All driven by 12-18V DC (including maybe 2x9V batteries). Power supply switched from the headphone jack so its only drawing current when actually in use.

All this is just template stuff, but I need help at the distribution stage. The ADA8000 has good DA output circuitry, so I'll just be taking feeds off the back XLRs. However, each of those 8 outputs has to be fed to (insert number here, I'm think 12) satellites. My gut feeling was to transformer isolate each send, but that would be 8 signals x 12 satellites = 96 transformers! The receiver at the satellite is pretty basic -

IN+ ---•----[   10K   ] ---•----[   6K8   ] ---•---- Opamp +
             |                               |                              |
       [   3K3   ]               [   220pf   ]            [7k8 + 5KVR]---GND             Opamp Out ---•---
             |                               |                                                                                              |
N-   ---•----[   10K   ] ---•----[   6K8   ] -------- Opamp - ------[   10K   ]-----|

I'm wondering whether I can get away with just a low value resistor at the distribution stage (say 50R) in series across each of the outputs, or whether I need to have a unity opamp buffer. 96 resistors is fine, I can probably manage 96 buffers (its only 24 quad opamps) but 96 transformers is out.

Or is there a way to just have 8 transformers and split the signal post transformer? I couldn't see how that would work, but I'm not expert.

[Kurt]

Probably be cheaper and easier to buy an old foldback desk off eBay.

[rob]

flesh out your circuit for the H/P driver to begin with
get rid of the 3K3, it doesn't do anything worthwhile, change the 7k8 + 5KVR to 10K to get the input balancing correct.
then you'll need some other bits to make it all work reliably.

if each H/P driver has an input Z of something in the order of 30K ( as your design currently has ) then each channel of the ADA8000 will be able to drive around 14 channels of H/P driver.

you don't need transformers, but you might need to be careful with the grounding / earth floating of this whole system

[chris p]

Probably be cheaper and easier to buy an old foldback desk off eBay

Er, except it would be 12 old foldback desks off ebay plus all that cabling, and you'd still need the distribution done right! - that's not the idea, and I'm not sure it would be cheaper. It may well be easier, but then what would I learn from the process?

change the 7k8 + 5KVR to 10K to get the input balancing correct.

A straight 10K achieves around 40db of CMR rejection. Its not bad (and likely enough for this purpose), but in theory the 7k8 plus trimpot, together with a multimeter and battery, allows for better figures, hence its inclusion - what you can get to is really only limited by the accuracy of your meter. If it has no benefit in real world, I can always replace the 7k8 with a 10K and just bridge the trimmer. I like if possible to design with these features in and then not use them, rather than try to kludge a board afterwards.

I agree that the 3k3 is optional, I have it in the design so I can try it out - if no benefit, again I can just leave it off. The idea of course is to help balance any variations in the cable impedance between the individual wires - I've not used CAT5 for audio before so I'm a little cautious. It also helps (a little) with the unequal input impedances ref to ground problem inherent in the single opamp receiver design - but I'm reluctant to go with the 3 opamp per line design as the satellites need to work within a current budget for battery power.

if each H/P driver has an input Z of something in the order of 30K ( as your design currently has ) then each channel of the ADA8000 will be able to drive around 14 channels of H/P driver.

That's great news, thanks Rob. So its just send each channel directly to all 12 units? Too easy.

but you might need to be careful with the grounding / earth floating of this whole system

oh yeah - current plan is to ground only at the ADA8000 end, each satellite running off its own floating ground (-9V 0V +9V). Need to check how this might impact that 10Kish input balancing resistor, but that's the only link between the satellite power scheme and the ADA8000.

[Kurt]

Er..
8 out of the ADA8000, 8 xlr leads, 8 inputs on a desk, each capable of routing at any level to any/all 12 places. Kind of exactly what foldback desks are built for.

Probably be cheaper and easier to buy an old foldback desk off eBay

Er, except it would be 12 old foldback desks off ebay plus all that cabling, and you'd still need the distribution done right! - that's not the idea, and I'm not sure it would be cheaper. It may well be easier, but then what would I learn from the process?

[chris p]

Kurt

OK, I see what you were getting at now - but the idea here is that each h/phone user does their own mix at their own position, rather than the mixes being done someone at a foldback desk who has to manage all those "can I have the drums louder and less guitar" complaints. Its the old "the beer tastes better once I've spit in it" - people are less likely to fuss over their headphone mix if they were the one who set it up (even if only in a limited sense of balancing 4 stereo stems).