Modifying M@ckie AD/DA Converter

[ej_whyte]

I do a lot of work on M@ckie desks, and the other day I found it interesting to find that the FW I/O functionality on their 1640i is all contained on one board, with basically just audio I/O and power on ribbons. I was recently looking in to getting a ADA8000 to expand my current setup to 16 channels, but as the M@ckie card has 16 channels this seemed like a cool way to make a custom converter exactly as I want it, especially as I could get the board pretty cheap.

If I got an ADA8000 I would want to do some pretty major mods anyway, take the pres out, change the IO to Dsub, and add some LM339 meters, maybe PSU. It would be advantageous that I could have all my I/O on Dsubs and in a single 1U box, not just the modified ADA8000 on Dsub and my other converter on TRS.

The M@ckie board would need to be modified slightly, just bypassing/removing a couple of un-needed components. The standard board has the ability to switch between a 2 sets of inputs via CMOS switches, but I think i would just rather bypass this and do it on a per-channel basis on a patchbay.

It has balanced outputs as well which is cool, but I need to look in to them more, how much they can drive, as i've just been looking at the inputs really so far.

Obviously the quality won't be absolutely amazing, but I've heard good reviews of those desks, and I dare say it will atleast equal or better the Behringer's efforts. I don't know much about converters, but they use CS5368 AD chips which Cirrus say are 'Premium performance audio converters'.

Anyway, looking at modifying the inputs to bypass the CMOS switches and anything else un-needed. The attached schematic is from just after the switching, could I remove IC1A and just put a balanced input directly in to R3 & R4? In any case, I guess i could still remove IC1A and just put in a line receiver, keeping R4 to ground.

The other thing is I'm slightly confused on the 0dBFS figures that are added on. As I interpret it, the converters will clip at -22.38dBu, which when you factor in the -20.17dB of the IC1B & IC2A stage comes out at -2.21dBu. However, -2.21dBu is 1.7Vp-p, not the 2.825Vp-p that is noted, which is +2.2dBu. Anyone have any ideas on this?

Cheers

EDIT: ADC_IN+ and ADC_IN- are the wrong way round on the schematic, should just be swapped over.

 

[millzners]

I looked over the schematic, and IC1A is half of the dual NJ4580 opamp, so to remove it without remove ICA1B (because they're each half of the same chip) would likely require cutting traces or pins of IC1... You probably already knew that, but I thought it would be worth mentioning just in case.

It looks like even if you do remove ICA1A, you couldn't just send a differential signal to R3/R4.  If my initial take on that schematic you posted is correct, it looks like that circuit is made to take an unbalanced input and create a differential signal by spliting it and inverting it with IC2A to create ADC In -.

A true balanced input to the ADC would be something similar to the circuit on page 28 of the CS5368 schematic which is more typical. 

So if you wanted to hack into what you've got to get a circuit like that you would pull R3 and R4, send your + (after the load and blocking cap) to pin 5 of IC1.  Then pull R6, and R8, and C4, send your -  (after the load and blocking cap) to pin 3 of IC2, then kluge in the input network of 100k loads, blocking caps, and the two 10k resistors from pin 5 of IC1 and pin 3... 

Honestly that seems like more work than it's worth, not to mention you could find better chips than the NJ4580, but I've done worse kluging to boards I've made...  It's never fun.

Anyway, take what I say with a grain of salt, I could be way off here.

 

[ej_whyte]

That is a schematic I have drawn myself, so the part numbers dont actually relate to the board. I'm not sure where the other half of "IC1A" is, maybe a buffer before the CMOS switches, but I'm not too bothered to be honest.

I'm thinking the easiest way would be just line receiver such as THAT 1200, straight in to R3, and keep R4 to ground. Its all SMD on the board but that'll just be tacking on one wire per channel. It probably isn't the most technically brilliant way of doing it, using a line receiver to unbalance the input and then IC1B & IC2A to rebalance it for the ADC, but I don't really fancy messing with the gain setup and source impedance for the ADC.

Any ideas on the 0dBFS confusion?

Thanks