That 1646 balanced line driver PCB

[ds23man]

I made a pcb design for the That 1646 balanced line driver. I posted it on the dutch Zelfbouwaudio forum and Kostas alias TheShaman found it and contacted me if I had some spare pcb's.

I will post here some pictures, the schematic and gerber files so everybody can order the pcb's at for example Itead.

[ds23man]

pcb and gerber

[ds23man]

Pictures

[TheShaman]

Thanks mate!

[linuxworks]

can folks who have used this chip, check this design?

I think the relay is wrong. the idea is that you can short one leg of the balanced output to gnd to make it drive unbal devices. however, it *looks* like its shorting both outputs of the line-driver together. this will cause the chip to overheat and also output a nice steady 0 volts

or, am I missing something?

[linuxworks]

edit: I think I mis-interpreted the relay function.

it seems the relay will mute the output (the next stage will see the same value on both wires, so its a null input).

I was thinking this was a mode-setting relay, that converts the chip from bal into single-end drive mode (shorting the minus-output to gnd). but that's not what this relay is about.

[ds23man]

The drill file is missing in the zip, so here it is:

[AndrewT]

the relay short across the output is a bit severe. Linux agrees.

Where are the output resistors to isolate trace and cable capacitances from the chip?
Use exactly matched 50r to 100r on each pair of outputs.
These need the same matching accuracy as the other resistances in the balanced feed.
Read Jung and Cordell about the need for these resistors.

[ds23man]

Originally Posted by AndrewT

the relay short across the output is a bit severe. Linux agrees.

Where are the output resistors to isolate trace and cable capacitances from the chip?
Use exactly matched 50r to 100r on each pair of outputs.
These need the same matching accuracy as the other resistances in the balanced feed.
Read Jung and Cordell about the need for these resistors.

Read the datasheet first........ No need for external resistors, outputs are shortproof

[dmills]

My review notes (Not intended as an attack):

You have a "pin 1 problem" in that the external cable screens connect via pin one to the internal single ended reference injecting noise and RFI, pin 1 should connect directly to the chassis and never to the internal reference (Which should be connected to chassis ground at a single point).

Where are the RFI measures? At present any RF picked up on the cable can get directly into the feedback pins, even ignoring the pin one issue. A common mode choke and some small NPO caps help here.

Where is the transient protection? Think P48 being present on a stage snake or similar, some diodes would be a good plan maybe?

Just what I see on a first look.

Regards, Dan.