[Nerull]

Just a quickie, do I need to tie anything to the RST pin or any external circuitry for the RST pin, or can I just leave it there and it'll be fine?

Thanks!
~Tom

 

[Garbz]

It is cricitcal that the chip doesn't come out of reset before the rail voltage for the digital section reaches 3.3V

This can be done via a simple resistor / capacitor to delay the pullup of the RST pin or the way I implimented it on my DAC, tying it to a voltage supervisor like the TPS3125 from TI. This will bring the Receiver and possibly DAC out of reset when the supply rail reaches a predefined voltage.

http://www.garbz.com/amp/webzeb/reciever.pdf
Ignore the bits in white it's not the most recent version of schematic. But the reset stuff works just fine.

 

[Nerull]

I have quite a few more questions that I need answered.

I just need a basic switch circuit for switching between optical and coaxial (RXP1 and RXP0 respectively, with RXN to ground). Could that be done by a switch like in this schematic?
http://5577.freewebpage.org/PCM1798.jpg
If so, that'd be great! (or a suggestion if it's not a good idea).

Another thing, are voltage supervisors (like the one in Garbz schematic) also needed for other parts of the DAC (like the DAC itself, PCM1794, and ASRC, 1896)? I'm guessing probably.

Thirdly, do I need power going to the TX pin if I'm not using balanced inputs (differential inputs)?

Fourthly, do I need a digital transformer (to filter noise) on the optical input? I know I do for the coaxial, but I'm not sure about that optical one.

Thanks if you're able to help. I really appreciate it.

Heh, one more thing. Seeing as how these boards are quite expensive, if anyone's interest...possible group buy of these things?

~Tom

 

[Garbz]

The schematic selection switch would work like that. The problem is that you will limit the device to 2 possible inputs. Hook one of the RXSEL channels to ground permanently places the bit at 0. Applying a switch to the otherone between GND and VLL allows you to select the status of that bit. However that schematic has a 47k resistor to VLL but shorts directly to ground. I'd put a resistor inbetween the switch and the ground too to limit the current.

In theory the other main chips should be voltage supervised as well. For instance most DAC chips require 1024 system clocks before they come out of reset. This happens in an instant so it's usually safe to link the DAC chip to the voltage watchdog. The main problem is that when the signal lock is lost the PLL looses the clock as well. When this happens the DAC normally goes nuts as in my case. The DAC put out a very rusty 15khz tone when this happened. You can either add another clock which the CS8416 uses when the signal dies or run the DAC and all other chips via a Transistor NOT gate connected to the CS8416's NV/ERR pin. When it looses signal lock NV/ERR goes high, the DAC's and other chips get pulled low into reset state, and everything is good. It even saves power

The TX pin is an output pin. It allows you to use the CS8416 as an input selector via the TXSEL pins. If you ever want to add an output I suggest you tie the TXSEL pins to RXSEL so the output will match the input selected. I simply placed a cap and a jumper in series so I could add this option later.

The transformers don't filter noise. They decouple devices which are linked together. You don't need on as such on the coaxial input but it does help. That said there's a whole array of fancy circuits you can put on your input such as buffering chips to make the input section better. I ignored all of these on the optical input simply because the TORX part I used had nasty jitter specs anyway. I'll only really use the Coax inputs.

Group buys are no nos here at the moment. I had my board printed at www.batchpcb.com and it worked out very cheaply. $40 inc shipping for a 9x9cm board.

 

[Garbz]

I was just double checking something on the TX pin and it occured to me that you may wish to pull this up to VL via a 47k resistor before the cap. The reason being that the PCM1794 is very similar in operation to Delta Sigma DACs and will perform better with Phase Detection Update Rate set high. The downside to this is that you will not be able to lock onto a 192khz S/PDIF signal, mind you the majority of devices out there don't support this frequency on their S/PDIF output anyway.

 

[Nerull]

Yet again, hehe. I just have to know these things. If I were to use a differential input (using RXP0 and RXN of course) from S/PDIF from the circuit attached and that alone (no more optical, seeing how it is crap anyway) how would I configure the RXSEL's? Would I just tie them to ground as I don't have to select anything anymore?

And another thing. Is the NV/ERR pin directly connected with a NOT gate to the RST pins of the other chips? Is there anything else in between? And would I still need to the voltage supervisors on them?

Thanks again, CS8416 expert. (None other than Garbz)

~Tom

 

[Garbz]

I've been working on a way to use multiple inputs differentially without interference between devices. The inital test showed that I could select inputs in a differential setup but if two cdplayers were on at the same time I would loose the clock. I'll let you know if my plan B works next week

If you are only going to use a single differential input RX0 and RXN you would tie them to ground. May I sugest though that you tie them to ground via jumpers. Simply put 2x2 pin SIPs there and jumper to ground. Do the same with the inputs. That way if in the future you ever decided you may want to use RX1 you can lift the jumper and attach an input selection circuit. That said you can only listen to 1 source at a time anyway.

If you are going to go balanced take a look at Jocko Homo's input scheme for CS841x chips. http://www.diyhifi.org/forums/viewtopic.php?p=5707#5707 It does a few things like properly terminate, transformer isolate, convert to differential, and convert to TTL voltages, to the signal all of which I've tried and yield significant improvments for the CS8416.

The NV/ERR pin is low during normal operation. When the power first comes on the voltage supervisor will delay the turn-on off the CS8416. At this point it is possible for the rest of the chips to turn them selves on and cause errors. However as soon as the CS8416 comes on NV/ERR goes high which forces all the other chips into reset state. When the CS8416 has finished it's startup cycle and locked onto an incoming signal the NV/ERR pin goes low again and all the chips connected to it will then go through their startup cycles again.

The transistor NOT gate I'm talking about can be seen in here: http://www.garbz.com/amp/dac/final/dac.pdf Ignore J151 and J152 it's so I can tap and override the signals. I can't remember if it was you I sent the schematic for the CS8416 to, if not let me know and I'll email it to you. I don't wish to post it here until it is fully working as I intend it to. A few things remain untested.