Interfacing to a digital IC on a separate 5v supply

[markiemrboo]

Hi all,

I am designing a fairly simple PIC16F / PGA2310 based preamp and have the PIC always running on it's own +5v supply, in order to switch on / off separate 5v and +/- 15v supplies for the rest of the circuit (i.e it also controls mains power to the rest of the stuff).

This of course means that the PIC and PGA2310 / LCD etc are on separate 5v supplies, though their grounds are joined. Is there anything I need to know in order to make this work?

Suppose the PIC supply sits at a steady 5.00v as it's not heavily loaded, but the PGA2310 supply sags down to 4.50v for whatever reason (it shouldn't, but who knows). Now lets suppose I try and set the volume and send +5.00v signals to the PGA2310 control lines. Will any damage be done because the PGA2310 supply is only at 4.50v but the control signals are +5.00v? The datasheet says digital input voltage is -0.3v to +VD, so I assume it's going to be a problem. How might I work around it, if it is?

Any advice would be much appreciated!

[KSTR]

Use series isolation resistors on all digital lines to the PGA. I assume that you don't clock the data in at full rate (6MHz, IIRC) so the roll-off from input capacitance will not be an issue. I'd try series R in the range of 3k...30k. They'd current limit the high level voltages enough to keep any backpower issues out of the picture. The leakage current is to small for any significant drop in steady state (10uA will give 0.1V with 10k). If the PIC puts out rail-to-rail (and can source some current), that is true 5V with 5V supplies, you could also use R dividers, as the PGA's high level threshold is at 2V. Design for a 3V input level and you are safe. These series R (or divider), especially when combined with ferrite beads, also do a great job to prevent any HF from the PIC entering the PGA's analog circuitry.

The absolut safest way would be additional open-drain/open-collector drivers (made of discretes, e.g.) pulled up to the PGA's 5V. And the best isolation (HF-wise and DC as well) would be acheivable with optocouplers, which would allow seperate GNDs also, thus total isolation. Just take care not to use a too fast clock rate.

- Klaus

[markiemrboo]

Thanks. I will try series resistors.

I do have another question though, sorry. Pull up and pull down resistors on control lines..... pull down to ground doesn't matter too much I guess, but pull up to 5v... which 5v supply should I pull up to?

The PIC has protection diodes on its pins to clamp the voltage apparently, but I don't know about the PGA2310, 74HC595 and anything else connected to the second 5v supply. I guess I should pull up to the second supply then, as if it happens to be a higher voltage then the series resistor and protection diodes will save the day... right?

For example:

[markiemrboo]

Actually, the only pin I think I will need a pull up on is RA4 (I will be using a PIC18F) as the datasheet says it is open drain. I will just take that to the HC595 5v supply. The rest of the pins interfacing to the HC595 / PGA will be outputs at the PIC, always.... so I don't think pull up resistors will be necessary?

[KSTR]

If the PIC had open-drain outputs (GP0 and GP1 in your schem) that would be perfect. Just use pullups to the PGA's 5V, then a series R to the PGA inputs (not the other way round as in your schem).

[markiemrboo]

Thanks alot Klaus!