On Thu, 4 Dec 2003, Eisermann, Phil [Ridg/CO] wrote:

> Alessandro Queri wrote:
> > Also notice that the switching approach would
> > lead to noise
> > also on 5Vcc, and definitively on the input (before the 5v
> > regulator).
>
> hmm, so the 5V is generated locally on the board? in
> other words, a voltage higher than 5V is available on
> the design?

Well, let's start from our setup: take a Power supply @10-12Vcc and put in
parallel a spectrum analyzer (dont't know if it's really "in parallel" to
be *extremely precise* :-P). You want to build a device which has a
18F8720 talking to a 12bit i2c ADC which is fed by a 4-20mA converted
powered @+/-12v.

* You cannot radiate on the INCOMING supply from 150KHz to 30MHz
* you cannot have (much) ripple on the +/-12Vcc line (due to ADC
sensitivity which would lead to erroneous reading on the 4096 step-ladder)

>
> > Of course the whole thing has to cost less (doesn't
> > matter how  much less is less, just
> > accept "LESS" :-)) ), because it should be cheap when "in
> > production",
>
> less than what? you are going to have to add components
> to achieve what you are trying to do, and so it will cost.
> Or are you really saying you already have a design but it
> generates too much noise?

It's a typical italian approach ;-): it has to cost less. Don't matter how
much it will cost, since you'll have a discount then. I was obviously
kidding...

>
> and when you say "production" are you talking 50 units per
> month or 500/month or more?

Should be a commercial product, let's say 10000/yr.

> > The real matter is to be VERY low from
> > 150Khz to 50MHz and with a PIC (18F8720) running @ 20MHz (...x4!)
>
> But the PIC isn't going to be running off the 12V! Are you saying
> there's already too much noise in your 5V supply?

From last test we have already some noise both from the PIC itself on the
INCOMING (12V) supply (the pic runs @5V), and from the switching power
supply feeding the GSM module (did I ever mentioned that? ;-) )

> If the 5V is locally regulated, and you indeed have a higher
> voltage available, then power the switching supply off that.
> place rf beads or a suitable LC filter before the 5V regulator
> if noise is getting in that way.

This should be too near to INCOMING SUPPLY UNDER TEST and not foolproof at
all (input range may vary depending on foolness, and 5V are 5V in the end)

> You do have access to an oscilloscope?

Unfortunately yes ;-(

:-)

Ale

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics