esd diodes as diodes

Depends on far "range" ranges :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

It's messy because his traceable calibration reference is no longer where the I/O is.

ot

how about something like this?:

/tpd4e004.html

4 channels, 1.6pf, 1nA

-Lasse

That's nice. 1 nA typical leakage I could maybe tolerate. It would constrain me to 5 volts p-p swing, but that might work, too. Thanks.

John

How do you intend to do the signal conditioning ?

The anti-aliasing filtering needs to be done individually for each channel before the mux (it is no use of doing between the mux and the ADC :-).

Adding two protection diodes for each channel should not be a big issue.

Sometimes, there is just no space on the PCB. In our case, we have a

26 pins connectors with signals coming from all sides (top, bottom PCB, left and right routings). The ESD diodes (0201 SMD) have to sit in the middle. But the assembler is cursing us (behind our back, of course). We are seriously thinking about building the ESD diodes inside the connector (26 pins 2.54mm zig zag). The rest of the board does not need such high density anyway.

see:

Geez, if I use 0402s my production people come after me with pitchforks and torches. Tombstone Territory.

The Brat, who does our board layouts, used to work in production summers. The last board she did, she used 0805s and up.

John

so

a

My point is to prebuild it separately, so the PCB assembler does not need to deal with it. We can build them on the connector or a separate PCB overlay. The single piece can then be wave soldered. We can build 80 pins stripe just as easily as 26 pins.

Size is relative. 0201 is tiny for the PCB assemblers, but huge for the wire-bonders.

I guess that depends on the cable length then. If the analog cables have to be tens of meters just having the input connector handy for calibration is a poor decision, but if the cables are reasonably short digitizing locally makes no sense since the entire system is local.

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

No, other people know better than to even try such a dumb stunt.

Maybe you could have Jim Thompson design a clamp chip for you. ;-P

Good Luck! Rich

Reread what he's said. His calibration reference is muxed into each input. If you distribute the input functions you have to distribute the calibration signal as well, negating your advantage.

We just convinced the production Neanderthals to allow us to go to 0402s. ;-) We just ran out of bag and had a couple of more pounds to add. 0402s are about all our pick-n-place machines will do, so we're not likely to go smaller anytime soon.

We've used 0603s as the standard parts for years (since we've been doing our own board stuffing, apparently). We only have a few values of larger resistors.

er

ear

/o.

h

er

put. If

It certainly does not negate the advantage of having the _analog_ input cables an order of magnitude shorter. Having them that long will degrade the quality of the sampled signal, there is no reasonable way around that unless the signals are DC, perhaps.

Dimiter

------------------------------------------------------

Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Ok, let me try this another way... How do you propose to do the muxing of this traceable standard? Cable *it* around to each point. Place one at every node? Of course it's a trade-off.

Some people have no sense of adventure.

I don't deal with people who deliberately insult my wife. Or with people who are obviously senile rednecks.

John

Why! BU Pimp, What is it you mean ?:-) Are you really that stupid, or are you pretending? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

                   Spice is like a sports car... 
     Performance only as good as the person behind the wheel.

:

y per

e

un,

e gear

d i/o.

with

ok

as

ither

input. If

ion

of

at every

It is a trade-off indeed, it is just that I would have made it in favour of the acquired signal integrity rather than the calibrating signal source. One can do things about the latter (average multiple measurements etc.) which cannot be done with the input signal.

Dimiter

------------------------------------------------------

Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

of

at every

The secret is, to pipe around only the standards that can be measured digitally. e.g. times and frequencies and counts.

e.g. instead of measuring current, simply count electrons flowing past.

I'm only half-joking. Any measurement that can be done digitally without reference to analog standards ought to be done digitally. In one case the traditional instrumentation for wheel speed consisted of a tachometer, followed by a frequency to voltage converter to make a voltage representing RPM, that we piped back to the DAQ system as an analog voltage, then converted back to a digital number. We would then integrate this analog voltage times wheel diameter to arrive at distance traveled. We would also used the reading from a fancy-pants

20-turn precision potentiometer which had been carefully set using precision test equipment to get the wheel diameter, a multiplication factor in the above calculations. Zero offsets were super-duper-extra- critical and a PITA.

Now we just count turns of the wheel and use landmarks to measure wheel wear dynamically.

Tim.

gear

i/o.

input. If

That may be the right decision, maybe not. It certainly is not an obvious one to make on a newsgroup, no less.

Maybe. But it's not always possible to make a choice that simple. I'm quite sure John has considered the issue.