Atmel ADC S&H switch on-resistance

[Johann Klammer]

Hello,

Does anyone know a ballpark value for Atmel's ADC S&H switch
on-resistance? It is not in the datasheet, it seems.

 

[Neon John]

Hello,

Does anyone know a ballpark value for Atmel's ADC S&H switch
on-resistance? It is not in the datasheet, it seems.

No set value. The mux switches the sample capacitor directly onto the
input. Thus the resistance is initially very low, rising to near
infinity. I mostly use the A/D converter in 8 bit mode and with that
resolution can get away with a 10k source impedance. Lower is better.
Buffer amps are best.

If you want to see this yourself, write a little program that samples
a pin every millisecond or so. Feed the port 2.54 volts through a 1
meg resistor. Observe with a X100 scope probe. That's how I
determined its characteristics.

John
John DeArmond
http://www.neon-john.com
http://www.fluxeon.com
Tellico Plains, Occupied TN
See website for email address

 

[Tim Williams]

Yeh... what part?

Tim

 

[Johann Klammer]

Yeh... what part?

Tim

atmega8
Or any of their 8 bit microcontrollers..
(I think they're all mostly the same ADC-wise)

 

[Johann Klammer]

atmega8
Or any of their 8 bit microcontrollers..
(I think they're all mostly the same ADC-wise)

nevermind, I've found something..

http://www.openmusiclabs.com/learning/digital/atmega-adc/

 

[Tim Williams]

nevermind, I've found something..

http://www.openmusiclabs.com/learning/digital/atmega-adc/

I recall something like 10k + 4pF from the datasheet... good to know
someone made their own measurements and found similar results. Thanks,

Tim

 

[Johann Klammer]

I recall something like 10k + 4pF from the datasheet... good to know

They do have a diagram in their data sheets that gives the resistor as
1..10k Ohm, which leaves some space for guessing...

 

[Johann Klammer]

1..10k Ohm, which leaves some space for guessing...

Oops, wrong. it says 1..100k which is even more off

Atmega 8(L) datasheet Rev. 2486Q-10/06 page 202

 

[Tim Williams]

Oops, wrong. it says 1..100k which is even more off

That's probably just shitty work on their part, but high ohm silicon paths
are notoriously inconsistent; they are matched on chip (hence precise
stable ratios, e.g., the R-R-R divider in the '555), but the absolute
value has something like 3:1 uncertainty (10k might be 3 or 30k, say).

Tim

 

[Tim Williams]

That's probably just shitty work on their part, but high ohm silicon
paths are notoriously inconsistent; they are matched on chip (hence
precise stable ratios, e.g., the R-R-R divider in the '555), but the
absolute value has something like 3:1 uncertainty (10k might be 3 or
30k, say).

For instance, the pull-up resistor is guaranteed 20-50k ohms, which is
relatively tight as these things go IIRC. Depends on technique too (long
trace vs. pinched trace).

Tim