What factors affect the input voltage threshold of a prop pin? (and why?)

Tubular

I measured an input voltage DC threshold of 1.41v on the proto board TQFP.

This is about 42.8% of the supply rail, rather than 50% (Vdd/2).

Does anyone have any experience with what affects this threshold? Someone recently mentioned the packages have a different threshold (why should that be?).

What else affects this threshold - Temperature? Doping? Supply voltage? Lunar phase?

ManAtWork

I think the threshold is not exactly 50% because the P and N chanel fets at the inputs are not perfecty symetrical. The exact voltage changes with temperature and time (drift). Supply voltage can also have a small influence due to non-inear effects. I didn't test it but the package type shoult theoretically have no effect. But it could be that the chips with different packages just are from different wafers that differ slightly in doping.

I use the propeller inputs as sigma/delta ADC in my servo project. To compensate offset I calibrate the ADCs during each power on cycle.

Mark_T

There is another effect that will cause drift of the threshold - ion migration - the field across the gate oxide causes impurity ions in it to migrate (very very slowly at normal temps) which changes the threshold voltage. Thus the history of the input will have some effect on the threshold. This same mechanism causes DRAMs to gradually 'burn in' their contents such that after power cycling the normally random start-up state is somewhat correlated with the pattern stored before power-down.

Rayman

I'm a little surprised it could be that far off... Are you sure it's not noise? Did you try measuring the threshold coming down from Vdd too?

Do you have enough capacitance on the Prop pin being tested and the Prop itself?

Try using identical 100nF ceramic caps from pin to ground and pin to Vdd... That's supposed to eliminate Vdd noise issues....

Phil Pilgrim (PhiPi)

Rayman wrote:

I'm a little surprised it could be that far off...

In my experience, 1.41V is not atypical.

-Phil

Tubular

Many thanks ManAtWork and Mark_T. I suspected much of this has to do with the semiconductor manufacturing process, but the change over time is something I will have to deal with.

Rayman I was a little suprised too. My measurement method is as follows - use a 500 ohm 25 turn trimpot that has 0.1" spacing across P12-P14 of a proto board powered by a 9v battery to eliminate noise, ground loops etc. P12=Output Low, P13 = sense (wiper of 25 turn pot), P14 = Output High. So that part is all at DC. Admittedly a cap (or two) is not a bad idea as I have seen noisy wipers.

I use a counter to count how many 1's on P13 in a 5,000,000 sample bucket and display the result. I'm trying to adjust the pot to get a count of ideally 2,500,000.00, which is impossible because the threshold is very sharp (also indicating low noise) . Then I measure the wiper voltage to ground at about 1.41 volts. While measuring the wiper voltage with the Fluke DVM the count goes wild due to the multimeter leads, but thats ok.

I'm definitely going to have to measure across some different Props to get a feel for the range of values... the operating time / ion migration doesn't make this any easier!

Rayman

It's an interesting measurement. I look forward to hearing what you find.

I've done a fair amount of sigma-delta ADC testing with 10% resistors and always seemed to get close to what I expected. But, I don't think I've ever actually measured the threshold voltage directly...

Rayman

Actually, now that I think about it some more, the actual threshold level doesn't matter in sigma-delta ADC. So, I haven't really looked at the threshold voltage at all...