finding short on motherboards with a shorty (with display)

[kripton2035]

the shorty is not a precision device. what you need to measure is two relative points to see at less than a milliohm if the resistance is lower or higher between the two points. the absolute ohm value is not important, and not to be trusted to the last digit !

[wacd]

Okay, thanks for the replies. I understand that it's not a precision device but it has very good components to be one, right? Maybe there is still room for improvement.

[kripton2035]

I will see if I can find the good function for the devices you measured, but the current may be not constant for the measure so that's why the value is not linear compared to the resistors.

[wacd]

Improvement on the software side would be good, what else can be done to make the circuit physically better than it is now? Constant current source maybe?

[aeg]

Make sure to null out the probes before measuring shorty_measured_res, and again before making your confirmation measurements.

If that's not good enough, change "Rx = Rx * calibration_factor;" to "Rx = (Rx + 0.039) * 0.916;" for linear two-point calibration.

[wacd]

I Appreciate all the inputs.

[kripton2035]

Yes I'm confident that a constant current source would improve the device accuracy...

[wacd]

Today i tried supplying constant 50mA from a lab power supply , removing the transistor and the following resistor network.
Results are promising. Funky readings went away but need a little more special care.

Edit: i didnt supply constant current but limit the current Max at 50ma.

[kripton2035]

you could try something like this

[wacd]

How abour this CMJA5050

https://my.centralsemi.com/datasheets/CMJA5050.PDF

[kripton2035]

why not, but for me not a drawer component I have around, easier to get 2 ex of some basic silicon diode ?

[Kean]

How abour this CMJA5050

https://my.centralsemi.com/datasheets/CMJA5050.PDF

Did you review the datasheet for that part?

It has a 10% tolerance, takes over 10 seconds to come into regulation, and has a TC of -0.3mA/degC.
It is not a precision part, and it also seems to need a supply of 8V or so to regulate.

[wacd]

Okay, how do i limit the current to 50mA then?
These are the results when i use the lab psu limited at 0.05A its almost perfect.

[Kean]

Okay, how do i limit the current to 50mA then?
These are the results when i use the lab psu limited at 0.05A its almost perfect.

If you want precision, I would suggest changing the design to use an opamp controlled current source.  https://en.wikipedia.org/wiki/Current_source#Op-amp_current_sources

So, looking at the original Shorty schematic, R3 would be the sense resistor and selected for higher precision, and R1 would be replaced with a transistor controlled by the opamp.  R2 and the DUT become the load.  Assuming R3 stays at 1 ohm, you would need a voltage reference (the 5V or 3.3V rail may be OK) and divider to get the 50mV reference.  Opamp would have to be selected for a low offset voltage.

[wacd]

Thank you for providing information. I'll go and spend some time on this to learn it better so i can implement it , hopefully

[Kean]

Thank you for providing information. I'll go and spend some time on this to learn it better so i can implement it , hopefully

If you don't have experience with opamps, then a much simpler but slightly less accurate solution would be using an LM317L as the current source (in place of R1).
You can get the 50mA using just the LM317L IC and a 25 ohm 1% or better resistor, and calibrate the final result in the firmware.
This will perform much better than that part you suggested earlier.

Note that 25 ohms is not a standard resistor value, so just use a 24.9 ohm, or parallel 3 x 75 or 4 x 100 ohm.  Or some other combination like 10 + 15 ohm in series.

[kripton2035]

if you feed 50mA constant, I'm afraid you can't measure resistors above 0.5Ω
it would be better to stay with 30mA.

[wacd]

Oktay, Lm317L looks easy enough , i'll give it a go.

[wacd]

Hello again. I've been playin with the calibration settings and i think i finally arrived at a point that the results satisfy me.
Lower values are absolutely on point. It reads slightly lower value above 1ohm.
Ordered a couple mcp3421's , waiting for their arrival. The one i've set up on breadboard will stay for further tinkering.
And i made this pcb ordered from oshpark, hope it works as i'd like it to.