AB resistance is not close to calibrated reading and measured Rw is higher than max spec

Hello mbratch,

We'll take a look at this and get back to you soon.

vsethia,

Can you help address the query above?

Thank you.

Regards,
Paul

Hi,

Offset is likely the fixed internal end resistance which is additive to the calibrated value, while the high wiper resistanceis common when using standard multimeters with low test currents. Could you please share the raw 16-bit hex value from the tolerance register and clarify if you are measuring with a standard 2/4-wire setup?

Regards,

Den

Hello dendalisay 

I have added the raw hex tolerance register values to the table in my original post.

I'm not sure I follow what you mean by "standard 2/4-wire setup". Could you please clarify?

You mentioned, "Offset is likely the fixed internal end resistance which is additive to the calibrated value". The data sheet doesn't mention anything about this resistance, but based upon my measurements and study of the datasheet, I assumed that there was such an additional "end resistance" between the B terminal and the top series resistance Rs, as well as possibly between the A terminal and the bottom series resistance. This is because the calculation given in the datasheet, Rwb(D) = Rab*(D/1024) + Rw, uses a 1024 factor, although there are only 1023 Rs resistors between A and B. Also, on page 21 of the Rev F datasheet, it shows a formula indicating Rs = Rab/1024 along with a simple diagram. These further imply that the resistance between terminal A and the 1023 position of the wiper is Rs, although I don't think measurements bear this out.

If there is such a resistance at either end or both, the variance of such a resistance is fairly wide between ICs and digipot channels based upon the measurements. What I found I need to do for accurate calibration is to measure and store Rwb(0) and Rwb(1023) for each digipot channel, then use the formula Rwb(D) = Rwb(0) + (Rwb(1023)-Rwb(0))*(D/1023).

Hi,

That behavior is might be expected, the factory 'Cal' value reflects only the internal resistor ladder, while your meter reads the full path including internal lead resistance and package parasitics. Regarding your question: a standard '2-wire' measurement adds your test probe resistance to the result, whereas a '4-wire' (Kelvin) setup uses separate sense lines to cancel that error out. Could you share the dmm model/test current, whether parts were measured in-circuit or off-board, and confirm VDD was 5.0 V?(since wiper resistance depends on VDD and temp)

Br,

Den

Hello Den

Thanks for the explanation. I am using a 2-wire setup. The DMM is a Fluke Model 172.

I do understand that there is additional internal resistance and package parasitics as well ass some inherent resistance in the test probes. The test probe resistance is negligible compared to the resistance values I'm seeing that are in the table I have shown. I measured the parts in-circuit. The VDD is 5vdc and I measured Rab right at the lead on the part, so there was no path between that lead and the probe. The wiper resistance I measured at a terminal on my circuit board. The path from the wiper lead on the component and the terminal only passed through an ADG715 analog switch with the path on the switch turned on. Per the data sheet for the ADG715, this should introduce < 10 ohms of resistance.

Mark

Hi,

Fluke is fine, but a handheld dmm in 2-wire mode will include any series elements such as board traces, solder joints, the ADG715, the meter’s test current etc. the factory 'Cal' value is the ladder resistance, so the ~130 ohm offset you see (~0.5% on a 25 kΩ ladder) looks consistent with additional end/package/board series resistance, and the tight spread across 18 units kinda ponts to a probable measurement/path effect than device failures. I do apologize, but can you try these quick checks:  try measuring with the ADG715 bypassed by probing directly at the AD5235 wiper pin with the device powered and set to midscale (code 0x200), using either a force-current/measure-voltage method or a 4-wire setup if available (recommended). For RAB verification, a 4-wire measurement directly at the IC pads would eliminate lead resistance. If still the same, please provide also device date/lot codes so I can on my end as well.

Br,

Den