digital potentiometer with Low Rwb

Hello hello,

I'm new here but was hoping someone could help me out. I'm in need of a digital potentiometer with SPI control, 10k end-to-end (Rab), and a minimum wiper resistance (Rwb) in the single digits. Does such a device exist?

The best i can seem to find on the market is Rw in the 70 Ohm to 120 Ohm Range, there
must be something lower.

thanks in advance,

Colle

perhaps you could use one that has more than one "circuit" per chip, then you can parallel them to reduce the affect of wiper resistance. E.g. six 100K pots in parallel, for a 16.7K pot, or six 50K for 8.3K, etc. For a 70 ohm wiper resistance, that will be an effective Rw of around 11.7 ohms. Still not in the single digits but on the way ;)

Also, these are typically CMOS devices, so the higher the supply voltage, the lower the "wiper resistance". Also, check out the "Wiper vs Voltage" diagrams. That should allow you to select an 'active' region for optimally low Rw. For instance, the AD5206 has an Rw around 43 - 45 ohms with a VDD of 5.5V and a Common Mode voltage in the range of 0 to 3.5 volts. Divide that by 6 and you get something more like 7.5 ohm wiper resistance (though the specs say that Rw can go as high as 100 ohms at VDD = 5V, but that probably takes into consideration the full temperature range, as I doubt Rw is temperature compensated, or perhaps they are lumping in the MIN VDD specs in that number).

Also, the AD5206's Absolute MAX VDD to GND is 7 volts, so consider running it at higher than 5V to get the lowest possible Rw. Probably good to add extra filtering on the supply line so spikes don't violate the MAX voltage.

In matters such as this, you can always contact the manufacturer and ask for recommendations. I've done this (even by email) and received all kinds of very useful information (including things I had never even considered)

Thanks a lot Steve i knew there was a simple solution that i was over looking! (I'm a little ashamed i didn't think of it myself.. )

No worries, that's why two heads are better than one

I've had plenty of V8 moments of my own!

In thinking about the implications of paralleling digital pots it occurred to me that there may not be a way to get them to all change at once. There will always be a kind of ripple effect, and that might be a source of noise or other mal-effect. Just something to consider (just playing devils advocate against myself .

But, then, on further thought, that could also play to ones favor, as it might also be possible to achieve finer gradations of adjustment by changing only one pot at a time. What...like 6 times the resolution for a 6-gang system? Or, since there are 6! * 256 permutations, how about a possible 184,320 resistance values (though, there might still be a lot of duplication of values -- I haven't really thought this out).

And that suggests a possible other way of swamping out Rw (but this would only work if you plan to use the digital pot as a voltage divider): Since you will have something like 6 times the resolution, perhaps you could limit the wiper range to values above some limit, such that said limit would swamp out Rw. But, you would need to tie the lower leg of the pot to a negative voltage, so when the pot reached the limiting value, the output of the voltage divider would be zero.

BUT, the transfer curve would only be an approximation of linear. It would be a "jagged" line, but the errors might be negligible enough to not matter.

Example (simplified to the 3 pot case) [¦ means "parallel-sum"]:

30K ¦ 30K ¦ 30K = 10.000K
30K ¦ 30K ¦ 29K = 9.8864 Δ = 0.1136
30K ¦ 29K ¦ 29K = 9.7753 Δ = 0.1111
29K ¦ 29K ¦ 29K = 9.6667 Δ = 0.1086
29K ¦ 29K ¦ 28K = 9.5529 Δ = 0.1111

With this technique, the upper and lower resistances as divided by the wiper position would be 6.6667K and 3.3333K which is much higher than even the worst case Rw (33.3 Ω -- for the 3 pot case) Because, to achieve 256 steps you would only have to move the wiper 2/3 of the way (i.e. it starts at 10.000k | 0.0000k at 0 steps, and is at 6.6667K | 3.3333K after 256 steps).

You could also start at 6.6667K | 3.3333K and stop at 3.3333K | 6.6666K to swamp out Rw at both the top and the bottom, but you would have to adjust the voltages accordingly (i.e. voltage translation required).

In my application the added resolution is not needed. (but it could allow uses of a cheaper chip with less taps) I will be using it in Rheostat mode and more of a concern to me is the resistor noise voltage 9 nV/sqrtHz for the AD5206 and the tolerance 1% between channels is expectable but 30% device to device could be an issue.

Also i was considering using a window comparator and some logic connected CS to only allow it to update it value when the incoming voltage nears zero. I haven't thought this thru yet but it could eliminate most of the noise form switching.. but it could make it harder to control.

Thanks Again for your great posts !!!!!