Homemade digital potentiometer

[wa9w00d]

This idea is so simple, I’m sure it’s been done before, but I’ve never seen it.  The idea is to use fixed resistors in parallel, with all but one in series with switchs (a dip type would be my preference).  You could have whatever resolution you wanted, and no real need for a meter, because your switch settings would tell you the output voltage.

For instance, if you were building an OKR or Raptor 10 amp, you could have 16 steps of .2 volts each; a range of 2.8 volts to 6 volts with a four position switch and five fixed resistors.  For this example, the unswitched resistor would be 535 ohms, and the least significant bit 5,826 ohms.  After that, 2,956 ohms, 1,481 ohms, then 737 ohms for the most significant bit.  I gave close to ideal values here, but of course close IS good enough in this case. 

The limitations are that you must pick your resolution in doublings of 2, you need to be able to read binary numbers (not hard to do, with a little practice), odd sized steps would require mental exercise (not a bad thing, in itself), and if you wanted a whole lot of resolution, the number of switches would get out of hand.

If anyone wants to do this, I found it much easier to work with mhos instead of ohms.  That’s because mhos (the reciprocal of ohms) are additive is parallel.  So you start with the highest resistance you want (that will be the unswitched), find the next resistive value you want, turn both into mhos, subtract the smaller fraction from the larger, then the reciprocal of your answer gives the resistance needed to create that step.

Anyway, when I get around to my first Raptor mod, I’ll be trying out this idea.

Wayne

[Breaktru]

Hey Wayne,
I've seen something similar about 4yrs ago. Nuck's Fistpack. He used a SP4T slide switch with fixed resistors for selective output voltage.
Fistpack v2.0 Walk-Through

[wa9w00d]

Yup, similar, but just not as efficient as it could have been. Since he used a 4050, there isn’t even a need for an unswitched resistor.  You could have done the same thing with two resistors and two contact points (switches).  With three resistors, eight increments were possible.

Oh, and for anyone that didn’t want to learn binary, I *think* that Cherry’s PAGA-3100 series “thumbwheel” switch would work. It would be neat, but expensive.  I read four digit binary as easy as decimal, so I’ll stick with dip switches.

Wayne

[wa9w00d]

OK, so I was bored this evening – I bread boarded my example with the closest resistors at hand, measured the actual values, and calculated the resulting voltages per the Raptor data sheet.  If you’re bored too, you can look at the chart.  If I’m still bored tomorrow, I might breadboard in a Raptor chip to look at actual output.  Doesn’t look too bad, except for the 670 ohm sub of the 737 ohm ideal.  Luckily 741 ohms is fairly common, just don’t have one in hand.

Posting as a pic 'cause my table gets scrambled

[wa9w00d]

Just a little update to this.  I finally found some cheap rotary dip switches, and am in the process of making a “pass-through only” mod for use in stationary situations.

I’m using a 10A Raptor and a single hex rotary dip switch. I’ve settled on 4.4 to 5.9 volts in 0.1 volt increments.  If I find that too coarse, next unit will either use two rotary switches for 256 increments, or one rotary with a high/low switch for 32 steps.

Anyway, I slapped together a spreadsheet to do the math for me as I was playing with voltages.  I would post it, but don’t see a way.  If anyone wants it though, PM me an email address and I’ll send it.  It’s rough, and only set up for 16 steps at the moment, but there’s a parallel resistor calculator at the bottom that makes tuning for a target resistance easy.

Wayne

[david4500]

Tool here for 16 steps: http://www.diystompboxes.com/analogalchemy/emh/emh.html

[wa9w00d]

Hey David, good site there, and they have a parallel calculator too.  But the trimpot cheater tool is for a series configuration instead of parallel.  Would work just as well, but might make resistor tuning more critical in the raptor application (low ohms).

Wayne

P.S. - this made me think a little, though (ouch).  The reason I've toyed with this is that a pot being scratchy could cause large jumps in output. But if you used a hybrid, switched fixed for the more significant change, and a pot to fine tune, that might be the better solution. will have to sleep on that idea.

[wa9w00d]

Argh! I never even noticed how non-linear the chip is! and in the wrong way for our application.  How does anyone live with a straight analog pot?!  Anyway, I'm now thinking of  using fixed resistors switched in series, in parallel with a pot.  That way the pot has less effect the lower the resistance of the fixed array ( should help offset the non-linear curve of the board).

This does point out one thing in favor of a pure fixed parallel setup, though.  the non-linear curve gets corrected with each step up the switch series.

Wayne

[Visus]

Wayne the 12 turn and 24 turn pot swings them linearly never had any issue with jumpiness other than grounding to load.

Now a couple of singles/all lol I have are jumpy as heck gotta be really light handed on the turn. 

Seems like a lot of work/space taken for no mcu, then again its a hobby..

Cool way to dial em in tho..  It is crazy how 4.2v is not 4.3v etc.,  so much so;  our mouths pretty cool on sensory. 
Until been drinking

[wa9w00d]

Thanks for the info Visus.  My previous post is a prime example of "if you want to think, don't drink".   

I'm gonna do some breadboarding, but the math tells me that to go from 3 to 3.5 volts takes almost 90 ohms, but from 7 to 7.5 it's only a 13 ohm change.

Need to let the cobwebs clear first, though.

[wa9w00d]

Well, been kind of busy, but I've realized that this prolly isn't gonna work that great, with the curve of the raptor anyway.  The lower bits are gonna have a lot more effect as you move up in the voltage range.  I'm still gonna play with it some, but can't recommend it to anyone who is looking for "fast and easy".

Wayne