3-bit folding flash ADC uses only 3 comparators

[Andon]

Thanks for that, Tom. Your last post before the most-recent one also brought it into perspective, insofar as the circuit being tailored to the IC - good points! I know that user slacker (reply #15 on page 1) was able to get it working with an LM324, and even then they mentioned that there were some dead zones at the extremes of the pot turn, which I also noticed. I also tried taking the initial voltage values that user earthtonesaudio mentioned for each word (4.07-4.93V) and tried voltage dividers on each side of the pot to really narrow it down, with limited success.

As I said, I can get some of the words (four or five, depending) and they all see to be "major" words, just not all of them inbetween.

[slacker]

Hi

This is what I built. The pot is connected across 9 volts (or whatever the opamp supply voltage is) and ground. AGND is half the supply voltage. Pot was 10k but the value probably isn't that important
R1 = 10k
R2 = 50k
R3 = 470k
R4 = 1Meg

I used an LM324, I think to work this needs an opamp whose inputs can go to the negative supply.

[ElectricDruid]

Thanks Slacker. I wonder if my LTSpice has the LM324 op-amp. It'd be good to try it if so.

It *should* be doable to get it running even with something hopeless (for this application!) like the TL07x that goes nowhere near the rails too, though, right? Tweaks the thresholds to compensate, off we go?!?

[slacker]

Yeah it should work with a TL074 so long as you keep the inputs in it's range. I'd forgotten that the reason I used an LM324 was so I could run it on 3.3 volts.

[Andon]

Posting an update to say that I was able to get this working when I tried it again on an LM324 - thanks y'all! Now to figure out how to limit/spread the sweep of the pot, as all of the words/numbers occur in the middle third of the pot's rotation.

[ElectricDruid]

Posting an update to say that I was able to get this working when I tried it again on an LM324 - thanks y'all! Now to figure out how to limit/spread the sweep of the pot, as all of the words/numbers occur in the middle third of the pot's rotation.

That shouldn't be too hard. A resistor above and below the pot equal to the pot's value will limit you to the middle third. In practice, you'll probably have to tweak them a bit, and the annoying thing is that it's all interactive. Altering one changes the other end too.
Still, it'll get you much closer.

[Andon]

Posting an update to say that I was able to get this working when I tried it again on an LM324 - thanks y'all! Now to figure out how to limit/spread the sweep of the pot, as all of the words/numbers occur in the middle third of the pot's rotation.

That shouldn't be too hard. A resistor above and below the pot equal to the pot's value will limit you to the middle third. In practice, you'll probably have to tweak them a bit, and the annoying thing is that it's all interactive. Altering one changes the other end too.
Still, it'll get you much closer.

Yup. I've found that around 5K on either side of the pot gives a pretty even spread across the entire sweep, save for numbers 5 and 6 which are pretty close together. Tweaking, indeed.

[PRR]

https://www.jameco.com/z/GTF311221-R-Jameco-ValuePro-Switch-Thumbwheel-Binary-Coded-Decimal-BCD-10-Position-28-Volt-DC-15-Amp-End-Plate-Part-26771-_26818.html

[ElectricDruid]

https://www.jameco.com/z/GTF311221-R-Jameco-ValuePro-Switch-Thumbwheel-Binary-Coded-Decimal-BCD-10-Position-28-Volt-DC-15-Amp-End-Plate-Part-26771-_26818.html

Awww! Where's the fun in that?!?

This whole thing is an exercise. Even if we stick with the pot-to-binary outline, a $0.50 processor is cheaper and easier, and could even be done on a 6-pin device. But you've got to try and keep your braincells fit, a little bit!

[Andon]

A final thought/question on this for now: would tying the outputs to a cap to VREF (as in reply #26 on page 2) - using say 1uF - also work if using this as an insert for a 4051 to smooth the transitions between words/outputs?