High value Digital potentiometer

[oracacle]

I have gone beck and looked at a long running project (I keep getting side tracked with work and other things).
In an ideal world I would need 11 M/ohms, however I think the 1M/ohm will be fine. The current prototype uses a rotary switch and 11, 100k/ohm resistors.
The application will be in an audio detector circuit (https://learn.sparkfun.com/tutorials/sound-detector-hookup-guide will replace r3 and r17).
I found this: http://www.farnell.com/datasheets/1934966.pdf?_ga=2.130267023.1813097283.1540600845-2089259871.1533396072 which I hope will work just fine.

Another option would be have a resistor ladder controlled by a Picaxe via some from of switch (opto, bilateral or something else).

I am also leaning towards using i2c if using a digital pot or port expander (to control switches if needs be), its normally fairly easy to use, and will keep pins spare. I also intend to use another smaller digital pot (5k) for control of a another sensor which can go on the same bus.

Currently the bread board version uses a 20X2, but will need to move to at least a 28X2, not just for pin count, but for programme space as well. hardware serial is being used for communicating with the display.

Is there anything I obvious I have overlooked? Is there a better option for a digital pot? Will the picaxe be able to deal with the i2c of the digital pot? Any suggestions on the smaller 5k pot? I've not even started looking for one yet.

 

[premelec]

Not sure if you could reconfigure the PCB op amp for + input and resistive divider on output with tap going to OP amp - pin ... then you could use a 5k or whatever digipot to control gain --- there are also log amps that cover a whole lot of variation [from ADI for instance].

 

[oracacle]

Even if I could, I don't think it would help as it will adjust the fly back resistance.
With that said, the + is already a resistive divider. Its in the first part of the circuit as an audio amplifier, which then runs into a peak detector rectifier, buffer amplifier and finally a Schmitt trigger.

the log amp looks like a good avenue to investigate further.

 

[premelec]

In what I suggested R2 would be from OP amp + to V- setting input impedance - feedback from the digipot would vary from about 1/255 setting high gain or even more if you put a resistor in series with the top end of the pot. Some stability problems might occur depending on required bandwidth etc...

 

[oracacle]

oh right, I see now. Not sure how that will effect things, my understanding (admittedly basic understanding) is that R2 was to control the discharge rate of C2 which act as an AC couple to the Op-Amp.

Bandwidth, in theory shouldn't be too much of an issue. The circuit is going to be used to detect sudden noises, providing the output of the amp is long for the peak detector to do it thing then it should be fine.

the circuit is going to build on a custom board so modification to the circuit are not going to be an issue. I will try and build a bread board version to test thing out.

 

[techElder]

I used a circuit, in my industrial life, that held a DC level of the noise floor. It acts like a threshold comparing circuit in that the signals of interest (essentially low frequency signals) could be seen on top of that DC level without feeding the DC level through.

Its called an integrator circuit, and is made by placing a capacitor/resistor combination in the feedback of an opamp that had a long period.

Solved lots of problems with variable sensor resistance and temperature drifts, but component values were chosen to allow for lower frequency sensor signals to pass through to a strip chart without the low frequency offsets getting through.

 

[hippy]

Rather than try and bodge something which requires mega-ohm digital pots, which I am not sure are that common, I would consider altering the circuit to add an op-amp amplifier which can have its gain adjusted using a more commonly available value of pot, or even redesigning the whole thing so it can work as you require.

It's basically an amplifier; taking a microvolt microphone input, boosting it to something in the volt range, and then into a peak detector. It should be easy enough to add an additional variable gain stage in that, and even possible to adjust the peak detection level.

 

[oracacle]

I have been looking at changing the amplifier part of the circuit.
It has occurred to me that R5 and R6 are a potential divider for a reference point. From previous testing getting rid of R13 and varying the value of R17 does what I need. Gain is calculated would then be calculated from 1+(R17/R6). Replacing R5 and R6 with 50 Ohm resistors and using a 5k pot for R17 should give the same results as the 2 10k and in 1M pot. If that works it seems like the simplest option.

 

[premelec]

I'm not completely sure but it looks like you are still referring to an inverting op amp circuit... values not quite right ... if you use the non-inverting [+ input] configuration and put the voltage divider on the op amp _output_ with tap to minus input that works...

 

[AllyCat]

Hi,

If you reduce R2 by a factor of 20, then you may need to increase the value of C2 by a similar factor to maintain the same frequency response. Furthermore, the gain calculation assumes that the output impedance.of the microphone is "low" (or at least comparable) to the input resistor, which might not be the case with an electret microphone. Note that in this configuration the input pin of the Op-Amp appears as a "virtual earth".

It might be better to take the feedback from a potentiometer configured as a voltage divider (to earth). But you would need to pay attention to the dc bias conditions, for example by putting a large capacitor in the low end of the divider chain.

Cheers, Alan.