Building a quantized sequencer without programming (MC)

[Waxxdaddy]

First post here. Looking forward to working with everyone here!

I keep being told that trying this is a dumb idea and I'm making it unnecessarily hard for myself. I totally get that but I'm stubborn and I'm having fun with this project so far! I'm enjoying circuit design but have no interest in learning programming at this point so I'm trying to find a way of building a quantized sequencer without using a microcontroller. The idea I came up with is a ciruit that is supposed to generate precise voltage references of 1/12V (for half steps) and 1V (for octaves) that can then be added together in various configurations to give a full scale of musical notes. (I'm not too worried about getting all octaves as long as I can get a few) The problem I keep running into is that my voltages are too unstable so the more of them I add together the further off I am from being in tune. (After adding 1/12V twelve times I'm over 150% off)

I'm adding the schematic as well as a flow chart to illustrate the intention of how it should work. Currently I'm working on debugging the original voltage reference IC (I'm using an LM4040) I cannot get it to be stable enough because as I turn RV3 to simulate the CV voltage coming from the sequencer the 12V supply voltage drops and with that drop the LM4040 voltage goes up by several mV. Aside from that I'm sure there are other places in my circuit where voltages are changed. (overall if I measure the output it keeps increasing as I turn up RV3 to open up the switches)

I'm very open to suggestions of doing what I'm trying to do here with other ICs as long as they don't require programming. I've looked into using ADCs and DACs but they are quite expensive. (At least the ones I can find)

On the flow chart I have not added the part that adds octaves since I haven't added it to the circuit yet either. But it would basically just be another chain of comparators and switches hooked up to a second variable voltage divider.

 

[Waxxdaddy]

One issue not discussed so far is no hysteresis is used on the comparators,

On another note. Looking at the datasheet for the comparator I'm using, is it correct that I can simply use the values for resistors shown here or do I need to adjust them based on my application? Can I use different values for the resistors as long as the ratios are the same?

 

[danadak]

The hysteresis window essentially keeps the compartor from switching on
noise centered around its value. So no hysteresis there is a V wiggle due
to noise when comparator hits the DC trip point.

As Vin rise not until it hits the upper trip level does it flip to logic low, and then
not until it drops from there to lower trip level does it flip to logic 1 (this is an invert-
ing comparator implementation. So width of window in V is usually set for pk to pk
noise level.

Comparator circuits with Hysteresis Design Tool

Regards, Dana.

 

[AnalogKid]

I had a quick look at what hysteresis is. By no means enough to fully grasp all the implications. So far it seems like it would only cause a "voltage wiggle" as the CV passes the threshold and would not explain the permament increase in voltage of the references? Am I understanding this correctly?

It will not explain a permanent increase, but that "wiggle" might be larger than you think.

At its core, a comparator circuit is a very high gain linear amplifier with no negative feedback to control its operation. Somewhere around the middle of the input voltage range is a small zone where the output stage is not saturated. For example, for a 5 V part with an open-loop gain of 1 million, somewhere around 2.5 Vin there will be a 1 microvolt region where the output will move 1 V. If there is 10 microvolts of noise, the output will try to produce 10 V of noise, banging up and down between the rails. Depending on what the circuit is supposed to do, this can be somewhere between annoying and disastrous.

ak

 

[rjenkinsgb]

Just checking because I know it's written a bit small. R27-R42 (the 10k resistors for the summing mixer and post switch) are currently 0.1% tolerance.

That's more than adequate!

What is a low value preset?

Something probably 5% to 20% of the value of the resistor you are "adjusting", depending on the circuit.
(Not needed for the 0.1% resistors in the divider or summing chains!)

eg. If you needed an "adjustable" 10K resistor to fine tune the gain of a circuit, you could use a 9K1 + 2K preset, or 9K53 + a 1K preset, so the preset adjustment allows going both above and below 10K.


The feed to RV3 and R4 (both linked together still) is what I think needs reducing a bit to at least 2V less than the LM339 supply voltage, either with a series resistor or a zener regulator, if you have not already done this.

 

[Waxxdaddy]

eg. If you needed an "adjustable" 10K resistor to fine tune the gain of a circuit, you could use a 9K1 + 2K preset, or 9K53 + a 1K preset, so the preset adjustment allows going both above and below 10K.

Okay so just to make sure I got this right: What you are describing is NOT a lower value fixed resistor + a trim pot like I have in the precision voltage source circuit for example. What you are saying is to have 2 fixed value resistors but one is most of the resistance and the other is the rest of the way? But then also youa re saying if I have 0.1% tolerance on a resistor that the low value preset is not needed?

The feed to RV3 and R4 (both linked together still) is what I think needs reducing a bit to at least 2V less than the LM339 supply voltage, either with a series resistor or a zener regulator, if you have not already done this.

I forgot to update the schematic but I actually added a 2.5V reference to RV3 and R4 and it improved the stability a lot! I updated the schematic now if you want to check how I set it up. (the only voltage reference I had lying around was 2.5V that's why I went so low. If I was doing the final build I would probably go for 5V so it can work with 5V CV outputs from other modules.

 

[Waxxdaddy]

From the Datasheet of the DG411 analogue switch IC I'm using. Where it says "Input Low Voltage......0.8V (Max). Does that refer to the signal going through the switch or the voltage required to activate the switch? Because my signal going through the switch is much lower at 0.0833V.

 

[Nigel Goodwin]

The voltage required to work the switch, obviously - as it's a logic specification (low is below 0.8V, high is above 2.4V, anything in between is out of specification and either could apply).

 

[rjenkinsgb]

What you are saying is to have 2 fixed value resistors but one is most of the resistance and the other is the rest of the way?

One fixed resistor (most of the resistance) and One preset resistor for the smaller value to to make of the total, that gives that total when it's somewhere around mid setting, so you can adjust the total value some amount above or below the theoretical value to "fine tune" the end result.

This is an example extracted from the schematics of an industrial servo - part of the speed setting input, which uses a differential amplifier so it is not influenced by ground noise (very much like professional audio, using balances signals via XLR connectors).

The R12 + P4 combination allow the non-inverting side to be matched or balanced exactly against the feedback resistor R18, with adjustment from around 9K to 11K in that case.
P4 is a 20 turn type on those boards.

(The asterisked parts are empty locations for special purpose adaptions).

 

[liamjerr]

First post here. Looking forward to working with everyone here!

I keep being told that trying this is a dumb idea and I'm making it unnecessarily hard for myself. I totally get that but I'm stubborn and I'm having fun with this project so far! I'm enjoying circuit design but have no interest in learning programming at this point so I'm trying to find a way of building a quantized sequencer without using a microcontroller. The idea I came up with is a ciruit that is supposed to generate precise voltage references of 1/12V (for half steps) and 1V (for octaves) that can then be added together in various configurations to give a full scale of musical notes. (I'm not too worried about getting all octaves as long as I can get a few) The problem I keep running into is that my voltages are too unstable so the more of them I add together the further off I am from being in tune. (After adding 1/12V twelve times I'm over 150% off)

I'm adding the schematic as well as a flow chart to illustrate the intention of how it should work. Currently I'm working on debugging the original voltage reference IC (I'm using an LM4040) I cannot get it to be stable enough because as I turn RV3 to simulate the CV voltage coming from the sequencer the 12V supply voltage drops and with that drop the LM4040 voltage goes up by several mV. Aside from that I'm sure there are other places in my circuit where voltages are changed. (overall if I measure the output it keeps increasing as I turn up RV3 to open up the switches)

I'm very open to suggestions of doing what I'm trying to do here with other ICs as long as they don't require programming. I've looked into using ADCs and DACs but they are quite expensive. (At least the ones I can find)

On the flow chart I have not added the part that adds octaves since I haven't added it to the circuit yet either. But it would basically just be another chain of comparators precision worldwide logistics and switches hooked up to a second variable voltage divider.

Can I use different resistor values, as long as I maintain the same ratios?