Building a Precision Adder - With a question regarding generating a reference voltage

Ok, so I am making a precision adder to allow me to create larger sequences from my #2001 Keyboard Sequencer and Evil Eye sequencers.

The idea being that I can take the CV out of one, tuned to a root note, and add the CV from the other (clocked at a different division) to provide musical arppegios around that root note. As such my adder needs to be nice and precise.

I also want to duplicate the functionality of my Doepfer precision adder that I use in Eurorack which has a channel of attenuversion with a normalled voltage of precisely 1v being supplied to each input when they are not in use. However I would like to have that normalled voltage just supplied to the attenuverter input and selectable as 0v, 1v or 2v as I often find by the time I have plugged in a pair of sequencers and a modulation source (usually controlled elsewhere) that I can only ever add or subtract 1v from the final output. This is so I can shift my sequence up or down a pair of octaves separately from their original source, allowing me to derive, say, a lead from a bassline etc.

Did the obvious googling and found the information I need to actually create a adder section, no problems there, and know full well how to create a nice precision attenuverter, added some buffering to the input of the attenuverter channel (to keep all my input impedances the same, not sure if it would cause precision issues later on, but it didn’t go pop on the breadboard and signals did made it through - I just havent measured them properly yet so I still don’t know if it was the right thing to do), then through the attenuverter stage and off to the adder and output.

Here is the schematic I have so far…

Now, what I want to do is supply the Input 1 jack socket tip normal (which is currently connected to ground in that schematic - J1TN) to a switch, that would allow me to connect either a 1v, 2v or ground the input (giving me my 0v) to then buffer and attenuvert.

Would Sam’s voltage divider ladder as employed in the #1222 Performance VCO work here, or should I attempt something using a proper reference voltage regulators? Would it be a case of using both methods in combination - say configuring a 3v reference regulator and then deriving the 1v and 2v from that using voltage dividers and buffering the output? I have had references in many a project over the years, but never actually tried to design something myself that has actually got them in, so I am a bit bamboozled with regards to how I should go about this.

Also, would it be a good idea to reconfigure the attenuverter to use a B10k, rather than my goto B100k, so that the rest of the resistors would come down to 10K and match everything else. Well, except the load resistors, they would end up at 4.7k, but thats a whole different thing. I don’t think it should make a difference here, but I am not sure, and it’s too late to go rummaging round with a multimeter.

Any input from you guys would be most appreciated.

EDIT: I am aware I haven’t actually connected the output jack to ground, and that the main power conditioning caps have no value - I’ll rectify that when I am not half asleep. There may be other little drawing errors too, but I wanted to get across what I was doing quickly.

1 Like

The latter’s exactly what happens in the 1222. There’s a 5 V reference, trimmed down to 4 V, and then divided down to 3, 2, and 1 V.

(But there are voltage references and voltage regulators, but not “voltage reference regulators”. A voltage reference is what you want.)

Yes. It’s typical to use 100k input resistors on an input stage, to provide the usual 100k input impedance, but (sayeth Matthew Skala) aside from that, it’s better to use lower values to improve the noise floor.

What’s R2 doing? It looks to me like it’s making a voltage divider with RV1 so you’ll never get your full input voltage at U2 pin 10. Is that intended?

2 Likes

If you want a really nice reference voltage, there is Yusynth’s CV standard :
http://www.yusynth.net/Modular/index_en.html
No need for precision resistors, but ton’s of trimmer pots.

2 Likes

As you have probably found out by now, there are many ways for creating references :wink:. If you want to stick with your current method of using precision parts (rather than trimming), I’d suggest you use something like an LM4040-3 and divide it down from there.

Also, BIG RED FLAG :checkered_flag:, your power connector is reversed. Pin 1 should -12V.

2 Likes

Thanks for the responses, it’s always good to get some feedback, especially on stuff I have done late at night.

@analogoutput I hadn’t even clocked on to the whole accidental voltage divider at R1/RV1, no that is not supposed to be there at all, so I’ll drop that resistor. I’ll drop down the attenuverter to a 10k, and hopefully not screw with it’s non-liniarity too much.

Ta on the power connector @TimMJN , hadn’t even looked at that, just whacked one in on the schematic. As you can see I have gone with an LM4040AIZ-2.5 (because I had them in the parts draw) and a trimmer to get my initial 2v reference, and divided from that.

@eric Since everything else in the unit is requiring precision matched resistors, it’s not exactly much of a problem to continue a theme here, lol.

Here’s my updated schematic:

I haven’t bothered adding an extra buffer stage between the switch and the jack normal, as it’s getting buffered immediatly afterwards, and getting hold of a dual precision buffer is tricky at the moment because the LT1013’s seem to be hard to find, out of stock, expensive, and have long lead times. Hopefully this won’t cause any problems.

I also have a feeling that R10 may be superfluous, so I might drop that too.

2 Likes