External cv inputs and attenuverters board

Status
Not open for further replies.
Hi,

It's some time i'm studying and developing a modular synth, but as i'm mainly a musician i still have some doubts .

Could you please have a look at the schematics?

This board is directly connected with the panel of the synth and is meant to process the cv inputs and basically to provide data input/output
Teensy 3.6 is on a secondary board together with the rest of the hardware.
On the right side there are all panel knobs, on left side all CVs.
There are also 15 switches whose value is meant to be red by an expander and some of them are also connected to a LED.
2 PWM LEDs and a CV output.
Do you see anything wrong?
View attachment Schematic_601REV3_Sheet-1_20190426084849.pdf

Thanks a lot for your time
 
Wow, that's a giant schematic sheet. Even on my 42 inch 4K monitor, it doesn't fit unless I shrink to the point the text is barely readable.

Do you see anything wrong?

Other than missing another (also huge?) sheet with the rest of the circuitry?
 
Thx for your attention and patience Paul.
i'm sorry is huge, and i'm a huge noob... ehm.. that was my first schem :)

the other schem is a bit more refined, i omitted the connections to cs42448 (i referred to the forum for those).
TI people helped me with part of this one.


View attachment Schematic_BOARDII_Sheet-1_20190427140159.pdf

here are the components used
switches
http://www.massuse-relay.com/details/MTSSeries.pdf
knobs (10k / 100k)
https://electricdruid.net/wp-content/uploads/2016/12/Taiwan-Alpha-16mm-pots-datasheet.pdf
adc
http://www.ti.com/lit/ds/symlink/ads8668a.pdf
expander
http://ww1.microchip.com/downloads/en/devicedoc/20001952c.pdf
multiplexers
http://www.ti.com/lit/ds/symlink/cd74hc4067.pdf
dac CVout
http://www.ti.com/lit/ds/symlink/dac7571.pdf
regulator 1A 3.3v
https://www.mouser.it/datasheet/2/389/ld1117a-974076.pdf
regulator 1A 5v
https://www.mouser.it/datasheet/2/389/l78-974043.pdf
LED
prod. code WP710A10SRD/D red 640nm
 
Looking at this again, but only briefly.

Near the center of the 1st sheet, I see 5 input jacks and 100 resistors, which go to a 8 pin connected called "To EXT ADC2".

On the 2nd sheet, I see what I'm assuming in the mating connector (for someone who doesn't know your design, how can they figure out which connectors mate?) near the words "ADC2 INPUT FILTERS". Looks like there's only a 0 ohm resistor between the pins of that connector and the input of a ADS8868 chip.

Have you considered whether the ADS8868 can handle the full voltage range of the types of signals which are likely to be plugged into those jacks? Usually modular synth gear needs protection against full +12V to -12V signals. Unless I've missed something, it sure looks like there's nothing but a 100 ohm resistor and wires between the jacks and those ADS8868 pins.
 
Hi Paul,

I know, is messy.. i will do better. Is my first schematic!! Sadly this baby is super complicated.

Anyway, ads8668 has internal protection up to +/-20V and a 1Mohm input impedance, i link datasheet http://www.ti.com/lit/ds/symlink/ads8668.pdf

All CVs , with or without attenuverter, go to a RC filter on second board before getting to ads8668 pins.
On the RC filters, at first i placed (i was trying to do a 320hz lp filter) 10K res and 24nF caps for RC filters but people at TI said that that resistance was too big compared to the ads8668 internal one and would produce drifts and errors. So after this hint i did set the RC input filters as in schem 2 also to progress on the pcb design..
What you think about RC filter setup?
Thanks a lot
 
i also have difficulties figuring out what's going on, but fwiw ... keep in mind that the ADS8688 inputs can't float* (because of the way the integrated front end works). that would happen though, if i see it right, everytime you patch an otherwise unconnected cable into one of those switching jacks (PJ301M).

a/the solution would be using a voltage follower and pull down the inputs to 0V. that would also help in terms of impedance matching; the "convention" re input would be 100k or more. (i'm assuming that's eurorack)

*
edit: well, they can float, but you'll basically see 1/2 * VREF or so at the *input* in this case.

edit: also fwiw, the DAC (U2) seems to lack any buffering / output stage?
 
Last edited:
Hi mxxx,

thanks a lot, the system uses a cs42448 card for audio input/output, so the dac is meant to be used only for a CVout (0-5v) and yes, this is meant to be eurorack compatible,
anyway all cvs go to a RC filter on second board before getting to ads8668 pins.

If i put 100k after all cvs, how would you setup the RC filter on second board?

do i have also to take into account that internal ref. is 4.096V on ads8668 ?

The CVout DAC is lacking that at the moment, yes. could it be done with an opamp and a 1k res at the DAC output?
 
If i put 100k after all cvs, how would you setup the RC filter on second board?

sorry for being unclear: i didn't say put 100k after all the CVs, i said keep in mind the ADS8688 inputs behave weird when floating (which they will, under certain circumstances, in your schematic as is). that might bother you or not. using a voltage follower + (say) 100k *pull-down* resistor at the + input is just one way of dealing with it, ie trying to make sure the inputs always stay at ~0V. you can always put a RC filter after the op amp, if one is needed. alternatively, you could, i guess, use an inverting op amp / mixer (with one input connected to 0V) and put a/the suitable cap in the feedback path (re filter)

do i have also to take into account that internal ref. is 4.096V on ads8668 ?
with respect to what?


The CVout DAC is lacking that at the moment, yes. could it be done with an opamp and a 1k res at the DAC output?

no. 1k at the output will do no good: it'll degrade the DC accuracy. the correct approach is called "in the loop" compensation. take a look at e.g. Mutable Instruments Yarns for details.
 
Last edited:
On floating CVinputs, you think ads8668 (not 88 so 12 bit) readings will be so crazy that cannot be managed via software, like, with a threshold?
Thanks for link and the input on dac.
The dac used here is i2c, rail-to-rail and fed with 5V . so i shouldn't do any scaling right?
 
On floating CVinputs, you think ads8668 (not 88 so 12 bit) readings will be so crazy that cannot be managed via software, like, with a threshold?

they won't be crazy; the input will go up to 1/2 VREF and sit there (also see here); so yes, there might be ways of dealing with this in software, though doing it in hardware seems to me the better/easier approach. it's a borderline case anyways, it can be avoided simply by not patching unconnected cables, ie first patch the output/source, only then the CV input.

Thanks for link and the input on dac.
The dac used here is i2c, rail-to-rail and fed with 5V . so i shouldn't do any scaling right?

it depends. 5V is just five octaves, and there's no guarantee you won't end up with a little offset; scaling/offsetting on the other hand isn't a big deal: it's just one op-amp you need, so not a lot of parts. but you'll end up with more range (say 10V); you can make sure to cancel the offset (= get exactly 0.000V); and amplifying/buffering the output offers some protection if you accidentally patch a signal into that output.
 
Status
Not open for further replies.
Back
Top