MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Dual Precision Adder - with schematics
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Dual Precision Adder - with schematics
Gibran
Hey guys ,

My dual precision adder with 0.002V tolerance , available as DIY too:

https://github.com/PantalaLabs/precision-adder

mskala
I think it'll be very hard to guarantee 2mV accuracy with that circuit. There are out-of-the-loop series resistors on the outputs, which means that the output voltages will vary depending on the impedance of the inputs they're connected to. The inputs are at 20k impedance, considerably lower than is customary for Eurorack, and that means if you plugged one of the channels into the other you'd have about a 5% voltage droop right there. There also doesn't seem to be any provision for trimming out the offsets of the LM324 amps, which are 2mV typical (up to 7mV max, depending on version) each for a total of up to 14mV per channel in that alone.

Putting the output resistors into the feedback loops would be a significant improvement.

I'd also have some concerns about stability. The op amps in the reference generator circuit have their outputs plugged straight into large capacitors, and there are no feedback capacitors on the op amps driving the module outputs, into which someone might plug capacitive cables. Those things create risks of parasitic oscillation.
efluon
Love the panel!
av500
more like a transposer than an adder, no?

the 7805 is not the best thing to use as a precision voltage reference, better to use something like an LM4040 and an opamp buffer if needed
CLee
Also, your trimming isn’t centered on the target resistance. You have 20k resistors and then a 20k with a 1k trim... no way to go below 20k
Grumble
You have to add the opamp output resistance to the 1st 20k...
CLee
Grumble wrote:
You have to add the opamp output resistance to the 1st 20k...


Ugh, Euro
Haralds:Werk
If you are going for precision with such a tolerance you need more stable CV Source. Like REF102 or similar. And precise OpAmps like LT1024 or such. Putting the output resistor in the feed back loop like mentioned above is a must. May be this one can help for orientation:
https://www.haraldswerk.de/CV_Sources/Scaled_Voltage_Reference_0_8/Sca led_Voltage_Reference_0_8.html
Nantonos
Sorry, this is a terrible circuit.

    * voltage regulator used as a voltage reference
    * 1k output resistors
    * no trimmable make-up gain on inputs
    * trim range does not cover range required
    * no precision specified for resistors
    * no overall error analysis
    * no accuracy specification for the module


Plenty of better circuits have been posted to this forum in the last few years.
Gibran
Hi all.

Just changed to the 100k resistors and 2k trimmers.

The every loop is managed by a 100kH (hight) and 100kL (low) values.
With a bunch of 100K , wecan select that with higher values and lower values to sum with every trimmer to reach the balance on each loop

MSKALA:
Quote:
Putting the output resistors into the feedback loops would be a significant improvement.

COuld you draw it where do I have to put then ?

Quote:
there are no feedback capacitors on the op amps driving the module outputs, into which someone might plug capacitive cables. Those things create risks of parasitic oscillation.


Could you tell me where do I have to put these caps ?

AV400
Quote:
the 7805 is not the best thing to use as a precision voltage reference, better to use something like an LM4040 and an opamp buffer if needed

I though that use a 78l05 and then trim it to +1 and -1 was OK.
Yes , I can change to LM4040 !

GRUMBLE
Quote:
You have to add the opamp output resistance to the 1st 20k...

Where ?

HARALDS:WERK
Many thanks for the circuit!!
I´ll study it NOW!!!

NANTANOS
Quote:
* voltage regulator used as a voltage reference

Even trimmed after it isnt a stable source ?

Quote:
* 1k output resistors

Thought that it was a rule on Euro , what do you suggest ?
Quote:
* no trimmable make-up gain on inputs

All loops are trimmable to reach the exact voltages on output
Quote:
* trim range does not cover range required

This info was cleared , the Highs and Lows range from the same resistor pack.
Quote:
* no precision specified for resistors

No precision needed, because they are all trimmed
Quote:
* no overall error analysis

I´ll try to post some tests here
Quote:
* no accuracy specification for the module

My readings came from a fluke multimeter

Thank you guys !!!!
av500
Gibran wrote:
Hi all.

The every loop is managed by a 100kH (hight) and 100kL (low) values.
With a bunch of 100K , wecan select that with higher values and lower values to sum with every trimmer to reach the balance on each loop


why select resistors when you can simply buy 0.1% specified ones? how is the average DIY builder supposed to select them?
Gibran
Because I can reach a better precision with for example a 100.1x for a "high resistor " and a 99.9x (low resistor) +trimmer

I thought that a simple DIY guy have an ordinary multimeter to select then

Am I wrong ?
mskala
Gibran wrote:

mskala wrote:
Putting the output resistors into the feedback loops would be a significant improvement.

COuld you draw it where do I have to put then ?

mskala wrote:
there are no feedback capacitors on the op amps driving the module outputs, into which someone might plug capacitive cables. Those things create risks of parasitic oscillation.


Could you tell me where do I have to put these caps ?


From my Web log article at https://northcoastsynthesis.com/news/understanding-stabilization-capac itors/:



Note the curent-limiting resistor (which also helps stability on long cables) is in the loop. The op amp drives the output jack to the desired voltage, rather than driving its own output to a voltage that will be dropped by the attached load. The 33pF capacitor kills the gain at high frequencies, preventing parasitic oscillation.

This is the standard circuit topology used in most carefully-designed modules that handle pitch CV. It would ideally be better to have smaller resistor values for noise reasons; the specific circuit shown is from a module where I had other reasons to use 100k. The optimal value of the stabilization capacitor will vary; in that particular circuit it was chosen by experiment, as described in the Web log article.
CLee
Here is a quick video I did a while ago showing why the 1K output resistor is a bad idea and why the circuit mskala shows should be used



your output voltage will stay stable as the load changes (you plug and unplug modules) on the modules output
Gibran
Thank you guys ,

I´ve done some improvements this night.

MSKALA
I was answering the post on mobile , and after I committed the post I remembered what was the feedback caps.
Thank you a lot for this info.
And yes , improved my final stability a lot.

I did not moved YET the output resistor to inside the loop.

CLee
Thanks a lot for this video!


Made some other changes:
-raised all 20k range to 100k range resistors + trimmed
-replaced the lm78L05 by a AMS1117 3.3V (I had some pieces here) , of course , changed the resistors to new values on #G1
-replaced the ICs to TL074 and 072

With these little changes , the output have a little oscillation when I connect to a VCO , for example. Just a little!

Tomorrow , I´ll move the output resistor to FB loop.

Again , thank you guys!
clorax hurd
Gibran wrote:

-replaced the lm78L05 by a AMS1117 3.3V (I had some pieces here) , of course , changed the resistors to new values on #G1


but AMS1117 is still a regulator and not a voltage reference. For example it has around 0.5% deviation just between 5°C and 35°C.
The part you want to use is called voltage reference. Even if trimming, you need something with a small temperature drift... and once you're getting such nice part, you can as well find one good enough so that even the trimming is not needed.
Gibran
Hi Clorax

Thanks for this info and to clarify the difference between regs and refs. :-)

I already made a little order for the LM4040s but it will take a little time to arrive from china. Here in Brazil they cost 9 to 11 times each one.

As soon as it arrive to me I´ll share the results !!

Thank you !!!!
djthopa
Hey Gibran,
Thanks for your work and for sharing it with the comunity!
Following your projects with lots of interest on github Guinness ftw!
Gibran
Thank you djthopa !!!
Feel free to ask anything about the projects !!!
Gibran
Sorry to take you so long guys.

To MSKALA and CLee.
Finnaly I did the change on my board including the 1k resistor inside the loop.

MAGICALLY the voltage did not fluctuate when I plugged to my oscillator!!!
Not even a 1/1000V

WOW!!!! Many thanks to share your knowledge!

As soon as the 4040s arrive , ´ll post the final schematics and board !
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group