Nonlinear Circuits Dividers incorrect?

[martb]

Hi,
Looks to me like "Divide and Conquer" is dividing wrong. . . at the top of the module,
3 is dividing by 4 (& 3/2 divides by 2)
5 is dividing by 6 (& 5/2 divides by 3)
7 is dividing by 8 (& 7/2 divides by 4)
Thanks!

Seems to be working correctly here. The lfo is running at 1 hz

dividers.jpg (66.38 KiB) Viewed 12119 times

[Steve W]

Hi,
Looks to me like "Divide and Conquer" is dividing wrong. . . at the top of the module,
3 is dividing by 4 (& 3/2 divides by 2)
5 is dividing by 6 (& 5/2 divides by 3)
7 is dividing by 8 (& 7/2 divides by 4)
Thanks!

Seems to be working correctly here. The lfo is running at 1 hz

I remember some of the modules were incorrect, and that they were fixed. Is it possible the D&C was never fixed?

[Steve W]

This is a clock divider with more outputs than you could ask for. It includes powers of 2 from 2 to 256, along with 3, 5, and 7. On top of that, there are 1.5 (3/2), 2.5 (5/2), and 3.5 (7/2). On these dividers, the half divisions are calculated when the incoming clock goes low.

The main clock input (on the bottom section) is normalled to the top three clock inputs. If nothing is plugged into the top three input jacks, they will receive the input connected to the main input.

I found my old Click-Burst-Counter preset (see above) and adapted it to use the Divide and Conquer Module. For comparison, I again used the clock dividers to count manual clicks X 32 bursts. For 10 x 32 bursted clicks (aka 320), clock divider method results are as follows

320 / 3 = 106
320 / 5 = 64
320 / 7 = 46

With the D&C Modules the results are as follows for the 320 bursted clicks:

Main in [320] divided by 3 = 80
Main in [320] divided by 5 = 53
Main in [320] divided by 7 = 40

Main in [320] divided by 3/2 = 160
Main in [320] divided by 5/2 = 107
Main in [320] divided by 7/2 = 80.

On the bottom half, I get clicks divided by 2, 4, and 8 respectively.

But, in terms of dividing pulse waves and doing a frequency measurement, divisions of 1KHz do work out (+/-).

It would seem then that as a clock divider, the math works out, but for counting its doesn't.

[Steve W]

100 Hz Divided and Conquered:

[schoekah]

Hi,
Thanks for the replies all -
When using clock triggers, Div&Conc doesn't work the way I expected it to - /3 & /4 both divide by 4. . .
When using gates, Div&Conc works the way I expected . . .
note: I was using trigger b/c drum modules say "trigger in"
(EG is to hit reset on the counter)

Thanks for checking it out everyone!

[Steve W]

When using clock triggers, Div&Conc doesn't work the way I expected it to - /3 & /4 both divide by 4. . .
When using gates, Div&Conc works the way I expected . . .

Thanks for posting the screen shots and for explaining the different results you got based on triggers v. gates.

[schoekah]

Thanks for posting the screen shots and for explaining the different results you got based on triggers v. gates.

you bet! very helpful to have others take a look

[trickyflemming]

Posting a reply kind of late on this, but regarding triggers vs. clocks, the most accurate/predictable behavior for Divide & Conquer is to use gates with 50% pulse width. Most clock dividers only take into account the rising edge, but the falling edge matters here for determining the odd divisions.

The one thing that I haven't patched is back in March I pushed an update to the VCV version to fix the relative phases of the outputs:
https://github.com/mhetrick/nonlinearci ... ec3af719d9

and it appears that it isn't fixed on the VM version:
https://github.com/mhetrick/nlcvoltagem ... nquer.java

I'll get a fix for that published soon.