MI Ripples theory - Can someone explain the Resonance path?

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, Joe., luketeaford, lisa

Post Reply
User avatar
Treefingers
Learning to Wiggle
Posts: 10
Joined: Sat Jun 09, 2018 11:37 pm
Location: NYC + London

MI Ripples theory - Can someone explain the Resonance path?

Post by Treefingers » Thu Nov 28, 2019 2:00 am

Hey everyone,
I recently breadboarded the MI Ripples filter from the schematic provided on their website and have a few questions about how the resonance works.

First, I noticed that the incoming signal gets dropped down to about 1/3rd of it's original volume - and then amplified 3x back to original value after the filter. Why is that? is that just to leave headroom for the resonance or is something else going on there?

I also don't quite understand the resonance path. It seems pretty straightforward, taking the 4p output thru the OTA and bringing it back directly to the input. But the original input signal also goes into the opposite input of the OTA, why? and why thru the 220n cap?

Lastly, I also noticed that the output on the 4p output has a DC bias to it that changes with the cutoff frequency. Shouldn't that be blocked by a capacitor before going to the output, or is it assumed that the next module it is going into will be AC-coupled?

Here's the schematic: https://mutable-instruments.net/modules ... es_v40.pdf

User avatar
devinw1
Super Deluxe Wiggler
Posts: 1705
Joined: Tue Aug 07, 2018 11:20 am
Location: Portland, OR
Contact:

Post by devinw1 » Thu Nov 28, 2019 2:22 am

Have you checked out AN071 (Filter design for the 2164) yet? Sorry I dont have a link but can post tomorrow if nobody beats me to it!

jakobprogsch
Common Wiggler
Posts: 103
Joined: Tue Jul 31, 2018 7:48 pm
Location: Switzerland

Re: MI Ripples theory - Can someone explain the Resonance pa

Post by jakobprogsch » Thu Nov 28, 2019 3:37 am

My guesses are:
Treefingers wrote:Hey everyone,
First, I noticed that the incoming signal gets dropped down to about 1/3rd of it's original volume - and then amplified 3x back to original value after the filter. Why is that? is that just to leave headroom for the resonance or is something else going on there?
The simplest explanation is that this way the input has 100kOhm impedance which is the "pseudo standard" for euro. While using 33k on the intermediate stages is just following the recommendations of the 2164 data sheet.
I also don't quite understand the resonance path. It seems pretty straightforward, taking the 4p output thru the OTA and bringing it back directly to the input. But the original input signal also goes into the opposite input of the OTA, why? and why thru the 220n cap?
The feedback signal is out of phase with the input and therefore partially cancels it. This is what causes the "loss of bass/fundamental" people often ascribe to ladder filter resonance. The same thing would happen here as well. By mixing in the input signal you can compensate for that. This compensation scheme also works for ladder filters btw. but since everyone seems to just clone the (mini)Moog filter that doesn't compensate for this, people seem to assume it's a ladder filter inherent thing.
Lastly, I also noticed that the output on the 4p output has a DC bias to it that changes with the cutoff frequency. Shouldn't that be blocked by a capacitor before going to the output, or is it assumed that the next module it is going into will be AC-coupled?
By that you mean the output with the VCA I assume? If you read the LM13700 datasheet you will notice that R19 and R21 are often shown as a trimmer with the wiper connected to GND. That trimmer can be used to trim out the offset of the VCA. I guess this was skipped here for simplicity. I can only guess at the reasoning but I would do it the same way arguing that any module that relies on the input not having a DC bias should be AC coupled on the inputs. Being fully DC coupled means Ripples can be used to "filter" slow moving unipolar signals like envelopes etc.

Edit: I guess the last part doesn't make sense since there is actually a coupling capacitor before the VCA? :despair:

User avatar
paperCUT
Veteran Wiggler
Posts: 704
Joined: Sat Mar 12, 2011 11:13 pm
Location: Stockholm

Re: MI Ripples theory - Can someone explain the Resonance pa

Post by paperCUT » Thu Nov 28, 2019 7:31 am

jakobprogsch wrote:My guesses are:


The feedback signal is out of phase with the input and therefore partially cancels it. This is what causes the "loss of bass/fundamental" people often ascribe to ladder filter resonance. The same thing would happen here as well. By mixing in the input signal you can compensate for that. This compensation scheme also works for ladder filters btw. but since everyone seems to just clone the (mini)Moog filter that doesn't compensate for this, people seem to assume it's a ladder filter inherent thing.
The MC-202/SH-101 method works really well, just feed some of the resonance into the output.

User avatar
forestcaver
Ultra Wiggler
Posts: 811
Joined: Tue Mar 14, 2017 4:41 pm
Location: UK
Contact:

Post by forestcaver » Thu Nov 28, 2019 9:43 am

Have a read of this:

https://mutable-instruments.net/archive ... alysis.pdf

(there are other useful analyses in the shruthi-1 GitHub)

User avatar
Treefingers
Learning to Wiggle
Posts: 10
Joined: Sat Jun 09, 2018 11:37 pm
Location: NYC + London

Post by Treefingers » Sat Nov 30, 2019 9:06 pm

devinw1 wrote:Have you checked out AN071 (Filter design for the 2164) yet? Sorry I dont have a link but can post tomorrow if nobody beats me to it!
Found it: http://soundsemiconductor.com/downloads/AN701.pdf

Never saw that document before.. it's great. Will start building some of the suggested circuits. As for Ripples.. I think I understand it now. I've messed with some of the resistors to keep even more bass while the resonance is up - it's great to be able to tweak it to taste.

Post Reply

Return to “Music Tech DIY”