Project development: AquaMan Delay

madbean · Post by **madbean** » 18 Aug 2009, 04:55

So, in my spare time I'm putting together an analog delay that marries the Aqua Puss and Deluxe Memory Man feature set (with a few caveats) , and I'm hoping to get a little feedback and show my work in progress as it comes together. Here is a list of the design goals:

(1) Aqua Puss (DM2) serves as the basis for the design.
(2) Integrate DMM features of modulation and input gain overload.
(3) Double the delay time (or maybe 1.5 times the delay of the AP) ~ shooting for around 4-500ms.
(4) Design for 3205 layout.
(5) Perhaps borrow the "Tails" feature from the Magnus Modulus??

The obvious caveat being that if it's based around the 3205 it's not going to sound like a DMM. But, that got me thinking: would there be any benefit to running the input and output gain stages at 18v rather than 9v to simulate the headroom of the DMM? Would it be necessary to change the input buffer to 18v as well, or does that even matter (I should probably know the answer to that)?

Any help I could get here would be much appreciated. Assuming I finish this project, it will of course be released here

I've got the skeleton of the schem worked out already.

madbean · Post by **madbean** » 18 Aug 2009, 05:04

One other thought is to increase the analog portion a bit (maybe not double it, though) and then couple that with a digital hybrid as discussed previously in other threads.

cpm · Post by **cpm** » 18 Aug 2009, 08:02

i will follow this topic closely.
i am breadboarding too a DMM with 3205, 9v. Havent got much time so its not very functional yet.

From the info that i've compiled, and many different thought:

I would suggest working the clock around the 4047 ic, like DMM, so you could plug the whole clock and chorus circuit in there. 4047 seem like an easier to find part, and cheaper than a 3102
Input levels for the 3205 are very narrow, so special care must be put into designing the compander bits. Also this means that theres is no need for big gains along the path, so 9v would be fine.
To avoid distortion, noise, and misbiasing along several 3205s, independent trimpots for each chip would be a benefit. These are the usual bias, output balance, gain recovery... DMM for example skips some of them
...

I think that opamp filters sound better than transistors, and more convenient to design, more opinions here?

An interesting hack, from the old MXR delay, is using a tracking filter. The low pass filters are variable with clock speed, so in shorter delays there are more highs through because clock noise is not so close to audio frequencies.
I think this mxr does by means of a clever clock configuration, sort of a pwm, so filter variation is continous. Other delays use a switch for two fixed filter values for each long and short delay setting.

One can dream... and the icing on the cake would be to use a PIC for clocking, then implementing tap tempo easily, andd pwm for filter tracking.

Compiling, several circuits to have a look, for different approaches:
- Deluxe Memory Man: -15v, 2xMN3005, 4047 clock, opamp filters. Also 4xMN3008 version.
- Stereo Memory Man: +15v, 1xMN3005, 4047 clock, opamp filters.
- DOD 585: +18v, 1xMN3005, opamp filters
- DOD fx90 / Ibanez AD80 AD100 / Boss DM2 / Aqua puss: >=+9v, 1xMN3005, bjt filters
- Ibanez AD9 / CD10: +9v, 1xMN3205, bjt filters
- Behringer VM1: +9v, 2xMN3205, bjt filters (functionally looks like a DMM)
- more...?

Interesting units to study:
- Ibanez AD999: lots of chips inside ¿is there a schematic available?
- old MXR analog delay: 6xSAD1024(!), tracking filter(!)

soulsonic · Post by **soulsonic** » 18 Aug 2009, 08:25

I am currently tracing a Maxon AD-900... might have some interesting ideas in there.

DWBH · Post by **DWBH** » 18 Aug 2009, 15:32

Stereo capability would be sweet.

puppiesonacid · Post by **puppiesonacid** » 18 Aug 2009, 17:46

DWBH wrote:Stereo capability would be sweet.

an effects loop would be better ( for me at least ) would LOVE to have this option...

Jack Deville · Post by **Jack Deville** » 18 Aug 2009, 23:56

cpm wrote: One can dream... and the icing on the cake would be to use a PIC for clocking, then implementing tap tempo easily, andd pwm for filter tracking.

i think the PIC clock solution will the the cheapest and most stable...

madbean · Post by **madbean** » 19 Aug 2009, 00:55

Cool! Lots of good ideas here. I am getting excited about this project. I will have some progress to show soon.

cpm · Post by **cpm** » 19 Aug 2009, 04:27

i was looking at theese chips, integrated filters, for example:
http://www.national.com/opf/MF/MF6.html

that would make a very convenient tacking filter, variable along with clock fequency.
down side is that these seem to be quite expensive ics.

swt · Post by **swt** » 19 Aug 2009, 12:13

maybe you were talking about ibanez ad99? i coulnd't find a 999 only the maxon one....
http://www.dirk-hendrik.com/temp/ibanez_ad99.pdf
that's traced by dirk, the ibanez master...

cpm · Post by **cpm** » 21 Aug 2009, 08:57

i got on the breadboard a stripped down DMM, using one 3205, 9v
the audio path goes preamp->compressor->prefilter->3205->postfilter->expandor

the key point here is the input voltage that suit 3205, much smaller that 3005. About 0.5vpp is a sensible amplitude to avoid clipping.
The pre-filter on DMM is amplifiying, so now for the 3205 must be at least gain 1. Also, post-filters are preceeded by a gain reduction on DMM, while here for 3205 must be at least some positive gain

The schematics for delays using 3205 usually shows the chip powered at 9v. I find at 9v there's a lot of background hiss. This gets much quieter if run off 5v. So both clok and delay circuits are now on 5v.

Filters look like they are too agressive, dark sounding, My fear is 3205 is inherently darker sounding than 3005, so getting relatively bright clean repeats will be tough. One more reason to benefit from a tracking filter. Also a faster clock and additional delay chips would help, since these are cheap.

right now i find aliasing a greater problem than clock filtering when going into long delay times. a sensible 150ms is a good compromise for one 3205 stage.

cpm · Post by **cpm** » 23 Aug 2009, 23:48

this is what i have now on the breadboard, 3x 3205, no modulation yet
its still a bit dark, need more work on the filters

sound:
http://www.soundclick.com/bands/page_so ... ID=8010515
starts at shorter delays (~250ms), then medium (~500ms) and longest (~680ms)

quarara · Post by **quarara** » 25 Aug 2009, 09:35

[first post after an year or so of lurkin']

that sounds great!! yes, it's true, the repeats seem a little bit dark, but it is a great delay indeed, way better than the ad3208 (with 2 mn3205) that i've built. i'm sorry i don't have enough knowledge to contribute creating this project, but i'll follow every step of this.

cpm · Post by **cpm** » 25 Aug 2009, 22:43

i am thinking...

Each 3205 has a insertion loss dependant on input's frequency, thus creating a low pass filter on the way through the BBD chip. I guess this is where the main "analog" character of the repeats come from.

Instead of stacking chips in series, which will give longer delay times, but a stronger low pass as a trade-off...
what about paralleling chips, on a differential operation?

each chip gets its bit of phase inverted from the input, running at complementary clocks, and the outputs get combined in a differential amplifier.
Theoretically this allows for clock and harmonics cancellation, and Nyquist freq now equals clock rate. Then clock freq could be slowed to achieve same times like a two-in-series operation but only with 4096 stages which lessens the mentioned low-pass effect.

does it make sense?
anyone wants to try? my breadboards are busy at the moment...

cpm · Post by **cpm** » 26 Aug 2009, 13:56

so far, but still working on it

cpm · Post by **cpm** » 31 Aug 2009, 09:01

i have tried the "differential" configuration.
Its a balanced split going into each 3205, and then combining the outputs into one opamp with differential inputs.
It sounds ok, the performance is not that good as expected theoretically, i guess because of all the mismatch on two separate ics and surrounding components.
Good thing is the whining noise can be cancelled easily, but the post-filters still need to be strong enough to avoid the aliasing

For the delay time of about 500ms i think prefer 3 3205 in series (1+1+1), with a higher clock rate and higher order filter, than 4 parallel (2+2).

I am putting this aside, until i get bigger breadboards,

cpm · Post by **cpm** » 31 Aug 2009, 10:01

whay youve got madbean?
anything worth sharing?

madbean · Post by **madbean** » 31 Aug 2009, 17:16

cpm wrote:whay youve got madbean?
anything worth sharing?

Yeah, I've been pretty lame not contributing to my own thread here! Anyway, I've got a lot of eggs on the skillet right now, but I have managed to get the AP in Eagle and get a working layout for the MN3005 version. I'll post that up here in a bit. It's a sweet little layout.

Next step is to set up for the 3205 and look at adding the double delay time. I've got it set up for the MN3102 clock right now, but also I think trying the 4047 would be good, and integrating some of the ideas you've put forward.

Thanks for keeping the thread alive!

~Brian

madbean · Post by **madbean** » 31 Aug 2009, 17:49

Okay, here's the first volley: the Aquaboy. MN3005, MN3102 clock, set for 125B enclsoure. Or, if you use the components I listed below then you should be able to get in in a 1590B with no hassles. This is kind of my baseline, to which I will add all the goodies.

Schematic:

Aquaboy.gif

Project File:

Aquaboy.pdf

Small components I used
Trims:
http://www.mouser.com/Search/ProductDet ... 2P-1-105LF
http://www.mouser.com/Search/ProductDet ... 2P-1-103LF
http://www.mouser.com/Search/ProductDet ... 2P-1-223LF
Caps:
http://www.mouser.com/Search/ProductDet ... L50V1.0-RC
http://www.mouser.com/Search/ProductDet ... -L25V47-RC
http://www.mouser.com/Search/ProductDet ... L25V100-RC
http://www.mouser.com/Search/ProductDet ... -L25V10-RC

cpm · Post by **cpm** » 31 Aug 2009, 21:41

isnt AP already one mn3005?

for a second 3005 i guess you can just take the output from the 10k balance trim, bias again and go into the 2nd delay ic. Just make sure you have about 100k impedance load on the MN outputs
I used one opamp stage like the DMM, to adjust gain and bias, and a potential filter.

the main difference i found in going DMM's 3005 to 3205 is that in the 3005 design, signal is boosted before the 3005 stages, and then attenuated after, while with 3205 is the opposite, due to the reduced input levels.

for the clock, all that mn3102 can be replaced with the DMM's 4047 stuff. Note the trick of using the bias trimpots to get Vgg for the MNs

Project development: AquaMan Delay [documentation]