Developing a powerful digital looper project for the DIY community

[Hides-His-Eyes]

DOD DFX9/91/94s are pretty common too.

[cloudscapes]

those old digidelays are my favorites too. I've got a DFX94, a few PDS delays, Aria DD-X10 (is amazing!)..
I've been trying to plan/build a looper for a couple months. truncating and realtime speed control will be necessities. overdub not so much.

I messed around with a couple 16bit DACs I found this morning. no luck with the TDA1543. the WS pin (left/right switch) is used to "finish" the LSB but it calls for it to be triggered before the last bit. I haven't figured out how to do that with an AVR or dsPIC. the offset options can only go so far. bitbanging it doesn't even give me random voltages. it just sits still dumbly because I guess the clock doesnt line up perfectly with WS.

a bit more luck with the PT8211 I found, but it completelly ignores the MSB no matter how I offset everything. so I just have 15bit depth on that one. it's strange, one would assume pulling the data line completelly low or completelly high would give me the min voltage and max voltage, but it instead goes to the "middle". for both of them. yet I still get that 15bit precision if I send the first 15 bits or last 15 bits.

[Taylor]

What's truncating in a delay? Does that mean that you can move the read pointer around without changing pitch? I could see that being useful in a glitch context. I'll see if I can do a mode like that in my looper.

By the way, it's quite easy to bit reduce on the FV-1, so there's no reason we can't have a 12-bit version of this looper - it already sounds rather crunchy due to the low sampling rate.

[Taylor]

Whoa. 

Today I added a delay time control to the looper. Instead of acting nice and polite like I expected, it turned this thing into a completely schizo glitch machine. With the time pot stationary, it works just as before, and you can dial in different times just as I expected and wanted. But when you play something into it and let it loop, and then change the time control, it glitches out in a fantastic way.

Instead of just repeating a snippet of sound, the snippet that you hear changes from moment to moment. So let's just say I recording myself saying abcdefghijklmnop into the loop. When I adjust the time parameter to a shorter time, I expected to hear

abcd abcd abcd abcd

(this I think is the truncation you guys were talking about above) but instead I get

abc nop def abdef abdeg fghijk pppp mno

all without playing with the knobs. So you can record a longer loop, then let it spit out random snippets of your loop in a jumbled and ever-changing manner.

By setting the time super-short, I get a very nice "playing inside of a rusty steel cube" kind of pseudo-reverb. I will try to record some clips today.

[cloudscapes]

That sounds fantastic! any chance for clips?

[Strategy]

great job Taylor. The sample-crumpling "truncation"  - i.e., sample time does not affect pitch but crumples up the data as you modify the delay time- there ARE some precedents for this, but they seem to always get programmed out of the line. I once borrowed an early Boss pitch shifting delay (possibly the sought after PS-2 - can't recall now, its 10 years ago) and one of the delay settings did that, I used it extensively. The later PS series did not feature this sample crumpling. IIRC, the EH "#1" delay has a nice crumple to it, and it can be really nice to be able to modify delay time without causing the huge pitch bends.

I really look forward to this project, once you have coded it up, it will get added to the queue.

Re: minor CV delay times, it's hard for me to conceptualize the lag in audio terms, without hearing it. If its glitch/delayed signal anyways, it might not be very noticeable, but on a filter setting, that may result in a kind of subtle stair stepping. It's worth a shot in any case, adding CV input jacks, esp being able to modify delay time sample crumple-glitching with my modular synth or analog sequencer.

The CV attenuation could be a pot after the input jack. Although I'm not sure what the ideal variable resistance would be for attenuating incoming CV's. Typical synth CV would be 5V, but they go higher. During the trouble shooting phase on the Tau, we actually cooked the CA3086 (ouch) giving it LFOs that were too hot from a CGS psycho LFO. But I did not measure the voltages. I will have to research what the hottest theoretical CV outputs are.

The recently issued Tiptop Z-DSP modular synth module (Eurorack format) uses FV-1, I think. A couple emails to Tip Top might clear up how to implement voltage control effectively with FV1.

- Strategy

-
Stratey

[Taylor]

RE: the lag, it's not stair stepping, it's just that the parameter changes 1/10th of a second after you move the knob. It's not something I hear moving knobs with fingers, but if sweeping with an expression pedal, or if you had a tempo-synced LFO going into it, it would be out of sync with whatever else is going on.

[Rectangular]

@Taylor

regarding your glitch "time control" ; you've essentially created a simple granulator. that's a whole other project I've wanted to do for a long time.

I forget exactly how the truncation was implemented in the older samplers. I recall a diode array and a switch matrix, but obviously you're trying to keep things simple and in software.

what makes the samplers from the early 1980s so special is that there wasn't really a precedent for the ideal sampler back then. so you got a lot of engineers experimenting with various options they thought the musicians might like. strange circuit designs, usage of local or proprietary japanese ICs, and constant partnerships between various (bubble-rich) Japanese manufacturers also helped cultivate this.  it wasn't until pop music and akai took over samplers in the late 80s early 90s, when you got a solidified set of expected parameters and controls for every sampler. they're all pretty faceless these days. 

[slotbot]

sorry i am late to this thread but i am sort of working on something similar. liek a looper/grain sequencer. i will try to make some clips tomorrow.

i made a test on breadboard that is just a sampler with a sort of "zero crossing" playback rate pot so say once your left of center it starts playing the loop backwards. you can screw with it in real time so it sounds like:

helllloooooo mmmmmmmmyyyyyyyyyyyyyyyyyyyy  nnnnnnnnnnnnnnnnnnnnnnnnaaaaaaammmaaaaaaannnnnnnnnnnnnnnnnnnnnn yyyyyyyyyym ooooolleh. and also a pot to truncate the beginning and end loop points of the sample once its recorded.

[Taylor]

What hardware are you using?

[slotbot]

dsPIC 33

but for what im using in it you could just use a pic24. Im not actually using any dsp functions. its more that i bought something high end to prototype with.

[cloudscapes]

"ooolleh" 

which dacs/adcs are you using scott?

[Strategy]

those truncate/chop/start/end knobs sound like my Akai S-612- one slider for start, one for end, and when you reverse them you are reversing the sample. Awesome hardware sampler,  I read somewhere that the engineer on it had gone from Electro-Harmonix to Akai, hence the pedal like nature and hands on functionality...

- Strategy

[slotbot]

oh man so weird i had a very similar idea to originally just have two sliders and a record button, maybe i saw this akai thing once and dont remember. but instead i decided to make the rate and direction in one knob. hmm.

i was thinking also somehting along the lines of having 4 sliders that act more as markers. then having different play modes for example one could have a pattern like, play from one marker to the next, but then skip from that marker to the next, then play from that marker to the next. etc... to rearrange the sample by chopping and reordering it.

or even having 1 slider and a button to "add a marker at the position of the slider" and a button to clear all markers or something like that. something simple.

also for hardware i am not using anything fancy right now. it was mostly about developing the algorithm before dealing with finer details. adc is built in and dac is some microchip spi dac.

[slotbot]

http://www.youtube.com/watch?v=FhuDJ2jMUr0

[Rectangular]

those truncate/chop/start/end knobs sound like my Akai S-612- one slider for start, one for end, and when you reverse them you are reversing the sample. Awesome hardware sampler,  I read somewhere that the engineer on it had gone from Electro-Harmonix to Akai, hence the pedal like nature and hands on functionality...

- Strategy

you're thinking of David %^&*rell. he worked for EMS in england, then went to design for EH, and finally Akai, where he worked on their first MPCs. absolutely wonderful engineer, I love all of his work.

[Taylor]

you're thinking of David %^&*rell. he worked for EMS in england, then went to design for EH, and finally Akai, where he worked on their first MPCs. absolutely wonderful engineer, I love all of his work.

Ha, that would be Co.ckrell. Censoring software...

[cloudscapes]

seeing how precise and flexible filters are needed for those of us doing ada conversions, this tool is super useful:

http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en010007

introductory video
http://techtrain.microchip.com/media/filterlab/filterlab.html

[slotbot]

yes that software is nice. There are a lot of excellent TI application notes too that have similar information.

also maybe you could use somethign like the switched capacitor filters maxim makes because then you can have up to 8th order in an 8 pin chip which saves some board space/soldering time . maybe TI makes somethign similar too since thay love signal processing so much.

just go to the maxim website and go products->analogfilters, neat stuff!

sort of side track:

in terms of 16 bit sampling dont you think it is excessive for soemthing like this? i think about 50 uV for the LSB (wiht a 3.3 volt chip). not to be pesimestic but your not likely to beat that noise floor without some nice filtering/pcb layout. and plus you are taking up 25 percent more memory (than 12 bit, boo). Maybe you could use an ADC with nonlinear quantization with ~10 bits and still get a nice dynamic range.

[Rectangular]

you're thinking of David %^&*rell. he worked for EMS in england, then went to design for EH, and finally Akai, where he worked on their first MPCs. absolutely wonderful engineer, I love all of his work.

Ha, that would be Co.ckrell. Censoring software...

haha, da.mn, I forgot this forum auto-censored things. god co.cking da.mnit.  anyway, Co.ckrell is a great engineer.