Xaoc Jena, Drezno, Lipsk: Leibniz Binary Subsystem update!

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: Joe., lisa, luketeaford, Kent

Post Reply
guestt

Post by guestt » Fri Jun 29, 2018 1:02 pm

I don't know about that.

I'm still not convinced it's doing anything it shouldn't be...

MSB will be triggered in the upper half of the saw wave, continuously at the clock rate, i.e. very fast if running on the internal clock, until the saw disappears below the half way threshold again.

That the triggers seem random on maths is because maths can't keep up with such fast trigger rates and only picks up one here and there.

At least i think that's what's going on... we need Xaoc dudes!!

guestt

Post by guestt » Fri Jun 29, 2018 1:04 pm

xaoc_tech wrote:
huffnPuff wrote:I’ll explain the 1st test further: if you are taking the MSB out you should be getting 1 pulse per cycle, next bit down 2 pulses per cycle etc.
That is only if you are asuming an ideal, textbook converter in a noiseless environment.

However, in the real world you need to take into account that each transition between digital states of the ADC is a result of a comparison of the instantaneous value of your signal with at least one threshold - that of the closest midpoint between two levels that represent an approximation of your value: from above and from below. Now, if you do the comparison 2 milions times pers second, and your value/signal has even a slight amount of noise, there will be random results of crossing the threshold, even two milion times per second. These yield at least two different binary numbers and they may differ on several bits, including the MSB. Slow down your clock, and you will experience more predictable results.

Btw, if you can get your original waveform with ADC+DAC linked, it's a proof both halves of your Drezno are working correclty.
Yes - this!! Thank you :tu:

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:05 pm

xaoc_tech wrote:
huffnPuff wrote:I’ll explain the 1st test further: if you are taking the MSB out you should be getting 1 pulse per cycle, next bit down 2 pulses per cycle etc.
That is only if you are asuming an ideal, textbook converter in a noiseless environment.

However, in the real world you need to take into account that each transition between digital states of the ADC is a result of a comparison of the instantaneous value of your signal with at least one threshold - that of the closest midpoint between two levels that represent an approximation of your value: from above and from below. Now, if you do the comparison 2 milions times pers second, and your value/signal has even a slight amount of noise, there will be random results of crossing the threshold, even two milion times per second. These yield at least two different binary numbers and they may differ on several bits, including the MSB. Slow down your clock, and you will experience more predictable results.

Btw, if you can get your original waveform with ADC+DAC linked, it's a proof both halves of your Drezno are working correclty.
I’m not saying you did a bad job with the Drezno! In fact I think Xaoc is way way above the average compared to eurorack vendors.

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:08 pm

Baddcr wrote:I don't know about that.

I'm still not convinced it's doing anything it shouldn't be...

MSB will be triggered in the upper half of the saw wave, continuously at the clock rate, i.e. very fast if running on the internal clock, until the saw disappears below the half way threshold again.

That the triggers seem random on maths is because maths can't keep up with such fast trigger rates and only picks up one here and there.

At least i think that's what's going on... we need Xaoc dudes!!
These glitches will trigger Maths just as they trigger the Numberwang.

guestt

Post by guestt » Fri Jun 29, 2018 1:12 pm

Okies!! :tu:

So, it's just noise in the system that's causing the behaviour you don't like?

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:17 pm

It’s the gremlins!

I’m saying if you test the Drezno’s performance expecting a mathematically perfect textbook behavior you’re in for a surprise. I don’t think Xaoc is debating that.

guestt

Post by guestt » Fri Jun 29, 2018 1:20 pm

Yes, but you should be able to get predictable results with a slower clock - yes?

I did exactly this and went from being very confused and wondering what the hell was going on to being very happy with it, for any non-audio work Clock it externally, that's just the way it is I think.

So I would like to know if your experiment works better with slower clock, start at 2MHz and slowly bring it down, see what rate it's at when it locks in and starts behaving correctly?

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:26 pm

I don’t have a clock (other than the Dreznos) that goes to 2MHz, the Synchrodyne’s goes to 400KHz which is an awful lot.

I’m done with my tests, now I know what scenarios will produce cleaner results and what scenarios will produce noisier results. I think this information should be available for anyone scratching their head.

With slower clock speeds you will still be getting the occasional glitch I think. It depends on how noisy the environment is, how long you wait for this to happen, how critical this is for you etc.
Last edited by huffnPuff on Fri Jun 29, 2018 1:32 pm, edited 2 times in total.

User avatar
xaoc_tech
Wiggling with Experience
Posts: 268
Joined: Mon Sep 28, 2015 8:29 am
Location: Poland
Contact:

Post by xaoc_tech » Fri Jun 29, 2018 1:27 pm

It really helps to understand that quantization involved in A/C conversion is actually plenty of comparisons, and comparison is a process that always amplifies noise.

The lower bit in the ADC you are considering, the stronger is the amplification of the background noise. However, the presence of noise in a DSP system is not a totally bad thing, because it acts as a dither and helps to mask the quantization distortion, or stepness.

If you look at the analog signal at the DAC output, it's not that noisy afterall. While the flickering of lowest bits is sometimes confused with lots of noise, they actually represent a very small part of the signal.



Btw, the LEDs in Drezno are slowed down for a reason. Rapid switching of LEDs produces plenty of current spikes that could additionally spoil your power rails and the GND rail. We decided it's better for your system to smooth them a little bit. Also, they are supplied from the -12V rail in order to help balance the power consumption. We do this in most of our modules.

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:31 pm

:tu:

The Drezno sits on the same bussboard as all my solid state logic modules, so I guess they’re contributing to the noise I’m seeing.

User avatar
Shledge
Super Deluxe Wiggler
Posts: 2885
Joined: Sun Jul 10, 2016 5:47 pm
Location: UK

Post by Shledge » Fri Jun 29, 2018 1:38 pm

I asked about this on here a while back.

Drezno's "noise" is a feature. Each bit output amplifies more than the last, making this more apparent. It's not the fault of Drezno, it's the actual noise floor itself of the signal going into it. If it were purely digital then this would not be an issue. Drezno will also show any internal noise on lower bits since the amplification on the lower bits is so extreme.

Putting a sound through in link mode so you can output it directly, means you won't hear this noise at all. Lowering the clock rate also will make this less apparent.

guestt

Post by guestt » Fri Jun 29, 2018 1:45 pm

Sorry if I wasn't clear, I meant use an oscillator to clock your Drezno ;)

I just did a little test:

Two clock sources synced together, one clock at 1Hz triggering a sine that goes to the input of the Drezno ADC, the other clock at 256Hz clocking the Drezno

i.e I matched the periodicity of the sine with the maximum resolution of the DAC - it outputs a sine!

This is exactly what I would expect :)

Maybe you should get one of those fancy filtered busboards?

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 1:52 pm

I think you and I are seeing the same thing, the difference is the approach - I’m looking for the glitches and you are looking to mitigate them. It could also be the case that my system is noisier.

It’s a synth module not lab testing equipment, I’ll add a note to my OP.

guestt

Post by guestt » Fri Jun 29, 2018 1:56 pm

Heh! Probably... :nana:

I have a very good/bad habit (depending on perspective) of always looking for the best in things and if something doesn't behave as I expect it to, I always put it down to my user error. Consequently it can take me a while to spot things that are actually wrong and I have to really focus and test properly to make sure.

Anyway, it seems as though all is well, I love this thing, totally changed the way I make music, so many possibilities!

I'd still like to know what the big bonus is with having two of them :D

User avatar
huffnPuff
Super Deluxe Wiggler
Posts: 1500
Joined: Sun Jun 05, 2016 9:50 pm

Post by huffnPuff » Fri Jun 29, 2018 2:05 pm

There, I toned my OP down a bit and added a comment.

Baddcr doubtless you’re a much more aggreable person than I am!
:chug:

guestt

Post by guestt » Fri Jun 29, 2018 2:09 pm

huffnPuff wrote:There, I toned my OP down a bit and added a comment.

Baddcr doubtless you’re a much more aggreable person than I am!
:chug:
Hahah... mostly yes, I live a very peaceful and quiet life!! But like anyone, I can be very disagreeable too - I think it's called being an actual real life human being who has values and won't stand for any shit!

Really enjoyed these discussions about this fantastic module - thank you!!

:guinness: :drinking: :guinness:

User avatar
maltemark
Veteran Wiggler
Posts: 709
Joined: Tue Dec 16, 2014 11:39 am
Location: Gothenburg

Post by maltemark » Sat Dec 22, 2018 8:19 am

Is it just me, or are most of the demos of the drezno + lipsk mostly users applying nothing more than a bit of a glorified bitcrushing on their sounds? I'm more interested in more exotic uses, analog computing, or other things I haven't even thought about... can anyone direct me to any threads or demos of uses like this? Disappointed / happy comments after having owned it for a while?

User avatar
mdoudoroff
Super Deluxe Wiggler
Posts: 3352
Joined: Fri Jul 17, 2015 8:10 am
Location: New York City

Post by mdoudoroff » Sat Dec 22, 2018 11:17 am

Agreed, maltemark. Intellectually, Drezno seems to carry a lot of potential. It feels deeply innovative. But so far, that potential feels unrealized.

User avatar
Swain Voorman
Common Wiggler
Posts: 66
Joined: Tue Feb 14, 2017 1:18 pm

Post by Swain Voorman » Sat Dec 22, 2018 11:49 am

Anyone know what's next in the Leibniz subsystem? Or at least how much HP I need to save?

User avatar
xaoc_tech
Wiggling with Experience
Posts: 268
Joined: Mon Sep 28, 2015 8:29 am
Location: Poland
Contact:

Post by xaoc_tech » Sat Dec 22, 2018 1:38 pm

8HP 8-)

iheartmodular
Common Wiggler
Posts: 153
Joined: Thu Aug 09, 2018 8:00 am
Location: Planet Earth

Post by iheartmodular » Sat Dec 22, 2018 2:29 pm

xaoc_tech wrote:8HP 8-)
ooooh!

give us a clue?

:cloud:

User avatar
mgscheue
Veteran Wiggler
Posts: 660
Joined: Wed Jul 30, 2014 9:34 pm
Location: Michigan

Post by mgscheue » Sat Dec 22, 2018 3:38 pm

xaoc_tech wrote:8HP 8-)
:nana:

Hmm, what could it be?

User avatar
MvK
Wiggling with Experience
Posts: 411
Joined: Mon Nov 24, 2014 8:46 am
Location: Berlin

Post by MvK » Sat Dec 22, 2018 3:41 pm

I've been in the Schneidersladen Xaoc workshop where they made a very detailed presentation of the two modules. It was recorded on video but wasn't able to find it. maybe it still needs editing.
We are the Borg. Resistance is equal to voltage divided by current.
https://soundcloud.com/logic_not
current 9U: https://www.modulargrid.net/e/racks/view/428567
Skiff: https://www.modulargrid.net/e/racks/view/818039

User avatar
Swain Voorman
Common Wiggler
Posts: 66
Joined: Tue Feb 14, 2017 1:18 pm

Post by Swain Voorman » Sat Dec 22, 2018 4:31 pm

xaoc_tech wrote:8HP 8-)
:tu: Cheers!

williamjturkel
Common Wiggler
Posts: 234
Joined: Thu Jul 06, 2017 10:50 am
Location: London, Ontario, Canada

Post by williamjturkel » Sun Dec 23, 2018 12:18 pm

maltemark wrote:Is it just me, or are most of the demos of the drezno + lipsk mostly users applying nothing more than a bit of a glorified bitcrushing on their sounds? I'm more interested in more exotic uses, analog computing, or other things I haven't even thought about... can anyone direct me to any threads or demos of uses like this? Disappointed / happy comments after having owned it for a while?
Running it with a slow LFO (or chaotic signal from sloths) as input and a mixer for some of the gate outs allows you to create stepped voltages for generative melodies. Put through a quantizer for musical results. You can also get sample and hold type effects by clocking input and output DACs.

The lowest order bits also provide a source of digital noise.

Post Reply

Return to “Eurorack Modules”