Author |
Message |
rosch
Joined: Oct 03, 2009 Posts: 164 Location: germany
|
Posted: Wed Mar 17, 2010 5:41 am Post subject:
LF398 rate reducer / bit crusher problem |
|
|
hi! in order to not keep on hijacking the bugbrand thread i've started this new topic, referring to my post here
i've built the rate reducer posted by modularkomplex:
http://electro-music.com/forum/topic-11369-54.html
but have following problem: the cv from 40106 is fed through to the output.
well i'm not at all sure anymore that the effect works fine besides the feedthrough. i thought maybe the signal of my bass wasn't strong enough but the behaviour stays the same when plugging in a casio vz-1 which should bring enough without further amplification.
i also checked for solder bridges, with eyes and a knife- apparently nothing.
while having plugged in the vz-1, the cable of the pot disconnected again, and so the sound of the synthesizer was totally normal, like bypassed. putting the cable back to the pot brought back the (detuned-style) pitch modulation with the cv square instantly.
any ideas more than welcome- and sorry for the thread salad...
thanks! |
|
Back to top
|
|
|
rosch
Joined: Oct 03, 2009 Posts: 164 Location: germany
|
Posted: Sat Mar 20, 2010 7:09 am Post subject:
|
|
|
i breadboarded the circuit to be sure to have no solder bridges etc.
the result was just the same like before.
maybe it is a problem with some parts? i think if i had the 40106 put in the wrong way it wouldn't function at all, but when the pot is turned i can hear the pitch of the square change, so this should not be the problem.
but maybe there are differences between LF398's from different manufacturers?? i have a LF398N, purchased from Conrad.
i think i'll try Reichelt next, maybe that helps.
i have no idea what i could have overlooked here |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Wed Aug 08, 2012 7:59 am Post subject:
|
|
|
i read that bugbrand wrote :
Quote: | You should be able to decrease any background clocking noise by narrowing the trigger pulses (though I don't know particularly with the 398 'cos it maybe does things internally unlike the AD781) |
do i need to narrow trigger pulses with the LF398 or not?
if yes,how do i do that?
what should i do?
cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
J3RK
Joined: Jun 05, 2006 Posts: 123 Location: Seattle
|
Posted: Wed Aug 08, 2012 10:26 am Post subject:
|
|
|
isak wrote: | i read that bugbrand wrote :
Quote: | You should be able to decrease any background clocking noise by narrowing the trigger pulses (though I don't know particularly with the 398 'cos it maybe does things internally unlike the AD781) |
do i need to narrow trigger pulses with the LF398 or not?
if yes,how do i do that?
what should i do?
cheers,
Isak E. |
Take a look at Rene Schmitz' YASH (Yet Another Sample and Hold) circuit. He uses gates from a 4093, a cap, and resistor to get the pulses the LF398 needs. http://www.uni-bonn.de/~uzs159/yash.png
You could also do a simple passive circuit with a cap, pair of resistors to ground, and a diode. Like this maybe: http://forums.adafruit.com/viewtopic.php?f=13&t=5060 (about half-way down)
These are just general ideas though, as I haven't built the circuit you're working on. |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Wed Aug 08, 2012 12:17 pm Post subject:
|
|
|
Thank you for replying.
The circuit workes very well, I'm just getting a background sound that I'm trying to get rid off.
You think from the links you gave me it will work?
Cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Wed Aug 08, 2012 5:47 pm Post subject:
|
|
|
Quote: | The 40106 is missing a 22pF capacitor from pin 2 to 3 and a 22k resistor from pin 3 to ground |
these 2 parts with the following 40106 section form a one shot circuit to provide the narrow pulse that is required to tell the lf398 when to take a sample and are definitely required.
as far as the bleed through noise goes does it follow the clock rate ?
this circuit may be better on an etched board rather than stripboard
as then you could provide for some ground plane around the lf398 pins
to prevent inducing noise to the input / output from the clock , i think i remember that being mentioned somewhere . _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Wed Aug 08, 2012 11:24 pm Post subject:
|
|
|
thank you for replying.
Quote: | as far as the bleed through noise goes does it follow the clock rate ? |
when nothing is connected to the input and i move the pot to the right there is a sound, like vco sound but at very low volume.
when i play with the pot left and right the vco sound changes the pitch, up and down.
When i turn it all the down to the left the sound is gone.
i put the 22pF and the 22k resistor in the circuit and the noise was almost gone.
but the sound was crappy.
when with 22pF when i move the pot all he way to the left the effect is still working.
when its with the jumper like the stripboard layout shows its sounds amazing and when i turn the pot all the way to the left the effect went off, like it should.
i try few values of caps, non of them worked with the background noise/vco.
when try less then 22pF the sound started to whistling. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Thu Aug 09, 2012 2:44 am Post subject:
|
|
|
Quote: | when i play with the pot left and right the vco sound changes the pitch, up and down. |
OK so the bleed through is from the clock that is being induced into the output.
Quote: | i put the 22pF and the 22k resistor in the circuit and the noise was almost gone.
but the sound was crappy. |
Interesting way of putting it. I thought that was the idea of a bit crusher
it is supposed to remove a variable amount of the information in the signal going into it.
but then again there is good crap and there is bad crap
without the cap and resistor the pulse that is sent to the clock input of the lf398 is 50 % duty cycle rate this would normally be a bad thing for a sample and hold circuit as the sample would not be very accurate the reason that the cap and resistor are there is usually to make the sample taken as short as possible so that the specific voltage at the audio input at the exact time of the sample can be taken and reproduced at the audio output and then held at that level until the next sample is taken this is the normal wanted behaviour for a sample and hold circuit , however having said all that in the case of the lf398 the clock pulse leading edge only is used .So the clock pulse width has no effect , now the clock rate determines how often a sample is taken of the input over a given time period, the faster the clock rate the closer the output signal will be to the signal at the input . less information is lost. slow the clock rate and you lose more information as less samples are taken over the same time frame
So that brings me back to the noise level decreasing when the cap and resistor are in place. this is simply because the average level of the clock pulse is less -.... 50 % of the time at 2/3 the supply voltage compared to a very short pulse at 2/3 the supply voltage. There is simply less signal to be induced into the audio path
So how to get rid of the noise , not easy with strip board.
on a PCB the clock pin would ideally have a grounded guard ring around it and the track leading to it would be surounded by a ground plane and kept well away from other signal traces , cant really acheive that on strip board
I would be inclined to leave in the cap and resistor to keep the noise level down but perhaps fiddle with the clock rate by playing with the value of C1, R2 and P1 this will affect the clock rate, A faster clock for more samples per second and a less affected signal, slower for less samples per second for a more affected output signal not sure from your post which way you are wanting to go . good crap - bad crap is very subjective. _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Thu Aug 09, 2012 4:14 pm Post subject:
|
|
|
thank you for the detailed answer
i play around with c1 and r2 like you suggested, didnt got good results like with c1 - 4.7nF and r2 - 10k.
with the 22pF and the resistor its very bad sounded.
with jumper wire instead the 22pF its sound much much much better!
but yet again with the noise issue.
i found out something, when i connect leg 3 of the freq pot to ground (leg 3 is not connected in this scheme) the noise is gone and appear to be in limited section of the pot.
i mean the noise is from 7 o'clock to 12:00 o'clock pot position, and only at this limited area the bit crusher works, other sections of the pot is not working.
maybe it will give yo an idea
thank you for your help!
cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Fri Aug 10, 2012 1:03 am Post subject:
|
|
|
If you check out the schematic of bugbrands you linked earlier
pins 2 and 3 of the pot should be connected together
pin 3 definitely should not go to ground as that will stop the 40106 section
which makes up the clock signal from oscillating once the pot is turned past a certain point after which the circuit will do nothing
try connecting pin2 and 3 of the pot together and then try adding a 1 meg resistor to ground from pin 2 of the 40106
that i think may get rid of the noise and still allow the use of the full rotation of the pot. _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Fri Aug 10, 2012 3:46 am Post subject:
|
|
|
You mean instead of r3 (22k) from bugbrand schematic?
I removed the 22pF and put a jumper instead, I did it cause I love the sound very much without the 22pF.
but I left the 22k to ground from pin 2 of 40106.
Ill just replace it with 1M as you suggests, I hope ull right and it will work.
I'll let you know later on tonight.
Thank you for not letting go
Cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Sat Aug 11, 2012 1:27 am Post subject:
|
|
|
Hi diablojoy.
ok that didnt helped.
this is the module situation...
i removed the 22pF, put a jumper instead, like in the stripboard layout shows.
i put 22k from pin 3 of 40106 to ground.
i put 1M from pin 2 of 40106 to ground.
leg 2 and 3 of the pot is shorted.
still the same results, background vco noise.
a pic with the changes...
any idea?
cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Sat Aug 11, 2012 10:23 pm Post subject:
|
|
|
I had assumed the 22pf to still be used however honestly i am running out of idea's for the moment
I can try to replicate the circuit myself just need to get some lf398
i have some strip board here somewhere
perhaps a pull down resistor to ground on the 40106 pin 6 say 10k -100k
but i am really grasping at straws with that one, oh and you may as well lose that 1 meg i said to try as its acheiving nothing seemingly. _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Tue Aug 14, 2012 11:24 pm Post subject:
|
|
|
Ok had some more time to look over all the info previously put out there
bugbrands original schematic used an AD781 which relied on a very short pulse to determine the sample time so the values that were borrowed from that circuit [C2-22pf and R3- 22k Roughly 480 nano seconds ] may be totally wrong when teamed with an LF398 I could be way off with this but I think the pulse is simply too narrow , so its possibly trying to go into hold mode early before a proper sample has been taken [LF398 minimum acquistion time - 6 micro seconds ] and thus giving you crap at the output
perhaps try replacing C2 and R3 with 100pf and 100k which should give about a 10 microsecond pulse width .
I think that should work a lot better .
But if not you could try increasing the value of C2 still further perhaps as far as 1nf but if there is no improvement at that point then i would think i have lead you down the wrong track yet again. _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Wed Aug 15, 2012 3:02 am Post subject:
|
|
|
OK built the circuit as per the strip board but with 100pf and 100k
to provide 10 microsecond pulse seems to put out what i expected.
quite cool really, in a low fi kind of way.
there is a noticable freq change on the overall output that follows the
pot however i think it should be there as it bears a relationship to the quantizing of the signal
noise is not all that noticeable
scope shows what i would expect to see, stepped version of the input signal
becoming more stepped as the pot is turned _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Thu Aug 16, 2012 1:52 am Post subject:
|
|
|
tried this circuit with strap in place of C2
not working as a sample and hold really at all just a very distorted output waveshape one flat spot that maybe equates to the sample point , most of the wave gets through sort of but not nice.
I then tried original values for C2 and R3 of 22pf and 22K
low output, crap waveshape and no sampling evident. sounds bad
with 100pf and 100K the circuit operates correctly in theory
however the sampling frequency is very evident as it is in the audio range.
so not to everyones taste
If the output is fed through a filter things do improve somewhat and with carefull adjustment the sampling freq can be minimised _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Thu Aug 16, 2012 2:34 am Post subject:
|
|
|
thank you very very much for all of your effort!
Quote: | tried this circuit with strap in place of C2
not working as a sample and hold really at all just a very distorted output waveshape one flat spot that maybe equates to the sample point , most of the wave gets through sort of but not nice. |
strange, it worked very well for me and sound very good.
i did what you did, the 100pF and the 100k is working great, less bleed is going through, yet again still there.
Quote: | If the output is fed through a filter things do improve somewhat and with carefull adjustment the sampling freq can be minimised |
how can i do that?
what should i do?
i tried to connect the MFOS SV filter through the bit crusher, the sound was bad.
the amount of the hiss was stronger (i know the SV filter haves a lot of hiss in general).
when the cutoff pot is all the way up the crusher sounds crappy, only if the cutoff is close or open half way its sounds good, any ideas why?
i also notice that if i play on midi keyboard and press hard on the keyboard (strong velocity) the crusher sound crappy, if i press a soft press it sounds amazing, i guess the input of the crusher should be low for good sound results?
thank you again for your effort.
cheers,
Isak E. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Thu Aug 16, 2012 3:06 am Post subject:
|
|
|
Quote: | i tried to connect the MFOS SV filter through the bit crusher, the sound was bad.
the amount of the hiss was stronger (i know the SV filter haves a lot of hiss in general). |
haha I used the same filter
went sinewave out from an mfos VCO in to [well i suppose we will call it the bit crusher]
out of the bit crusher and into the mfos SV filter take low pass out then on to a VCA set to be open so no EG in the mix yet
Low pass out seemed the best choice to me to take out a lot of the top end and then for interest I put a cv from a sequencer into the filters cutoff input
Quote: | i also notice that if i play on midi keyboard and press hard on the keyboard (strong velocity) the crusher sound crappy, if i press a soft press it sounds amazing, i guess the input of the crusher should be low for good sound results? |
hmm on the scope i see unity gain . The input and output are of equal levels I am sending in a 10v peak to peak sine wave so it should be ok at least to that level. though i am powering of +/-15volt supply so maybe i have a bit more headroom. _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
Posted: Thu Aug 16, 2012 3:18 am Post subject:
|
|
|
we use the same filter for checking hahahaha
Do you suggest to put the crusher before the filter and not after?
I mean in the sound chain. _________________ http://www.myspace.com/mgmtrance |
|
Back to top
|
|
|
PHOBoS
Joined: Jan 14, 2010 Posts: 5591 Location: Moon Base
Audio files: 705
|
Posted: Thu Aug 16, 2012 8:22 am Post subject:
woggle S&H |
|
|
ok this might be interesting for me aswell. First of all I always wanted to build that bugbrand bitcrusher,
but I can't get the AN781 (or it's just to darn expensive). so S&H seems like an alternative that's worth a try.
besides that I'm currently working on wogglebug design (modding it to my own liking with -5/+5V levels).
which also uses a LF398 S&H. The Richard Brewster design I'm mostly working with actually has 2 S&H's.
I was wondering is the bleed through you get a soft ticking or just the fact that you're sampling at audiorates.
If it's ticking it could be that the design richard brewster uses reduces it a tad more. I have yet to breadboard
that section myself which will also give me a change to listen how it sounds as a rate reducer/bit crusher
anyway you can find the schematic for that section here (PDF)
it has some more stuff involved with the LF398 in/out. but you could ignore that (or test how it sounds).
O and maaaaaybe it is possible to clean the sound a bit more with some sort of envelope stage but that's
probably just adding to much complexity. _________________ "My perf, it's full of holes!"
http://phobos.000space.com/
SoundCloud BandCamp MixCloud Stickney Synthyards Captain Collider Twitch YouTube |
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008 Posts: 809 Location: melbourne australia
Audio files: 11
|
Posted: Thu Aug 16, 2012 3:14 pm Post subject:
|
|
|
Quote: | Do you suggest to put the crusher before the filter and not after?
I mean in the sound chain. |
Yes exactly that
Quote: | I was wondering is the bleed through you get a soft ticking or just the fact that you're sampling at audiorates. |
pretty sure it is the sample rate being in an audible range
at the extreme high end of the pot the frequency approaches the upper limit of my old ears not that that is a particularly high frequency these days
way too much rocknroll . just as a guess i would think the range to be from around a few 100hz to about 12khz
Quote: | O and maaaaaybe it is possible to clean the sound a bit more with some sort of envelope stage but that's
probably just adding to much complexity. |
Well it's definitely not perfect but i was thinking if it was desired to remove maybe a tracking filter something along the lines of what is used in the blacet time machine but that maybe problamatic and with just using a normal LP filter after i kinda like some of the overtones you get at certain settings
I have started to draw up a dual board by the way.
but i want to experiment a little further with this, so nothing is finalised.
edit: ah AD781 = monolithic sample and hold chip not AN781 i think
typo perhaps ? _________________ In an infinite universe one might very well
ask where the hell am I
oh yeah thats right the land of OZ
as good an answer as any |
|
Back to top
|
|
|
PHOBoS
Joined: Jan 14, 2010 Posts: 5591 Location: Moon Base
Audio files: 705
|
|
Back to top
|
|
|
isak
Joined: Dec 13, 2009 Posts: 847 Location: Israel
Audio files: 18
|
|
Back to top
|
|
|
|