Author |
Message |
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Wed Aug 13, 2008 9:29 pm Post subject:
|
|
|
You could replace the two CD4516 with a PIC 16f84 and a 4067 16-1 multiplexor. Then you could do quantize to scale. I'm going to give it a try this weekend.
Here's my idea...
Pseudo code:
//setup the i/o directions
set porta as outputs expect ra1
set portb as outputs
//two variables..
counter_a //counter for scanning switches 0..11
counter_b = 1 //coutner for feeding DAC 0..256
start:
//increment counters and wrap
counter_b = counter_b - 1
if (counter_b = 0 ) then counter_a = 10
counter_a = counter_a + 1
if (counter_a = 11) then counter_a = 0
//read the relevant switch from the multiplexor
output counter_a to port_a
//if switch is set then output new value to DAC
if (RA_1 is HIGH) then output counter_b to port_B
//need some delay here depending on clock speed of the PIC
delay
goto start Last edited by widdly on Thu Aug 05, 2010 10:09 pm; edited 1 time in total |
|
Back to top
|
|
|
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Thu Aug 14, 2008 3:34 am Post subject:
|
|
|
I found an old post on sdiy from Harry Bissell about the arp quantizer circuit...
"The Chris List design works OK, but it has some problems
imho
The best way to do it is to add a D flip-flop to the comparator
output that clocks on the opposite edge of the counter advance signal.
This way, the analog voltage is stable for 1/2 clock cycle before the
S/H is clocked. This works MUCH better. I have built this and used
it for a while."
Sounds like an idea worth checking out. Last edited by widdly on Thu Aug 05, 2010 10:09 pm; edited 1 time in total |
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005 Posts: 761 Location: Utrecht, NL
|
Posted: Thu Aug 14, 2008 3:42 am Post subject:
|
|
|
Quote: | Sounds like an idea worth checking out. |
That is, of course, more or less what Scott's design already does, albeit not with a flipflop, but his delayed clocking of the S&H is similar. _________________ http://www.casia.org/modular/ |
|
Back to top
|
|
|
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Thu Aug 14, 2008 3:58 am Post subject:
|
|
|
Not quite though. In Scotts circuit, the comparator is testing the value when the DAC hasn't settled...then sampling later.
With Harry's flip flop, the comparison and sampling occur after the DAC has settled.
Not sure if that makes a difference though. Last edited by widdly on Thu Aug 05, 2010 10:09 pm; edited 1 time in total |
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005 Posts: 761 Location: Utrecht, NL
|
Posted: Thu Aug 14, 2008 4:04 am Post subject:
|
|
|
Probably none at all. The critical point is clocking the S&H and both circuits delay that. It doesn't matter what they do before the S&H is clocked... _________________ http://www.casia.org/modular/ |
|
Back to top
|
|
|
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Thu Aug 14, 2008 4:20 am Post subject:
|
|
|
Before the DAC settles, there will be some movement around the expected output value. If the CV signal is slightly below the DAC output, the comparator may go high incorrectly. Over a number of cycles you could get the s&h toggling between two adjacent steps. Last edited by widdly on Thu Aug 05, 2010 10:08 pm; edited 1 time in total |
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005 Posts: 761 Location: Utrecht, NL
|
Posted: Thu Aug 14, 2008 4:22 am Post subject:
|
|
|
Yes, but not if you clock the S&H after the DAC settles, which both designs do. _________________ http://www.casia.org/modular/ |
|
Back to top
|
|
|
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Thu Aug 14, 2008 9:05 am Post subject:
|
|
|
Yes I know they both S&H with a delay. What I am saying is the Scott's comparator compares the un-settled DAC value with the cv input value . While a DAC is settling the output value moves around. So there is potential for the comparator to trigger the S&H when it shouldn't. There will be less errors at the comparator stage if the DAC output has settled. It may improve the toggling Scott was seeing with slow moving CV inputs. Last edited by widdly on Thu Aug 05, 2010 10:08 pm; edited 1 time in total |
|
Back to top
|
|
|
Scott Stites
Janitor
Joined: Dec 23, 2005 Posts: 4127 Location: Mount Hope, KS USA
Audio files: 96
|
Posted: Sun Aug 17, 2008 11:11 am Post subject:
|
|
|
I've been thinking of putting a DAC into my ribbon controller - not necessarily for quantizing (though I think it will offer that, at least for half steps), but for very precise infinite hold - quantizer takes over from S&H with the closest quantized level.
A DACO8 actually works very well for half step (0.083) intervals - my TH keyboard uses one, and I'm perfectly happy with that. $3 or $4 bux apiece beats the hell out of expensive unobtainium, AFAIC. That's why I let the quantizer project languish, for the most part. It doesn't have to be that high precision AD stuff, to be honest. It was just handy, more than anything else. And free.
Anyway, 8 bits is fine for keyboard voltages (IMO), but won't cut the mustard in inserting values for high resolution "hold" voltages that I'm looking for in my ribbon controller.
So, why not just use two DAC08s and get 16 bit resolution, which is more than I'll need anyway?
DAC08 (or a variant thereof) is easy to get hold of, and does not require the sale of one's first born to obtain (if it can even be found) some fly-by-night endangered DAC IC anyway?
Just a thought.
Cheerios,
Scott _________________ My Site |
|
Back to top
|
|
|
Tim Servo
Joined: Jul 16, 2006 Posts: 924 Location: Silicon Valley
Audio files: 11
|
Posted: Wed Aug 20, 2008 5:12 pm Post subject:
CV Quantizer PCB Layouts |
|
|
I think you won't get 16-bit accuracy with two 8-bit DACS. I've heard this tried before, and you start to run into LSB errors with even a really good 8-bit part. Those first few LSBs on a 16-bit part are working with really small voltages, and you start to see why 16-bit DACs are so expensive.
You will however, be able to obtain something like 12 bit resolution, or maybe slightly better. The Oberheim Xpander uses a similar trick by using a single 8-bit DAC and making two samples. The first sample uses a large reference voltage (something like 10V), and stores the output with an analog S/H. The second sample is made with a much smaller ref voltage, and the two voltages are summed. I'm not exactly sure, now that I try to explain it, but it might be worth a little research. If you latch the digital input value, then you can keep refreshing the analog s/h and avoid any droop.
Might be worth a shot anyway. Seemed to work nicely for the Xpander and M12.
Tim (working to avoid any droop) Servo |
|
Back to top
|
|
|
Scott Stites
Janitor
Joined: Dec 23, 2005 Posts: 4127 Location: Mount Hope, KS USA
Audio files: 96
|
Posted: Thu Aug 21, 2008 8:27 am Post subject:
|
|
|
Actually I wasn't shooting for anything better than 11 bits. I didn't think the extra resolution would be worth the effort (or be even possible to pick out using a comparator on a staircase). 11 bits would give you 5 mV resolution at 10V, which I think is plenty for quantizing or grabbing those values for the ribbon controller. I sincerely doubt the ear will hear a rise or fall of 2.5 mV when the DAC takes over from the S&H.
Let me know if you figure out any more about that technique! _________________ My Site |
|
Back to top
|
|
|
fluxmonkey
Joined: Jun 24, 2005 Posts: 708 Location: cleve
|
Posted: Mon Aug 25, 2008 1:04 pm Post subject:
|
|
|
hmmm... stumbling around music from outerspace site, i stumbled on this
i figured either ken or ray would get around to this sooner rather than later....
b _________________ www.fluxmonkey.com |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Mon Aug 25, 2008 1:22 pm Post subject:
|
|
|
This is great news! Last edited by ericcoleridge on Mon Aug 25, 2008 2:36 pm; edited 1 time in total |
|
Back to top
|
|
|
Scott Stites
Janitor
Joined: Dec 23, 2005 Posts: 4127 Location: Mount Hope, KS USA
Audio files: 96
|
Posted: Mon Aug 25, 2008 2:16 pm Post subject:
|
|
|
Quote: | Before the DAC settles, there will be some movement around the expected output value. If the CV signal is slightly below the DAC output, the comparator may go high incorrectly. Over a number of cycles you could get the s&h toggling between two adjacent steps. |
Somehow I missed this part of the thread.
At the time I was jacking with this thing, I'd read Harry's comments on the D Flipflop. In fact, I'd had a few exchanges with him about it. The real point was lost on me though - but your explanation makes perfect sense to me. Sure, I was giving it time to settle...but there's no point in letting it settle if it's the wrong step to begin with!
Thanks for that!!!
Cheerios,
Scott _________________ My Site |
|
Back to top
|
|
|
funkyfarm
Joined: Jan 21, 2007 Posts: 583 Location: France
Audio files: 3
|
Posted: Sun Jan 11, 2009 2:22 am Post subject:
|
|
|
I'm still dreamin about that. |
|
Back to top
|
|
|
numbertalk
Joined: May 05, 2008 Posts: 992 Location: Austin, TX
Audio files: 5
|
Posted: Tue Jan 13, 2009 1:08 pm Post subject:
|
|
|
bbob wrote: | hmmm... stumbling around music from outerspace site, i stumbled on this
i figured either ken or ray would get around to this sooner rather than later....
b |
Anyone build the MFOS circuit yet? Opinions? |
|
Back to top
|
|
|
guitarfool
Joined: Feb 26, 2007 Posts: 160 Location: Maryland
Audio files: 8
|
|
Back to top
|
|
|
numbertalk
Joined: May 05, 2008 Posts: 992 Location: Austin, TX
Audio files: 5
|
Posted: Wed Jan 14, 2009 8:02 am Post subject:
|
|
|
Great - thanks! |
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Tue Feb 24, 2009 9:19 am Post subject:
|
|
|
today i wanted to build scotts quantizer on breadboard. it is the 2nd try. 1st try has been in 2006
i stuck at the same point as back in 2006: the schmitt triggers won't oscillate, wether the CD4516s are connected or not. pins 1 and 4 are always high. i use a 40106, all unused inputs are pulled to GND, and i used a 100n bypass cap.
did someone else get the clock to work following this schematic? i don't believe it would work on perfboard insted of breadboard - i don't know, but i just don't believe
_________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source |
|
Back to top
|
|
|
Scott Stites
Janitor
Joined: Dec 23, 2005 Posts: 4127 Location: Mount Hope, KS USA
Audio files: 96
|
Posted: Tue Feb 24, 2009 11:37 am Post subject:
|
|
|
I'll check and see if there were any changes. Take heed of Widdly's observations - I believe he's right, and it could be made more precise on the decision points with the comparator. _________________ My Site |
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Tue Feb 24, 2009 4:23 pm Post subject:
|
|
|
thank you scott.
BTW i used another oscillator -mickey mouse , and i got the dac doing it's thing now. a nice staircase going down from +5V to -5V. now it's time to go to bed for me here. will do the comparator/s&h on weekend or next week...
and then? ... we might be going for a nice professional manufactured PCB? scott? are you fine with that? it will cost something, i think - just because only a few people will be willing or be able to get hands on this DAC. i don't think i will struggle with replacing it...
[spelling edited: staircase instead of starecase ] _________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source Last edited by fonik on Wed Feb 25, 2009 11:54 pm; edited 1 time in total |
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Tue Feb 24, 2009 4:35 pm Post subject:
|
|
|
fonik wrote: | i don't think i will struggle with replacing it... |
or should i? DAC08? what would be the difference? less resolution? i will have to take a look at the datasheets to understand what realy is going on! could someone show up with a sketch? _________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source |
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Wed Feb 25, 2009 2:07 am Post subject:
|
|
|
i will try an DAC0800, an cheap high speed DAC. that would require an additional reference voltage, but hey, that would be no problem at all.
as far as i understand it (and i might not understand it all), the AD7245 just provides some extra functions which are not used in the quantizer design: control logic inputs (for the integrated latches) and output offset. _________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source |
|
Back to top
|
|
|
funkyfarm
Joined: Jan 21, 2007 Posts: 583 Location: France
Audio files: 3
|
Posted: Wed Feb 25, 2009 1:57 pm Post subject:
|
|
|
fonik wrote: | AD7245 just provides some extra functions which are not used in the quantizer design: control logic inputs (for the integrated latches) and output offset. |
under control ? pot or cv ? transposing something ?
With this IC, the whole project (panel, parts...) would reach 120/130$ ?
for four channels.
Thank you for your ideas and devotion to this project alive. |
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Wed Feb 25, 2009 3:02 pm Post subject:
|
|
|
funkyfarm wrote: | With this IC, the whole project (panel, parts...) would reach 120/130$ ?
for four channels. |
120$? i got my 7245 for free as a sample from analog devices
i envision a two channel version using a cheaper an available DAC. and maybe an precise internal mixer, so one CV could control the base frequency...
we will see. _________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source |
|
Back to top
|
|
|
|