Wanna know what does interest me? MIDIFied analog FX chips. MMMmmm, MBFX 
Anyone in?
iām working on a mbdelay pcbā¦
that is; a delay with vc delaytime/feedback which could be controlled with cv from midibox
keep you updated.. 
You using some delay IC or custom circuitry?
Iāve been thinking about a box with a bunch of SSM/Curtis/BBD ICāsā¦ Just found a rediculously cheap lot of ssm2044 tonight actually - lots of them, but cheap
And Iām working on a distortionunit. Right now itās not even prototype phase, but the schematic I cooked up looks promising. Will need some adaptions to make it MIDI controllable, and to adjust the imput from guitar level (1MOhm) to SID/Synth level (1KOhm). Might even put a switch to select between both input levels, to make it more versatile.
As a start Iām going to make it a standalone FX unit, with normal analog pots and switches, and a few different clipping options to select. Then Iāll test them with the SID, and check what options sound best. Lastly, Iāll <try> to swap the analog pots for digital pots, and make it controllable from a PIC.
But thatās all Iāll say right now, it might make this completely dull thread intersting all of a sudden, that would be a shame now wouldnāt it? ;D
I would like a dedicated topic on the fx / midi thing, and am also very interested in Illogikās cookings!
Cheers, Alex.
Definitely sounds like we need an MBFX thread. Would be good to get our heads together on this, and a shame for each of us to do this independently when we could all team up Maybe TK will tell us about his mysterious VCF box?
I had some thoughts on the last post but Iāll save that for the new thread if you guys are keenā¦
I had some thoughts on the last post but Iāll save that for the new thread if you guys are keenā¦
Here ya go, new thread!
Very interesting thread hijack BTWā¦ ;D
Best
Smash
Iāll take that as Smash-speak for āKeep talkingā
Iāll take that as Smash-speak for āKeep talkingā
hehe yessir!
OK, Iāll share some more of my thoughts now this is a new thread
Iām not sure how many digital pots can be driven from 1 core. If thinking from a quad-SID standpoint, maybe the slave cores could each drive the digipots attached to their respective FX units. If thatās not possible, a dedicated FX core might be necessary, this core could then run an adapted MBCV firmware so you also have CV control over other stuff.
Iāve read digipots introduce zipper noise. For my current project, the earlier mentioned distortion, this is not really a problem, as you usually donāt do distortion-sweeps. Only the initial setting when changing a patch would create noise, and if you do asjust when playing I doubt the noise would be identifyable from the distorted signal anyway.
Anyone here thatās toyed around with digipots before? Iām ordering some samples from microchip next week, something to toy with until Smashās reopens.
Cheers, Alex.
hi jacks! ???
as you usually donāt do distortion-sweeps
iām not usual; i sweep em all
hey alex; iām interested in your approach but wanted to keep control as much in the analog realm as possible so im more into vactrols, jfets and OTAās
:-[ this is for the people who have seen me saying this for years but not coming up with anythingā¦ itās not so easy to understand for me as all my electronic experience is coming from practice/making (other peoples) diy projects, not doing complete studies
i have to start reading books about this stuffā¦ a lot..
; so keep it up so we can compare; vc-resistors VS Digipots!! one dayā¦
cheers
You using some delay IC or custom circuitry?
a digital (iām sorry, but i fink itāll sound good) delay ic and some opamps
thing is i have to ask permission first; not my design.. ;D 
had a bad day so put in lots of smilies and some jokes
Maybe āthe Adventures in digital to analog interfacingā should be WIKIfied?
Illogik is the man here. HE has learned and is still learning a lot. I respect him a lot for his trouble preventive way of dealing with things, like using time to study - Yup, many external resources, but I guess heās not that helpless to use forum search too?
Once I saw a post in this very forum by a man who had tried using digital pots in a circuit that consists many analog effects in one box. It wasnāt really encouraging about using digital pots in a high gain circuit like guitar distortion. Maybe itās just because guitar FX input impedance and the gain factor - or maybe not?
His name was Captain Hastings. http://www.midibox.org/forum/index.php?topic=2556.msg16762#msg16762
So, keep on experimenting and Have Fun! like Thorsten says - even if it means hearing databus switching in the audio output!
M
Maybe āthe Adventures in digital to analog interfacingā should be WIKIfied?
Illogik is the man here. HE has learned and is still learning a lot. I respect him a lot for his trouble preventive way of dealing with things, like using time to study - Yup, many external resources, but I guess heās not that helpless to use forum search too?
Once I saw a post in this very forum by a man who had tried using digital pots in a circuit that consists many analog effects in one box. It wasnāt really encouraging about using digital pots in a high gain circuit like guitar distortion. Maybe itās just because guitar FX input impedance and the gain factor - or maybe not?
His name was Captain Hastings. http://www.midibox.org/forum/index.php?topic=2556.msg16762#msg16762
So, keep on experimenting and Have Fun! like Thorsten says - even if it means hearing databus switching in the audio output!
M
Hi Moebius, I actually already read that topic, but I havent seen Hastings around lately, hence my question. The link that Smash provided to give a LIN pot a LOG feel is also in my āgathered electrical goodies docsā folder
Iāll give it another study though, might find something I missed initially. thanks.
/edit
What do you know!
Analog Devices does samples too! Sampled myself some AD5206BN10, AD5204BN50 and AD5204BN100 a few minutes ago.
Because Hastings has already done some prototyping, to me these seemed favorable above the digipots microchip has. Also, the AD digipots are 6way, 4way and 4way, while MC only has 1 or 2 way specimen.
I am very curious if my PSU design can make this combination of Digital and Analog circuitry quiet enough. Perhaps Iāll use a separate Analog ground. Also, I believe Hastings used guitar-rated impedance for his circuit, and I recently learned (thanks Mendelt ) that high resistors bring high noise, so I suppose thereās still room for improvement.
My test distortion circuit will have a switch to select Guitar rated impedance (typically 1MOhm) or Synth rated impedance (for example SID has 1KOhm). The test circuit will probably end up as a stand-alone effectunit in a 19" rack. My current test design can switch between 4 different clipping options, one of which is no extra clipping. I will evaluate the sound, and then develop a much slimmer distortion board with probably only one clipping option. People can then stuff the diodes they find best sounding.
then, if I find some more time ( ;D ) Iāll want to digitally control the distortion settings via the digipots. but that phase will probably be a few months from now.
Hey illogik, hope youāre having a better day
I was looking at some of those Bucket Brigade Delay ICās ā¦ Iām not so choosy, I put more weight in the value of composition than whether Iām using this thing or that to make it I just think it would be good to have a midi controlled delay, then I can play with different types.
As far as MBFX goes, I had thought that it would be good if there were some kind of framework that could be used for lots of different effects, so people could, for example, go to diystompboxes.com and make an effect, then midi it up using MIOS, or MIDI up their favourite bought stompbox or amp or somethingā¦
The trick with this is making a single solution that would work well in many different applications. I would like to hope that we could all work out a way to do this which would be cheap, electronically quiet, and easy to do. If we can all work together, it will be a good way to go because we have strength in numbers. But the question is, can we find a way that is flexible enough to be suitable for us all?
It seems at present we have three options. Iāve summarised as much information as I was able to find by literally studying all day and part of last night, I hope it will help us all to compare the options at hand by summarising the proās and conās of eachā¦
Mechanical
Motorpots attached to analog pots - This way uses motorised pots or rotary motorfaders (depending on what you want to call them) to move analog pots like an abolute servo. It is completely noiseless because the MB and FX circuits are electrically isolated and only coupled mechanically, but could be expensive, as motorpots arenāt cheap, and it could also be harder to do mechanically. This way would mean that the price would vary depending on how many knobs you need but we would need the additional MF module for this. It would also mean that if you were midifying existing gear, you could keep the existing controls in place, so the MIDIBox wouldnāt have to be active to use the gear - this is of course assuming that there is room inside the case of your gear to add the motorpots. This method would also introduce support issues with MBFX users coming here for mechanical support on how to link up their pots etcā¦ Also Implementing switches/buttons with this method would probably involve either analog or digital solutions which might defeat the bonus of total electrical isolation, unless using a servo to flick the switchā¦
Cheap -
Quiet + (Doesnāt get any quieter than that, itās a robot hand turning the knobs for you, just like your hand would do)
Easy to Design -
Easy to Build -
Easy to Support -
Flexible Implementation -
Digital
Digipots/DCRās, CMOS - This method means desoldering and replacing existing pots with digitally controlled variants, or switching between digitally controlled and physically controlled resistors, and switches would be CMOS controlled switches. It seems a fairly easy solution at first and digipots are not quite so expensive as motorfaders, but as we know, the digital signals can make things a bit messy. There are concerns about noise from the digital control buss leaking into the audio signal, but that might be worked around by heavy shielding and perhaps a separate ground plane. The separate ground could be tricky to implement though, and that implementation would vary greatly from one device to another, increasing support requirements. One upside with this is that little additional hardware is required.
Cheap +
Quiet -
Easy to Design + (already done by Capt. Hastings)
Easy to Build +
Easy to Support -
Flexible Implementation +
or if we try to find a quiet way
Cheap ? (Need to see what the design is like)
Quiet -
Easy to Design -
Easy to Build ? (Need to see what the design is like)
Easy to Support -
Flexible Implementation +
Analog
Vactrols/FETs/OTAs, Relays - This way converts a control voltage to a resistance or uises it to switch a relay. It is analog so it eliminates the noise problems, but it introduces other problems - cost, because you need the AOUT with the expensive MAX525 to control a number of pots. This method is basically about converting existing designs to CV control, and using the existing MBCV hardware, so means no design issues to deal with from the MIOS end, itās a matter of using any DIY effect design with CV control, or adding CV control to existing gear. When adding to existing gear, you could build the CV controls, and use a switch to select internal control (wired to pots) or CV control (wired to the CV Ins)
I notice the MAX5250 is available and about half the price, seems to be the same thing, but it is 10 bit resolution, not 12-bit. This would reduce the price fairly well, but Iām not fantastic with hardware and havenāt read the entire datasheet yet (just the specs and features) so it may not work. Iāll look into that further shortly but if anyone else knows or is interested please let me know what you think But I donāt see the harm in 10-bit controlā¦ 1024 steps is a fair amount, I canāt imagine the zipper noise being a problemā¦ Meanwhile thereās the AOUT_LC option which uses reduces the cost, but only allows for 2 CV channels.
Cheap - (+? what about the 5250?)
Quiet +
Easy to Design +
Easy to Build +
Easy to Support +
Flexible Implementation ? (Can we make a simple and flexible CV-to-resistance circuit?)
So going by the information I have been able to find, hereās the nutshell:
Mechanical is the best possible way audio-wise, but not much good elsewhere.
Digital is great but noisey, so it will not fit the needs of some people (like illogik and I who sweep everything). We could work on methods to reduce noise, but that will always be something of a struggle and effective methods of reducing the noise will vary from one circuit to the next, which increases support efforts.
Analog is good but has two drawbacks. The first is cost, but I am confident that the MAX5250 will get around that. The second is flexibility of implementation. We need a way to get CV into the devices and controlling resistance. It seems that if we could make a simple, cheap and flexible CV-to-resistance (VCR) and relay switching circuit, we would make this cheap while keeping a good number of CV outs, and an easy way to implement it in various devices, then have the best of all the options.
A few issues have arisen with VCRās - they need to be able to match the various different types of pots in the analog gear weāre converting, so they need a variable range of resistance. Also, they would need to have a variable resistance curve characteristic - linear, log, neg log, etc etc. I would imagine that we could design a board that has all these characteristics and the ability to select which ones fit your needs, or a series of cutdown boards which would each have a different characteristic each eg Linear VCR, Logarithmic VCR etc etc
I was thinking that I could try adapt some existing designs I have seen, but Iāll probably need some help as my strength is not hardware, so I wonāt start that unless I know that others are interested. I do know that vactrols are less accurate and they are slow, so I think that FET or OTA would be the way to goā¦
Please add to this or correct me, Iām by no means an expert, more of a librarian in this case
Just while Iām typing massive posts hereās some more about why I mention this What I was thinking of doing was making a set of analog effects, and stacking them all into a rack-mount case (you could do desktop just as easily) with a small patchbay of inās and outs, so they could be run independently or in series, or sidechained or whichever, and you could repatch them manually as required in the audio chain, and use the MIDIBox to save presets and use MIDI automation for each module. A screen and encoders could be used for editing the FX parameters, as well as a feature to enter the configuration of the manual patch panel for saving an āimageā of it with a set of FX presets to a bankstick or sysex. Iād really like to do an audio switch matrix instead of a patch bay but that gets a bit complex So I figured stick with a manual patchbay and make a record of itās connections.
aaaand last but not least, a collection of links if you feel like an information junkie and a summary is just not enough
Capt Hastings Digipot Driver: http://compiler.kaustic.net/fm_offline/fm_006/aout_AD5204.inc
http://www.midibox.org/forum/index.php?topic=2556.0
http://www.midibox.org/forum/index.php?topic=2186.0
http://www.midibox.org/forum/index.php?topic=3132.0
http://www.midibox.org/forum/index.php?topic=4065.0
http://www.midibox.org/forum/index.php?topic=3373.0
http://www.midibox.org/forum/index.php?topic=5749.0
http://www.midibox.org/forum/index.php?topic=6245.0
FET as VCR: http://freespace.virgin.net/ljmayes.mal/comp/vcr.htm
http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/transistors/vcr2n4n7n.pdfā>http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/transistors/vcr2n4n7n.pdf
Aaron D Lanterman ā ECE Facultyā>http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/transistors/vishay_fet_cvr_an.pdf appnotes for above device
http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/ also mentions āAaronās note: It seems to me that this āFET as variable resistorā trick really only works easily if one side of your āresistorā is tied to ground, or at least a constant voltage; otherwise your V_GS will move around as V_S moves around, even if youāre keeping V_G constant, and taking care of that requires some thoughtā Good links there!
OTA as VCR: http://et.nmsu.edu/~etti/winter98/electronics/grise/wrg.html
http://www.uni-bonn.de/~uzs159/ota3080.html
Vactrol (LED->LDR) as VCR: http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/vactrol/perkinelmer_vactrol.pdf
http://users.ece.gatech.edu/~lanterma/sdiy/datasheets/vactrol/vtl5cx2.pdf
So what do you think guys? Could the MBFX be a module which does configurable analog CV-controlled-resistance and analog gate-controlled-relay? Could the MAX5250 be used in the MBCV to reduce costs?
A Friggin Big Ass Post Iāll just take out the things I want to comment on ;D
Digipots/DCRās, CMOS - This method means desoldering and replacing existing pots with digitally controlled variants, or switching between digitally controlled and physically controlled resistors, and switches would be CMOS controlled switches. It seems a fairly easy solution at first and digipots are not quite so expensive as motorfaders, but as we know, the digital signals can make things a bit messy. There are concerns about noise from the digital control buss leaking into the audio signal, but that might be worked around by heavy shielding and perhaps a separate ground plane. The separate ground could be tricky to implement though, and that implementation would vary greatly from one device to another, increasing support requirements. One upside with this is that little additional hardware is required.
Cheap +
Quiet -
Easy to Design + (already done by Capt. Hastings)
Easy to Build +
Easy to Support -
Flexible Implementation +
or if we try to find a quiet way
Cheap ? (Need to see what the design is like)
Quiet -
Easy to Design -
Easy to Build ? (Need to see what the design is like)
Easy to Support -
Flexible Implementation +
Iām pretty confident that I will get a quiet digipot solution. As I am really only interested in MIDIfying new circuitry (as opposed to modding existing stompboxes) desoldering pots is not an issue for me. I will work with SIL headers and female SIL connectors, so as soon as a circuit works with standard analog pots, I can swap the pots for the digital ones, and test away.
Analog
Vactrols/FETs/OTAs, Relays - This way converts a control voltage to a resistance or uises it to switch a relay. It is analog so it eliminates the noise problems, but it introduces other problems - cost, because you need the AOUT with the expensive MAX525 to control a number of pots. This method is basically about converting existing designs to CV control, and using the existing MBCV hardware, so means no design issues to deal with from the MIOS end, itās a matter of using any DIY effect design with CV control, or adding CV control to existing gear. When adding to existing gear, you could build the CV controls, and use a switch to select internal control (wired to pots) or CV control (wired to the CV Ins)
I notice the MAX5250 is available and about half the price, seems to be the same thing, but it is 10 bit resolution, not 12-bit. This would reduce the price fairly well, but Iām not fantastic with hardware and havenāt read the entire datasheet yet (just the specs and features) so it may not work. Iāll look into that further shortly but if anyone else knows or is interested please let me know what you think
Cheap - (+? what about the 5250?)
Quiet +
Easy to Design +
Easy to Build +
Easy to Support +
Flexible Implementation ? (Can we make a simple and flexible CV-to-resistance circuit?)
AOUT_LC is also an option. I wonder if 12bit isnāt something of a requirement, as TK does also design this board to have a 12bit (and a 4 bit) DAC. Vactrols need to be shielded from light, or they will behave differently during dy and night. Also, when including more than one vactrol in a circuit, optical isolation is a big issue because else the LED from vactrol 1 could affect the LDR from vactrol 2, right? I believe implementing vactrols instead of pots is very, VERY tricky.
Digital is great but noisey, so it will not fit the needs of some people (like illogik and I who sweep everything). We could work on methods to reduce noise, but that will always be something of a struggle and effective methods of reducing the noise will vary from one circuit to the next, which increases support efforts.
I was surprised by Hastingsā audio sample, I couldnāt hear any artifacts at all while he was sweeping the settings. After this audioclip, I became confident digipots are the answer. Moreover because they can be interfaced via SPI, no need for additional (expensive) hardware like DACs. The Analog Sollutions IC also used by Hastings is in effect 4 potmeters in one. Surely for new designs, the costs for 4 analog potmeters would be the same or more.
Just while Iām typing massive posts hereās some more about why I mention this
While my current design plan is aimed at providing digitally controlled distortion for the 4xMBSID, I have also played with that idea. I believe that digipots via SPI is by far the cheapest sollution for FX gear that is not yet equipped with CV inputs. My setup for a MBFX unit would be like a modular synth, but with no knobs and buttons on the outside of the āmodulesā. A Core equipped with a digipot board and an AOUT should be sufficient for controlling everything you can dream of. A panel with some rotary encoders can be optional, if no other midibox provides this functionality externally already. Example: Iām not planning on making a MB64e myself, but if you already own one it would be perfectly suitable to control stuff. I would opt for a less extended control surface. But LEDrings/bars and stuff like that could be part of the module frontpanels, very good oversight what gets tweaked at what time.
A patchbay is not something I would make, Iād just add the input and output jacks on the front of every module. That way you can patch any way you like, and itās always clear what youāre attaching where.
So what do you think guys? Could the MBFX be a module which does configurable analog CV-controlled-resistance and analog gate-controlled-relay?
No, I think MBFX is more like a firmware to interface with FX modules, not a MBHP module by itself. The digipot or vactrol circuits could become modules in the āMBHP programā if they work good. Thatās also the path Iām going to take: a separate board with the ICās to interface with the analog gear. Only some cables run from the FX circuit pot connectors to the interface board connectors, to keep digital noise and stuff far away.
Could the MAX5250 be used in the MBCV to reduce costs?
Donāt know, I think not. But sorry: donāt care too much either. AOUT_LC is available to reduce cost, AOUT classic to provide all options.
Cheers, Alex.
Yes, the F.B.A.P. has taken me many hours to master Observe, my FBAP skills in action! ā¦
Iām pretty confident that I will get a quiet digipot solution. As I am really only interested in MIDIfying new circuitry (as opposed to modding existing stompboxes) desoldering pots is not an issue for me. I will work with SIL headers and female SIL connectors, so as soon as a circuit works with standard analog pots, I can swap the pots for the digital ones, and test away.
Sorry I didnāt mean to imply that desoldering was a problem, just trying to explain what is involvedā¦I actually added a bit of stuff in after I was finished so that the concepts could be more comprehensible to the recently arrived or uninitiatedā¦ I mean, I tried to make it more noob friendly Like, you and I know what a digipot is, but not everyone will, and not everyone will understand what the purpose of the digipot is, so I figured hey, with all that work and typing I might as well add in 3 more sentences to help explain to noobs wtf us geeks are toying with this time ;D
My post was not so much about doing what we need for our particular applications, because in that case I am in the same situation as youā¦ Iām just trying to think about flexibility so people who come along in future have an easy way to MIDIfy FX, no matter what the effect they want to useā¦ Itās very often requested, it goes way beyond just this handful of us I thinkā¦Of course that doesnāt mean we all have to do it the same way, we could all try different methods which suit our needs and then people could apply whichever one they decide would suit them by adapting the notes on the forum/wiki to their particular designā¦ But of course all of us teaming up has advantages for development because many hands makes light work, and a single solution might reduce support requests on the forumā¦
Iāll keep talking in the context of searching for a universal solution for the time beingā¦ I believe that it should be possible to find a solution where it will meet everyoneās needs, maybe Iām just being an optimist but hey
AOUT_LC is also an option. I wonder if 12bit isnāt something of a requirement, as TK does also design this board to have a 12bit (and a 4 bit) DAC. Vactrols need to be shielded from light, or they will behave differently during day and night. Also, when including more than one vactrol in a circuit, optical isolation is a big issue because else the LED from vactrol 1 could affect the LDR from vactrol 2, right? I believe implementing vactrols instead of pots is very, VERY tricky.
AOUT_LC is an option but I think 2 CV outs is a little too limiting for many FX machinesā¦ 12bit is 4x as many steps, maybe itās to avoid stepping so sweeps are smoother? Maybe itās something to do with the capabilities or existing code for his CEM3378, thatās what his is connected to (or it was in the doco) But of course thereās always the jumper to split the resistor ladders evenly over the 595ās and use 8 + 8 bit, so who knowsā¦ At the end of the day, itās just resolution, Iām sure MIOS can handle it, so I think the only ill effect would be the steppingā¦ But I think those digipots are only 256 steps which is only 8-bitā¦ I only just thought of that, if anything the digipots will be more subject to zipperingā¦ And you said theyāre fine in capt hastingsā mp3, so 10bit shouldnāt be an issueā¦ Although I just listened and it sounds OK to me, I donāt know if a distorted delayed filtered guitar is the best specimen for checking for quiet noises and steppingā¦
Agreed, I am not keen on vactrolsā¦ Too slow, too inaccurate, too sensitive, just too messyā¦ Just trying to be non-biased and present all the available options butā¦mm, nah. yuck. All signs point to no. go directly off the drawing board, do not pass go, do no collect $200 heheh
I was surprised by Hastingsā audio sample, I couldnāt hear any artifacts at all while he was sweeping the settings. After this audioclip, I became confident digipots are the answer. Moreover because they can be interfaced via SPI, no need for additional (expensive) hardware like DACs. The Analog Sollutions IC also used by Hastings is in effect 4 potmeters in one. Surely for new designs, the costs for 4 analog potmeters would be the same or more.
I do really prefer the digipots ideaā¦ If we can just find a cheap and easy way to implement them with no noise or stepping, that doesnāt involve altering the FX circuitry, it is definitely the way to goā¦ But if we have to mess with the ground busses in the FX then it would be no good for some people - I donāt think thatās the case by the sounds of things which rocks 
For those who want to modify existing gear, they could keep the manual pots and install digipots, and switch between the two sets of connections, so it works there too, and the digipots are pretty small so they should fit into small spacesā¦
I believe that digipots via SPI is by far the cheapest sollution for FX gear that is not yet equipped with CV inputs.
Definitely, and possibly cheaper than those with CV inputs cause of the DACs.
My setup for a MBFX unit would be like a modular synth, but with no knobs and buttons on the outside of the āmodulesā. A Core equipped with a digipot board and an AOUT should be sufficient for controlling everything you can dream of. A panel with some rotary encoders can be optional, if no other midibox provides this functionality externally already. Example: Iām not planning on making a MB64e myself, but if you already own one it would be perfectly suitable to control stuff. I would opt for a less extended control surface. But LEDrings/bars and stuff like that could be part of the module frontpanels, very good oversight what gets tweaked at what time.
A patchbay is not something I would make, Iād just add the input and output jacks on the front of every module. That way you can patch any way you like, and itās always clear what youāre attaching where.
Thatās exactly what I mean All those in and out jacks, Iād just line them up near each other so they could be easily patched into one another, which is why I said āpatchbayāā¦Sorry for being confusing.
No, I think MBFX is more like a firmware to interface with FX modules, not a MBHP module by itself. The digipot or vactrol circuits could become modules in the āMBHP programā if they work good. Thatās also the path Iām going to take: a separate board with the ICās to interface with the analog gear. Only some cables run from the FX circuit pot connectors to the interface board connectors, to keep digital noise and stuff far away.
If we can get the digipots to work without digital noise or zippering/stepping and without having to do strange things to the FX circuitry then definitely thatās the way to go. You sound pretty confident man, is there anything you need for the prototype? Help? Parts? Moral support?
Donāt know, I think not. But sorry: donāt care too much either. AOUT_LC is available to reduce cost, AOUT classic to provide all options.
Well there wonāt be much point in toying with the CV stuff for our purposes if youāre right about the digipots.
Although for the sake of curiosity, I do wonderā¦ I mean, the MAX525 is a barrier for some people against having CV control, it would be great for the community if we could find a full-featured DAC for AOUT which is also cheap.
TK would you mind advising us on this oneā¦ Can DACs of a lower resolution be used? Why is it that you have used 12bit for AOUT_LC and your 3378, is it there a need for 12-bit signal resolution or is it for compatibility reasons or was it something else like 16bit was too much and 8bit not enough orā¦?
Assuming for a moment that lower resolution DACs are OK, this opens some interesting opportunities. the 5250 at 10-bit is interface-compatible, pin-compatible, code compatible (check out the datahseets, itās like a game of āspot the differenceā), is half-price of a 525ā¦ and even cheaper still, if you can deal with lower resolution and surface mount packaging you could go the 5258, at 8-bit and get 8 channels, twice as many, for 1/4 the price of the MAX525ā¦ Now 8-bit, 256 steps, immediately conjures images of 1983 or something, but if you think about the fact that much of the time we are using MIDI CCās or Note Numbers (0-126, 7 bit) to control these things, weāve still got 1 bit to spare. Could be food for thought, AOUT at better than MIDI resolution for 1/8 of the current cost per channel And thatās the same resolution as the digipotsā¦ I should add that the 5258 is listed as being QSOP packing but the 525 is listed as SSOP and it didnāt look like it to me. Maybe Iāve gotten my packages confused but anywayā¦
Hopefully we can get the digipots sorted and it wonāt be a big deal, but it might be of interest to someone, and if we canāt get the digipots working, well it could be very handy!
Some days ago I wrote in the german part of this forum, that it would be great if the community would develop a third AOUT module as alternative solution to AOUT and AOUT_LC (http://www.midibox.org/forum/index.php?topic=6333.msg39275#msg39275).
The MAX525 was selected after some discussions in the forum because of the best price/quality ratio. The focus was on the number of channels, interface type, latency, cascadability and last not least linearity (important for controlling OSC frequencies, not so important for controlling volume or cut offā¦)
TK would you mind advising us on this oneā¦ Can DACs of a lower resolution be used? Why is it that you have used 12bit for AOUT_LC and your 3378, is it there a need for 12-bit signal resolution or is it for compatibility reasons or was it something else like 16bit was too much and 8bit not enough orā¦?
It really depends on the usecase. If the cut off frequency of a filter should be controlled over a large range, 8bit resolution would lead to some unwanted steppiness. If the range would be reduced (e.g. by using a gain and offset pot), you are maybe happy with 8bit, but the reduced range means less flexibility. It should also be considered, that once the resonance is set to a point, where the filter tends to self oscillation, you definietly want to finetune, and this just requires a high resolution.
12bit for CV is enough in my eyes, higher resolutions are difficult to handle without special layout and PSU measures. With v_max = 5V, each step adds only 1.2 mV to the output voltage, now add the background noise of your PSU and the digital circurity, and you see that it doesnāt make much sense to increase the resolution (this statement is not true for audio signals).
Another point which should be considered is the use of S&H in order to multiplex DAC outputs. You could use a cheap single channel DAC and a cheap S&H chip in order to get 8 channels. S&H requires special software measures (timer driven DAC output), which loads the CPU more than a multichannel DAC, but for applications like MBCV or MB64, where the CPU is mostly idle, S&H is a good alternative. Sidenote: if MBCV currently doesnāt support S&H, this doesnāt mean, that itās difficult to implement this!
A last general tip when searching for an alternative DAC: search for samples at different vendor webpages (Maxim is not the only one), prefer chips, which are easily available, check with www.findchips.com, if the IC is available in low quantities.
Best Regards, Thorsten.
Thanks TK
So as far as MBFX would be concerned, it looks like those digipots we looked at may not cut the mustardā¦ Because if 8-bit is not enough for good analog filtersweeps, then we would need a digipot with more steps that 256. There are digipots with more steps (1024) but they are more expensive and 2-channels is the most available on 1 ICā¦ but they could be chained of course.
Iāve looked at the MAX5250, it is more widely available than the 525, half the price, and completely compatible, it just ignores the 2 bits that it canāt convert, so we wouldnāt even have to change the existing code (not that it would be hard anyway). This will give us 1024 steps too, but more channels going outā¦ And if you want 12-bit, you just drop in a 525
I think that the digipot vs CV race is neck-and-neck at the moment
For digipots to work, we need a quiet solution that does not involve modifying the FX circuitry (eg ground planes)
For CV to work, we need a cheap OTA or FET based linear voltage-resistance circuitā¦ Illogik, I was wondering if you might pipe up about the LM13700 Iāve looked at the datasheet and itās got example schematics for a VCR in there already which is great, do you know if anyone has tried to adapt it yet? Those ICās are under $2 eachā¦ But the digipots still work out much cheaper and simpler because you donāt have to build the AOUTā¦ But then again CV allows for better resolution.
It seems thereās two degrees of resolution with each methodā¦ with a really rough price estimate:
6 X 256-step (8bit) SPI Digipots $10 <$2 per channel
2 X 1024-step (10 bit) SPI Digipots $10 $5 per channel
8 X 1024-step (10 bit) 5250 DAC+VCR $40+20=60 $7.50 per channel
8 X 4096-step (12 bit) 525 DAC+VCR $60+20=80 $10 per channel
I donāt think the 2nd one is worth itā¦ For a little more money the CV option could be used, which is upgradeable too, and because itās only 2-channel youād probably need a lot of them.
Edit: Fiinished nowā¦ I hit save instead of preview hehehe
Hi,
if CV can be used instead of digipots: fits me fine. I already have a couple of MAX525 lying around, And itās more universal than digipots too. Now all you need to think up is an easy to use way of converting CV to resistanceā¦
Another option: use both! A single core is able to support an AOUT board and digipots. Use the pots where 8 bits is enough, use aout where you need 12. In general, 12 bit is used for filtersweeps and stuff, and those filters are already CV controllable most of the time. 8 bits is 16 times the resolution of 4 bits (the second channel of AOUT_LC) and that should be more than enough for everything not filter related in my opinion.
In that case the stuff to develop is still a digipot board (sorry to be so stubborn. ) as the CV outs are already in place on the AOUT. What might be interesting is a new S&H board, with ICās that arenāt so hard to come by.
Cheers, Alex.
Now all you need to think up is an easy to use way of converting CV to resistanceā¦
And thereās the tricky part
Another option: use both!
Yeh I was starting to think that myself MY aim was to attempt to find one solution that has all the benefits rolled into one, but if that wasnāt possible, to find the most effective hybrid. The facts weāre uncovering here are starting to make a clear pictureā¦To exemplify the comparison:
Digital = quicker, easier, cheaper, lower sound quality
Analog = Time consuming, lots of soldering, expensive, higher sound quality
If we could write a framework application (MBFX) which supports controlled resistance via SPI -and- AOUT, along with the hardware side of MBFX, a DCR board (SPI Controlled Digipots) and a VCR board (Voltage controlled OTAās), then people could pick and choose according to their needsā¦ If they want a cheap and easy but 90% functional device they can go the digital route, if theyāre noise-conscious or want smoother realtime control, then they can choose the analog routeā¦
In general, 12 bit is used for filtersweeps and stuff, and those filters are already CV controllable most of the time. 8 bits ā¦ should be more than enough for everything not filter related in my opinion.
In that case the stuff to develop is still a digipot board (sorry to be so stubborn.
Oh man itās SO not like that, I donāt feel youāre being stubborn at all, and I really hope you donāt feel like I am! ā¦ this isnāt an argument or anything When I say things like the Analog vs digital āraceā Iām not referring to you vs me or anything like thatā¦ Iām just taking the facts and using them to find the best solutionā¦ And the facts are, digipots are easier and cheaper and have a great deal of other factors that mean that there are a lot of reasons to lean towards them. Itās just kinda convenient really that youāve been looking into digipots a great deal, so youāve been telling us a lot about the digital way, and so I am leaving that to you and talking about CVā¦ But please donāt think Iām trying to steer away from digipots, or trying to steer or argue or debate at all, just trying to get all the facts together
Disclaimers aside Yeh I know what you mean about the 12-bit resolution only really being necessary for smooth sweepsā¦I wouldnāt say itās just filters though (I sweep loooots of things), but I do catch your drift.
I didnāt realise that the SHX8 board used rare ICās either, thanks for the heads-up on that. It did bring up a matter which I havenāt discussed yet, which is expandability. That is one of two matters we havenāt really discussed yet, the other being switchingā¦ But both are fairly straightforward of courseā¦
Now, before I bust my brain trying to create a VCR board and a more easily sourced SHX8 boardā¦ Thereās something that I thought of just now which could eliminate the issues with the large steps between values with the DCRāsā¦
DCRās in series
Think about it.. Resistors in series add themselves. We can send simultaneous commands to two resistors at onceā¦One resistor for coarse steps, and one for the āfine tuningā inbetween them
If we have a 10KOhm digipot with 256 steps, thatās about 400Ohm per step rightā¦ We could put another DCR in series with that DCR which holds a varied rsistance between 0 and 400Ohm. Given that the lowest range digipot I can find is 1K, we would only ever use 4/10 of the resistance on that one, thatās 100 or so steps inbetween each of the 256 steps of the 10KOhm pot, thereās about 25600 steps, thatās better than 14 bit!
Iāve literally only just thought about this, so there may be issues, I havenāt thought it through at allā¦
Like for example there may be some delay between the SPI commands which would set the two resistors, so weād have to code some smarts into the firmware so that it didnāt make the resistance values jump about or somethingā¦ I dunno if we could make this workā¦ Maybe we could screw with the SPI clock signals, so we could load the values into the DCRās and then clock them both simultaneouslyā¦Maybe TK will know what kind of latency would be involved in sending SPI commands from MIOS (given that AFAIK nobody has used SPI yet), if itās not too great then we might not have any issues (because of the latency in the DCRās)
What do you think? Is this idea worth playing with? Is it even possible?
Edit: Thereās another way, we could bit-bang dual channel SPIā¦
If we could write a framework application (MBFX) which supports controlled resistance via SPI -and- AOUT, along with the hardware side of MBFX, a DCR board (SPI Controlled Digipots) and a VCR board (Voltage controlled OTAās), then people could pick and choose according to their needsā¦ If they want a cheap and easy but 90% functional device they can go the digital route, if theyāre noise-conscious or want smoother realtime control, then they can choose the analog routeā¦
I have no idea what OTAās are, do you have a link with some info? I wouldnāt say a finished digipot design would be 90% functional, just lower resolution.
Oh man itās SO not like that, I donāt feel youāre being stubborn at all, and I really hope you donāt feel like I am! ā¦
I was not implying any argument at all, just making a little fun, no worries!
I know what you mean about the 12-bit resolution only really being necessary for smooth sweeps ā¦ I wouldnāt say itās just filters though (I sweep loooots of things), but I do catch your drift.
What I meant was that you can sweep away anyway you like, and itās still possible in 8-bit resolution, which produces no audible artifacts with some FX. The more crude FX like distortion for example, you wonāt notice any zipper noise between the rest of the signal, Iām sure.
I didnāt realise that the SHX8 board used rare ICās either, thanks for the heads-up on that. It did bring up a matter which I havenāt discussed yet, which is expandability. That is one of two matters we havenāt really discussed yet, the other being switchingā¦ But both are fairly straightforward of courseā¦
Maybe the original SHX8 ICās are still available somewhere, but Reichelt doesnāt stock them and theyāre not available anymore at mouser. Digital switches have to be easy to implement. Do you pehaps already have a particular IC in mind?
Now, before I bust my brain trying to create a VCR board and a more easily sourced SHX8 boardā¦ Thereās something that I thought of just now which could eliminate the issues with the large steps between values with the DCRāsā¦ DCRās in series
If we have a 10KOhm digipot with 256 steps, thatās about 400Ohm per step rightā¦ We could put another DCR in series with that DCR which holds a varied rsistance between 0 and 400Ohm. Given that the lowest range digipot I can find is 1K, we would only ever use 4/10 of the resistance on that one, thatās 100 or so steps inbetween each of the 256 steps of the 10KOhm pot, thereās about 25600 steps, thatās better than 14 bit!
Iāve literally only just thought about this, so there may be issues, I havenāt thought it through at allā¦
Like for example there may be some delay between the SPI commands which would set the two resistors, so weād have to code some smarts into the firmware so that it didnāt make the resistance values jump about or somethingā¦ I dunno if we could make this workā¦ Maybe we could screw with the SPI clock signals, so we could load the values into the DCRās and then clock them both simultaneouslyā¦Maybe TK will know what kind of latency would be involved in sending SPI commands from MIOS (given that AFAIK nobody has used SPI yet), if itās not too great then we might not have any issues (because of the latency in the DCRās)
What do you think? Is this idea worth playing with? Is it even possible?
I have no idea If potentiometers work in series, let alone digital ones. Sounds tricky though.
Edit: Thereās another way, we could bit-bang dual channel SPIā¦
Bit gang banging dual SPI?.. sorry you lost me. What have you conjured up this time? ;D
Cheers, Alex.