New Low-Cost AOUT (MBHP_AOUT_LC) under development

[TK.]

This is a suggestion from KD (member of the SynthDIY list) for a low-cost DAC interface as a cheap replacement for the MAX525:

http://www.midibox.org/users/kd/KDdac.pdf

I'm refering to the 8bit discrete DAC module which is uses one 74HC595, one OP amp stage and resistor ladder --- the signal quality is possibly adequate enough for most of our applications - and the availability of the parts should be superb! I will try it :)

Any thoughts?

Best Regards, Thorsten.

Update: the final schematic of the MBHP_AOUT_LC module can be found here: http://www.ucapps.de/mbhp/mbhp_aout_lc.pdf

[vintageuser]

I think that is a good idea. Twenty years ago I developed the CGX - Interface ( Midi-CV-Gate Interface 1HE 19"" Interface with blue panel), which was distributed by JMS. It´s also based on a resistor ladder. It is a very fast one and was very successful. If you need any help, get in contact with me. I have ordered the max525 and will build up a cv-box within the next weeks. I will compare it to the cgx-interface and let you know.

[TK.]

a comparison between the two solutions would be great!

Best Regards, Thorsten.

[moldava]

The different solutions look really interesting...i was reading an old pdf of the PAIA PROTEUS,from 1981,a programable analog synth...the patches were stored by multiplexing cv pots and rotary switches,wired as voltage dividers...stored in ram,and then demultiplexed to generate cvs to control the analog circuitry...i've been dreaming of using this aproach in a more modern version,with an lcd

I supose midibox or mios would be the perfect platform...reading the pots,storing in banksticks,generating the cvs,and full midi control...wow!

Tell me if this looks like an interesting project...

Thanks for all Your hard work and inspiration TK !!!

[seppoman]

Hi Thorsten,

if 8 bits are enough, have you had a look at the TI TLC5628? It´s an 8x 8bits DAC with serial interface, so there would be really few parts needed.

for 12 bits, I´ve found the Microchip MCP4922 DAC. It´s 2x 12 bits, also serial interface.

They´re both really cheap (TI about 3.50 EUR, MC 1.60 EUR) and also available as free samples.

Seppoman

[Duggle]

hi,

For a single channel converter using stock parts its a good solution.

However, with 16+ resistors per channel the effort is transferred to construction time and board size.

As suggested, there must be low cost, integrated solution for such a generic requirement.

cheers

[TK.]

Seems that I have to add some words why I favour the resistor solution - I want to realize an alternative board which is always madeable without the need for special ICs which are rarely available or hard to get in low quantities.

Remember how difficult (and/or expensive) it is to get special parts -

Maybe not in your country if you've the luck that a national distributor offers the chip...

Maybe not for people who are ordering electronic stuff once or twice a month so that shipping costs don't matter...

Maybe not if you are doing batch orders....

Maybe not for people who order a small number of samples (who knows how long this will be accepted by the manufacturers, I already got complaints...)

But whats about the people who just want to play a little with analog outputs *immediately* and without much costs?

And what should happen once the chips, which have been suggested in the last months, are discontinued?

And where did you lost your DIY spirit? ;-)

Ok, seriously: this is a very generic solution which won't require any change in the application code if somebody wants to connect a DAC with parallel interface (which are mostly cheaper and easier to get than DAC with serial interface).

So long the device isn't laser-trimmed, the output voltage of the resistor ladder can be more adequate, especially if hand-selected resistors are used. The output voltage is not limited (you need balanced or high voltage outputs? No limitation here).

7-bit resolution is mostly sufficient so long the output is controlled by MIDI notes, velocity, CCs, after touch, etc. --- higher resolutions become interesting with frequency or filter sweeps, but they are mostly so fast that a gain or offset error cannot be regognized.

If somebody wants to have a high-quality solution, he can spent more effort in selecting the resistors, he can still use the MAX525 design, he can choose another DAC and adapt the software driver, or he can connect a lasertrimmed integrated DAC with parallel interface

Ok, these are my motiviations :)

Best Regards, Thorsten.

[KD]

Hi folks!

Im KD, im the name of the KDdac.pdf design, i tought that it could be of help if i spoke

directly to the forum regarding the kddac.pdf, i saw that some are concerned about the

amount of resistors, you can buy ready matched R2R nets in SIL and SO package,

Farnell carries some last time i pooked in the catalog.

Reg

KD

[moebius]

Ok, seriously: this is a very generic solution which won't require any change in the application code if somebody wants to connect a DAC with parallel interface (which are mostly cheaper and easier to get than DAC with serial interface).

Shift registers, eh? Should have guessed ;D

Actully, this is a very smart move. A generic solution for parallel DACs, that really doesn't "break" anything, just gives us more choices and easier expandability - remembering that one of the MAX525s real advantages is daisy chaining. The most other serial dacs don't have serial outputs and would need some (unnecessarily complex) glue logic for that.

Bye, Moebius

[Martin_Haverland]

i'm really excited! this cheap cost solution is really what i'm looking for since i've alot of analog gear not really needing this high resolution cv... and the max is really a very high cost solution hard to get anywhere. i'm looking forward the first midibox-ready pcb designs, maybe one for standard metal film resistors and one for an array...

:D

[smashtv]

i'm looking forward the first midibox-ready pcb designs, maybe one for standard metal film resistors and one for an array...

So am I!

Hehe just kidding! ;D

I'm tinkering with a layout to support both discrete resistors or an array in the same real estate.

Trace tetris is on!

Smash

[chriss]

hello all,

just checked the maxim homepage. price for the 525 is 11,95 US dollar, what is about 9-10€ at the moment. so this price seems pretty reasonable for me, for this  quad 12bit DAC. and i never had any problems with this chip, although it came in ssop package (at least i know that i don´t have to fear the Midibox FM). maybe smash TV wants to take them in stock so all non US diyers don´t have to pay taxes and stupid high shipping costs.

and we all know the advantages of analog filters, but analog oscillators are huge as well, and as far as i know is the difference between 10bit and 12bit quite huge, if you want accuracy for the pitch.

but i like the resistor ladder solution as well. very simple and very undestroyable.

chriss

[TK.]

I'm tinkering with a layout to support both discrete resistors or an array in the same real estate.

please wait for the first schematic.

One feature I'm planning is to allow an on-board configuration for 4*8bit outputs or 2*16bit outputs (however, I think that more than 12bit are not really required, but it's more than 8 bit ;-).

Another thing which isn't clear yet is the number of OP amps (I guess one TL074 is enough for 4 outputs per module, but maybe I've to add another for the voltage reference) - this could affect the layout

Best Regards, Thorsten.

[Martin_Haverland]

@chriss: i've a max 525 laying around here, payed nothing for it... but...last time i was looking for to order one from germany the max 525 + shunt were around 45€ + shipping, so maybe prices changed in the meantime...

tl074 and resistors are almost at anytime in my stock, so one could try out and experiment without soldering a (maybe not so expensive) max smd.

after all it is just an option, there's nothing wrong with the max 525 at all...

i'm looking forward at the things to come ...

[smashtv]

please wait for the first schematic.

Whoop I should have been more specific, I'm working on just the plot of real estate with the resistor ladder itself, not the rest of the design (yet).  Just making options not compromises for when it is time for the whole layout.....  ;)

Best

Smash

[TL]

What's the actual state of this module?  ;D

[TK.]

It's built, but requires some improvements

Best Regards, Thorsten.

[TL]

Thank you Thosten  :)

[Davo]

Sorry if this has been asked before, but what's so special about the max525 chip?  I perused Jameco and found the following DACs that I think might be suitable to build a lower-cost AOUT.  Would someone who knows better please check this out?

DAC1222LCN (#14980) 12-bit $9.95

7109CPL (#43764) 12-bit $7.49

TLC5615CP (#239425) 10-bit $3.49

IMHO, this lower-cost AOUT module ought to be able to support more than just two 16-bit outputs at once.  Or at least have an extended version of the 2-output one.

[TK.]

Hi Davo,

you've listed single channel DACs, the MAX525 provides 4 channels at once, this makes the chip so special.

The Low-Cost AOUT module is just a design idea and will be provided "as is", everbody is free to use different DACs. The reason why I want to use resistors instead of ICs is to avoid the endless discussions about the "best DAC", which you will possibly get today from your favourite distributor, but which vanishes from the list tomorrow...

Best Regards, Thorsten.

[Davo]

So this is back to the earlier discussion of resistor ladders controlled by a DOUT module?

[TK.]

yes...

Best Regards, Thorsten.

[Synthhead]

Hi.

Watching this space with great interest! 8).

One feature I'm planning is to allow an on-board configuration for 4*8bit outputs or 2*16bit outputs

That sounds like a very useful choice.The major problem I've had with midi-cv converters is zipper noise when controlling filter cutoffs -  particularly with resonance, and at low cutoff freqs. The Paia midi2CV8 (which I think is 8-bit) was very "zippery" when controlling an EFM diode ladder VCF . Problem solved by passing the CV output through a simple lag circuit I  found on the net. When adjusted, this gave adequate control speed when waggling a fader to control cutoff, resonance etc.

Applying this fix to the Pitch Wheel wasn't so successful - adjusting the lag speed to completely remove zipper noise made the response of the wheel WAY too slow - so I guess there's a definite case for high resolution when controlling pitch.

Is it planned for the SHX8 still to be compatible with the new AOUT board? I suppose any smoothing lag on an 8-bit output would have to be post-SHX.....??

Very exited about this development. Nice one TK :)

Dave

[Tanstaafl]

in my book... the r2r ladder thing is the best way... but has some caveats.

1. space... (it takes a lot of space)

2. space... (it takes a lot of space)

3. space... (it takes a lot of space)

the circuit board would be large...unless we use smd technology

smd is a pain in the tuckus

(because it's small)

I know I sound like a stuck record... but

PAIA has done all of this on a 8051 / 8031

it seems to me that MIOS has the horsepower to do all of this stuff.

maybe..... a consult with them?(PAIA)  to see if they are willing to get this

into  the diy community?

their way of doing D2A conversion is simple and elegant.

just my 2 cents worth...

gb

[TK.]

Yes, the output is currently very zippering, and yes, the SHX8 will help :)

The reason why there is no progress yet is not, that I'm having problems where I have no idea how to solve them, it's just because of the missing time to get more experiences, to improve the circuit and to write a documentation

Best Regards, Thorsten.