The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> 12 bits R-2R DAC design
https://designers-guide.org/forum/YaBB.pl?num=1314669361

Message started by mists on Aug 29^th, 2011, 6:56pm

Title: 12 bits R-2R DAC design
Post by mists on Aug 29^th, 2011, 6:56pm

I want to use poly resistor to design a 12 bits voltage mode output dac based on R-2R structure. in spite of resistor matching, I found that the voltage coefficient of ploy resistor affect dac accuracy much, any good way to remove this effect or what other structures for 12 bits voltage mode output dac design? thanks!

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Aug 30^th, 2011, 4:03pm

Sigma Delta methods? PWM with weighted summation?

R-2R are pretty slow, more modern methods which are less process dependent are more commonly used.

Do some research into audio quality sigma-delta DACs

Title: Re: 12 bits R-2R DAC design
Post by mists on Aug 31^st, 2011, 10:56pm

but I have not clock and the DAC setup time should less than 10us which contain output amp setting time. :'(

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Sep 6^th, 2011, 4:40pm

I would give serious consideration to oversampling methods, even if you have to create the clock oscillator yourself.

The resistors associated with R-2R are going to take up a lot of space.

Many of the other architectures are going to be sensitive to offset and matching problems at 12 bits.

Title: Re: 12 bits R-2R DAC design
Post by harpoon on Sep 7^th, 2011, 1:25am

What process geometry are you using and what are your constraints (besides no clock and 10us settling time) ?

it would be interesting to know why you have chosen R-2R for a 12bit DAC (you'll probably need to see ~14bit accuracy in sim to get 12bit real life)

Title: Re: 12 bits R-2R DAC design
Post by RobG on Sep 7^th, 2011, 9:39am

Mists,
Following a 6-bit tapped R with another (buffered) 6-bit tapped R might suit your needs if you can tolerate some INL. See Johns/Martin page 468 (Multiple R-string D/A converter).

A Thermometer+binary weighted current source dac might also suit your needs. You can use a current splitter to get the last few bits. Lin/Bult, JSSC, Dec 1998 has an example of a 10 bit binary+Thermometer dac.

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Sep 12^th, 2011, 10:22am

some of the weighted approaches are going to need alignment and calibration to make it a true 12 bits.

The john-martin structure with the op-amps will have problems at 12 bits due to op-amp offsets.

As you get beyond 8 bits in resolution (plus or minus)
you get into a lot more issues with process dependency, bias variance and mismatch. Geometry scaling to meet matching needs does not get it done a lot of times and you need to include alignment and calibration methods.

Whatever you use, make sure that you include mismatches, PVT variance and bias variance into your simulations to get close to what the real device looks like.

Is 10us setup time critical? Is "no clock" critical?

Title: Re: 12 bits R-2R DAC design
Post by RobG on Sep 12^th, 2011, 10:36am

loose-electron wrote on Sep 12^th, 2011, 10:22am:

The john-martin structure with the op-amps will have problems at 12 bits due to op-amp offsets.

You had me going for a second, but looking closely at the circuit I don't think it is a problem. He uses the same opamp for each tap. You'd get INL, but it will be monotonic as long as the offset is less than the first dac LSB (6 bits in my example).

However, if he simply walked the opamps down the first resistor string so the "top" opamp was always used for the "topmost" tap the opamp offset would have to be less than 12 bits.

rg

Title: Re: 12 bits R-2R DAC design
Post by mists on Sep 13^th, 2011, 7:07pm

I just want to design a 12 bit voltage output DAC with no external component based on cmos process, I had thought to use current structure, but how to convert current to voltage?if we use resistor, it will have voltage coefficient too. now I only know R-2R or R-string structure can be used, but R-2R has resistor voltage coefficient issue and R-string structure is too complicate. I want to know any other better structure for this dac spec or some method can decrease voltage coefficient of resistor. thanks!

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Sep 14^th, 2011, 12:07pm

RobG wrote on Sep 12^th, 2011, 10:36am:

loose-electron wrote on Sep 12^th, 2011, 10:22am:

The john-martin structure with the op-amps will have problems at 12 bits due to op-amp offsets.

Hmmm, maybe we are talking about different circuits - I am thinking of the one where you mux the LSB on an R-string and buffer it with a pair of op-amps

That thing will have Offset, and as you get up into 12 bit land you will start have more subtle stuff play in like offsets that change slightly due to the common mode voltage.

Thing is the OP is asking for a lot - 12 bits with no alignement and calibration is not easy. The oversampling world takes a lot of the problem out of analog issues and makes it a more robust duty cycle set of problems.

But, they don't want a clock and they want a fairly quick response characterisitic.

Have not even asked the question if R-2R can settle out in the desired time, and if you can get good enough geometric matching.

Opinion? Anything beyond 8 bits should be calibrated. Some could argue 10 bits, and others would argue 6.

Title: Re: 12 bits R-2R DAC design
Post by carlgrace on Oct 29^th, 2011, 5:32pm

loose-electron wrote on Sep 14^th, 2011, 12:07pm:

Opinion? Anything beyond 8 bits should be calibrated. Some could argue 10 bits, and others would argue 6.

Put me in the 10-bit camp. I've gotten away with both 10-bit ADCs and DACs without calibration.

The OP doesn't mention the feature size of the process. If its deep submicron the MOS matching is pretty decent (actually improved from older technologies). I did the 10-bit current-source DAC using 65 nm. I strongly suggest the OP should do a current-source DAC if for some reason oversampling is not available (I concur with loose-electron here).

As for voltage coefficient, generally resistors made of n and p materials have opposite signed voltage coefficients. So, once you choose a value of resistance, you can do a couple of resistors in series to get nominal zero voltage co. Of course this isn't perfect in practice but it goes a long way. Most modern processes have both PPOLY and NPOLY resistors.

So, my advice is do oversampling. If you don't want to or can't, do a current-source DAC. And don't be afraid to segment if you're in an older technology. If its deep submicron, brute-force that sucker.

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Oct 29^th, 2011, 5:44pm

Well, the OP seems to have gone away - but nonetheless, if you do a current steering DAC its going to take a lot of area and a lot of power. Bet money you can generate an oscillator and do oversampling in a smaller space, and consume less power.

If it has to be really fast, then current steering is the way to go, I do agree, but even there you are well advised to include alignment and calibration. If you don't need it, that's great, but good chance is that you will need it to get the linearity good enough to go if its segmented. Pure thermometer code, you can survive without calibration, but the area gets pretty big.

Title: Re: 12 bits R-2R DAC design
Post by Dan Clement on Oct 31^st, 2011, 7:58pm

I think 12 bits is a stretch unless you really do some expensive calibration or error correction of some kind. If you could do some dynamic element matching that would be nice but it sounds like you don't have suitable clock references. Good luck!

Title: Re: 12 bits R-2R DAC design
Post by loose-electron on Nov 5^th, 2011, 1:03pm

yeah, the oversamapling methods are not going to require lots of fancy calibration, and you can create a crude clock to get the job done.