Hi, Is there a good book discussing the effects/advantages/disadvantages of the *type* of converter used inside a digital control loop? Most books on digital control will model the ADC/DAC as a black box with delay, and usually do some analysis of the effect of the sampling rate chosen. I wonder if there exist something that goes a little more into the details of the converters, like SAR versus sigma-delta etc, but with emphasis on control loop design. Thanks! Dave _____________________________ Posted through www.DSPRelated.com
Control system litterature on the type of A/D D/A converter used
Started by ●July 4, 2013
Reply by ●July 4, 20132013-07-04
Hi, this may be obvious, or maybe it isn't: The digital sampling rate demands a minimum latency for the implied lowpass filtering. A higher sampling rate enables faster response, because the group delay of the filter is smaller (time-bandwidth product). This delay may be hidden in analog components. This can become a limitation, especially if you manage latency otherwise carefully (i.e. use multi-phase clocks for ADC, registered logic and DAC). _____________________________ Posted through www.DSPRelated.com
Reply by ●July 4, 20132013-07-04
>Hi, > >this may be obvious, or maybe it isn't: >The digital sampling rate demands a minimum latency for the impliedlowpass>filtering. A higher sampling rate enables faster response, because the >group delay of the filter is smaller (time-bandwidth product). This delay >may be hidden in analog components. > >This can become a limitation, especially if you manage latency otherwise >carefully (i.e. use multi-phase clocks for ADC, registered logic andDAC).>Hi, Yes I understand the basic trade-offs, but I wonder if a more complete and rigorous treatment was done somewhere. Different converter architectures have quite different characteristics in terms of latency, precision, linearity, noise etc...For example, the anti-aliasing analog filter for a SAR or sigma-delta is quite different - but you do need the digital decimating filter in the sigma-delta etc... Dave _____________________________ Posted through www.DSPRelated.com
Reply by ●July 4, 20132013-07-04
On Thu, 04 Jul 2013 14:09:36 -0500, gretzteam wrote:> Hi, > Is there a good book discussing the effects/advantages/disadvantages of > the *type* of converter used inside a digital control loop? Most books > on digital control will model the ADC/DAC as a black box with delay, and > usually do some analysis of the effect of the sampling rate chosen. > > I wonder if there exist something that goes a little more into the > details of the converters, like SAR versus sigma-delta etc, but with > emphasis on control loop design.Well, my book does. But if it spills off of the one page with the table I'd be surprised. The reason I didn't go into great detail is because I felt that I'd put as much detail in there as I could without having the book go obsolete in a half a year. Well, that and I'm lazy. I suppose I should have taken one or two data sheets and gone over them, to show people how to pick out the good stuff. Do you have a specific project in mind that you're trying to select an ADC for, which is causing your concern? In my experience, unless you're using a sigma-delta converter or closing a loop around something that's screaming-fast (like, loop bandwidths in the 100's of kHz or MHz), ADC delay and filtering effects aren't an issue. Noise is an issue, but in a screwy sort of way it can be your friend, by adding dithering which allows you to average out quantization noise. It depends on the ADC, and the application. (Sample your ADC as fast as you can, and average the resulting values down to the sampling rate of your loop. You'll pick up some precision, but not accuracy, and your loop will be quieter for it. The required filter also has nulls around all the frequencies that could alias down to DC, meaning the noise is lowest right where you want it to be low). I don't have to mess around much with DACs, because I nearly always use the PWM outputs of a micro to do that job. There's not many ways they can screw you up, unless you're using a sigma-delta (why, oh why, would you do that?). -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●July 4, 20132013-07-04
Tim Wescott <tim@seemywebsite.really> writes:> [...] > In my experience, unless you're using a sigma-delta converter or > closing a loop around something that's screaming-fast (like, loop > bandwidths in the 100's of kHz or MHz), ADC delay and filtering > effects aren't an issue.100's of kHz is screaming-fast? I guess this is like "how long is this piece of string". -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
Reply by ●July 4, 20132013-07-04
On Thu, 04 Jul 2013 19:59:48 -0400, Randy Yates <yates@digitalsignallabs.com> wrote:>Tim Wescott <tim@seemywebsite.really> writes: >> [...] >> In my experience, unless you're using a sigma-delta converter or >> closing a loop around something that's screaming-fast (like, loop >> bandwidths in the 100's of kHz or MHz), ADC delay and filtering >> effects aren't an issue. > >100's of kHz is screaming-fast? I guess this is like "how long is >this piece of string".For a loop bandwidth, yeah, that's enormous.>-- >Randy Yates >Digital Signal Labs >http://www.digitalsignallabs.comEric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●July 4, 20132013-07-04
>On Thu, 04 Jul 2013 14:09:36 -0500, gretzteam wrote: > >> Hi, >> Is there a good book discussing the effects/advantages/disadvantages of >> the *type* of converter used inside a digital control loop? Most books >> on digital control will model the ADC/DAC as a black box with delay,and>> usually do some analysis of the effect of the sampling rate chosen. >> >> I wonder if there exist something that goes a little more into the >> details of the converters, like SAR versus sigma-delta etc, but with >> emphasis on control loop design. > >Well, my book does. But if it spills off of the one page with the table >I'd be surprised. > >The reason I didn't go into great detail is because I felt that I'd put >as much detail in there as I could without having the book go obsolete in>a half a year. Well, that and I'm lazy. I suppose I should have taken >one or two data sheets and gone over them, to show people how to pick out>the good stuff. > >Do you have a specific project in mind that you're trying to select an >ADC for, which is causing your concern? In my experience, unless you're >using a sigma-delta converter or closing a loop around something that's >screaming-fast (like, loop bandwidths in the 100's of kHz or MHz), ADC >delay and filtering effects aren't an issue. > >Noise is an issue, but in a screwy sort of way it can be your friend, by >adding dithering which allows you to average out quantization noise. It >depends on the ADC, and the application. (Sample your ADC as fast as you>can, and average the resulting values down to the sampling rate of your >loop. You'll pick up some precision, but not accuracy, and your loop >will be quieter for it. The required filter also has nulls around all >the frequencies that could alias down to DC, meaning the noise is lowest >right where you want it to be low). > >I don't have to mess around much with DACs, because I nearly always use >the PWM outputs of a micro to do that job. There's not many ways they >can screw you up, unless you're using a sigma-delta (why, oh why, would >you do that?).The frequent misguided postings about control loops suggest you should have gone into the ADC + anti-alias filter tradeoffs in considerable detail. Steve _____________________________ Posted through www.DSPRelated.com
Reply by ●July 4, 20132013-07-04
On Fri, 05 Jul 2013 00:56:52 +0000, Eric Jacobsen wrote:> On Thu, 04 Jul 2013 19:59:48 -0400, Randy Yates > <yates@digitalsignallabs.com> wrote: > >>Tim Wescott <tim@seemywebsite.really> writes: >>> [...] >>> In my experience, unless you're using a sigma-delta converter or >>> closing a loop around something that's screaming-fast (like, loop >>> bandwidths in the 100's of kHz or MHz), ADC delay and filtering >>> effects aren't an issue. >> >>100's of kHz is screaming-fast? I guess this is like "how long is this >>piece of string". > > For a loop bandwidth, yeah, that's enormous.The fastest loop that I've ever closed around a mechanical assembly sampled at 9600Hz, and closed at something less than 1000Hz. It was a piezo stage. I've certainly closed faster loops than that in analog circuitry, but those don't count. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●July 5, 20132013-07-05
eric.jacobsen@ieee.org (Eric Jacobsen) writes:> On Thu, 04 Jul 2013 19:59:48 -0400, Randy Yates > <yates@digitalsignallabs.com> wrote: > >>Tim Wescott <tim@seemywebsite.really> writes: >>> [...] >>> In my experience, unless you're using a sigma-delta converter or >>> closing a loop around something that's screaming-fast (like, loop >>> bandwidths in the 100's of kHz or MHz), ADC delay and filtering >>> effects aren't an issue. >> >>100's of kHz is screaming-fast? I guess this is like "how long is >>this piece of string". > > For a loop bandwidth, yeah, that's enormous.Proof by assertion? Really, Eric, if you could present evidence not based simply on your arrogance, it would be more helpful. I'm not even a controls guy, but I worked with a loop bandwidth of around 100 kHz on a LLC resonant converter project at Eaton. -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
Reply by ●July 5, 20132013-07-05
On Thu, 04 Jul 2013 23:20:58 -0400, Randy Yates <yates@digitalsignallabs.com> wrote:>eric.jacobsen@ieee.org (Eric Jacobsen) writes: > >> On Thu, 04 Jul 2013 19:59:48 -0400, Randy Yates >> <yates@digitalsignallabs.com> wrote: >> >>>Tim Wescott <tim@seemywebsite.really> writes: >>>> [...] >>>> In my experience, unless you're using a sigma-delta converter or >>>> closing a loop around something that's screaming-fast (like, loop >>>> bandwidths in the 100's of kHz or MHz), ADC delay and filtering >>>> effects aren't an issue. >>> >>>100's of kHz is screaming-fast? I guess this is like "how long is >>>this piece of string". >> >> For a loop bandwidth, yeah, that's enormous. > >Proof by assertion? Really, Eric, if you could present evidence not >based simply on your arrogance, it would be more helpful.What's arrogant about agreeing with a previous post? Seriously, Randy, chill. I'm just indicating that my experience is in line with Tim's. If you have a problem with me, I suggest you stop reading my posts.>I'm not even a controls guy, but I worked with a loop bandwidth >of around 100 kHz on a LLC resonant converter project at Eaton.I still think it's unusual. Evidently Tim does, too.>-- >Randy Yates >Digital Signal Labs >http://www.digitalsignallabs.comEric Jacobsen Anchor Hill Communications http://www.anchorhill.com
