ADC for vibration controller

Curious of using delta-sigma ADCs for vibration control.  For instance, the
ADS1271 has a sample rate up to 102kHz, but has a group delay of 39
samples.  I'm probably missing something, but the group delay seems to make
it not viable for use as a controller of up to 10kHz.	 

_____________________________		
Posted through www.DSPRelated.com

Try ad7725, you can program the decimation filter to whatever you want. 

Bob

On Thu, 05 Dec 2013 09:17:23 -0600, smc123 wrote:

> Curious of using delta-sigma ADCs for vibration control.  For instance,
> the ADS1271 has a sample rate up to 102kHz, but has a group delay of 39
> samples.  I'm probably missing something, but the group delay seems to
> make it not viable for use as a controller of up to 10kHz.

What leads you to thinking that a sigma-delta ADC should be tried?  
What's the matter with a good modern 16-bit SAR sampling at 100kHz or 
faster?

-- 
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

>Curious of using delta-sigma ADCs for vibration control.  For instance
the
>ADS1271 has a sample rate up to 102kHz, but has a group delay of 39
>samples.  I'm probably missing something, but the group delay seems t
make
>it not viable for use as a controller of up to 10kHz.	 
>
>_____________________________		
>Posted through www.DSPRelated.com
>

From a math point of view, I would imagine you could account for thi
latency in your system transfer function and determine if you still hav
proper gain/phase margins........	 

_____________________________		
Posted through www.DSPRelated.com

On Fri, 06 Dec 2013 09:17:57 -0600, jacobfenton wrote:

>>Curious of using delta-sigma ADCs for vibration control.  For instance,
> the
>>ADS1271 has a sample rate up to 102kHz, but has a group delay of 39
>>samples.  I'm probably missing something, but the group delay seems to
> make
>>it not viable for use as a controller of up to 10kHz.
>>
>>_____________________________
>>Posted through www.DSPRelated.com
>>
>>
> From a math point of view, I would imagine you could account for this
> latency in your system transfer function and determine if you still have
> proper gain/phase margins........

If you want to achieve decent loop closure from DC to 10kHz, you couldn't.

-- 
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

Is it correct to assume that in this application aliasing is not really an issue? If not then as Tim suggests you can't beat a SAR. The sigma-delta part I mentioned has downloadable FIR coefficients so you could load it with an iir- like truncated impulse response for min delay. 

Bob

Thank you for the responses.  I agree that the 16 bit SAR would be a good
choice.  I was curious about the delta sigmas because the data sheets for
some vibration controllers claim they achieve 120dB dynamic range using
delta sigma converters.  I can see a control loop of a few hundred Hz with
these, but  some claim to go to 5kHz.	 

_____________________________		
Posted through www.DSPRelated.com

Yes, of you want those kind of dynamic range numbers then sigma-delta is the way to go. If you have control over the decimation filter then you can optimize the filter for low group delay.  If you know what you're doing you can co-design the decimation filter along with the overall feedback control filter so that any large peaks in the group delay of the decimator will happen when the loop gain is less than 0 db. The trade- off is that some of the sigma-delta noise might leak through at high frequencies. Whether or not this is a problem depends on the application. 

Bob

On Sat, 07 Dec 2013 10:49:09 -0600, smc123 wrote:

> Thank you for the responses.  I agree that the 16 bit SAR would be a
> good choice.  I was curious about the delta sigmas because the data
> sheets for some vibration controllers claim they achieve 120dB dynamic
> range using delta sigma converters.  I can see a control loop of a few
> hundred Hz with these, but  some claim to go to 5kHz.

We are talking about controllers for shaker tables, yes?

If the plant is well behaved (which mostly means that it's good and 
linear) then you could use a sigma-delta in an adaptive loop that 
acquires the table transfer function and then essentially drives it open-
loop to get the desired result.

This would work particularly well if the user could stand having a 
calibration step before the "real" shaking, and if what was being 
promised was that the power spectral density of the shaking, but not 
necessarily the time-domain waveform, was on the money.  You could 
probably still approximate the time-domain stuff reasonably well, though.

I'm kind of dubious about that whole 120dB dynamic range, unless they 
mean they can do 10dBg (how does one measure vibe in dB, anyway?) at one 
point, and then -110dBg later on.  Even then the shaker table would have 
to have a very good suspension system -- if the bladder were made out of 
the wrong material, or if it were aged too much, then the thing may stick 
in one place when you're trying to achieve the low-amplitude vibe, even 
if it rattles enthusiastically enough at the high amplitudes.

-- 
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

Yes, it is in regards to shaker tables.  Thank you for the additional
insight.  I haven't got a specific requirement.  I was looking to construct
a simple system, but got curious about the descriptions of systems with
high resolution claims.  	 

_____________________________		
Posted through www.DSPRelated.com