THD and DC performance

Here's 4 PPM

in a "24 bit" dac. You could do some software predistortion to get that down, conditions permitting.

There are some claimed "32 bit" audio dacs!

John

Look thru the Cirrus logic web site

Cheers

"Ban"

So I am not the only one to who can see the BLEEDING OBVIOUS !!!

... Phil

I have thought of audio DACs. Unfortunately, they have abysmal performance as far as zero shift and gain accuracy. I will probably have to calibrate those parameters at every measurement; that will slow down the things considerably. As for THD, a good analog filter is would be better for this application. The 32-bit interfaces to audio DACs/ADCs are quite usual, leaving the 32-bit performance claims on conscience of marketing.

Tandeming two DACs (one for DC, the other one for AC) could be an option, however it looks like a heavyweight solution; I would rather do everything in one DAC.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Unfortunately, audio DACs and ADCs have poor accuracy of zero shift and gain; because those parameters don't matter for audio applications. Otherwise there wouldn't be a question.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

I use 16 and 20-bit audio-type DACs in NMR gradient drivers, where a few PPM of zero offset gets attention. They seem to be pretty good at DC, except that they don't have super-good internal references. I fix that by ovenizing them. You just have to experiment when the parts aren't spec'd for DC performance.

But how can you measure 1 PPM distortion? And where does it matter?

John

if you think an active analog filter will give you -120dB THD performance, go on dreaming. Some fixed frequency LC filter maybe.

I have seen better posts of yours. ciao Ban

State of the art geophysic instruments specmanship is ~130+ dB of SNR and ~120+ dB of THD (in the bandwidth of few hundred Hz). That's true numbers; we have them already. However I need to do a self test circuit that could confirm the performance numbers as well as a bunch of other parameters. It doesnt't have to be super accurate; it has to be simple. Board space is a premium, too.

VLV

But that is a totally different piece of cake. Neither you need -120dB THD nor 10uV accuracy nor 20bit DAC, you just need a stable passive attenuator and reference. ciao Ban

Maybe:

Their Geophysical parts are awesome (I've characterised them beyond their datasheet specs):

Dave.

Pretty much, welcome to the world of Geophysical measurement. I spent more than decade in the industry, and I can assure you the specmanship and brandmanship is real and *must* be met, otherwise you can kiss your contracts bye-bye.

Analog Device are or were in the Geophysical sector too, but because Cirrus Logic have the rep with the grey beard crowd, your products can also live or die by which brand you choose.

I'm surprised Vladimir is asking about this stuff, as usually the decison is already made for you :->

And I'm not sure why he needs several kHz bandwidth either, it's usually several hundred Hz for this stuff.

Power consumption is the main killer in geophysical ADC/DAC systems, as you have to typically power many thousands of channels over many km of cable.

Dave.

Vladimir wrote, "However I need to do a self test circuit that could confirm the performance numbers as well as a bunch of other parameters." I assumed the why of it is to be found in some interpretation of "self test" and "confirm the performance" parts. But you probably know how to "read" into that a lot better than me.

Jon

That implies an ADC good to better than 1 PPM. Do they exist?

John

The ADCs with that kind of performance are available. Look at ADS128x series from TI. However, it doesn't really require 1ppm performance: the fundamental frequency can be notched out by a good analog filter; or lock-on amplifier techniqies can be used.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Good point.

(speaking about THD)

Another good point.

I can pre-calibrate DACs for DC and THD compensation wrt temperature, however I am not certain if this calibration will last.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

us

or

is

ou

log.com

I noticed Cirrus has a line of parts for geophones, but the chopper amp isn't any better than the LT1128, and the gain seems to low to be useful. And the same line of parts is supposed to do hydrophones too, which are very different beasts.

"John Larkin" "Phil Allison"

..... Phil

About distortion? Pretty much nothing.

John

In the seismic industry it's common (and pretty much standard actually) to have a test DAC associated with every ADC channel in the system for self testing when deployed in the field. If you've got a 5,000 channel system, you need 5,000 test DAC's too. That's why Cirrus offer combined ADC/DAC's like the CS5373A.

Test signals are typically less than less than 200Hz IIRC, but Vladimir may be working on a different beast with more unique requirements.

The "bunch of other parameters" would usually refer to other system parameters like DC offset, hydrophone capacitance, noise floor, and a few others I can't recall. THD is just another one of these standard tests run on every channel in the system before they start aquiring data. The customer demands (and often supervises) such tests in the field to ensure the quality of the seismic data.

Dave.