Looking for an IC or module for instantaneous floating point amplifier

Thread Starter

yanjunyang

Joined Mar 25, 2020
4
Hi, I am planning to design a dynamic signal acquisition system, which requires an instantaneous floating point analog amplifier, so a gain set up is no longer needed. anyone can recommend a IC chip or commercial module that I can buy?

Thanks!


Yang
 

OBW0549

Joined Mar 2, 2015
3,306
Hi, I am planning to design a dynamic signal acquisition system, which requires an instantaneous floating point analog amplifier, so a gain set up is no longer needed. anyone can recommend a IC chip or commercial module that I can buy?
I've been designing for many years, and I've never heard of any such "instantaneous floating point analog amplifier." The concept makes no sense, as "floating point" is clearly a digital concept and amplifiers are inherently analog. Without taking a wild guess, I have no idea what you're talking about. None.

Instead of inventing mysterious terms like "instantaneous floating point analog amplifier," try describing, in basic terms, exactly what you would want this circuit to do. Perhaps then we could help you.
 

Papabravo

Joined Feb 24, 2006
13,533
Many 1st time posters come on with requests of a phantasmagorical nature. Ho-hum. Maybe the real culprit is bad translation algorithms.
 

Thread Starter

yanjunyang

Joined Mar 25, 2020
4
Hi, I am planning to design a dynamic signal acquisition system, which requires an instantaneous floating point analog amplifier, so a gain set up is no longer needed. anyone can recommend a IC chip or commercial module that I can buy?

Thanks!


Yang
I've been designing for many years, and I've never heard of any such "instantaneous floating point analog amplifier." The concept makes no sense, as "floating point" is clearly a digital concept and amplifiers are inherently analog. Without taking a wild guess, I have no idea what you're talking about. None.

Instead of inventing mysterious terms like "instantaneous floating point analog amplifier," try describing, in basic terms, exactly what you would want this circuit to do. Perhaps then we could help you.
Thanks for your reply! Please allow me to try to explain in another way.

For a A/D converter, its best performance occurs when an input of analog signal falls in a range, where the signal amplitude is smaller than the full range of the A/D converter and larger than half that range, if so, the best accuracy can be reached.

In every time cycle of A/D conversion, the gain of amplifier can be automaticaly set in order to let the analog input signal amplify appropriately, and the set gain is recorded as "floating point number" (like digital binary number) that comes together with the results of A/D conversion.

What I try to find is this type of amplifier as the form of commercial IC or module.

Thank you very much!

Yang
 

MrChips

Joined Oct 2, 2009
20,904
You are attempting to optimize the resolution of the ADC, i.e. get more bits out of the ADC.
Have you looked at a sigma-delta ADC?
 

Thread Starter

yanjunyang

Joined Mar 25, 2020
4
it is not only optimize the resolution, but also automatically gaining, means that you don't need to set amplification each time when you start signal collection, because some time you have no idea about the signal amplitude.

A sigma-delta ADC can still not meet my expectation, because of a signal-noise ratio, if the signal is too small, it might sink into the noise background.
 

Analog Ground

Joined Apr 24, 2019
364
Last edited:

Thread Starter

yanjunyang

Joined Mar 25, 2020
4
it is not only optimize the resolution, but also automatically gaining, means that you don't need to set amplification each time when you start signal collection, because some time you have no idea about the signal amplitude.

A sigma-delta ADC can still not meet my expectation, because of a signal-noise ratio, if the signal is too small, it might sink into the noise background.
A "programmable gain amplifier" (PGA) is used to optimize the dynamic range of an ADC. The gain of a PGA is set to keep the full range of the analog signal near the full range of the ADC. Is this what you have in mind? If so, what is the frequency range? How fast is the ADC?
I know PGA, that is not what I want.

what I expect is that the amplifier can adjust its amplification with respect to the input analog signal amplitude for each cycle of AD conversion.
 

MrChips

Joined Oct 2, 2009
20,904
You are confusing gain with dynamic range.
Suppose that you have an input signal of 9V being digitized by a 10-bit ADC with a reverence voltage of 10V. How can you improve on resolution by changing the gain?

You need to know the dynamic range of the input signal before you can apply an AGC block.

A sigma-delta ADC achieves higher resolution by subtracting the sigma signal so that you are now digitizing only the delta.
 

Analog Ground

Joined Apr 24, 2019
364
I know PGA, that is not what I want.

what I expect is that the amplifier can adjust its amplification with respect to the input analog signal amplitude for each cycle of AD conversion.
I think I understand what you want. The closest thing would be a circuit with a PGA which also has a lower resolution but faster ADC to control the PGA before each cycle of the main ADC. No one does this type of processing (that I know of). The approach is to use a higher resolution ADC.
 

MrChips

Joined Oct 2, 2009
20,904
Let me put it to you differently with what you have in mind.

Suppose your input signal is 10mV into a 10-bit ADC with a full-scale input of 10V.
Of course that would only register as a count of 1.
How much would you amplify the signal?
Why don't we amplify the signal by 1000 so that we get 10V and make maximum use of the ADC's 10V range.
In other words, we amplify the signal with just the right gain so that we just reach the maximum range of the ADC.

That is what happens in a sigma-delta ADC. It uses a 1-bit DAC that essentially maximizes the range of a 1-bit ADC.

Look up delta-sigma modulation.
https://en.wikipedia.org/wiki/Delta-sigma_modulation
 

danadak

Joined Mar 10, 2018
4,057
What is the needed sample rate of the signal input ? Its possible to do a
simple two conversion sampling such that the first conversion is used
to set the PGA G and then next conversion is result of new G setting.
This would implement a form of AGC, easily done in any micro that has
both PGA and A/D. Note of course there is a latency involved, but that
can be minimized with a SAR for AGC setting and a DelSig for higher
res/accuracy of final signal digitized to float.

An example, single chip, IDE and Compiler free, dev Board to use $10.

Note most of chips resources unused, see right hand window.


1585345596914.png


Basically SAR looks at signal level, result modifies PGA G, then DelSig is triggered to do
the hi res conversion. Again, one chip. I estimate a max of 15-20 lines of code would handle
this. Keep in mind this is a chip with Analog + Digital + ARM core in it. The SAR, DelSig, PGA
are called components, in this chip these are internal resources. A list of components internal
for this chip in attached file.

Note the DelSig has a PGA in its front end, with G's available of 1,2 4, 8. Whereas the PGA_1
has Gs of -

1585346486854.png

The G accuracy of the DelSig internal PGA is capable of 20 bit accuracy, whereas the
PGA_1 is on the order of 1 - 2 %. You could add an internal analog mux to design
and do a cal routine. to error correct for PGA_1's error. The onboard reference for
the A/D's is good for +/- .1%.

Note both A/Ds have ability to operate differential, to help getting rid of common mode
signals, if so desired. One would add another PGA internal to handle the other diff channel.
also onboard chip. Although the PGA to PGA G variance would substantially lower the
accuracy of the solution, and the common mode behavior of the diff signal path.



Regards, Dana.
 

Attachments

Last edited:
Top