R-2R ladder DAC for controlling the gain of the conventional instrumentation amplifier

I'd call it more of a binary weighted feedback network, but yeah. If you can stand some code-nonlinearity then parallel banks can be better than serial switched links, put the switches against gnd / outputs so their C is up against a low-Z source.

dick_freebird said:

I'd call it more of a binary weighted feedback network, but yeah. If you can stand some code-nonlinearity then parallel banks can be better than serial switched links, put the switches against gnd / outputs so their C is up against a low-Z source.

Click to expand...

Thank you freebird for your answer,
yes it doesn't matter me the code non-linearity,

It will be helpful to make me understand your explanation by referring to the basic r2r ladder below and how to connect it with regards to the inamp. I am working with single supply operation

Thank you

There are digital programmable gain IA's, just one example :




Oer use digital pot :

https://www.analog.com/media/en/news-marketing-collateral/product-selection-guide/Choosing_the_Correct_Digipot.pdf

What G resolution do you need ?


Regards, Dana.

Dear Dana

Thank you very much for the resources you shared,

Actually, I am designing in the transistor level with CMOS technology. I also used digital pot in my design for the gain programmability, but as mentioned in my first post the number of the taps or switches will be equal to the number of gain setting I want to achieve.

for example if I have 16 gain level then I need 16 switches for each of R2, make them in total 32 switch. I have found out is affecting my AC performance.

Regardless this effect, and if I could use the r2r princible, I will achieve the 16 levels using only 4 switches. and if I use 6 bit r2r I will be able to chieve 64 gain level, making it the better programmable gain amplifier.

My problem is that to incorporate the r2r

Thanks

SPDT switches are generally used to illustrate the principle of R2R DAC. As you build a stand-in for switches it would be convenient if you can reduce parts count.
Since transistors conduct in one direction, you might get by giving each bit one switching device...
if you provide a reference voltage through adding another resistor. It's a workaround. Adjust reference voltage as desired.

My simulation has diodes conducting in one direction from each binary output (in order to depict a device which can source but not sink current). However if your R2R network can't be driven that way...
then you can substitute transistors and turn them On-Off via biasing signals from the bit wires. As an alternative reverse diode orientations so they point toward the binary outputs (in order to depict a device which can only sink current, not source current).

Hi,

May I ask what this "programmble gain" is used for.
In my many years of design I never needed that "high resolution" gain.

If we knew more about the signal source and where the signal goes to we maybe can give better assitance.

Klaus

R-2R ladder is specifically for a linear current-mode DAC
but that is not what you're building. Right?

You want a weighted ladder that gives you a "good enough"
outcome at the end of it all. Don't fixate on the form so early.
You don't care about code-linearity as long as you have the
step resolution across the span (which, worst-case bit bin size,
is maximized by dead linearity, but maybe 1 extra switch and
segment is better than trying more elaborate things).

Opportunistic things like turning a divider shunt R into a
parallalel shunt RDAC bank won't be easily linear, but can be
"plenty good enough" for next to no effort or risk.

Dear friends,

Thank you very much for your reply,

Beacause Klaus askem me about the application, I want to answer it here,

it is not for specific application rather than general purpose PGA. In my old design I have gain setting of

1,2,4,8,16,32,64,128

You know we amplify to fit the signal to the full-scale voltage of the ADC, so having more gain adjustment give me better control of the signal.

The r2r DAC can give me a finer step with fewer digital control switches/bits, but I am not sure in principle if I can use it as a voltage feedback attenuator to set the value of the feedback resistor and define the gain.

Thank you

Best Regards