R-2R DAC resistor network as part of op-amp feedback loop?

[jeff miller]

Hi, I want to make an op-amp circuit with digitally programmable gain
over a wide range. Is there an arrangement of the R-2R DAC resistor
network topology that will work?

Thanks!

-Jeff

 

[Tim Wescott]

Hi, I want to make an op-amp circuit with digitally programmable gain
over a wide range. Is there an arrangement of the R-2R DAC resistor
network topology that will work?

Thanks!

-Jeff

Do a web search on "multiplying DAC" -- that's essentially what's happening.

 

[John Larkin]

Hi, I want to make an op-amp circuit with digitally programmable gain
over a wide range. Is there an arrangement of the R-2R DAC resistor
network topology that will work?

Thanks!

Sure; just use it to replace Rf. Use a multiplying DAC for the same
effect but less work. Beware the added capacitance, though... this can
make an opamp oscillate.

John

 

[Fred Bloggs]

Hi, I want to make an op-amp circuit with digitally programmable gain
over a wide range. Is there an arrangement of the R-2R DAC resistor
network topology that will work?

It will not work out well for you. Use a digital pot(s).

 

[Larry Brasfield]

Hi, I want to make an op-amp circuit with digitally programmable gain over a wide range. Is there an arrangement of the R-2R DAC
resistor network topology that will work?

One option is to use a DAC having a R-2R ladder
and CMOS switches, but exchange the role of the
input network (usually the ladder) and the feedback
resistor (one end avalaible separately). This yields
a 1/N gain characteristic which can provide a better
distribution of resolution over the control range for
some applications.

As always with DAC circuits, you need to be sure
that stability is not a problem. And the app notes
will likely not cover that unconventional topology.

 

[Pooh Bear]

Hi, I want to make an op-amp circuit with digitally programmable gain
over a wide range. Is there an arrangement of the R-2R DAC resistor
network topology that will work?

You don't mention the end application.

I've used MDACS with excellent results.

If it's audio - you might find THD a bit high though. In that case I've
seen *discrete* R-2R networks used with jfet switching ( much more linear
). High component count though !

There are 'digital pots' available now too. Often something simple like
I2C control.

Supply more info on the app - and more ideas may result.


Graham

 

[jeff miller]

Sure; just use it to replace Rf. Use a multiplying DAC for the same
effect but less work. Beware the added capacitance, though... this can
make an opamp oscillate.

John

Hi,

Yes, I'm thinking about multiplying DACs as well. I'll probably try that
first.

I thought it was pretty obvious how to use an R-2R network in the
feedback loop: instead of putting logical 1's on the bit inputs, just
put the output of the op-amp. But a lot of references suggest that if
you don't put a 1 on an input, it must be tied to ground. So in this
case, where do I tie the "low" inputs? To ground? That would seem to
imply a digitally programmable woltage divider on the output, I don't
recall incorporating voltage dividers into feedback loops though I have
a vague feeling it will work. I used to be very handy with op amps,
guess I just have to put my thinking cap on.

I was just hoping there might be some clever configuration I was missing.

-Jeff

 

[Nicholas O. Lindan]

Sure; just use it to replace Rf. Use a multiplying DAC for the same
effect but less work.

Works great. Connect the op-amp output into the DAC's Vref input and the
(-) op-amp input into the DAC output.

This makes a good front-end to an ADC for applications with lots of
dynamic range: 12 bits of resolution with 18 bits of dynamic range
is easily achievable. Absolute accuracy, is of course, degraded.
Works best when tracking signal change/unit time or making ratio
measurements. It is possible to software calibrate the DAC across
gain steps.

 

[Jeroen Belleman]

Sure; just use it to replace Rf. Use a multiplying DAC for the same
effect but less work.

Works great. Connect the op-amp output into the DAC's Vref input and the
(-) op-amp input into the DAC output.
[...][/QUOTE]

Some years ago, I worked out the values for a resistor ladder that
affords logarithmic gain steps (linear in dB). I went as far as
making one with 8 gain steps (3 bits). It's only an approximation,
of course, with the steps accurate to 0.15 dB or thereabouts, but it
worked quite well up to 30MHz or so.

It didn't make it into the final design. Using an AD603 was easier.

Jeroen Belleman

 

[jeff miller]

jeff miller wrote:




You don't mention the end application.

I've used MDACS with excellent results.

If it's audio - you might find THD a bit high though. In that case I've
seen *discrete* R-2R networks used with jfet switching ( much more linear
). High component count though !

There are 'digital pots' available now too. Often something simple like
I2C control.

Supply more info on the app - and more ideas may result.


Graham

In fact I was concerned about THD of multiplying DACS, some of them
(mostly the ones not advertised as multiplying DACS, but which claim
"can be used as a multiplying DAC") seem to have really horrible
characteristcs. Others advertised as multiplying DACS seem to have "bad"
characterisitcs. I kind of extrapolated to assume that, especially for
low signal levels, multiplying DACs were not a good approach. To be
honest with you, I'm thinking reed relays as the switching elements.
Linearity and low noise of signal are important in my app., though
absolute accuracy of gain and linearity of gain are not.

-Jeff

 

[Pooh Bear]

In fact I was concerned about THD of multiplying DACS, some of them
(mostly the ones not advertised as multiplying DACS, but which claim
"can be used as a multiplying DAC") seem to have really horrible
characteristcs. Others advertised as multiplying DACS seem to have "bad"
characterisitcs. I kind of extrapolated to assume that, especially for
low signal levels, multiplying DACs were not a good approach. To be
honest with you, I'm thinking reed relays as the switching elements.
Linearity and low noise of signal are important in my app., though
absolute accuracy of gain and linearity of gain are not.

-Jeff

Likewise - for a digital controlled mic pre-amp I also concluded that reed
relays were the only suitable answer.


Graham

 

[John Larkin]

Likewise - for a digital controlled mic pre-amp I also concluded that reed
relays were the only suitable answer.


Graham

LTC and others have 16-bit 4-quadrant mdacs with superb specs,
certainly good enough for audio. And cmos switches could do coarse
gain switching with roughly zero distortion relative to what's
inherent in any microphone.

John