This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DAC8820: Questions about Voltage Reference in 4-Quadrant Operation

Part Number: DAC8820
Other Parts Discussed in Thread: REF5010, OPA227

I am looking at Figure 42 from the datasheet, the EVM, and this Reference Design.

First, let me confirm my understanding on the operation. (Voltage Mode - 4-Quadrant Mode)
A reference voltage is applied to R1 and Rofs pins. (i.e. +10V)
Applying the +Vref voltage to Rofs allows the MDAC to be operated in 4-Quadrant mode; allowing +/-VREF output range.
An opamp connects to R2 and Vref and uses the MDAC internal resistors in an inverting buffer configuration that generates an inverse reference voltage to Vref.

I see from the RD documentation the "A1" reference buffer (circled in yellow below) is required if the RC filter capacitance on the output of REF5010 is "excessively large"

 From the Datasheet:

A1 is responsible for buffering the reference output, which is required with larger filter capacitors.
The buffer prevents the RC filter from being loaded by the feedback network of A2

From this statement, it looks like there could be an application that might not need the A1 reference buffer opamp.
I also understand the input impedance at VREF is constant; non-changing like in a SAR-ADC. (Reference Video from TI see 6:12min)
From that, it seems reasonable there is very little step-load or transients at R1 or Rofs requiring an input buffer.

So as long as the RC filter isn't too strong that it starves the MDAC of current, I don't think we need A1.
Is my understanding correct?


  • Hi Darren, 

    You're right, the impedance seen by VREF is constant, the output impedance is what changes on an MDAC. The VREF input impedance is 6kΩ typical, so if your reference has no issues driving this (along with your comments about the filter) then the buffer is not required. 


    Katlynne Jones 

  • Darren,

    I was looking at this post when Katlynne answered it. One thing that I would point out is you should make always make the calculation for the amount of current going into the device. Noting that R1 and Rofs are typically 12kΩ and the reference is 10V, you might expect about 1.6mA of current going into the two nodes.

    If your R from the filter is 10Ω, then you might drop a 16mV across the resistor, dropping the apparent reference voltage from 10V. If the R from the filter is 1kΩ, then the drop across the filter resistor is closer to 1.6V, which would give you a huge gain error.

    Joseph Wu

  • Hi Joseph,

    Just trying to clarify my understanding of your comment above.
    I'm looking at the Reference Input characteristics on p3 of the datasheet...

    Current into R1 Pin
    R1 and R2 resistance is listed as 12kΩ (typ).
    Is this the resistance of the internal resistors? (See Table 2 in TIDU964)

    If R1 and R2 were 12kΩ each, and the series resistance was 24kΩ; 
    then with +10V at R1 pin and -10V at Vref pin, the current is (+10V - (-)10V) / (24kΩ) = 0.8mA;
    and this current is being pulled from the reference into the opamp output (sink)

    Having -10[V] at Vref would also draw current through the internal R-2R resistor network...but doesn't matter for the reference.

    Current into Rofs Pin
    Basically the Rofs and Rfb resistors act the same way; they are part of the feedback network for our transimpedance amplifier.
    If Rofs = +10V and our trans. amplifier was outputting -10V, then we get the same 0.8mA as above.

    Total Current Draw from Reference Source
    So the max current draw would be 0.8mA + 0.8mA = 1.6mA being pulled from our reference; as you mentioned above.
    Did I get it right? (well, not max because I'm using typ. resistor values, but still)

    In this case, if a filter is desired for the reference, as you mentioned, depending on the resistance you are looking at a huge error.


    If all my understanding above is correct, then couldn't we use a cap between R2 and VREF pins (TI reference example) to create an active filter with A2 opamp and filter Vref voltage?

    This would give us a nice, filtered Vref voltage, but Rofs would still be susceptible to noise...thoughts?

  • Darren,

    The values that you mentioned for the resistors are correct (they are also listed in the datasheet as 12kΩ typical). With those values, you would see the currents that you mentioned as well.

    As for the filtering, I guess you could use a cap across R2 for filtering and that might give a moderately clean signal at VREF. However, you would still need the input for Rofs and that should also be a filtered VREF for the bipolar offset. In the end, I think that using the OPA227 at A1 would still be the easiest solution.

    Joseph Wu