DAC80508: Questions about the datasheet:

Erin,

Thanks.

I thought of using internal Ref due to its better stability, but this would ask for "1/1" setting. 

What does "slightly more accurate" actually mean?

So, what you are saying, I cannot simply connect 5V to the Ref input as I need to turn the Internal Ref Off first (which is what I suspected; it would be nice for it to be clearly worder in the Datasheet). And for me to turn it Off I need to power up and write to it. Meaning, I have to have switchable 5V external reference (like a P-MOSFET or something) to apply it only after having the internal Vref turned off? This makes me really want to know about the "slightly better" part, as using internal Vref would make it so much simpler, should the accuracy be acceptable.

PS. I have just found this discussion from 5 years ago. It clearly states that the chip was designed with the internal Ref usage in mind

e2e.ti.com/.../dac80508-using-an-external-reference-with-the-dac80508

(QUOTE:

"Finally, we do not really expect external references to be used on this device (though it is possible).  We invested a lot of design time to make the reference of the DAC80508 very good (with low drift, noise).

In addition, the REF-DIV and GAIN features gives the output range of 1.25V, 2.5V, and 5V, so that covers most ranges for low voltage applications.  

I suspect that if you are looking at an external reference, then you either want a very specific output range (eg. 4.096V, 3.3V) or you want even better drift performance (REF6225).).  

But if I want 2.5V and 5V outputs using the internal Vref, I have to use 1/1 setting. Was it not "not recommended" 5 years ago?

MIke

Hi Mike,

I found a more complete answer to the DIV/GAIN question, please see this thread: (+) DAC8568: DAC80508 setting DIV=1 & GAIN=1 are not recommended, why? - Data converters forum - Data converters - TI E2E support forums

1/1 is not recommended specifically for lower VDD voltages, as VREF can run into headroom issues when VDD is too low. Dividing VREF by 2 helps to eliminate these issues with VDD.

If your VDD input is sufficiently high (the thread recommends at least 5.5V), then you can use the 1/1 div/gain case with no issues. 

I agree that this would be a lot easier than using an external reference. You would unfortunately have to implement either a timing scheme or a switch to ensure that you have time to disable the internal reference voltage before you apply the external voltage.

Thanks,
Erin

My Vdd is 5V, and it is used all around the analog portion of the device, and in some devices (I also have DAC128S085 where I use it as both the Va and the Vref), so I would have to implement a separate 5.5V (the thread you showed says 5.2V is also OK) power supply.

Does this also mean that it is preferred that the Vdd is about 0.5V higher than the Vref? Is it also the case withh the external Vref if it is 5V?

Hi Mike,

Since the internal reference is 2.5V, there should be enough headroom at 5V VDD. A concern with 5V VDD is that when you are using the 5V output range, the DAC output buffers will be limited by VDD. Note the Output Headroom spec in the datasheet shows the required VDD headroom at different loads. If you intend to use the 5V output at full-scale, you may want to consider increasing your VDD voltage to accommodate your expected current load.

For an external reference of 5V, you would not require the any headroom according to the recommended operating conditions table. VREF can be up to VDD as long as you are using the reference divider. 

Thanks,
Erin

Thanks Erin.

I have searched some other chips from various manufacturers, including TI, and this seems to be the only one with 16 bits, 8 channels, simple enough control and configurable per-channel output range, so I think I will make a separate Vdd 5.5V for this chip, plus a 3.3V for Vio, and I will use the internal reference to achieve range of 2.5V and 5V for various channels.
If this sounds correct, I think this does it for me then.