ADS1248: ADS1248 problem

Thanks for the ideas, I am still trying to come up with the best solution.  On the one hand, power the strain gauge off of ±2.5V supplies and powering and referencing the ADC on the same ±2.5V rails would help drift and take care of the nominal near zero negative voltage that could appear.  Just seems expensive to have to do a bipolar rail configuration for just a few mV negative condition that will only be there in a no load situation.  Again, I am setting the ADC up in differential mode and my nominal gauge voltage will be VEXC/2  but could be slightly negative due to balance problems in the strain gauge at time off build.  Does TI have a trimmer circuit application note(s) for using a rail to rail opamp to trim out a slightly negative input voltage? It just seems it should be easier than this to force an always positive voltage off of the strain gauge configuration.  I do not think I have to run a lot of gain on this system with 24b resolution, so if I could always make sure my voltage was say around 200mV I would be ok with staying away from a 0 - 5V rail and not violating the 100mV limit and not introducing a negative voltage rail with a noisy DC to DC converter stage or killing half my range to a slightly negative voltage.  comments?  Ideas?

Hi Greg,

Maybe I misunderstood, but if you said you were running your ADS1248 off of +/-2.5V supplies in your initial post. Why could you not use these same supplies to power the bridge? I am not sure where the added cost would come from.

-Bryan

sorry for the confusion, they would be a whole lot easier on the phone.

If I power the bridge from 5V and 0V I get a nominal 2.5V output on each output of the wheatstone, I could then power the ADC from the same 5V and 0V supplies.  That is fine, BUT if say the negative side of the differential input to the ADC is more negative than the positive side, say the positive side reads 2.495V and the negative side reads 2.501 then will not the ADC see a negative voltage of -0.006V?  And that is not only below the 0V rail but within the 100mV rule of the supplies.  or am I making this too hard?  It would seem that I would need to level shift the voltage up so it is one: not negative and two: further than 100mV from the 0V rail.  I could do this with a level shift opamp and go single ended into the ADC.  If I do that do I compromise performance?  Do I need to be 100mV above the 0V reference at all times?

my expense comment was due to the noise a DCDC converter would add to the circuit and loosing nearly half the ADC range for a tiny negative offset at no load.

i am thinking a circuit like this may be applicable where I take the differential of the wheatstone bridge and convert it to single ended and use the REF pin to inject an offset to force an always positive voltage.  i would not need the range this has so the trim coming into pin 5 would need to be limited to say 200mV

Hi Greg,

If the differential voltage is negative, that is okay. In fact, the ADC has a bipolar coding scheme, so if you don't use the negative code range on the device you are not using half of the ADC's available codes (this is common for bridge measurements). It is the absolute voltage that you apply to the input pin that must be within 0-5V (if you use a 0-5V supply).

In the case you described in your previous post, both input voltages are positive with respect to AVSS (2.495V and 2.501V). Powering the ADC and the bridge from a unipolar 5V supply is acceptable, since the bridge common-mode will be ~2.5V at all times, which is what the ADC's PGA expects. This is a very common application for the ADS124x products.

Let me know if this is clear.

-Bryan

so I was making it too hard.  figured so.  So, I am going to use a 5V and 0V supply and excitation for the bridges and tie the same 5V and 0V to the +REF and -REF pins on the ADC.  I think it is ok to use an analog excitation that is higher than the digital VDD.  I am running my digital side at 3.3V.

Does TI have any application notes that site the ADC1248?  

Hi Greg,

One point about the ADS1248 specifically: this device has limitations on the external VREF inputs such that you cannot apply a voltage up to the supply. See table 7.3 for more information, but there is basically a 1V headroom requirement, which would make it impossible to use the excitation voltage as the reference.

If you are still able to select an ADC, the ADS124S08 is the next-generation version of the ADS1248. The ADS124S08 does not have this VREF limitation, so you can power AVDD, REF+ and the bridge with the same 5V excitation voltage (just make sure it is a clean supply voltage). The ADS124S08 is also smaller, lower power, faster, lower noise, includes more features, and is offered at a lower price point.

Let me know if this an option for you. If not, we can try to find some way to use the ADS1248, but a true ratiometric measurement would not be possible. You can check out this app note for more general information about using delta-sigma ADCs to take bridge measurements.

-Bryan

Specifically, I run the +REF at 5V the -REF at 0V DVDD at 3.3V and AVDD at 5V tying DGND and AVSS to the same 0V ground that -REF is at a single point that is the ADS1248 ADC chip.  I think this configuration would be ok.

Hi Greg,

Unfortunately this will not work. As you can see below from Table 7.3, the differential reference voltage must be 1V below the difference in supplies. So if you have AVDD = 5V, AVSS = 0V, then VREF (diff) max = 4V. So you could not then apply a 5V reference voltage to the REF pins.

You could always use the internal VREF of the ADC, but this will not offer better performance compared to a ratiometric reference. I will reiterate that the ADS124S08 does not have this limitation, and is more ideal for ratiometric bridge measurements, in case that is still an option. But I understand if you are too far along in your design to make a device change at this point.

-Bryan

yes I saw that as well in the data sheet and was concerned about it after I sent the post.  I will look at the other chip.  I have not done allot of work on the project as of yet.

Could I just do a voltage divider on the 5V that is running the strain gauges and get the reference down to 3.5V?  Then it would track changes in the system and I think minimize error.  

sorry I responded to the wrong section of the post  hope you still find my response.

ok thanks Brian.  I looked the chip up you mentioned earlier and it is shown obsolete on DigiKeys website now.

How should I proceed on a new thread title it 'ADC recommendations for ratiometric strain gauge application'

Hi Greg,

Which device did you look up? The ADS124S08 is here: https://www.digikey.com/en/product-highlight/t/texas-instruments/ads124s08-adc

If you looked up a different ADC, please let me know.

If you are going to consider this device, I would suggest getting an EVM and testing it out. The connections are fairly straightforward for a ratiometric bridge measurement as shown below (this is from the ADS1220 datasheet).

Then, if you have additional questions, please post them on our forum and we can answer them.

-Bryan

thanks, I was looking up the differential input version of the chip you recommended, it was obsolete, the single ended versions were available.  I am thinking that straight differential in is better?  That is why I was steering that direction, I asked a related question on the forum to kick off a better solution.  it is my understanding that the related question will link back to this thread for reference.