ADS8588H: Datasheet Questions: channel-to-channel match, external reference input range and conversion influence

Hi Sebastian,

For point one - if you were looking for the extreme worst case scenario, yes that is correct.  Having both MAX gain error and MAX matching error would be a very rare occurrence.

For point two - your opinion is valid, ideally we would expect an external 2.5V reference with at least +/-1% tolerance, the internal reference is +/-0.1%

For point 3 - when you look at the X scale of Figure 7-15, it is in LSBs with the Y scale in codes.  The height of each step is 1 code and the length of each step is 1 LSB.  Your 10V input example would have an LSB size of 10/65536 or about 153uV.  Internally, the conversion is based on +/-Vref so its actually 2x - at the 20V range, it would be 4x.

Hello Tom,

Thank you for your answers, this helps.

According the matching errors, is there any information on how rare this occurence could be?

During the reading of the Datasheet, I found another topic, you may be able to clarify:

The parameter "Input leakage current" mentions unter Test conditions only AIN_nP and refers to a formula: (V_IN - 2 ) / R_IN

• I suppose, the value 2 is in Volts. Else, the units won't match.
  But can you explain, where it comes from? I would like to understand the value of 2V.
• does this leakage current also apply to the AIN_nGND. The datasheet mentions on several places that the external impedances on both pins (P/GND) should be matched to help cancel any additional offset error contrbuted by the external resistance.
  For me, this matching can only be useful, when the Input leakage current is on both pins.

Hi Sebastian,

The Min Max and typical numbers in the datasheet are based on a gaussian distribution, so having two parts at the extreme limits is not likely to happen.  I'll see what I can find out about the leakage currents.  I suspect the equation is supposed to be (Vin/2) / Rin where Vin is the FS range.  The AINn pin is normally grounded, so leakage should not be an issue.  If you are doing an LPF filter as shown in Figures 7-3, 8-1 or 8-2, matching on the external resistors becomes critical.

Hi Tom,

when the Min Max numbers are based on a gaussian distribution, what is the cofidence level? Some of the histograms would let me assume, that it is +/- 3sigma while some other could point to +/-6sigma. 

Do you check if the leakage current is as you mentioned? (Vin/2)/Rin
For me too, this is a formula that fits better. It point to the direction, that the leakage current is dependent of the applied common mode voltage
But I dont think, that Vin is the FS range. I rather think it is the acutal value of AIN_nP - AIN_nGND. I wouldnt make any sense, to give an input-dependent formula for the leakage current. The variation of the current comes with the 15% tolerance of Rin...

Hi Sebastian,

They are 6 sigma.  I'm still waiting on a complete explanation of the leakage issue.  Apparently the equation is correct and its a subtraction of 2V from Vin.  I'll let you know when I have more details.

Hi Tom,

sorry, if I am annoying. 

To clarify the 6 sigma.
Is it either 6 sigma "to the left and right" from the peak value?
Or is it a range of 6 sigma meanig +/-3?

This picture shows a range of 6 sigma.
I hope it is understandable.

Nevertheless, I you provided me with the necessary information and I am able to do a reliable Analysis.

I am waiting for the more details, but this does not block me.

Thank you.

OK - on the leakage current, I still need to clarify a couple things with the design team.  I'll get back to you ASAP.

 The leakage current will be (VIN – 2V)/RIN, the RIN value is the typical Rin listed in the datasheet.