DAC53204: About IOUT tolerance (min/max)

Hi Makino-san,

The following specifications contribute to the total accuracy of the device when in IOUT mode:

You can use these to estimate a total unadjusted error which will give a very conservative estimate for the worst case error. TUE is calculated by the RSS of the INL, gain error and offset error. I do not have an absolute min/max accuracy spec to share because this was not characterized for this device, but the TUE will give a conservative estimate. 

Best,
Katlynne Jones

Hi Jones-san,

Thank you very much for your kind reply.

I would like you to answer the following additional questions.
-----
1. Can the worst tolerance of the DAC output be estimated in TUE?
2. I think the worst TUE in the IOUT mode is about -0.3%~+0.4% in Figure 6-38.
    Therefore, is the worst tolerance of IOUT mode about -0.3%~+0.4%?
3. Even if there is a variation in parts, is the worst tolerance of IOUT mode about -0.3%~+0.4%?
-----
I would appreciate it if you could answer.

Best regards,
Hiroyasu Makino

Hi Makino-san,

1. Yes, TUE is the best estimate. The alternative would be to add up the worst case errors for each spec, but because these specs are uncorrelated adding up the errors would give a much worse than would ever be seen in the device. 

2. The typical plots show the data measured for one device. This plot helps show the variation of the TUE for one device over temperature which certainly helps understand how much TUE can fluctuate over temperature.

3. Calculating TUE with the RSS like I mentioned in the last gives ~±1.6% error (the gain error is the highest contributor). If you take this as worst case, with the potential for ±0.4% additional error over temperature, you can estimate a worst case error over temperature of ±2%. With that being said, most devices will likely have a TUE closer to the plot in 6-38. 

Best,

Katlynne

Hi Makino-san,

I calculated the TUE by sqrt(INL^2 + GE^2 + OE^2). You need to convert the INL into %FSR first which comes out to 1LSB/2^8 = 0.39%. TUE = sqrt(0.39^2 + 1^2 + 1.3^2) = 1.68%.

The INL is a min/max spec, so this already includes the temperature variation, but this is such a small contributor to the overall TUE so this shouldn't throw off the results of the part to part TUE by too much. 

Best,

Katlynne