Settling time for multi channel measurements

[EricLarsen]

I'm hoping somebody can explain what the multichannel settling time specification means.  I'm using an NI 9205 in FPGA Mode, with a range of +- 10V.  According to the spec. sheet, for a 4 microsecond interchannel delay, the settling time accuracy is "+- 120 ppm of full-scale step, +- 8 LSB". 

If I have a 5 volt signal before a 0 volt signal in the scan list, what level of ghosting would I except to see on the 0 volt channel?  According to my probably wrong calculation, for a +-10 volt range, 120 ppm should be about 2.4 mV.  But I'm seeing about 60 mV ghosting instead. 

Either I'm not understanding the specification, or something else is wrong.  Anybody have any ideas?

 

[smithd]

You could trigger manually as shown here:

http://zone.ni.com/reference/en-XX/help/373197L-01/criodevicehelp/conversion_timing/

with a big delay between samples, and see if that improves at all. If you are using scan mode, you can change the min time setting here: https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z000000kGRtSAM

also please verify you have the module grounded: https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z0000019KrCSAU

and also if you are using scan mode: https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z0000019LBlSAM

And you should also verify that you have the module configured to single ended or differential as appropriate for your source.

Edited October 4, 2018 by smithd

[EricLarsen]

The ghosting goes away when the interchannel delay is 20 us.  I'm in FPGA mode, so I'm only scanning the channels specified in scan list, and the grounding appears to be correct.  That's why I'm wondering if I really understand the specification or if something else is wrong.  I can live with a 2.4 mV error to get 250 kS/s, but 60 mV is too high and appears to be far out side of spec.

[smithd]

your understanding seems correct to me, 1% error from a daq card is too damn high

Edited October 6, 2018 by smithd