Are you sure this isn't the bottom mounted chassis fans causing the EMI?
What input range are you using on the DAQ boards?
It's not recommended to use chasses with bottom mounted fans with these DSA cards, specifically when it comes to the low input range on the DSA cards (they can pick up EMI noise from the fans).
I don't think this is what you're seeing, since using a UPS to power the system hides the noise, but it's worth pointing out.
Craig, your discovery seems more appropriate in this case, as per the equipment setup, the chassis used is 1073 which is not recommended with 449x and 446x as knowledge article cites.
Thank you Craig
My EMI is not caused by the bottom fan of PXIe1073， Absolutely.
The EMI of bottom fan is about 25.6kHz. In Sound test we only need 20kHz Analysis bandwidth。
So I ignore the noise of bottom fan . And I also know how to avoid the botton fan noise——Only need to use the rear fan chassis, the more expensive ones.
This kind of EMI is full from 0~20k, not in 25.6k. And the buttom fan EMI is only near 25.6k.
Actually,Long time ago, When the NI staff suggested the list for me ,they didn't tell me the details of the EMI in buttom fan . I just found it myself accidentally when I used.
After that the NI staff bring in another Rear fan chassis to show me the difference of EMI. So I do know how the buttom fan EMI looks like in FFT.
Sorry my mistake. Because the NI system sent me an e-mail, I misunderstood, thinking that you are a NI staff member.
1 According to the information I consulted, the shell of MIC = BCN wire shell = MIC AI-Reference 0 = DSA AI-Reference 0, which is 50Ohm with the PXIe1073-Ground.
2 I also found out that if the Real-Ground is not clean, it makes other bad influences. So all of my project is grounded perfectly.
3 The EMI would coms from the BNC cable.
If i use 15m cable, it is worse than 10m cable.
If I use B&K double shielded cable，it is worse than Single shielded cable.
I also try to put another“dense copper mesh hose” to cover the BNC cable, the much worse effect.
Finally, I used a single shielded wire with a thinner shielding layer, the effect is better.
But the B&K instrument(About 2.5 times more expensive than NI) doesn't has these problems，at all.
4 So I guess, the EMC of PXIe is bad, and the BNC cable just like a antenna to collect EMI and let the EMI get close to the PXIe. The thicker shielding layer is ,the more EMI collected.
5 The NI staff gave another idea. They think the voltage level of BNC outer shield（the AI-Reference 0）is influenced by the EMI， so the FFT is bad.
6 There are two test Labs in my progect, for air conditioner. They are used for conditioner Outdoor side and indoor sied .These rooms have some EMC capabilities, so I used one side for Install NI temporary. It can't be keeped in that all the time.
7 I can not build another EMC room for these kinds of lab. It is expensive, and also make me seems unprofessional and stupit. Buying B&K system is much cheaper than build another room.
UPS is only a temporary way to find out the problem, but not the usual way for test.
8 I do have several ways to reduce this problem，but only cure the symptoms, not the disease.
I can take various shielding measures to reduce the transmission interference and radiation interference of Mitsubishi inverter.
In these projects , I provided the Mitsubishi inverter for the users. So I can do some shielding on the Mitsubishi inverter. But in a factory , you know, there are too many device，and some devices are large, some equipments are provided by other suppliers .
It ’s impossible for us to shield one by one.Even if we can shield all interference sources, the cost are also huge.
9 So the best solution is to shield the PXIe itself. But I can't find a way.
In addition, we once had a PXIe1073 sent for repair; at that time, the time was urgent, so the panel inside was replaced, resulting in the lack of speed regulation function for the bottom fan. When this function is cancelled, the interference of 25.6kHz also disappears.
So I am sure my EMI is not about the bottom fan.