I have been asked to "fix" the conducted emissions aspects of a Luminaires board that just touches the EN55015 average limit (47dBuV) at about 350kHz. The board is powered by a certified MAINS to 28Vdc supply. The board has an array of buck LED drivers and they are running at about 350kHz. The board itself only has a differential mode type filter, and apparently they made a umpteen thousand of these before checking for compliance (oops).
So I tried adding a common mode filter on the DC link and it drops comfortably below the limit line.
I'm using a 50uH LISN on the MAINS side, and on the SSA3021, I'm using the EMI filter with a RBW of 9kHz, average detection and the frequency step is 1kHz.
Initially I set up the SSA3021 to mimic the pre-scan results I was provided (9kHz to 30MHz).
However I now notice the amplitude of the 350kHz noise seems to change considerably depending on the frequency Span setting. Sweeping from 9kHz to 30MHz, I'm well under the limit, however as I narrow the span the amplitude of the harmonics of the switching frequency move above the limit line as the upper frequency limit of the span is set lower than 30MHz.
What am I missing?
Is the span setting changing the RBW? Changes in RBW can certainly have this affect.
From the manual:
Modifying the span in non-zero span mode may cause an automatic change in both CF step
and RBW if they are in Auto mode. Besides, the charge of RBW may influence VBW
(in Auto VBW mode).
Is there a RBW setting that mimics the levels you are seeing when you change the span back to 9kHz-30MHz?
Is the span setting changing the RBW? Changes in RBW can certainly have this affect.
From the manual:
Modifying the span in non-zero span mode may cause an automatic change in both CF step
and RBW if they are in Auto mode. Besides, the charge of RBW may influence VBW
(in Auto VBW mode).
Is there a RBW setting that mimics the levels you are seeing when you change the span back to 9kHz-30MHz?
I was wondering the same thing, however when I look at the RBW is still shows 9kHz. It looks like a bug
Make sure to measure with the same RBW as the conducted emissions test. Also be aware that your setup is likely to give different absolute levels so any measurement must be made relative. If you device is 6dBuV over the limit than a modification should give you at least that amount + some margin of improvement before attempting another official test.
If the SSA is using a 'proper' average detector according to CISPR 16, then the scan time would be quite long over .15 to 30MHz.
I'm not familiar with the Siglent EMI option that provides 9 kHz RBW and CISPR detectors. Not sure how these are implemented or how they recommend you use them.
Typically for QP and average measurements a peak detector sweep can be carried out and then spot frequencies are measured with either QP or average detector in zero span mode.
Quick look see average detection over broad scans can however be achieved using a small video bandwidth setting.
Refs here on CISPR 16 detector time constants. Don't think these have changed over best part of 50 years.
http://www.emcia.org/journal_article.aspx?artid=107Hope that's on the right track?
If the SSA is using a 'proper' average detector according to CISPR 16, then the scan time would be quite long over .15 to 30MHz.
I'm not familiar with the Siglent EMI option that provides 9 kHz RBW and CISPR detectors. Not sure how these are implemented or how they recommend you use them.
Typically for QP and average measurements a peak detector sweep can be carried out and then spot frequencies are measured with either QP or average detector in zero span mode.
Quick look see average detection over broad scans can however be achieved using a small video bandwidth setting.
Refs here on CISPR 16 detector time constants. Don't think these have changed over best part of 50 years. http://www.emcia.org/journal_article.aspx?artid=107
Hope that's on the right track?
Yes, scan time is really long (naturally).
(Optional) SSA3000X Measurement Receiver:
Resolution bandwidths: 200 Hz, 9 kHz, 120 kHz (6dB)
Detector is
Quasi-peak （following CISPR 16-1-1）
Dwell time adjustable ~0 to 10 s