Power Spectral Density Measurement Discrepancy?

[mikeh]

I am having trouble duplicating (with Spike) Power Spectral Density measurements made using the previous SH application (V2.18d), and verified with a Rhode & Schwartz analyzer. Below are screen shots of the three measurements of the same RF signal. Notice that the old SH app & the R&S measurement are very close to each other (+4.7dbm/3.2khz vs +5dbm/3khz). However, the Spike measurement is -.774dbm/3khz. Am I doing something wrong?

• This topic was modified 9 years, 4 months ago by mikeh.

Attachments:

You must be logged in to view attached files.

Hi Mike,

For the Spike software, to get comparable measurements, ensure video units are set to power, detector is set to average, and disable spur reject in the settings file menu. With these settings, you should see similar readings. For a wide band signal like this, ultimately, you will want to compare channel power readout.

Let me know if you have follow up questions.

Regards,
A.J.

A.J.

For the Spike software, to get comparable measurements, ensure video units are set to power, detector is set to average, and disable spur reject in the settings file menu. With these settings, you should see similar readings.

Attached is a snapshot with the settings you suggested. Note that the measurement result is very nearly the same as before. This is a serious issue. Any other thoughts?

For a wide band signal like this, ultimately, you will want to compare channel power readout.

I am trying to make a PSD measurement for FCC part 15 compliance which requires that you measure the peak power in a 3khz RBW. How does the channel power readout help?

Attachments:

You must be logged in to view attached files.

Hello Mike,

Can you try attenuating the input to ensure you are not in compression? Spike also sets the attenuator and pre-amp automatically based on reference level. With your ref offset active, set the reference level to +40 to see if this affects the measurement. I am concerned that the signal is possible being compressed in the Spike software.

Having said that…

I performed a similar measurement on both the Spike software and version 2.18. I was able to get similar results on both pieces of software. I have attached the measurements. In this case, I used a signal generator to create a wideband signal with an output power of -40dBm. Seeing a channel power output reading of -40dBm on both software versions lets me know I have configured the software correctly. If you have a generator, can you try a similar test?

Regards,
A.J.

Attachments:

You must be logged in to view attached files.

AJ

Can you try attenuating the input to ensure you are not in compression?

There is a 20db fixed attenuator on the input of the SH and the Ref Offset is set to 20db.

With your ref offset active, set the reference level to +40 to see if this affects the measurement. I am concerned that the signal is possible being compressed in the Spike software.

I have changed the Ref Level from +40 to 20dbm with very little difference in peak power measured.

If you have a generator, can you try a similar test?

I don’t have a generator other than the radio being tested. It’s output is +20dbm.

Also, FYI, I am using V3.0.12 software if that makes a difference.

[Andrew]

Mike, please call me at 1-800-260-TEST Ext 2 so we can work through the discrepancy. I’m available between 9-3 PST

Regards,
A.J.

• This reply was modified 9 years, 4 months ago by Andrew.

I’ll give you a call tomorrow AM your time.

Thanks!
Mike

Is there any resolution to this issue? I’m attempting to make the same measurement, and I believe I’m finding the same issue – 3 kHz PSD is several dB below the expected value.

I never followed up, but still have the same problem.

I assume you are doing FCC 15.247 “The maximum permitted peak conducted output power is +30 dBm (1 W). However, the power spectral density conducted from the intentional radiator to the antenna shall not be greater than +8 dBm in any 3 kHz band during any time interval of continuous transmission. ”

To achieve this, I would use the following settings:
1) Spur reject off
2) Video detector set to min/max
3) 3 kHz CISPR / Gaussian RBW, 3 kHz VBW (in Spike, when VBW = RBW and min/max selected, there is no video filtering)
4) Trace set to max hold
After accumulating for several seconds, peak search. This should give you the correct reading. Let me know if I misunderstood the test you are running.

Justin:

You’re exactly correct. I’m shooting for 15.247.a.2 and 15.247.e to achieve (or approach) the 1W limit allowed in 15.247.b.3.

I didn’t have spur reject off, so perhaps that’s the difference. I believe I had all other settings the same though. I’ll retest this afternoon.

In this gem of a document from the FCC: https://apps.fcc.gov/eas/comments/GetPublishedDocument.html?id=237&tn=158724

… Section 7.3.4 explains one accepted method of testing PSD. They suggest using RBW of 3 kHz (which I am) but a VBW of 9 kHz. Spike won’t allow me to set VBW > RBW in any case. Is this going to lead to any inaccuracies in measurement?

I don’t need cert-quality measurements, but I do want realistic numbers from this pre-scan, so I know how much margin we have before hitting the 8 dBm limit.

Thanks very much.

Chris

The goal of VBW > RBW is essentially to remove any significant filtering after the RBW. Our software is a little different in that we bypass VBW processing unless VBW < RBW or the average detector is used. So for Spike software, VBW equal to RBW is what you want.

Justin, Andrew,

This is still a problem. In fact, I can now reproduce it on two different SH units (SA124A & SA44B). Perhaps it could be the that the amplitude or the modulation format of the measured signal is affecting the measurement algorithm? My radio puts out +20dbm with a OQPSK signal. Your measurements above were made at -40dbm, not sure of the modulation. There shouldn’t be a difference, but there is. As stated earlier, I have verified my measurement with a Rhode & Schwartz spectrum analyzer at my FCC test lab. How do you propose we address this?

Mikeh,
Have you tried attenuating the signal to maybe -10 dBm (well into linear region of SA-series), ensuring detector is min/max, spur reject is off, gaussian filter. What kind of reading do you get?

The last picture had detector set to average. I’m just trying to figure out what the difference may be.

[mikeh]

Justin,

Yes, there is a 20db pad on the input to the SH (0dbm into SH), and it was not changed between the Spike and V2.18d app measurements. I am using peak hold for measurements since that’s what the FCC lab does.

• This reply was modified 7 years, 3 months ago by mikeh.
• This reply was modified 7 years, 3 months ago by mikeh.

And when you change the detector to min/max, ensuring spur reject off, you see no change at all in Spike?

[mikeh]

Please disregard these two screen captures (wrong settings). Not sure how to delete these on the forum..?

• This reply was modified 7 years, 3 months ago by mikeh. Reason: wrong settings

Well, I was finally able to delete the erroneous screen captures, but now I cannot edit my post….:(

Below are the corrected captures with spur reject off and min/max detector setting.

Attachments:

You must be logged in to view attached files.

Your VBW was less than RBW, so you were low pass filtering the power response. Did you try setting VBW to auto, or = RBW?

[Author]

Posts