Forums › General Discussions › Measuring the impact of SA124B’s internal OCXO
- This topic has 6 replies, 2 voices, and was last updated 2 weeks ago by
MAsgari.
- AuthorPosts
MAsgariParticipantHi.
I’ve got a SA124B with option 1 (OCXO) and a BB60D.
Currently, when I connect the output of an OCXO to these SAs, BB60D shows exactly 10.0000000MHZ. However, SA124B shows around 40Hz of offset (e.g. 10.000003709MHZ)
Now, I don’t care about that, but I am very curious to find out why.
I have 2 questions:
1. What could explain this difference? (Naturally, I’d assume that SA124B with OCXO would have a better frequency precision. But maybe I’m mistaking initial offset and stability over time)
2. How can I prove that SA124B is stable over time? For example, I can let them run for 12 hours, and read the frequency (or max amplitude marker) every minute and draw a plot. Is there a logging feature in Spike I can use for this purpose? If not, can I achieve that via SDK?Thanks in advance.
AndrewModeratorI’m assuming you are using the sweep mode + markers to perform these frequency measurements? This will have limitations, notably the frequency resolution will be limited to your RBW. If you wanted 1Hz resolution on your measurement you would need ~1Hz RBW. At higher RBWs, the freq bins will land on discrete steps that might give you the illusion that one device is more “accurate” than another, in this case it sounds like the BB60D has a frequency bin that lands exactly on 10MHz, and the CW energy happens to be closer to that bin than the adjacent.
We have other methods for measuring the frequency of a CW signal that are preferable. You can use either the frequency difference meter utility, or the analog demod analysis mode, which will provide you a high accuracy frequency measurements of a CW input with faster update rates than a low RBW sweep.
For logging, sweep mode + low RBW might be ideal. There is a sweep recorder which can log sweeps over long periods. Depending on how much time resolution you need, you can add artificial delays in the in the sweep recording toolbar so that you have adequate time and frequency resolution over your 12 hour period without a prohibitively large recording file.
The SDK will allow you to retrieve sweeps and I/Q data. All post processing such as frequency estimation must be performed by your application, but this gives you the most flexibility in how you log your measurements.
Keep in mind, the OCXO you are measuring will have a frequency error. In addition, as you mention, without a known reference, you don’t know what offset either of your SA’s have. A common approach to this is to use an external 10MHz reference, commonly a GPS disciplined one, which has a known frequency accuracy and is generally quite good, and provide this as a reference into the SA’s 10MHz input port. When an external 10MHz reference is connected to the SA’s, you can instruct the device to use it in the “Settings -> Reference” menu.
MAsgariParticipantThank you.
I redid the tests, with a GPSDO (output set to 100MHZ)In this test, SA124B showed smaller offset/difference:
SA124B -0.1Hz (BB60D: 34Hz)I also played with the Sweep Recorder. If I understand correctly, upon Playback, I can only check the spectrum and read the power by Marker, as other parts in menu are greyed out in the playback mode (e.g. the Analog Demod or Freq diff meter), but then as you explained, that won’t be an accurate method to check frequency reading accuracy/stability. Do you know what I can do for this?
Also, FYI I am only curious to test the SA124B’s stability over time and compare it against BB60D for experimentation, otherwise, as you mentioned, I will use an external reference for more accuracy/stability in my real world measurements.
AndrewModeratorThanks for the follow up.
You are correct that when you are in sweep playback mode, only the swept analysis is available. Unfortunately the other modes I mentioned only work with I/Q and not sweep data, so they are not compatible. Other than the sweep mode recording, I am aware of any other recording or logging features in Spike that would be useful for your specific task (long term logging/recording). I apologize for the inconvenience.
MAsgariParticipantOk, so I did another experiment, and the result was a bit surprising.
I connected the output of a GPSDO to SA124B and BB60D.
The initial drift measured by “Frequency Difference Meter” was 0.1Hz and 34Hz respectively.
I let them run for 2 hours. I expected some drift on BB60D, but to my surprise, they both still had the same drift as before!
Maybe I should measure over a much longer period like 24 hours or so.
AndrewModeratorDid you enable the use of the GPSDO? You have to go to the Settings -> Reference menu and tell the instrument to use the reference.
Also, is your DUT being disciplined by the GPSDO?
We calibrate our devices to a GPSDO. If you just put our instrument on it, there might be minimal change (basically whatever drift there has been since we adjusted it in our facility). Once our instrument and the DUT is using the GPSDO, then I expect nominally ~0Hz offset.
MAsgariParticipantHi Andrew. No, actually I was just comparing their built-in reference.
It was clear to me that SA124B with its built-in OCXO would have accurate reading after the warm-up, and also not much drift over time.
However, BB60D surprised me with not having any drift over the 2 hour period.
Thanks for your hints.- AuthorPosts
You must be logged in to reply to this topic.