Forum Replies Created
- Posts
This bug was resolved in a release of Spike today (4.0.8). https://signalhound.com/spike/
If you continue to have issues with this release, please contact us at support@signalhound.com
Thanks
mmark,
There are a couple scaling factors you might need to consider when FFT’ing the I/Q data from our API.
– If you are requesting 16-bit I/Q data from our API, and you want to measure dBm, then you will need to get the I/Q correction and apply it via the equations in our API manual. If you are requested 32-bit floating point data (recommended), then the data is already scaled.
– The FFT itself. I would recommend scaling the Fwd FFT by N. What additional scaling you need to apply is dependent on which FFT library you are using. Some use scaling by default, some dont.
– If you are using a window function, you want to normalize the window before windowing the data.
– Once you have the above resolved, the output of the FFT can be converted to dBm using the equation
10*log10(I^2 + Q^2)
The problem with the 2 script approach is that all the resources associated with the handle and device are in the working memory of the first script. The second script can’t access those.
A couple alternative solutions,
– Go to a 1 script approach where after the initialization occurs, the script goes into a waiting for user input loop, and responds to the requests for data, all within one script. This is the easiest solution.
– Change the first script to act more like a service, which stays open and communicates with other scripts/programs that request data. The first script is the only script that interfaces the device, and relays all data to the other script. This seems like a more complex approach unless you plan on being able to have several programs requesting data simultaneously.
It has been marked as spam. It got through our spam filters.
Hi Euan,
Good questions,
The marker “Set Freq” entry was designed as a one way control. It doesn’t show you the current marker frequency, but can be used to set the marker frequency. This is why you don’t see it ever update until you manually type something into the box. So things like moving the marker, or loading presets will not affect it.
As for why you don’t see the marker frequency report exactly what you set, is because the sweep bins do not always land on nice round frequencies. So when you set 1.83GHz, it will find the closest frequency point to your value. What frequencies bins the sweep lands on are hardware and configuration specific and are not predictable. RBW controls how close sweep bins are, so as long as your RBW is good for your application, using the set freq will always choose a sweep bin within your RBW, which would presumably be sufficient.
Let me know if you have follow up questions.
The SP145 would be the better choice for phase noise measurements due to it’s better absolute phase noise performance and higher frequency range.
The purpose of the preselector is to filter out of band energy. This is most commonly helpful when measuring over the air with an antenna, as communications signals such as broadcast radio, cellular, WLAN, are filtered, improving measurement performance such as dynamic range.
Phase noise measurements are commonly/ideally made directly connected to the DUT via coax, eliminating the presence of out of band energy unless generated by the DUT itself. You can always insert a filter if out of band energy was a problem for a particular measurement setup.
Yes, presets for these 2 devices should be compatible. Your client should be able to load your generated presets.
There is no road map or timeframes we are targeting. We are releasing updates as we can allocate engineering resources to them. I will be sure to reply to this thread when there are any updates to our ARM offerings.
Hi Jequin,
We have not yet built a LabView wrapper project for the SP145 API. It should be possible. One approach I could suggest now is to look at the SM LabView project and adapt it to the SP API. This may be difficult if you don’t have experience calling DLLs from LabView.
I don’t have a timeline on when we would have usable examples for the SP145 at this point.
If you attempt to build your own, and have any questions, please contact us at support@signalhound.com.
Thanks
I’m using this thread as the general update thread on our ARM progress since it has the most visibility and comments on our forum. We are very much interested in supporting the ARM architecture and have been slowly adding support since our last update.
In our SDK we now have API builds for both the SP145 and SM200/435 for both the Nvidia Jetson Orin AGX and MacOS (M4 CPU). These APIs support a subset of functionality, namely sweeps and I/Q streaming with decimations up to 4, which hopefully covers most use cases. 10GbE support for our SM devices is also not supported on the Mac build, just the USB variants for now.
We are continuing to develop capability on ARM to support a larger subset of our APIs and new APIs altogether, (I.E. the BB60 API).
We appreciate everyone’s feedback and patience.
Thanks Joe!
Pooja, we have also responded to your email. For this issue we will continue our support via email.
Regards
April 10, 2025 at 8:22 am in reply to: Can once PC control two SM200Cs streaming IQ at the same time using MatLab? Yes, our SM API will support interfacing 2 SM200Cs simultaneously. I would recommend using the API and C++ for initial testing and proof of concept work. MATLAB adds significant overhead for streaming I/Q especially at high data rates. Our MATLAB examples only interface 1 device so it will need to be extended to support multiple devices. Starting with C++ will also help you understand how you will need to extend the MATLAB examples.
We would also generally recommend Linux over Windows if you are going to run multiple devices at the highest sample rate (200MS/s).
Are you using GNURadio v3.10? If yes, our OOT modules were developed on v3.9 and will require porting to 3.10. We currently have this as a TODO, so it is left as a customer exercise until we can make time for this. There are some resources that might help, for instance, https://wiki.gnuradio.org/index.php/GNU_Radio_3.10_OOT_Module_Porting_Guide
A reference level of -20dBm on the SP145 achieves the minimum noise floor, but will limit input signals to ~-20dBm before IF overload occurs. On the BB60D, the minimum noise floor is achieved at -30dBm reference level.
Mehdi,
You specify an attenuation of 0dB, which for the SP145 I believe is overriding your reference level control. If you want reference level to control the sensitivity of the receiver we recommend leaving all gain/atten/preamp values to their defaults. The reference level control alone will properly set these controls behind the scenes for the best dynamic range. If you change attenuation back to Auto, with a ref level of 0dBm, your measurement should work again on the SP145.
The reason you probably don’t see the same issue on the BB60D is because you have to set both gain and atten before it override the Ref level. Again, we still recommend leaving everything auto and just using ref level, unless you have a specific reason not to.
If Spike is set to hidden mode, that dialog will not appear, but I also believe it will not automatically enter demo mode. There is not a way to automatically enter demo mode. The only way to bypass that dialog into an active state is with 1 Signal Hound device connected.
Galc,
Spike has a “hidden” mode in which you can launch the software and it only appears in the task list. You can read more about this in the manual. If the issue is that the software is visible and you want just your application showing, that might be interesting.
Hi Volgy,
Nice work for being new at this! You have largely properly characterized the functionality of the fast sweep mode. There is only one more thing to know about. To accomplish “arbitrary” RBWs, you can perform FFT zero-padding. This basically gets us the ability to configure for RBWs as if we have FFT sizes between 16 (or whatever we use as the smallest) and 16384, in increments of 1, and not just 512,1024,…,16384.
And to address your other question, you are correct, the detector is not relevant for the reason you listed.
If you need more flexibility than what fast sweep mode offers, take a look at the I/Q sweep list. This mode lets you preconfigure a list of freq steps and I/Q capture sizes, and executes them at the same speed as fast sweep mode. This gets you the time domain I/Q samples as opposed to the processed spectrum data. On the SM435C, the I/Q bandwidth for each “step” is 160MHz, so you can achieve > 1THz sweeps with the right settings. For the SM435B models, the I/Q bandwidth is only 40MHz, so it falls short of 1THz per second, and in that case the fast sweep mode is providing functionality that I/Q sweep lists cannot. The downside of I/Q sweep lists is that now you have to do your own processing, but it will let you reproduce most of the sweep capabilities offered through the sweep modes.
The concept of a preset does not exist in the API, that is only in the Spike software.
Also, you are correct, the API provides just the sweep and I/Q data from the instrument. Any measurements on the data need to be performed in your application.
As Roger noted, the SCPI commands support both loading presets and exporting trace data.
Let us know if you have follow up questions.
I have replied to the email that you also sent asking this question.