Loading a waveform in the RS500X's internal memory

Quote from AdminTC on 13/08/2023, 03:35

No thats not possible. A 5Gbit USB load is no problem at all for any of todays computers so i dont see any advantage.

Quote from hexium on 13/08/2023, 17:34

Thank you for the response. I see two major advantages:

• If you could also somehow autostart the waveform playback, someone could use RS500X as a standalone device without the need of a PC
• Reduce the CPU load on the host PC

But on top of these, playing a waveform without issues (underflows) has not been so straightforward for me. I am using relatively powerful PCs (Ryzen 5700G / i5-12400) and I am seeing occasional "input underflow" errors and 40-50% CPU utilization.

I am most likely doing something wrong so I would be grateful if you could check the attached mission file.

Thanks.

Quote from Sofon on 14/08/2023, 06:37

Quote from hexium on 13/08/2023, 17:34

Thank you for the response. I see two major advantages:

• If you could also somehow autostart the waveform playback, someone could use RS500X as a standalone device without the need of a PC

But what is the meaning to autostart the waveform playback, using RS500X as a standalone device without the need of a PC?

Quote from hexium on 16/08/2023, 09:39

Hi Sofon,

Please note that there are two issues here:

- The least important is to be able to autoplay a waveform, therefore enabling standalone operation of the RS500X. One scenario that this would be useful is for using RS500X as a transmitter of a known sequence (e.g. WiFi or Bluetooth packet) and use this to debug the receiver operation at the other end. For example, perform debugging of a DoA (Direction of Arrival) receiver device using a known transmit sequence. In that case you will only need RS500X to be plugged in a power bank and therefore it could be freely and easily moved around.

- The second and most important issue related to using built-in memory to replay a waveform is to:

a) remove the processing load from the PC and enable even a Raspberry Pi to be used as a host,

b) remove the need for a high-end USB cable (which are usually limited in length).

Would it be possible to review the existing setup that I attached in my previous email and let me know if there is somethign wrong? Is it expected that this kind of mission would consume ~50% of the CPU resources in relatively powerful PCs?

Thank you in advance.

Quote from mm_dev on 16/08/2023, 10:13

[quote]

Quote from hexium on 16/08/2023, 09:39

Would it be possible to review the existing setup that I attached in my previous email and let me know if there is somethign wrong? Is it expected that this kind of mission would consume ~50% of the CPU resources in relatively powerful PCs?

The CPU saturation comes from the modulator block, and there from the high sample rate. Reducing the samplerate (and effectively the span) there will lead to a linear reduction in CPU saturation, but of course may contradict your specific use-case. If you enable "Adapt Sample Rate" it will automatically be adjusted by the corresponding Spectran V6 settings (mainly IQ samplerate and Span Factor), similar for "Adapt Center Frequency". That way you won't have to worry about its settings at all, just control it from the V6 block.

Please note that CPU saturation is not the same as CPU load: For timing reasons most processing blocks in the RTSA only utilize a single CPU core, and CPU saturation just shows the maximum load of ANY CPU core. So if it shows 50% that doesn't mean that 50% of all your CPU resources are consumed, only that at least one CPU core is 50% loaded.

In some contexts it can therefore help to split processing into multiple steps across multiple instances of the same block, so the processing load is also distributed across multiple CPU cores. But that is always a case-by-case decision to make as there are of course also downsides to this approach (additional overhead and complexity, potential degradation of data quality).

And if you don't actually need live modulation of your input signal you could also record the modulated signal into a new file, and just feed that into the V6.

Quote from hexium on 22/08/2023, 09:12

Thank you for the response and the tips.

Please note that the use case that I am interested in requires such high sample rates (20/40/80 MS) for 20/40/80 MHz Wi-Fi signals. Therefore, reducing this is not an option.

And if you don't actually need live modulation of your input signal you could also record the modulated signal into a new file, and just feed that into the V6.

Regarding this, I recorded a modulated and IQ normalized waveform, then tried to use a simpler mission (File Source with new waveform, directly connected to Spectran V6 block). When Spectran V6 is in "Off" state, CPU Sat is ~30%, when it is "On", CPU Sat is 40-90%. Is this normal?