X4M200 problem with detecting breath in XeTruExplorer Windows 10

Discussion in 'Respiration Sensor' started by sten, Apr 8, 2019.

  1. sten

    sten New Member

    Hi,

    I using XeThruExplorer(windows 10) connected to X4M200 over USB to laptop (i7-7700HW + GTX 1060). Device have actual firmware 1.6.8 (updated from older firmware). On PCB i found this numbers to help identify specific board (290056-010; STM-5 94V-0 3217;10000018805 000167-007).

    I use Respiration_2 (tried other with similar result) profile with default configuration(in same case i limit distance to 0.4-2m) and i have problem with respiration detection. Device is unable detect consistently respiration. In best case scenario i manage get about 60s "Breathing" other 60s "Movement *". It is really different result than expected from your product video.

    When initialization was in-progress, all moving object were removed from detection zone in-front of PCB (this is mentioned in video, but i didn't find any information about this in other document).

    I tried in multiple environment with same result, all without any object between test subject and PCB.
    PCB pointer on test subject with side with RX a TX antenna. I try room without other test subjects and room with multiple test subject, result are about same.
    Distance between PCB and test subject is about 1m, XeThruExplorer confirm this distance.
    Test subject awareness about measuring respiration didn't have any noticeable impact on result.
    Test were conducted with test subject sitting on chair and working with PC or lying on couch.

    I also notice high CPU a GPU usage when XeThruExplorer doing Initializing (CPU +-35%, GPU 60%), after completing initialization utilization to half. It still seem too high for me, it should be simple GUI application, displaying value from device or do you get raw data from device and doing data analyzing inside XeThruExplorer?

    Screenshot from XeThruExplorer (both screenshot were captured in independent test session with their own initialization):
    Best case:
    https://ibb.co/L1y9HbJ

    Worst case:
    https://ibb.co/zRR1WtG

    Do you have any advice how to get reliable respiration data, or how verify PCB isn't faulty (PCB isn't visibly damaged)?

    Thanks.
     
  2. BenChong

    BenChong New Member

    I have same issue on my X4M200 too. It was connected to Win10 through USB to my laptop and we tried many time on different people and different environment, the result is so poor that we think it is unusable and have huge different compared to video demo. In short, it's initialization process is too....long and it is very hard to detect continuous respiration pattern (pattern lose with some minor movement). so please let us know what can we do to improve this. Thanks!
     
  3. Charlie Shao

    Charlie Shao Moderator Staff Member

    Hi,

    1. Please make sure that you send support requests to support@xethru.com.
    2. We have released one application note to explain how to use the sensor and which kind of application it fits. I am not sure if you have downloaded and read it. If your application needs real-time constant RPM output, X4M200 may not fit. XeThru radar sensor can detect very tiny motion like chest movement when people are still, but it means that the RPM detection can easily be interrupted by other body movements, so the sensor is more fit for the condition when people are laying down or sit still. From the document, you can see that the sensor can report RPM at 63.86% of the whole night sleeping time.
    3. XeThruExplorer does not process any radar data, but it plots almost all the radar output messages, especially radar raw data with 17FPS. The plotting process is resource-consuming and it is just used for demonstration. You need to switch to ModuleConnector for your custom application, which will need far less data message than XE.
    4. The initialization process now can achieve around 20s by using default noise map.
    5. Respiration pattern has high requirement on reflection signal, so it is hard to achieve stable output by far.
     
    Last edited: May 16, 2019
  4. sten

    sten New Member

    Hi,

    1. Thank for information about another way how contact your support, on website I find only support question form and forum. Because I didn't receive any response when I tried contact you via support question form, I put it here on forum.

    3. You are right, performance effectivity XeThruExplorer doesn't really matter. I just wanted be sure that all calculation are done on X4M200 side so when I use MCPW that I get same result. I currently using your MCPW to gain data from X4M200 with same problems as the one gained using XeThruExplorer.

    2. I read it. I tried replicated testing environment from video with result mentioned in first post.


    I cant find any mention about 77.88% in linked application note or datasheet. I only see there Fig. 6 (breathing 47.64%), Fig. 7 (breathing 63.86%) in application note.

    I don't need constant realtime RPM. What I need is +- same result quality as in video and application note. At this moment only thing that we can get reliable from X4M200 is information if something moving.

    When i tried recorded my sleep (forget click record so I have only record 04:00-06:00 of not very good sleep):
    Captured using XeThruExplorer with board mentioned in first post. Distance was 1.5-2.0m. Board was positioned according application note. XeThruExplorer used default setting and respiration profile 2.

    Breathing 35.39%,
    Movement 36.06%,
    Movement tracking 22,**%,
    No Movement: 6.45%
    sleep_record_graph.png
    Signal Quality is for some reason low compared to data inside application note zip.

    5. Any advice how to get better signal quality?

    X4M200 support Low(6.0 - 8.5GHz) and High (7.25 - 10.20GHz). In your documents i didn't seen any use for High band ("Default User Settings for the Supported Profiles" use for all profile Low). Is there any use-case when you recommended using High band?
     
  5. Charlie Shao

    Charlie Shao Moderator Staff Member

    Hi Sten,
    Sorry, you are correct, the breathing time is 63.86% on app note. I referred a wrong version.
    There might be many causes that you can not get the same performance as shown in the video, so I need to understand your setup better. It would be better if we can have a meeting to discuss this. The same with your one-night recording, I need at least some pictures to make sure there is no problem with your placement. Please connect us through support@xethru.com to provide your setup information.

    For X4M200, it is designed for EU (CE/ETSI), USA (FCC) and Canada (ISED, formally known as IC), the low band should be used for regulations compliance. High band is used by X4M202 for Korea (KCC) regulation compliance.