Page 5 of 5 FirstFirst ... 345
Results 41 to 45 of 45

Thread: WS2811 flicker but only at low brightness - Help please?

  1. #41
    Join Date
    Jun 2020
    Location
    UK
    Posts
    17
    Post Thanks / Like

    Default Re: WS2811 flicker but only at low brightness - Help please?

    And the last attachment missed from post above.

    Kind regards,
    Steve
    Attached Images Attached Images

  2. #42
    Join Date
    Dec 2011
    Location
    UK S80 postcode
    Posts
    1,488
    Post Thanks / Like

    Default Re: WS2811 flicker but only at low brightness - Help please?

    Hi Steve, that is a fantastic piece of work and shows what you can do without all the latest gear.

    To review this data to be fair and objective let’s look at any potential issues that could have led to inaccurate results so that we can take them into account:

    The frequency of the sampling could have been too slow; this could introduce random lower or higher values to be recorded. With a sampling rate that was too slow a line could have been high for some time before a sample was taken resulting in a shorter high time being recorded. I’m not seeing any evidence of inconsistent measurements in shots 1 (F16v3 output) or 5 (good pixels) and am inclined to think the sampling frequency was more than adequate.

    An external influence local RF, ground noise or some sort of effect in relationship between the scope power supply and the pixel power supply could have affected the results. Again the shots 1 and 5 suggest not, unless variations in the data and external effects cancelled each other out to give what appear to be very consistent results.

    With that under our belts and noting that each one of these periods covers about 14 pixels worth of data. The value being sent to each pixel is R=1, G=0 and B=0, as such 24 bits are being sent;
    0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.

    It’s relatively easy to see each set of data that is being reproduced, with 1/3 being before the single ‘dot’ and 2/3 being after.

    Note:
    It is possible that depending on how you interoperate the data sheet you or the person in the video that you used as a reference might have been generous with the maximum and minimum timing values.
    The data sheet for example gives for a ‘1’, the time the line should be ‘high’ as 1200nS +or- 150nS and then states that “It is one half the time when high speed mode (reset time unchanged)”
    You are not using the ‘slow mode (400KHz)’, as you are using the ‘high speed mode (800KHz)’, “it is one half of the time”.
    I haven’t seen the video and haven’t got time to check tonight or I’ll not finish typing this, I’m just going on the values you have used.
    It appears (shot 1) you have used (from above) 1200nS / 2 (high speed mode) +or- 150nS as your values.
    The maximum ‘high’ time as 600nS + 150nS = 750nS
    The minimum ‘high’ time as 600nS – 150nS = 450nS
    The only way to check this is to email World Semi and check, in my book if you are halving you half everything including the tolerance:
    The maximum ‘high’ time as 600nS + 75nS = 675nS
    The minimum ‘high’ time as 600nS – 75nS = 525nS
    OK this raises an issue as it potently puts the puts the F16v3 well out of spec high on the ‘high’ side and low on the ‘low’ side.
    End of note.

    Back on to looking at the data:

    We only have two sets of data taken at the same pixel position to make a comparison of (before pixel 51 good and bad).

    Forgetting any tolerances and looking at the left hand side of you shots for ideal times taken from the data sheet:

    T1H 1200nS / 2 = 600nS and T0H 500nS / 2 = 250nS.
    Shot 5: Looking good, the 600nS – bang on and the 250nS – could be a little long but not by much and consistent.
    Shot 2: Not quite the same. There is an element of consistency in that most of the measurements have only consistent deviations.

    On to the right hand side of your shots, again looking at the ideal times taken from the data sheet:
    T0L 2000nS / 2 = 1000nS and T1L 1300nS / 2 = 650nS.
    Shot 5: Both are looking a little short the 1000nS could be 960ish nS and the 650nS looks more like 625nS to me. Both are consistent.
    Shot 2: On what should be 650nS timings, it does appear that there is some sort of ‘possibly sinusoidal’ influence on the timings. On the 1000nS, it’s hard to make an objective comment.

    My conclusion for what it’s worth:
    In as much as data is available to have done a one to one comparison, the two sets of pixels demonstrate different characteristics in the transmission of data.
    The problem that for me limits any conclusions beyond that is that the data you have provided suggests that the F16v3 is out of spec in the data it is sending out and whilst older pixels may have been happy with this the pixels you have recently purchased may require ‘the timing to be correct’.
    End of review.

    I am inclined to agree these pixels have issues which you have demonstrated not only by the fact that they don’t work correctly but the data you have captured. I’m not sure how much time the charting takes, it took me a while. If it was me, given the fact that the F16v3 could be an issue with regard to its timing, I’d throw in a single ‘good’ pixel to correct the timing of the F16v3 as their timing looks so solid from the previous tests and do the ‘before pixel 51’ test again after that pixel for the good and bad strings with just a single comparison, that’s all I think would be needed.

    It doesn’t really matter the particular reason, if the timing is out of spec and the pixels don’t work I think you have a valid reason to link Ray this thread to run past his tec.

    Disclaimer:
    I’ll be honest, I’ve purchased from Ray for over 8 years now and never had an issue but that doesn’t mean that stuff doesn’t go wrong. If the evidence is presented I’m sure he will see you right.
    Last edited by Barnabybear; 06-29-2020 at 06:53 AM.

  3. #43
    Join Date
    Dec 2012
    Location
    Newtown CT
    Posts
    4,292
    Post Thanks / Like

    Default Re: WS2811 flicker but only at low brightness - Help please?

    Hi Barnabybear

    Great analysis of the data. However, I do not believe you can claim anything about the F16 beyond the data sent to Pixel one in the chain. After that you are seeing the the pixels one interpretation of the data.

    Yes there is a propagation delay for the data across each pixel. I have not looked it up but from my measurements it looks like one bit per pixel. This is why any simultaneous measurement between two points would only be useful on both sides of a single pixel.

    My take:

    You have pixels that either have input sensitivity issues or output generation issues. The only way to tell is to identify the first pixel that is having an issue and measuring what goes into that pixel and what comes out. Any other pixel after that pixel (while still a potential perpetrator) is a victim of data corruption at the first bad pixel. Keep in mind that the addition of a scope prob also changes the characteristics on the system under test and may skew your results, making analysis of the data very difficult. You may need to use super low capacitance/high impedance probes.


    2020 Full sized show reworked for the new location. Only adding (famous last words) 13 RBLs that I finally got converted to using pixels
    2019 - Just moved into a new home (yet another change of plans). Will be dim but not dark. Too much to do at the new place to leave time for a show. Dim show (3000 pixels) had regular visits most nights.
    https://www.youtube.com/channel/UCyX...ttrsZNARkUce0Q

  4. #44
    Join Date
    Jun 2020
    Location
    UK
    Posts
    17
    Post Thanks / Like

    Default Re: WS2811 flicker but only at low brightness - Help please?

    Hello all,

    MartinMueller2003 - The one set of data I believe Barnabybear is referring to for the F16v3 timing is the data capture of the raw data coming straight out of F16v3 before it enters pixel 1. There is no one single point where I can identify the pixels 'go wrong'. It changes depending upon brightness levels and colours selected. Up to about 250-300 there is no visible problem whatsoever under any conditions i can test. From around 300 to 1000 pixels it progressively gets worse. If I set full brightness white, the problems go away all the way to 1000 pixels. All strings of 50 on their own work fine (but still exhibit timing issues, but not sufficient to affect performance) so I have tried progressively swapping out full strings trying to identify a 'bad' pixel.



    Barnabybear - Many thanks again for your thoughts and time understanding and explaining things to me. The plots are real quick to produce now - just read the USB scope data straight into the LabView analysis code and everything is automated.
    I've tried what you suggested with a good pixel (50 actually to save me cutting a string) before the bad pixel strings and probed after the first 50 good pixels and after the first 50 bad pixels (and then also after 500 bad pixels).
    Screen shots attached. Not much difference even though the bad pixels were getting much better timing from the good pixels.


    Update on Ray Wu:
    Ray Wu has agreed to send replacement pixels from a batch that has the chip de-coupling capacitors on the board. These are said to be from an older batch and an older pixel board design that still need the capacitor and resistor).
    Whilst there is no acknowledgement that there might be a problem with the new design pixel boards, he is now prepared to help resolve this problem, and stand behind his products. This is far more than I thought was going to happen when I started to discuss things with him, so whether this is as a result of this forum discussion or not (I don't know), it does appear his reputation reported on here and elsewhere is good and justified.

    I will report back when I get the replacement pixels and have tested them if anyone is interested in how this turns out.



    Thanks again to everyone who has offered advice and helped me through this.

    Kind regards,
    Steve
    Attached Images Attached Images

  5. #45
    Join Date
    Dec 2011
    Location
    UK S80 postcode
    Posts
    1,488
    Post Thanks / Like

    Default Re: WS2811 flicker but only at low brightness - Help please?

    Hi Steve, it’s great news that Ray is replacing these pixels. I do think it is credit to the work that you have put in that this has happened.

    I try to put myself in other people’s shoes when considering life in general. With the evolution of the internet electronics has been brought to the masses, with tutorials, suppliers and 100’s of projects listed. On the down side for suppliers you have no idea who you are selling to, what their skill set is, whether they are using the part within spec or if they have just purchased it and butchered it and then want a refund. I’m guessing Ray and all pixel suppliers have had quite a few emails stating that the pixels don’t work (possibly more 5V than 12V) and I’m sure someone must have tried it on and just said these don’t work when they do to get some freebies.

    With the process you have gone through not only have you demonstrated that you know how to make pixels work, you have proved that there is an issue and gone quite a way to diagnose it with testing that shows that the WS2811 regenerated timing isn’t to the manufacturers spec. Even if the exact cause of this hasn’t been identified it distinguishes you from the ill-informed or chancers and offers the opportunity for your supplier to also seek recompense from their supplier.

    I’m glad you appreciated the importance of at least one good pixel between the F16v3 and the ‘bad’ pixels, it was important that a set of data as close to the manufacturers spec as per the data sheet was used to provide validity to the results. I didn’t think it would make much difference but now we know it doesn’t.

    I for one will be interested in seeing how the replacements in similar tests if you don’t mind the time

    I’m sure this is not where you wanted to be at this time of the year but I hope you have gained knowledge from this that will be useful to you as you incorporate more pixels into your show.

Page 5 of 5 FirstFirst ... 345

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •