**** BEGIN LOGGING AT Thu May 05 02:59:57 2022 May 05 07:21:55 BBONE-AI May 05 07:22:23 Please tell if the extended or commercial variant of the AM5729 is used May 05 09:20:45 Hi, it's me again with the beagle booting MLC NAND. I think it works now.  Thanks @zmatt!! I do have to use 2nd option for BCH poly - and (how crazy is that), I forget to double write each sector. So  finally it creates out if 1kbyte code not just 2 (by adding BCH) but 4kbyte by double write that. omg!!! The real bad thing with that TI coding is, May 05 09:20:46 this will recognize that as bad block by all kind of flash tools, which make handling extremely difficult. Once written you can't even erase it without scrub  (delete bad blocks) and this is (in my uboot) affecting the complete flash. So it wipes even the correct bad block markers. May 05 09:22:46 Seems I need at flash_erase that forces to erase even blocks that are marked as bad. May 05 10:23:34 Guest61: lol May 05 10:24:54 yeah, what a mess.  but happy to see it booting again! May 05 10:26:26 tests with biterrors are not done yet, don't know whether correction really works May 05 10:28:07 yeah, you should probably first fix the issue of not being to erase these blocks normally May 05 10:28:33 if need be add a custom erase command that just ignores the bad block bits May 05 10:29:27 once you can easily reflash these blocks, you can more easily perform additional tests, e.g. with intentional bit errors May 05 12:26:47 back May 05 12:27:48 damn internet monopolies May 05 12:28:09 my service sucks but I have no other options May 05 12:29:10 mattb0ne: https://libera.irclog.whitequark.org/beagle/2022-05-04#1651692455-1651693276; May 05 12:30:13 just read all that May 05 12:32:41 I thinkk the code works well but I think I need to account for the load cell delay. When I say running a bit slow... the input signal "should" take a second. The code will take about 1.7 seconds to finish a period according to the messages being streamed to the ring buffer May 05 12:33:04 I can take the filtering off to boost the sample rate to 1200 samples per second just to test May 05 12:34:09 I don't understand what you mean May 05 12:34:22 or I can increment by two to see if that gets to .85 seconds a period. May 05 12:34:28 let me share a plot May 05 12:34:29 ???? May 05 12:35:11 your signal period will be signal length (number of data points in array) divided by load cell sample rate May 05 12:36:06 if you want your signal to have a 1 second period, then make the signal length equal to the load cell sample rate May 05 12:36:31 https://ibb.co/W2tCM6m May 05 12:37:19 uhh, "control" here is the signal input? May 05 12:37:42 or, what's what here? May 05 12:38:07 refrence that I am trying to match. Nothing to do witht he coding. Focus is on the Magnapress Input vs Magnapress Output May 05 12:38:25 what do those mean? May 05 12:38:29 Magnapress Input is the 1000 points I put into shared memeory assuming a constant dt of .001 seconds May 05 12:38:43 that assumption is wrong, so why would you assume that? May 05 12:38:49 Magnapress Actual is the position read from the encoder May 05 12:39:06 well the PID loop code on the PRU assumes a constant dt right? May 05 12:39:06 also, where do you even use that assumption? May 05 12:39:39 seriously, can you go back to irclog and read what I said _AGAIN_ ? May 05 12:39:47 That assumption is in how the Input curve is plotted. I assume my x axis (time) is 0.000, 0.001 etc... May 05 12:40:04 that assumption is wrong, you know the actual timing of the data PRU sends you May 05 12:40:15 WHICH I KEEP EXPLAINING TO YOU May 05 12:40:18 for the actual I do May 05 12:40:36 and that is what is plotted but when I am generating the input signal my assumption is what I told you May 05 12:40:42 which is wrong May 05 12:40:47 I can change that May 05 12:41:54 otherwise the code works great May 05 12:42:22 I just misinterpreted the dt thing being constant May 05 12:42:38 dt is constant, it's equal to the interval of load cell samples May 05 12:42:54 which is the period of your entire main loop May 05 12:42:56 so I can basically ignore all other operations, I do not need to account for any other instructions or is the PRU so fast May 05 12:42:58 which is the period of the PRU messages May 05 12:43:14 which is the period of EVERYTHING in your program.... it's all synchronized together May 05 12:43:27 right my call for a load cell value is blocking May 05 12:43:37 so it decides the overall speed May 05 12:44:31 exactly, so one load cell sample = one loop iteration = one signal sample used May 05 12:44:39 = one message sent to python May 05 12:45:56 so my dt ~ 1.67 ms May 05 12:46:01 the sequence number included in each message is the proper x-axis, measured in load cell samples May 05 12:46:02 which matches the graph May 05 12:46:06 damn you zmatt May 05 12:46:21 so basically to fix I need to shrink my list of 1000 to 600 May 05 12:46:25 sequence number modulo signal length is signal index May 05 12:46:37 if you want a 1 second period, then yes May 05 12:47:06 also, why are you back to 600 Hz sample rate? I thought you switched to 1200 ? May 05 12:47:29 maybe I misremember May 05 12:47:32 I could but then the data gets super noisy since it is raw. I will try it out May 05 12:47:36 ah May 05 12:47:49 i would need better equipment May 05 12:47:53 to do better May 05 12:47:55 or filter the data yourself May 05 12:48:24 I know even less about filtering than I do about the beaglebone lol May 05 12:48:39 :p so I think I will put that on the back burner May 05 12:48:56 plus wouldnt it slow everything down. I guess I could do it after the fact May 05 12:50:06 since I am not using it as a control value for anything just an output it does not have to be done real time May 05 12:50:51 brb May 05 12:53:56 which filter setting are you using? May 05 14:29:36 the IIR filter setting 2 May 05 14:36:58 I got to go do some lab stuff but zmatt don't you go and make a fancy filter on me lol be back later May 05 15:06:08 mattb0ne: oh I'm not, however I am going to show you how not-fancy the filtering in the EM100 is ;) https://pastebin.com/hwwVR6wT this is just meant for illustration (and also completely untested, didn't even check it compiles) May 05 15:06:44 for PRU it would need to be rewritten to avoid floats of course May 05 15:16:31 of course, since PRU isn't actually using the load cell measurement, you can just do the filtering in python May 05 15:22:50 also, I'm pretty sure the filter attenuation values at 200 Hz given in the EM100-G user manual are wrong, and I think I know how they messed up: they used the frequency response of the right filter spec (2nd-order butterworth with given cutoff frequency) but the wrong sample rate (600 Hz instead of 1200 Hz) May 05 15:25:49 (resampling to half the sample rate is done by first filtering at the original sample rate and then just discard every other sample) May 05 22:45:43 hello May 05 22:49:06 (hi?) May 06 00:00:17 @zmatt: You made it back! May 06 00:00:29 Where have you been? May 06 00:02:43 ??? May 06 00:03:01 No triple question marks today, sir. May 06 00:03:08 Where have you been? May 06 00:03:37 Sitting while chatting or walking in a pace mode being alleviated? May 06 00:03:48 Tell me! May 06 00:04:03 Please. May 06 00:04:19 what do you mean "where have you been?" ... I've been active in here every day since we last spoke, which was a few days ago May 06 00:04:27 Oh. May 06 00:04:30 Okay. May 06 00:04:38 When was the last time you went to Vegas? May 06 00:05:01 and was it for a BBB conference? May 06 00:05:20 I've never been in vegas, and the last time I was in the US was when I was still a teenager May 06 00:05:27 Oh. May 06 00:05:38 Did you travel to this lovely state of Louisiana? May 06 00:05:57 Or just being safe and going around? May 06 00:06:51 I am asking b/c some people, you know them, just seem to miss our state and then never come back if they have ever lived here. May 06 00:06:57 Anyway, no issue. May 06 00:07:00 I mostly remember New York, I think I've also been to Boston but I don't remember much, I may have been too young May 06 00:07:08 Oh. May 06 00:07:10 Okay. May 06 00:07:28 Well, Boston is known for pizza, the Red Socks...and etc. May 06 00:07:43 Oh! I dislocated my back taking off some red sock. Ha. May 06 00:07:45 No joke. May 06 00:08:03 I just healed today and went back to work. Now... May 06 00:08:32 Did you ever...did you ever reply to me asking about your resource for GPIO? May 06 00:09:07 dunno, check irc log? I'm not sure what question you're talking about May 06 00:09:28 You made a GPIO library for the BBB with sysfs. May 06 00:09:39 I think it is called Sysfs-GPIO. May 06 00:09:44 yes, what about it? May 06 00:09:57 Have you left it in the past or are you still working on it? May 06 00:10:14 what work does it still need? May 06 00:10:19 And! The kicker. Are you going to add PWM functionality to it? May 06 00:10:21 Oh. May 06 00:10:25 None but PWM. May 06 00:10:30 no since it's a gpio library, not a pwm library May 06 00:10:47 Oh. Okay. I accept it and lie in wait for something to make me understand. May 06 00:11:21 Your GPIO lib. does not need anything. It covers all bases. May 06 00:11:42 Up, down, pos, neg, and in, out. May 06 00:12:13 it provides the functionality I wanted it to have, which is pretty much all the functionality of the sysfs gpio interface itself May 06 00:12:15 More than I would have ever thought I needed. But, there are use cases for rising edge or falling edge. May 06 00:12:23 Right. I noticed. May 06 00:12:38 I even looked up die. May 06 00:12:47 I do not understand it. May 06 00:12:56 It is like, do it or die. May 06 00:12:58 Ha. May 06 00:13:07 And by die, error out. May 06 00:13:20 it prints a message and exits... what's there to not understand? May 06 00:14:00 Previously, I was reading the man pages on die. I must have read them over and over blindly. B/c I missed exactly how it does it. May 06 00:14:10 it has no man page, it's just a function in my code May 06 00:14:18 Oh. May 06 00:14:27 I looked up die in the man pages. May 06 00:14:33 This is where I took a wrong turn. May 06 00:14:48 I found it if you wanted to know. May 06 00:14:56 Let me see if I can find it again. May 06 00:15:06 then whatever you found is unrelated May 06 00:16:03 Okay. May 06 00:16:17 Well, I am hitting 0 for 0 now. I probably should quit it for now. May 06 00:16:36 @zmatt: Again, thank you for being there in chat to calm me down. May 06 00:16:39 I needed it. May 06 00:17:06 Although based slightly around the BBB b/c of sysfs and GPIO, my take on it went left and quickly. Sorry. May 06 00:17:33 my library is not BBB-specific May 06 00:17:38 Oh. May 06 00:17:43 Hmm. Okay. May 06 00:18:00 I figured it could work with other related types of boards w/ Linux and sysfs. May 06 00:18:40 it should work on any linux system that has gpio May 06 00:18:45 I found a lot of source these days. People just give random snippets out. Some work and some do not work. But...they work for a reason and they do not work for a reason. May 06 00:18:46 Nice! May 06 00:19:06 This is the excitement of my current state for now. May 06 00:19:15 and building w/ crosstool-ng. May 06 00:19:21 anyway, I'm off, zZ May 06 00:19:24 Fine. May 06 00:19:27 I love me too. May 06 00:19:44 Typo. May 06 00:19:57 I should have said, "I really love me too!" May 06 00:20:02 Ha. May 06 00:20:05 Later! **** ENDING LOGGING AT Fri May 06 02:59:56 2022