**** BEGIN LOGGING AT Fri May 19 03:00:04 2017 May 19 15:36:23 hii everybody May 19 15:38:05 ds2: , nerdboy: , zeekhuge: I have sent an email May 19 15:39:04 I was reading currently the technical reference manual of am335x May 19 15:39:50 and also discussed our project on PRUDAQ google group, May 19 15:40:59 https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!topic/prudaq-users/nWLYR0npKHo May 19 15:46:33 don't see anything May 19 15:50:06 ds2: is your email is this??hy-gsoc@hy-research.com May 19 15:50:23 thetransformerr: yes May 19 15:50:41 ah,I see a bounce from a .in address... is that u? May 19 15:50:49 yeah May 19 15:51:03 do you have a gmail account? May 19 15:51:10 I had registered in gsoc with this id May 19 15:51:20 doh May 19 15:52:24 it is getting bounced due to excess spam May 19 15:52:53 may I send it with another id?? May 19 15:53:27 try it May 19 15:53:45 gmail.com would work?? May 19 15:55:04 it should May 19 15:58:51 I have sent it, May 19 15:58:54 although email does not present the complete plan, though it contains the possibilities that we can implement in our project May 19 16:00:03 i see it May 19 16:00:08 let me digest May 19 16:01:02 pulse of 40Khz? that makes no sense... May 19 16:01:16 40KHz implies it is repeating... a pulse is well.... a pulse May 19 16:04:05 by pulse I meant trigger here, however I think I forgot to add its time period May 19 16:07:18 . May 19 16:08:38 you mean a 25uS pulse? May 19 16:10:27 a trigger signal that can generate 40 kHz sound wave when applied to sensor May 19 16:10:44 25 microseconds would do I guess May 19 16:11:59 This is s a smart sensor that just requires a Go signal? May 19 16:13:05 what HW do you have on hand? May 19 16:14:06 I cant find much info on input and pulse duration, but I think they can be modified later according to our needs, and in datasheet May 19 16:14:43 of sensor they applied a wave of 40kHz to sensor, and since sqaure wave is easisest to apply May 19 16:14:55 I havent received beagleboard May 19 16:15:09 k May 19 16:15:24 but I do have multimeter and access to oscilloscope in college May 19 16:16:43 by the way jkridner is there anything that can be done about those customs fees? May 19 16:21:50 thetransformerr[: are you going to be using the PRUDAQ and a dev board for the murata signal source? May 19 16:22:14 jkridner: are you in the SF bay area now? May 19 16:22:34 yes May 19 16:24:09 theransformerr[: so are you trying to figure out where to hook up the murata signal source? May 19 16:24:26 ds2: no. I got sick last week, had surgery and am not traveling. May 19 16:24:37 ds2: gallbladder was removed due to stones. May 19 16:25:51 yeah, I wanted to know how to connect sensor such that it can be used as receiver and transmitter May 19 16:26:21 jkridner: oh :( hope you are recovering ok May 19 16:26:51 by dev board you mean a custom soldered board or something like vero May 19 16:27:06 thetransformerr[: how many of them do you have? May 19 16:27:07 jkridner: Get Well soon :) May 19 16:30:15 currently I have no sensors, and actually I was thinking it would have been nice if you or nerdboy first implement the circuit, after I create the required code, dts, and other software stuff May 19 16:30:36 thetransformerr[: okay... how many do you plan to use? May 19 16:31:02 is this the part - http://www.murata.com/en-us/products/productdetail.aspx?cate=cgsubUltrasonicSensors&partno=MA40S4R May 19 16:31:03 because I think adc is a sensitive device May 19 16:32:00 ADC tend to indeed be sensitive May 19 16:32:44 yeah :) May 19 16:33:16 if you use 8, it could simplify the HW May 19 16:34:07 the link you sent may prove to be good for testing purpose, but firstly it is not waterproof and secondly they cant function as transmitter May 19 16:34:26 I also thought of such method May 19 16:35:29 it is nice and simple for initial stage I think and later we can replace them accordingly to our requirements May 19 16:35:39 do you know what the exact part is then? May 19 16:37:12 just to answer your Q, the EHRPWM can do 40KHz as the base freq May 19 16:37:25 but it may not be exactly 40.000KHz May 19 16:38:30 I guess there is range to which receiver is sensitive May 19 16:38:52 thus it can vary as long as it is detected by receiver May 19 16:39:29 in case of 8 sensors the part you sent link of is good with its transmitter counterpart May 19 16:40:06 and with 4 sensors of transreceiver type I want such type of sensor May 19 16:41:22 https://pastebin.com/K2vgrkfp <--- take a look at that May 19 16:41:49 8 sensors would require you to have a chirp or something to do ToF May 19 16:42:26 nevermind... it would not as long as you switch it May 19 16:43:13 hold one sec... better diagram coming up May 19 16:43:37 http://www.murata.com/en-us/products/productdetail.aspx?cate=cgsubUltrasonicSensors&partno=MA58MF14-7N May 19 16:46:09 and does ehrpwm generates signal with help of PRU or main proceesor clock, because PRU may not be available to pwm generator May 19 16:46:28 the EHRPWM runs by itself May 19 16:46:48 I have used it to generate PWM audio while using the PRU to generate video May 19 16:47:26 than I guess it is going to be easier since main processor and PRU both share a common memory May 19 16:48:01 we can have a nice way to identify the time period May 19 16:48:24 better diagram - https://pastebin.com/2mJE6CYu May 19 16:49:17 the 2:1 mux might have to be discrete fets for power handling May 19 16:50:03 the MA58MF14-7N is a 58.5KHz part May 19 16:51:20 sorry I sent you wrong link May 19 16:51:22 http://uk.farnell.com/multicomp/mcusd18a40s09rs-30c/transceiver-40khz-18mm-metal/dp/2362680 May 19 16:52:26 the files nerdboy shared with me in drive lists a waterproof sensor from murata of 40kHz but I cant find it on their official website May 19 16:52:41 anyways any generic transducer would do May 19 16:54:32 unless you find another supplier, you can't use this one May 19 16:54:45 only 2 available. est.arrival of more in July May 19 16:54:47 too late for GSoC May 19 16:55:28 I would try it in local market in my city, they are much faster May 19 16:55:40 I wonder if you can reduce this to 2 sensors May 19 16:55:48 and would also check it on indian websites May 19 16:56:16 I'll leave it to nerdboy to build a driver circuit May 19 16:56:21 they are a PITA May 19 16:56:51 atleast 3 sensors are required, that too if we change the structure to calculate wind in two dimension May 19 16:56:58 thetransformer[[: do you know what physical HW they used last year? May 19 16:57:02 3 sensors? May 19 16:57:04 explain May 19 16:57:10 I don't think nerdboy is going to build a driver circuit May 19 16:57:25 and if you are thinking about hc-sr04 that isn't going to work May 19 16:57:49 I would build circuit but I would require guidance about voltage handling May 19 16:58:09 that I think I would get easily :) May 19 16:58:13 nerdboy is mr_science not mr_hardware after all May 19 16:58:33 wormo: hmmmmm May 19 16:59:49 I am thinking of 2 of the murata sensors in the X and Y axies along with a echo surface for each May 19 17:00:14 assumption is - wind in one direction will increase ToF and in the reverse decrease ToF May 19 17:01:20 and how would you then calculate the time taken by sound to travel to surface and surface to sensor May 19 17:01:34 yes that's how it works, but without a well done driver circuit I bet it's going to be hard to get real data May 19 17:02:04 thetransformer[[: good point... was just thinking outloud May 19 17:02:17 let us assume s be sound velocity and a be a wind velocity May 19 17:02:26 okk:) May 19 17:02:41 problem is piezos want a high voltage May 19 17:02:55 the 58KHz part has a driver circuit in the datasheet May 19 17:02:59 note the boost May 19 17:03:36 however that may work with three sensors by some sort of phase analysis but that requires dsp May 19 17:04:02 let me share another circuit also May 19 17:04:49 wormo: would you like to build it? :D May 19 17:05:04 NOPE NOPE May 19 17:05:45 This is why I was lobbying for using more hobbyist sensors May 19 17:06:34 hobbyist sensors?? May 19 17:06:56 Something that people actually use with arduino projects etc May 19 17:07:24 not something that assumes hw engineer skills to get it taking real data May 19 17:07:55 it could be on an eval board or something to make it realistic for a *software* project to use May 19 17:09:37 but sonic anemometer is sophisticated device, and therefore it deserves little bit more efforts, May 19 17:10:09 we can build circuit using IC and other parts that were used by other projects May 19 17:10:20 to implement sonic anemometer May 19 17:10:33 would could hack up one of those distance modules May 19 17:11:08 the only problem with them is the analog output is not available. May 19 17:11:34 uptill I have now just found 3 completed projects for anemometer, that may indicate its complexity May 19 17:12:13 ds2: have you a read a blog where distance module was reverse engineered May 19 17:12:41 let me share link if I could find it in my history May 19 17:13:02 * thetransformerr[ uploaded an image: Screen Shot 2017-05-19 at 10.38.09 PM.png (231KB) May 19 17:13:04 this is another circuit May 19 17:15:20 In my experience hacked up boards rather than clean pcb are a frequent source of error in non-trivial projects, so prepare to spend a lot of time troubleshooting hardware unless you do a spin of a well-reviewed board May 19 17:16:01 https://uglyduck.ath.cx/ep/archive/2014/01/Making_a_better_HC_SR04_Echo_Locator.html May 19 17:16:43 yes, I was just about to send you that link May 19 17:16:52 we can implement few concept used in this board, they use rs-232 to convert voltage level May 19 17:18:24 that is a cheap way to get +/- voltages May 19 17:19:20 does it work for us?? :) May 19 17:20:33 probally May 19 17:22:17 the circuit you sent in paste bin is useful and may be completes our discussion for interfacing with bb and now I guess driving circuit is major task May 19 17:23:46 nerdboy: you here?? May 19 17:23:57 not yet May 19 17:27:00 thought he was suppose to be around by 10? May 19 17:27:23 He left me a msg at 9:40 that he was running a bit late... May 19 17:27:38 I'm guessing soon May 19 17:28:11 Thought he might be here about now actually, maybe he didn't leave instantly after that May 19 17:36:38 i think I have the answer - https://www.digikey.com/product-detail/en/maxim-integrated/MAXQ7667EVKIT-1-/MAXQ7667EVKIT-1--ND/4271390 May 19 17:37:01 build 1 axis (use a pair of those boards) May 19 17:37:13 you can pull out the analog data at the jumper headers May 19 17:38:03 wormo: does that satisfy your off the shelf requirement? :D May 19 17:38:08 Yay! May 19 17:38:17 Nice find May 19 17:38:58 ideally the controller should he put into Hi-Z and we drive it but we can work in steps May 19 17:39:09 use the onboard stuff to generate the bursts May 19 17:39:30 and just use PRUDAQ to read the data May 19 17:39:43 but is $68.85 not quite bit high :| May 19 17:39:57 if that works, figure out how to hi-z it May 19 17:40:06 No that's *good* for an eval board, it could save a lot of time May 19 17:40:22 I'll defer the issue of pricing and funding to the appropriate controllers of the purse :D May 19 17:40:24 jkridner: I hope you have a quick recovery. take care :) May 19 17:42:26 hmmm could be as simple as holding it in reset via the jtag... will leave it as an exercise for the student to confirm it in te datasheet ;) May 19 17:42:33 it has adc also May 19 17:42:35 5-Channel, 12-Bit SAR ADC with 250ksps Sampling Rate May 19 17:42:52 thetransfermerr[: let's stick with PRUDAQ/Beagle stuff for now May 19 17:43:17 I got a feeling we can do it all on the Beagle w/o the PRU May 19 17:43:50 okk I start coding assuming a driving circuit May 19 17:44:12 cuz we have a 250Ksps ADC (alas, horrible noise) and a way to generate the bursts (EHRPWM or even regular PWM) May 19 17:44:36 thetransformerr[: confirm this with the others May 19 17:44:44 but I am getting feeling that one axis can easily be done with beagle, 2 or 3 axis would be next to impossible May 19 17:45:07 do you need them at the same time? May 19 17:45:27 cuz I'd argue you can mux them May 19 17:46:15 there is enough PWM for individual control... but you can always drive all 3 axis and only listen to the one of interest...we have enough analog inputs exposed (this is the non PRU path) May 19 17:46:47 they are good also for discrete as long as we can reduce latency to low scale May 19 17:46:59 *nod* May 19 17:47:17 the problem is time... to piezos want a compartively high voltage May 19 17:47:26 note the xformer in the dev board May 19 17:47:39 or the max232 (which gives you about 5-10Vpp swing) May 19 17:47:52 some of the muratas want 120Vpp May 19 17:49:25 and I am not aware how does 250ksps onboard adc stores result, would non premptive nature of linux kernel cuse any potential trouble May 19 17:49:45 the ADC can be set to autosample at a fix rate May 19 17:49:51 and it has a FIFO May 19 17:49:58 I have used it as a microphone input May 19 17:50:02 hence hte noise complaints May 19 17:50:16 it was good enough for pocketsphinx to work May 19 17:50:45 I guess noise occurs due to low sampling rate May 19 17:51:05 as we cant recreate wave more closely?? May 19 17:51:10 no, it samples at up to 250Ksps; I did 8K May 19 17:51:39 no... there are just random noise so I effectively get like 9bits of data May 19 17:51:58 wish I had the plots handy... I fed in a nice 1KHz sine wave and I get lots of fuzz on it May 19 17:52:29 this project sometimes seems so much trivial and at next moment it throws you a complete new possiblity May 19 17:52:50 let's not get ahead of ourselves May 19 17:52:53 or problem ... May 19 17:53:20 there might be accuracy issues... donno how fast you need to sample to get a decent resolution on airspeed May 19 17:53:22 :) of course May 19 17:53:38 if that is a problem, then you do need to look at phasing May 19 17:53:45 can be done w/o a DSP May 19 17:54:33 not to confuse things but I wonder if it is more effective to run the input to a limiter instead of an ADC May 19 17:54:40 then just look at edge timing May 19 17:55:01 according to nyquist we are good with 80kHz,but good resolution requies about 10 times may be May 19 17:55:05 then you get almost 100MHz equiv sampling May 19 17:55:15 let nerdboy: come up online May 19 17:55:26 limiter?? May 19 17:55:30 I assume you are looking for the leading edge of the signal, right? May 19 17:55:49 send... wait for first cycle of signal, subtract times and call it a day May 19 17:56:13 problem there is at 80KHz, you have 12.5uS resolution... donno what that comes out in mph of wind May 19 17:56:37 think of a limiter as a very very very very large gain amplifier so it ALWAYS clips the signal May 19 17:56:56 so you are turning the received signal into "square" wave May 19 17:57:04 yeah it is the plan, but it is considered the least accurate method May 19 17:57:09 then you read it on a GPIO pin and just look for a transition May 19 17:58:36 if you want to research it - look at FM radio demo from the analog days... limiting was used to remove AM noise (yes, there is AM noise on a FM signal) May 19 17:58:38 give me time I get more knowledge about feasiblity of this and phase method May 19 17:58:52 s/demo/demodulation/ May 19 18:03:15 till then we are going ahead with PRU stuff and analog switch stuff May 19 18:08:34 isn't there a 4 channel analog switch on the PRUDAQ? May 19 18:18:32 yes it has May 19 18:19:06 that is to select input source in a input channel May 19 18:19:51 since it has two channel with four source, it can handle 8 in round robin fashion May 19 18:33:04 ds2: please review these circuits too May 19 18:33:05 m_w: pcbs looks great May 19 18:33:45 * thetransformerr[ uploaded an image: Screen Shot 2017-05-20 at 12.01.58 AM.png (442KB) May 19 18:33:48 * thetransformerr[ uploaded an image: Screen Shot 2017-05-20 at 12.02.28 AM.png (816KB) May 19 18:34:16 * thetransformerr[ posted a file: ProjectReport.pdf (1235KB) May 19 18:36:32 pmezydlo: now I just have to assemble them May 19 18:37:08 likely going to do the assembly tonight May 19 18:37:38 there resistors are very small May 19 18:38:02 you must have good eyes May 19 18:38:33 pmezydlo: I dealt with 0402 components when assembling the pocketbone May 19 18:38:59 luckily I can see fairly well up close May 19 18:39:18 can you take a photo after assembly May 19 18:39:40 pmezydlo: of course May 19 18:39:45 thetransformerr: what's that for? May 19 18:40:36 m_w: I can't do anything but to watch May 19 18:44:43 pmezydlo: I can make a video maybe May 19 19:25:39 m_w: did you managed to get grove connectors? May 19 19:26:20 pmezydlo: no, I did not get any response from seeed studio May 19 19:28:52 this is no problem, does not affect the project May 19 19:29:21 it is just annoying that they use a connector that you can only source in China May 19 19:32:01 I thought it was a very popular connectors May 20 00:25:39 https://www.youtube.com/watch?v=Dk-8eI10qhM **** ENDING LOGGING AT Sat May 20 03:00:02 2017