From nick at ucc.gu.uwa.edu.au Mon Feb 22 09:08:22 2021 From: nick at ucc.gu.uwa.edu.au (Nick Bannon) Date: Mon, 22 Feb 2021 09:08:22 +0800 Subject: [PLUG-AV] Found failed part on the dead Numato In-Reply-To: <167df0ba-0333-46c7-039f-483709878dfe@decisions-and-designs.com.au> References: <20210211014508.GC12269@ucc.gu.uwa.edu.au> <2ee3de46-c645-be48-aab8-036188be02a6@decisions-and-designs.com.au> <167df0ba-0333-46c7-039f-483709878dfe@decisions-and-designs.com.au> Message-ID: <20210222010822.GH12269@ucc.gu.uwa.edu.au> Heya, Harry! Sure, if you've got an order going in, grab 'em now. (I haven't done an order yet, though I'm sure I can get a minimum shipping order together, still happy to. Yes, PLUG has a credit card handy for online orders, or can reimburse promptly.) Thanks, Nick. On Sun, Feb 21, 2021 at 10:18:29PM +0800, Harry McNally wrote: > Ah ha! I see a PLUG + POSH do on March 14. Do you think we can fix the > Numoto by then ? > > The stock detail on that part is: > > "*197* available for 4 - 5 business days delivery" > > which means it is coming out of the UK. So that should be enough time. I > have some parts I may order shortly. If I do, I will give you a text or ring > to see if you have odrdered some and, if not, add to my order. Or let me > know if/when you have ordered. I realised PLUG may not have a CC which is > the easiest way to place a guest order. > > All the best > Harry > > On 11/2/21 11:18 am, Harry McNally wrote: > > Hello Both > > > > On 11/2/21 9:45 am, Nick Bannon wrote: > > > Heya, Harry, Patrick! > > > > > > On Wed, Feb 10, 2021 at 01:11:24PM +0800, Harry McNally wrote: > > > > Hi Nick > > > > See attached. U8 had failed in the 1.5 volt supply and was overcurrent for > > > > the 1A plugpack. Plugpack shut itself down after 45mS then retried to start > > > > after a second which is why the Numato LEDs were flashing. I think this may > > > > have saved the board; fingers crossed. > > > Fabulous! Was that all hardware diagnosis? > > > > Yes. I used my low impedance meter to see where on the board the > > resistance was lowest across the 12V supply and this led to that PSU > > circuit and the TI switcher was the only possible cause. Once removed, > > the impedance matched the good board (well .. "open circuit" from a low > > impedance perspective). > > > > > ? and/or have you tried talking to the firmware on either board over USB? > > > > You've got me. Many months ago I was shopping for a Xilinx JTAG > > programmer clone and settled in a Open PCB to build up (hasn't arrived > > even now) I did this assuming I had to bring up a bricked board (and to > > test the FPGA spoke back to me etc). > > > > Because of that expected entry point I didn't realise that there was > > also USB access to JTAG. So I haven't tried this access to either the > > good or the dead board yet. > > > > > > This is U8: > > > > https://www.ti.com/lit/ds/symlink/tps54625.pdf > > > ooh, up to 6.5A output...? > > It may not be used to deliver 6.5A although 1.5V supplies in FPGAs (like > > PC CPUs) can be quite high currents. > > > > I have removed it and that has cleared the 12V supply loading but I don't > > > > want to power the circuit in case 1.5V is necessary to prevent failures > > > > elsewhere. > > > > > > > > I'll will add some of these parts to another order (or freight is $15). > > > > Element14 have them: > > > > https://au.element14.com/texas-instruments/tps54625pwp/ic-sync-step-down-650khz-htssop/dp/3007368?ost=tps54625 > > > > > > > Lovely - and don't be out of pocket, we're happy to pay. > > > Thank you for your time so far. > > > > I'm a bit away from another order so, if you like, jump on the element14 > > site and order say three pieces with a CC as a guest (no difference in > > price). Sometimes element14 run specials for free delivery but I haven't > > seen that offer code come through recently. So freight will be $14 or so > > which doubles the price of three pieces unfortunately. I suggest three > > in case some intermittent short exists on the 1.5V rail. I can't see any > > now and the TI part should current limit and protect itself but not much > > point in buying one then "*smoke* oh, we actually need another one now" > > scenario. > > > > > I'm hoping we can get together for an AV hack session very soon, I'm sure > > > the new PLUG VP, Patrick, is up for it. We'd want to use a "working" > > > Opsis board, and probably a cheap black-box hdmi2usb widget too - but > > > debugging is really hard if every setup is a unique snowflake. > > > > Ok. Well what about if you get parts and I repair the PCB and see if we > > get the blue "loading" light (then off) like the good board. After that, > > have a good look at USB programming etc at the following AV meetup ? > > > > We can try the board access through USB and decide if we want to by an > > off-the-ebay Xiling JTAG debugger clone (around $30US vs $200US for the > > genuine Xilinx device). > > > > > We'd rather start with replicating someone else's system than figure it > > > all out for ourselves, or we just don't end up getting very far in a day: > > > > > > - https://debconf-video-team.pages.debian.net/docs/hardware.html > > > - https://debconf-video-team.pages.debian.net/docs/opsis.html > > > > > > Every Second Sunday - including this Valentine's weekend coming - > > > is an opportunity, but maybe we need something else to kickstart the > > > process again. > > Yes. I won't have the PCB ready by Sunday but happy to repair then > > attend the next meetup. > > > Had a lovely PLUG talk on Tuesday - but we were running around getting the > > > basics working at the last minute, like any kind of audio out, as usual... > > > https://meetings.ucc.asn.au/b/nic-4cd-3jg > > Listening and viewing now .. > > > > When I get them I need to use hot air to re-install (it has a heatsink tab > > > > underneath the package). But probing around the circuit, I cant see anything > > > > shorting the 1.5 volt supply rail so U8 alone may fix it. > > > > > > > > Smalls steps but progress none the less :-) > > > > > > > > All the best > > > > Harry > > > Thank you! > > > Nick. > > > > More than welcome Nick. It is a small contribution back to PLUG. > > Apologies I put this aside waiting for the JTAG PCB. > > > > All the best > > Harry -- Nick Bannon | "I made this letter longer than usual because nick-sig at rcpt.to | I lack the time to make it shorter." - Pascal From harrymc at decisions-and-designs.com.au Mon Feb 22 11:15:26 2021 From: harrymc at decisions-and-designs.com.au (Harry McNally) Date: Mon, 22 Feb 2021 11:15:26 +0800 Subject: [PLUG-AV] Found failed part on the dead Numato In-Reply-To: <20210222010822.GH12269@ucc.gu.uwa.edu.au> References: <20210211014508.GC12269@ucc.gu.uwa.edu.au> <2ee3de46-c645-be48-aab8-036188be02a6@decisions-and-designs.com.au> <167df0ba-0333-46c7-039f-483709878dfe@decisions-and-designs.com.au> <20210222010822.GH12269@ucc.gu.uwa.edu.au> Message-ID: <6e1f71d8-048a-ee85-a9fa-78e944a1be5b@decisions-and-designs.com.au> Cheers Nick I am prevaricating on the order until I get some more details from another supplier. So, if you are happy to use the CC and get some in, I will get one on the board and see if it will "boot" Just let me know if you get time to order so I don't order as well :-) All the best Harry On 22/2/21 9:08 am, Nick Bannon wrote: > Heya, Harry! Sure, if you've got an order going in, grab 'em now. > > (I haven't done an order yet, though I'm sure I can get a minimum shipping > order together, still happy to. Yes, PLUG has a credit card handy for > online orders, or can reimburse promptly.) > > Thanks, > Nick. > > On Sun, Feb 21, 2021 at 10:18:29PM +0800, Harry McNally wrote: >> Ah ha! I see a PLUG + POSH do on March 14. Do you think we can fix the >> Numoto by then ? >> >> The stock detail on that part is: >> >> "*197* available for 4 - 5 business days delivery" >> >> which means it is coming out of the UK. So that should be enough time. I >> have some parts I may order shortly. If I do, I will give you a text or ring >> to see if you have odrdered some and, if not, add to my order. Or let me >> know if/when you have ordered. I realised PLUG may not have a CC which is >> the easiest way to place a guest order. >> >> All the best >> Harry >> >> On 11/2/21 11:18 am, Harry McNally wrote: >>> Hello Both >>> >>> On 11/2/21 9:45 am, Nick Bannon wrote: >>>> Heya, Harry, Patrick! >>>> >>>> On Wed, Feb 10, 2021 at 01:11:24PM +0800, Harry McNally wrote: >>>>> Hi Nick >>>>> See attached. U8 had failed in the 1.5 volt supply and was overcurrent for >>>>> the 1A plugpack. Plugpack shut itself down after 45mS then retried to start >>>>> after a second which is why the Numato LEDs were flashing. I think this may >>>>> have saved the board; fingers crossed. >>>> Fabulous! Was that all hardware diagnosis? >>> Yes. I used my low impedance meter to see where on the board the >>> resistance was lowest across the 12V supply and this led to that PSU >>> circuit and the TI switcher was the only possible cause. Once removed, >>> the impedance matched the good board (well .. "open circuit" from a low >>> impedance perspective). >>> >>>> ? and/or have you tried talking to the firmware on either board over USB? >>> You've got me. Many months ago I was shopping for a Xilinx JTAG >>> programmer clone and settled in a Open PCB to build up (hasn't arrived >>> even now) I did this assuming I had to bring up a bricked board (and to >>> test the FPGA spoke back to me etc). >>> >>> Because of that expected entry point I didn't realise that there was >>> also USB access to JTAG. So I haven't tried this access to either the >>> good or the dead board yet. >>> >>>>> This is U8: >>>>> https://www.ti.com/lit/ds/symlink/tps54625.pdf >>>> ooh, up to 6.5A output...? >>> It may not be used to deliver 6.5A although 1.5V supplies in FPGAs (like >>> PC CPUs) can be quite high currents. >>>>> I have removed it and that has cleared the 12V supply loading but I don't >>>>> want to power the circuit in case 1.5V is necessary to prevent failures >>>>> elsewhere. >>>>> >>>>> I'll will add some of these parts to another order (or freight is $15). >>>>> Element14 have them: >>>>> https://au.element14.com/texas-instruments/tps54625pwp/ic-sync-step-down-650khz-htssop/dp/3007368?ost=tps54625 >>>>> >>>> Lovely - and don't be out of pocket, we're happy to pay. >>>> Thank you for your time so far. >>> I'm a bit away from another order so, if you like, jump on the element14 >>> site and order say three pieces with a CC as a guest (no difference in >>> price). Sometimes element14 run specials for free delivery but I haven't >>> seen that offer code come through recently. So freight will be $14 or so >>> which doubles the price of three pieces unfortunately. I suggest three >>> in case some intermittent short exists on the 1.5V rail. I can't see any >>> now and the TI part should current limit and protect itself but not much >>> point in buying one then "*smoke* oh, we actually need another one now" >>> scenario. >>> >>>> I'm hoping we can get together for an AV hack session very soon, I'm sure >>>> the new PLUG VP, Patrick, is up for it. We'd want to use a "working" >>>> Opsis board, and probably a cheap black-box hdmi2usb widget too - but >>>> debugging is really hard if every setup is a unique snowflake. >>> Ok. Well what about if you get parts and I repair the PCB and see if we >>> get the blue "loading" light (then off) like the good board. After that, >>> have a good look at USB programming etc at the following AV meetup ? >>> >>> We can try the board access through USB and decide if we want to by an >>> off-the-ebay Xiling JTAG debugger clone (around $30US vs $200US for the >>> genuine Xilinx device). >>> >>>> We'd rather start with replicating someone else's system than figure it >>>> all out for ourselves, or we just don't end up getting very far in a day: >>>> >>>> - https://debconf-video-team.pages.debian.net/docs/hardware.html >>>> - https://debconf-video-team.pages.debian.net/docs/opsis.html >>>> >>>> Every Second Sunday - including this Valentine's weekend coming - >>>> is an opportunity, but maybe we need something else to kickstart the >>>> process again. >>> Yes. I won't have the PCB ready by Sunday but happy to repair then >>> attend the next meetup. >>>> Had a lovely PLUG talk on Tuesday - but we were running around getting the >>>> basics working at the last minute, like any kind of audio out, as usual... >>>> https://meetings.ucc.asn.au/b/nic-4cd-3jg >>> Listening and viewing now .. >>>>> When I get them I need to use hot air to re-install (it has a heatsink tab >>>>> underneath the package). But probing around the circuit, I cant see anything >>>>> shorting the 1.5 volt supply rail so U8 alone may fix it. >>>>> >>>>> Smalls steps but progress none the less :-) >>>>> >>>>> All the best >>>>> Harry >>>> Thank you! >>>> Nick. >>> More than welcome Nick. It is a small contribution back to PLUG. >>> Apologies I put this aside waiting for the JTAG PCB. >>> >>> All the best >>> Harry From james at jamesh.id.au Tue Mar 9 10:23:00 2021 From: james at jamesh.id.au (James Henstridge) Date: Tue, 9 Mar 2021 10:23:00 +0800 Subject: [PLUG-AV] Another experiment in updating our AV workflow Message-ID: As an evening project, I've been having another go at trying to modernise PLUG's recording/streaming workflow. I think I've got enough working to demonstrate some of it at tonight's PLUG meeting, but thought I'd outline some of the details here. Benjamin had mentioned OBS Studio as a possible way forward, so that's what I used as my starting point. The main thing it's missing from our old dvswitch based workflow was an easy way to integrate remote cameras into the scene, so that's what I've been focusing on. The work in progress code can be found in this repo: https://github.com/jhenstridge/obs-rtsp-source It comprises two pieces: 1. an RTSP server to share a camera stream over the network 2. an OBS plugin that helps configure OBS Studio's standard "Media Source" to consume these streams. The RTSP server advertises each stream via Zeroconf using Avahi, so you can see all the cameras available on the network with "avahi-browse -rt _rtsp._tcp". I'm still working on adding the Avahi support to the OBS plugin so it can select cameras by published name rather than an rtsp:// URI. The eventual goal is to allow the different parts to self configure. If you turn the camera nodes on first, then the OBS machine should connect to them when it starts. If the OBS machine starts first, it will wait until the cameras come online, and plug them into the scenes you've set up previously. If everything works correctly, then the RTSP sender machines could easily be headless: the RTSP server would start on boot, and Avahi would advertise the camera once it gets a DHCP address. Given that tonight is our Raspberry Pi Jam, I thought I'd put this to the test, and have configured a Pi running the RTSP software. So hopefully we can give this a short demonstration. James. From james at jamesh.id.au Fri Mar 12 11:33:36 2021 From: james at jamesh.id.au (James Henstridge) Date: Fri, 12 Mar 2021 11:33:36 +0800 Subject: [PLUG-AV] Another experiment in updating our AV workflow In-Reply-To: References: Message-ID: On Tue, 9 Mar 2021 at 10:23, James Henstridge wrote: > The eventual goal is to allow the different parts to self configure. > If you turn the camera nodes on first, then the OBS machine should > connect to them when it starts. If the OBS machine starts first, it > will wait until the cameras come online, and plug them into the scenes > you've set up previously. If everything works correctly, then the > RTSP sender machines could easily be headless: the RTSP server would > start on boot, and Avahi would advertise the camera once it gets a > DHCP address. I think I've got this part of the plugin working now. The source is now gets you to choose a service name for the stream (e.g. "Camera-A" in the sample rtsp-sender.conf), providing an option list of all available cameras on the local network. When the camera drops from the network, the underlying ffmpeg_source is reconfigured with an empty URL, and when it appears again we configure it with the new URL. This means that OBS reconnects to the camera almost instantly when it becomes available. So that pretty much covers what I wanted from an auto-configuration stand-point. If you set up all the OBS scenes you want for a broadcast in advance, it would pretty much be a case of plugging everything in and having it self organise. There is no need for a display on the sender machines, so it could easily be a Raspberry Pi like I demoed this week. We could even store the Pi and capture card wired together, so setup would only involve connecting power, ethernet, and HDMI. James.