Igor Posted July 10, 2016 Posted July 10, 2016 I guess it must be a bug in displaying since there is no wifi tag within Nanopi M1 ... cant fix now. Wrote on mobile phone
tkaiser Posted July 10, 2016 Posted July 10, 2016 I guess it must be a bug in displaying since there is no wifi tag within Nanopi M1 ... cant fix now. It's no real problem, only wrong description and an unnecessary module: # H3 quad core 1Gb SoC Wifi ... MODULES_NEXT="brcmfmac" IMO it would be more interesting to add support for the NEO and find some people testing DRAM reliability (still have a hard time believing 432 MHz are necessary if every other DRAM parameter is the same as on NanoPi One or any of the Oranges). But since the NEO has neither HDMI nor CVBS connector it requires somewhat skilled people to test since when starting with the usual approach (see starting from post #4 here) no visual feedback is available so we could only rely on temperatures (measuring cpuburn/cpuminer temperatures before, comparing with NanoPi M1 and then deciding whether a lima-memtester without physically available display is working correctly or not)
vlad59 Posted July 10, 2016 Posted July 10, 2016 I should be receiving 2 nanopi Neo in 3 weeks hopefully. I'll make all the tests needed as soon as received. <off topic>I can confirm that my Nanopi M1 work fine even at 744MHz so no need to lower the DRAM settings for them</off topic> 1
trewq Posted July 10, 2016 Posted July 10, 2016 It's no real problem, only wrong description and an unnecessary module: # H3 quad core 1Gb SoC Wifi ... MODULES_NEXT="brcmfmac" IMO it would be more interesting to add support for the NEO and find some people testing DRAM reliability (still have a hard time believing 432 MHz are necessary if every other DRAM parameter is the same as on NanoPi One or any of the Oranges). But since the NEO has neither HDMI nor CVBS connector it requires somewhat skilled people to test since when starting with the usual approach (see starting from post #4 here) no visual feedback is available so we could only rely on temperatures (measuring cpuburn/cpuminer temperatures before, comparing with NanoPi M1 and then deciding whether a lima-memtester without physically available display is working correctly or not) I have 10 of the NEOs on the way, they will hopefully be here this week. If you can give me a little bit more guidance on anything additional that needs to be done other than the post you linked I can run any tests for armbian to be optimized.
tkaiser Posted July 11, 2016 Posted July 11, 2016 I have 10 of the NEOs on the way, they will hopefully be here this week. If you can give me a little bit more guidance on anything additional that needs to be done other than the post you linked I can run any tests for armbian to be optimized. Simply download the Armbian test image (containing a statically linkes lima-memtester version and a modified kernel to not reclock DRAM) and follow the steps (installation of RPi-Monitor first, then starting lima-memtester), let it run for a few minutes and check temperatures using RPi-Monitor (or in a separate terminal window using 'sudo armbianmonitor -m'). If temperatures get pretty high then everything should be ok and testing works correctly. If temperatures are rather low something's wrong. In either case please let the test run for at least one hour, then get back to us with a screenshot showing RPi-Monitor's temperature curves and the output from 'sudo armbianmonitor -u' 1
trewq Posted July 11, 2016 Posted July 11, 2016 Simply download the Armbian test image (containing a statically linkes lima-memtester version and a modified kernel to not reclock DRAM) and follow the steps (installation of RPi-Monitor first, then starting lima-memtester), let it run for a few minutes and check temperatures using RPi-Monitor (or in a separate terminal window using 'sudo armbianmonitor -m'). If temperatures get pretty high then everything should be ok and testing works correctly. If temperatures are rather low something's wrong. In either case please let the test run for at least one hour, then get back to us with a screenshot showing RPi-Monitor's temperature curves and the output from 'sudo armbianmonitor -u' Thank man! That's exactly what I was after. I'll let you know when they arrive and I should be able to do the tests the same day.
Per-Mattias Nordkvist Posted July 12, 2016 Author Posted July 12, 2016 I really enjoy this conversation. For me the big advantage of NanoPi NEO is not the price but the size. Leaving out unnecessary components and having connections for secondary and third USB makes a lot of sense. I would gladly pay extra for better power management but it sounds from this conversation that power is manageable with decent cooling right?
wildcat_paris Posted July 12, 2016 Posted July 12, 2016 I really enjoy this conversation. For me the big advantage of NanoPi NEO is not the price but the size. Leaving out unnecessary components and having connections for secondary and third USB makes a lot of sense. I would gladly pay extra for better power management but it sounds from this conversation that power is manageable with decent cooling right? @Per-Mattias I guess yes, but the huge heatsink or fan will make the size less interesting ok, sorry I am kidding
tkaiser Posted July 12, 2016 Posted July 12, 2016 I would gladly pay extra for better power management but it sounds from this conversation that power is manageable with decent cooling right? Consumption and improved heat dissipation aren't directly related. Only if you're interested in constant full CPU load you have to take care of heat dissipation (huge heatsink or even combined with a fan) but regarding consumption it's all about configuration and reliability testing (you don't need to do yourself if you choose Armbian since we wasted a few days already digging deep into this stuff). BTW: I think I found why FriendlyARM chose just 432 MHz DRAM clockspeed on Nano Pi NEO: It helps with both consumption and temperatures. See results here: http://forum.armbian.com/index.php/topic/1614-running-h3-boards-with-minimal-consumption/ NanoPI NEO should be able to run with just 160mA idle consumption while still being able to increase performance by factor 5 when needed (everything adjustable from user space so no reboot needed, the H3 device can operate in low consumption mode and dynamically switch to power mode when needed) 1
trewq Posted July 13, 2016 Posted July 13, 2016 Looks like I won't be getting my order this week. Apparently due to a high volume of orders they are unable to send out my order until next week. Kinda sucks but can't do anything about it.
Wazou Posted July 13, 2016 Posted July 13, 2016 Hi guys ! Kudos for your work and keeping up with the newest board. I'm currently drafting a POE Hat for the nanoPi Neo, but without the board in hand it's hard to do the math for the power requirement. Can I have someone opinion on the the avg power consumption of this board, so I can dimension the step-down DC/DC stage of my HAT. Cheers,
tkaiser Posted July 13, 2016 Posted July 13, 2016 Can I have someone opinion on the the avg power consumption of this board, so I can dimension the step-down DC/DC stage of my HAT. Please look through this thread http://forum.armbian.com/index.php/topic/1614-running-h3-boards-with-minimal-consumption/ to get an idea how to 'design' power requirements of H3 devices. So in case you don't need full performance you can keep consumption below 2W if needed (might require u-boot adjustments to lower cpufreq in early boot stage but according to my measurements yesterday 2W / 400mA should be ok). The upper limit is simply up2u since it depends on how many CPU cores you allow to be active at which clockspeeds and other consumers like connected USB peripherals. BTW: You know that every FriendlyARM board (at least NanoPi / NanoPC) features a 4 pin header that provides serial console as well as the ability to feed DC-IN (pins 1 and 2 are for 5V and GND)?
trewq Posted July 13, 2016 Posted July 13, 2016 Please look through this thread http://forum.armbian.com/index.php/topic/1614-running-h3-boards-with-minimal-consumption/ to get an idea how to 'design' power requirements of H3 devices. So in case you don't need full performance you can keep consumption below 2W if needed (might require u-boot adjustments to lower cpufreq in early boot stage but according to my measurements yesterday 2W / 400mA should be ok). The upper limit is simply up2u since it depends on how many CPU cores you allow to be active at which clockspeeds and other consumers like connected USB peripherals. BTW: You know that every FriendlyARM board (at least NanoPi / NanoPC) features a 4 pin header that provides serial console as well as the ability to feed DC-IN (pins 1 and 2 are for 5V and GND)? To add to this, from the schematics and the pictures of the board there is at least one pair free on the ethernet cable. It may just be as simple as soldering a couple of wires on the board side of things.
Wazou Posted July 13, 2016 Posted July 13, 2016 Thank for the info's! I'll look into that. Yup, I was planning to feed in 5v directly into the header, but for the POE I'd rather feed > 6v and step-down / split later hence the question on power rating so I can select the right step-down switching hardware. ... so the 5v/2A advertised on friendlyARM wiki is a total overkill ? Cheers,
Wazou Posted July 13, 2016 Posted July 13, 2016 To add to this, from the schematics and the pictures of the board there is at least one pair free on the ethernet cable. It may just be as simple as soldering a couple of wires on the board side of things. I did not consider to POE directly from the on-board RJ45 plug, that's actually a good idea However I'm going to run for about 20/30 devices, making a plug PCB/HAT is probably more user friendly approach for my use case. I'll share the schematics, if someone here is interested.
trewq Posted July 13, 2016 Posted July 13, 2016 Thanks, I did not consider to POE directly on the on-board RJ45 plug, that's actually a good idea However I'm going to run for about 20/30 devices, making a plug PCB/HAT is probably more user friendly approach for my use case. You can disregard my comment on feeding directly in. Just read the schematics for the RJ45 (HR911105A) and no pins pass straight through from the plug to the board. To answer your other question it's best to plan for the worst and hope for the best so while 2A may seem high you never know what you may want to do in the future but it won't actually get that high without additional peripheral devices.
tkaiser Posted July 13, 2016 Posted July 13, 2016 To add to this, from the schematics and the pictures of the board there is at least one pair free on the ethernet cable. It may just be as simple as soldering a couple of wires on the board side of things. BTW: I use passive PoE with RPi B+ (surveillance cameras) with 24V or 48V PSUs and utilizing the 2 unused Ethernet cable pairs for power with simple step-down converters at the 'camera' side (24V is ok since cables aren't that long, 48V are only needed when trying to get close to specification limits of 100m). When starting to play with Oranges I did some research about input voltage range: http://www.orangepi.org/orangepibbsen/forum.php?mod=viewthread&tid=751 So depending on cable lengths one might save the step-down converters and simply inject 6.5V to end up with 4.5V - 5.5V based on distance and unfortunately also load (the latter is important since load peaks will automagically lead to temporarely voltage drops). I also suggested to Steven/Xunlong to provide this passive PoE approach directly on board (connecting the 2 PoE cable pairs to VCC/GND directly) but he didn't liked the idea. The power requirements FriendlyARM wrote in their Wiki are reasonable since H3 doesn't come with a PMIC and therefore USB consumers might easily exceed power requirements according to specs (500mA per port -- at least with Orange Pi PC it works perfectly to attach one 2.5" HDD to each USB port that consume close to 3A at spin-up! Maybe that's the reason Xunlong provides a 5V/3A PSU for the boards) 1
tkaiser Posted July 13, 2016 Posted July 13, 2016 However I'm going to run for about 20/30 devices, making a plug PCB/HAT is probably more user friendly approach for my use case. I'll share the schematics, if someone here is interested. Please share schematics here. I use the PoE splitters and WaveRF PoE injectors from here https://www.i4wifi.eu/EU-230V-powering-1/ (tested the 8 and 16 port variants) and at the device side simply soldered cutted DuPont wires to the step-down converter (no drawbacks when used with Orange Pi or NanoPi, less protection when feeding DC-IN this way on RPi). The problem here is that individual ports can not be switched on/off so to do an emergency reset I have to power-cycle the one central 24V PSU connected to the PoE injector. That means that peak consumption of all the boards booting in parallel defines the PSU's power dimensions. BTW: Isn't missing CE/UL certification for Orange Pi and NanoPi an issue?
ahrlad Posted July 13, 2016 Posted July 13, 2016 Not sure if relevant, but the FriendlyARM wiki of both the M1 and NEO does state that input voltage through the pin connector should function in the range of 4.7V~5.6V
vlad59 Posted July 13, 2016 Posted July 13, 2016 Looks like I won't be getting my order this week. Apparently due to a high volume of orders they are unable to send out my order until next week. Kinda sucks but can't do anything about it. My order has been delayed too
trewq Posted July 13, 2016 Posted July 13, 2016 My order has been delayed too I ordered very close to when it went up for sale. I'm wondering if they are having supply issues rather than it being an issue with high volume as I don't see how having a high demand would change anything (it's what they want as a business) unless they didn't have the stock.
ahrlad Posted July 13, 2016 Posted July 13, 2016 I ordered very close to when it went up for sale. I'm wondering if they are having supply issues rather than it being an issue with high volume as I don't see how having a high demand would change anything (it's what they want as a business) unless they didn't have the stock. It could easily be both. I can see this board getting a lot of domestic demand.
Wazou Posted July 13, 2016 Posted July 13, 2016 Please share schematics here. I will ! Right now I'm just going for passive POE injector @ 9-12 V. The PCB to fit right on top of the NEO with POE splitter (2x RJ45 jack) + DC/DC step-down. But I'm also adding other functionalities for my project: 2.5W Audio amplifier, Microphone boost input and a bunch of LEDs. I will base on 1.5A power rating for my step down convertor. PS : my order is also being delayed
trewq Posted July 14, 2016 Posted July 14, 2016 my new USB disk Nice! Did you just get the one? I wonder if it's only the 512MB model that's delayed then.
usboot Posted July 14, 2016 Posted July 14, 2016 Nice! Did you just get the one? I wonder if it's only the 512MB model that's delayed then. No one is delayed. 1
trewq Posted July 14, 2016 Posted July 14, 2016 No one is delayed. Mine are delayed as are another posters here. You must just be one of the lucky ones.
tkaiser Posted July 14, 2016 Posted July 14, 2016 FYI: https://github.com/friendlyarm/h3_lichee/issues/4 According to https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=127210 being able to let the NanoPi NEO idle at ~100mA (while being comparable to RPi Zero from a performance point of view) isn't that bad. But we can provide an OS image that allows performance to exceed RPi 2 level easily. Simply by adjusting settings correctly. A NanoPi NEO can idle like an RPi Zero and instantly transform into an RPi 2 when performance is needed (or anything in between, if the user prefers minimal consumption over highest performance, configuring this is also possible) In case FriendlyARM ships developer samples we might add this to Armbian (including a configuration tool where you can configure performance vs. consumption) 2
Per-Mattias Nordkvist Posted July 14, 2016 Author Posted July 14, 2016 my new USB disk Quick question on this. Are you powering the board through the pin cables or how are you providing power?
tkaiser Posted July 14, 2016 Posted July 14, 2016 Quick question on this. Are you powering the board through the pin cables or how are you providing power? Both possible, FriendlyARM's H3 boards can be powered through the 4-pin header or through Micro-USB (if power through header exceeds the one available from Micro USB then devices can even be charged connected to the Micro USB port -- FriendlyARM unlike most other Chinese SBC vendors provides excellect documentation so simply look through their wiki)
Recommended Posts