1st Banana Pi M3 review


#1

Our beloved hardware manufacturer of choice sent review samples out and the crowd started to…: http://forum.armbian.com/index.php/topic/474-quick-review-of-banana-pi-m3/

Stay tuned…


(lionwang) #2

thank you for you cool test.

i know, you always criticize us , but your are right ,so many things we do not very well , but we try our best to do this .


(Joe C. Hecht) #3

Good Show, and thank you for posting the results.

The squeaky wheel gets the grease. I hope it does not work :slight_smile:

We requested a board (and offered to pay for it plus express shipping) to create reviews to feed publications. No surprise at no response.

We are going to purchase a board for review. I suppose that means at least we won’t get sent a hot board that may have been “hand picked”.

TJoe(h^);


(Bruce) #4

[/quote]

Hey thanks for taking the time to post that review.

I had the same problem powering the board using the Micro-USB… Oh it was a really bad design choice and they (SinoVoip) should have stayed with the 4.0mm DC-In connector. I eventually gave up and soldered wires directly to the board until I can get the correct DC Socket.

Also agree the lack of a heatsink on the A83T is problem and additionally some forced airflow helps to maintain higher CPU clock speeds. The minimum they (SinoVoip) should have supplied was a passive heatsink.

I’m still feeling my way around this little board but I’m liking it so far :smiley:


#5

Yes, covering DRAM and SoC. If they would’ve changed the board layout a bit it might be possible to use a large (and effective) heatsink covering also the PMU and eMMC. I’m curious whether others will also experience eMMC corruption when running at full speed. Maybe that’s the reason they don’t ship with an applied heatsink: To always throttle the SoC down :wink:

I still don’t know what to do with this board. Due to only 1 USB port being used, the ultra-slow USB-to-SATA bridge and Ethernet being CPU bound, the whole performance is useless for my use cases. For me a BPi-M1 easily outperforms the M3 at a fraction of costs. But YMMV as usual :smile:


#6

When ALLWINNER sent their test boards, they attached a heat sink, see pictures

DC-Jack, is an OPTION :slight_smile: :slight_smile: see posts here - if you had known you could have asked them.


#7

When you order a batch of 1000 (or even more?) pieces:

It’s useless to complain because the problems users are running into (instable system due to crappy USB cables and the crappy micro USB connector) don’t affect them. And they don’t care as usual. I would suspect they use only quality cables internally and can not even imagine how many crappy USB cables exist in the wild. Or they simply don’t care. We’ll never know…


#8

These connectors are THT (through-hole technology) and is mounted after the reflow process, if I remember correctly. And because they have samples, why shouldn’t they sell /send some with DC-Jack. In the posting was no comment about quantity.


(bpi team) #9

DC-Jack and Micro USB jact , just can choose one , for development board ,we choose micro USB , if you need DC ,we also can support to you. but just micro USB board in stock.


#10

Thanks for choosing the wrong one by default.

I’m currently trying to investigate whether these fixes for HW accelerated video (H3 / Orange Pis) also work with the A83T. And guess what: If I connect HDMI display, an USB keyboard and mouse and try to compile something the board simply powers off.

Again thanks for this great micro USB DC-IN “solution”.


#11

This a big advantage of the LeMaker Guitar - this device is only available with the reliable DC-Jack


(bpi team) #12

we have test all function , micro USB can support fine, so we choose Micro USB , it is easy to find the adapter for user.

if some one want use M3 to use as a product, we can support it with DC-jack.


#13

Yes, we all know and I’ve written it already before here in this thread:

I would suspect they use only quality cables internally and can not even imagine how many crappy USB cables exist in the wild. Or they simply don’t care.

So even if hundreds of users will complain about this crappy powering method you won’t change anything because internally with best quality connectors and cables you can not reproduce the issue. Useless to complain, only way to solve the problem is to avoid your products?


#14

Here you find the specs: http://mgvs.org/public/shema/datasheet/usb_20/Micro-USB_final/

The connector is rated 1.8A max. How should this work with a board with connected USB peripherals, a connected display and the SoC being busy on both CPUs and GPU? And is it really that hard to understand that crappy cables

  1. lead to voltage drops below an acceptable minimum where the PMU simply shuts off due to undervoltage protection

  2. are widely used out there at your customers


#15

And they used a DC jack and not that weird micro USB thing:


#16

I just added the consequences for every customer that bought the M3 because of “octacore @ 2GHz” to the linux-sunxi wiki:

http://linux-sunxi.org/Banana_Pi_M3#Sudden_shut_offs_.2F_maximum_consumption_.2F_cooling_vs._consumption


(BPI_Justin) #17

Good news for M3 :grinning:


#18

Nope, bad news. You need solder skills to be able to use the advertised CPU performance. That’s just weird.


#19

I fear this is a must-read. Unfortunately most of the M2/M3 customers discover this forum only after they already bought the devices.


#20

Do you still think so?

I’ve never seen on any photo or video you showed here that the SoC is used with a fan or heatsink? But according to the review at the start of this thread, the linux-sunxi wiki http://linux-sunxi.org/Banana_Pi_M3 and countless other posts here without a heatsink the CPU will not be clocked that high and then the power requirements are rather low. And problems occur when using peripherals or a heatsink due to increasing power consumption.

Can you please provide a clear company statement whether you still think the micro USB connector is an appropriate power solution?