From indiegogo campaign begin, I received a lot of emails, some people said they are very excited and can not wait for it, some people send us special thanks for making it. But also I find some people emailed me, they want it so bad but just can not afford it yet.
My target for VoCore is to dig electric physics limit and my skill limit, make it to the best and do it to the extreme, it is a test for myself so I do not care much about its cost, but their request light my brain again, is it possible to make a wonderful thing but it also can be afford by the budget limited people such as students?
RaspberryPi zero is very good, only 5USD, its price is low and everybody can afford it. I waited the exciting thing over six months, try to get one but never success, it is too hard to buy sometimes I just think it is not that real :p
VoCore2 is almost ready, production process is smooth for now, I am thinking if it is possible to use exists technology and production process make something even cheaper. After a lot of test, it comes, VoCore2 Lite!
PS: We have limited production capacity, but I promise one people have one piece at least.
Compare VoCore, VoCore2 and VoCore2 Lite:
|CPU||RT5350, 360MHz||MT7688AN, 580MHz||MT7628AN, 580MHz|
|Memory||32MB SDRAM||64MB DDR2||128MB DDR2|
|Storage||16MB NOR||8MB NOR||16MB NOR|
|Ethernet Port||x5||x1 / x5*||x1 / x5*|
|USB 2.0 Host||√||√||√|
|USB 2.0 OTG||√||×||×|
|GPIO||> 30**||>=40**||>= 40**|
People may ask, is that price lower than the cost? The answer is it is not, but it is very very very close to the cost, and only for us with the experience of production VoCore2(we reuse the VoCore2 tech, test jig, test app, to save cost etc…). Just hope students and other cost sensitive people will have a chance to play with the most advanced IoT board. Feel happy just like the first time I made VoCore.
Final, please help us spread the news! Good luck!
Using it as a router… Not many settings, all default but the username/password.
VoCore2 is now on indiegogo.com again https://igg.me/at/vocore2
Compared with VoCore v1, we added a beautiful shell(the picture is only a 3d print sample, final shell is much much better) and more features, also improved the LuCI interface to make it easier to use. Hope everyone will like it, same as usual, fully open sourced.
Many curious people have asked me such question (PS: we did not kill any cat yet). This is always a little secret of VoCore, but today I will talk. It is not only about its looks better, also it is part of quality control.
PCB mainly has six colors, green, blue, red, yellow, white, black. (the truth is you can choose any RGB 24bits color you like, somebody even choose purple)
Most of the factories love black and green, that is because the first pass yield around 95% ~ 98%, the not good ones will send back to workers and repair, during the repairing, the pcb boards have to be heated to 260C again, green and black pcb is good but white pcb will change to light yellow one.
Most repaired ones are good, but the quality might come low. Sn after heat and heat again will partly become SnO2, SnO2 is not conductor and have a high chance broken the connection.
So we decide we only ship the first pass VoCore to users, and use white PCB to make it easy to check every VoCore we shipped is first class quality.
PS: another fact is I am lazy, keep them 100% good rate so I can have more time on bed but not shipping the packages again and again. :p
Last month, I spend a lot of time on DMA and ALSA driver, fight with the noise, optimized every feature(I also made a fork on github.com/vonger, welcome everyone).
Really sorry for the delay, but my only target is to make it to the best I could, not the cost or the deadline.
PS: I heard many companies even big company, they just make their products usable but not extreme, because there are deadline, market, pressure from investor, lack of professional engineer etc…They are telling me, the end user do not know the much tech, so who care…
But I want VoCore2 not only to be the tiny computer but also the best router, the best embed wireless platform, if I can not make it to what I want, why waste the time? There are thousands of such boards already.
Now I am still looking for possible flaw on VoCore2, once I am happy about the sample, it will go to production process. It should be very soon
Ethernet Transfer Speed: >= 95Mbps(iperf3), maximized.
Wireless 802.11n Transfer Speed: >= 100Mbps(iperf3), maximized too. —- Most routers same test condition only reach around 90Mbps.
USB Transfer Speed: maximized.
SDXC(SD card) Transfer Speed: maximized.
The Power Consume: 75mA.
Almost there Just finish DMA driver and I2S
driver, committed two patch to my github.
(clone and checkout VoCore2, update feeds, then it will be ready to compile.)
PS: I planed to reuse exists dma/i2s code for mt7620/mt7621, but none of them are real compatible with mt7628…
It is really hard work for Linux kernel develop. Code is a simple part, but understanding Linux arch is really a time-cost work(also every linux kernel has some slightly difference). I learned a lot from reading the source of linux, and really love the flash wisdom between the line of the code.
I am going to do:
audio codec driver: this driver for the dock part, codec wm8988 or es8388.
spi dma driver: current spi driver just works but without dma the transfer is slow and takes full cpu time. There is some example in linux kernel source such as spi-ep93xx.c. Write this driver will be skillful, need to understand DMA arch and SPI arch.
clock driver: standard linux kernel should have clock driver for every device on the bus, current openwrt driver just like a “hack”, using fixed clock frequency in the source. If we want our code in real linux kernel(is that LEDE going to do?) must follow their rules.
Hello every beta tester, current openwrt(or lede) default wifi driver is not very well.
My test result of iperf is around 12Mbps.
Thanks to nobblepepper test result:
1. With the image(mtk driver) that came on VoCore2 I get: $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.0 sec 191 MBytes 53.4 Mbits/sec $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ 3] local 192.168.1.177 port 39644 connected with 192.168.1.1 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.0 sec 212 MBytes 59.3 Mbits/sec 2. With lede with mt76 driver I get: $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ 3] local 192.168.1.177 port 39712 connected with 192.168.1.1 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.0 sec 44.5 MBytes 12.4 Mbits/sec $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ 3] local 192.168.1.177 port 39713 connected with 192.168.1.1 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.0 sec 38.4 MBytes 10.7 Mbits/sec 3. When I disable cal_free data like this (which gave me BIG improvements in speed on MiWiFi and Linkit7688) https://github.com/openwrt/mt76/issues/28 I get: $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ 3] local 192.168.1.177 port 39775 connected with 192.168.1.1 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.7 sec 7.75 MBytes 2.12 Mbits/sec $ iperf -c 192.168.1.1 -t 30 -f m --snip-- [ 3] local 192.168.1.177 port 39776 connected with 192.168.1.1 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-30.5 sec 7.00 MBytes 1.92 Mbits/sec
Here is a firmware from mtk openwrt, wifi works very well.
default password is 12345678
This firmware is only for one antenna. I disabled another one.
It is easy to enable by update factory setting section. update 0x0034 of factory setting to 0x3422(old should be 0x3411)
The answer is simple: the price is much lower than its cost.
Currently in market 256MB DDR2 cost around 3~4USD, but RPi Zero have 512M DDR2(price should be over 5USD), Broadcom do not make DDR2, they have to buy it from somewhere else, even they can make it to 2.5USD, the other costs still over 2.5USD.
The total cost is impossible to lower than 5USD.
It is a good advertising, but we have to forget it, it won’t product anymore.
PS: … Hope I am wrong. :p