Other

Technical Guitar Review

43 3799
tkaiser  
tony_zhang replied at Mon Oct 12, 2015 19:33
We have evaluted the USB micro B and the normal USB. If we use normal USB, it can not be automatic ...

According to this Wikipedia article only a Micro-AB plug receptacle is allowed for OTG use. On the Guitar Micro-B is used instead.

If I understand the article correctly then I would need an adapter cable violating the USB specs to be able to use normal USB3 peripherals like disks and so on?
An OTG cable has a micro-A plug on one end, and a micro-B plug on the other end (it cannot have two plugs of the same type). OTG adds a fifth pin to the standard USB connector, called the ID-pin; the micro-A plug has the ID pin grounded, while the ID in the micro-B plug is floating.
You're always talking about OTG (the need to flash the eMMC, I know). But I'm not interested in this OTG stuff, I want to use the Guitar in host mode with standard USB3 peripherals. So please don't tell me about OTG requirements, these are not the issue...

tkaiser  
tony_zhang replied at Tue Oct 13, 2015 21:34
I have updated the Guitar GPIO wiki page to the latest base board rev.b

Do you refer to this change? http://wiki.lemaker.org/index.ph ... 2068&oldid=1994

If yes, then the Wiki gets more and more questionable to be considered an information source (I won't call that specifications at all). If the information regarding the power-plug in the wiki is right ("outside positive, inside negative") I wonder why I'am able to power the Guitar with my Olimex PSU which features the standard center-positive 2.1x5.5x11mm barrel connector https://www.olimex.com/Products/Power/SY0605E/

tkaiser replied at Wed Oct 14, 2015 00:46
A wiki without a 'recent changes' link is totally useless! Or do you want us to check every few ho ...

It seems the history permission is closed. And I also will ask the web developer team to open all the permission.Mine account can see all the history.

tkaiser replied at Thu Oct 15, 2015 01:54
Do you refer to this change? http://wiki.lemaker.org/index.php?title=LeMaker_Guitar:Hardware_speci ...

It is our fault.

tkaiser replied at Wed Oct 14, 2015 03:27
According to this Wikipedia article only a Micro-AB plug receptacle is allowed for OTG use. On the ...

For example, on a mobile phone, it has a microUSB. We can use a cable like this http://item.jd.com/1333522.html, to connect the phone to the comupter, the phone is in device mode. If we want to connect a usb mouse to the phone, we need a OTG cable like this http://item.jd.com/1746769227.html, then the phone will be in host mode.
So the same for Guitar, if we just use a cable like this http://item.jd.com/1333522.html to connect guitar between PC, it acts as device mode, then we can flash OS into emmc. And if want to connect a hard disk or a usb mouse, we need a OTG cable like this http://item.jd.com/1170469.html, so the guitar will act as master mode. But such a cable have two different pin definition that we found when we are testing, we need buy a right one.

tkaiser  
Edited by tkaiser at Thu Oct 15, 2015 12:27
tony_zhang replied at Thu Oct 15, 2015 08:01
For example, on a mobile phone, it has a microUSB. We can use a cable like this http://item.jd.com ...

Sorry, what's this all about?

There's no such thing as "microUSB", there are tons of different "micro" variants. Two of the 3 cables you showed are just USB 2.0 Micro-B. Why would I need them (I STILL don't want to use USB OTG since there is exactly no need for me to flash the eMMC using Actions Semi's ADFU mode)?

If I want to use USB 2.0 peripherals I use one of the 2 USB 2.0 type A ports. If I want to use the OTG mode (which I clearly don't want and not even can since here doesn't exist a single x86 machine not running OS X) then I could also use just use a simple USB 2.0 Micro-USB to USB 2.0 type A since in this mode the Guitar only behaves like a USB 2.0 highspeed device. Also no USB 3.0

I'm still asking

  • How to use USB 3.0 peripherals?
  • Where are informations regarding your USB 3.0 implementation?


Regarding the latter: if the information in Actions Semi's S500 manual is correct then only one of the 2 USB2 ports can act as OTG port? How did you manage to provide OTG functionality on the USB3 port? Then regarding the OTG controller:
The Actions OTG (AOTG) controller is designed to support all tasks specified in the OTG Supplement.
Either AOTG module is composed of AOTG controller and UTMI+ transceiver. AOTG supports hardware implementation of the Host Negotiation Protocol (HNP) and Session Request Protocol (SRP). Special Function Registers are provided to control the HNP and SRP.
Either AOTG can be used as a dual-role device and can act as a USB host or a USB peripheral device. Each of the ID input pins controls the default role of the relative controller. If the 1D=1, it means that the mini-B plug is connected and the AOTG becomes a B-device. When the 1D=0 it means that a mini-A plug is connected and the AOTG becomes an A-device.
For details, see <On-The-Go Supplement to the USB2.0 Specification Rev. 1.0a>

That's all about Mini not Micro (Mini doesn't exist any longer with USB3.0. But no wonder, the whole manual only speaks about USB 2.0 together with OTG since OTG is only supported by the USB2 ports)

If this should work with USB3.0 then we would need not Micro-B but Micro-AB instead? According to the aforementioned wikipedia article:
USB 3.0 introduced a backwards compatible SuperSpeed extension of the Micro-AB receptacle and Micro-A and Micro-B plugs. They contain all pins of the USB 2.0 Micro and use the ID pin to identify the A-device and B-device roles, also adding the SuperSpeed pins.

"Micro-AB" not "Micro-B". To be able to insert Micro-A plugs to let the device as an A-device based on the ID detected. How should this be possible with the Micro-B receptable you use on the baseboard?

You say I should use a cable like this http://item.jd.com/1170469.html and the problem is "But such a cable have two different pin definition that we found when we are testing, we need buy a right one." Which one? The one I bought looks identical (it's also just a plain USB3-Micro-B to USB3-type-A-female like the one you've shown):



BTW: Can you please tell Actions Semi that it's a bit braindead to release a technical manual with applied PDF security restrictions that don't permit copy&paste from? OMG, this is so weird...

tkaiser  
Edited by tkaiser at Fri Oct 16, 2015 04:40

Now you released some informations at http://wiki.lemaker.org/LeMaker_Guitar:USB3.0 (hooray, links containing colons now work!)

Regarding "DEVICE:When we use USB3.0 as device interface, we can connect desktop PC to LeMaker Guitar via USB3.0 Standard A Plug to USB3.0 Micro B Cable as above."

Based on my testings (BTW: am I'm really the only one trying to get USB3 to work? How many Guitars did you sent out for review?) this is not true since the Guitar always connects as an USB 2.0 highspeed device:

  1.     USB 3.0 Hi-Speed Bus:

  2.       Host Controller Location: Built-in USB
  3.       Host Controller Driver: AppleUSBXHCI
  4.       PCI Device ID: 0x9cb1
  5.       PCI Revision ID: 0x0003
  6.       PCI Vendor ID: 0x8086
  7.       Bus Number: 0x0a

  8.         Composite Device:

  9.           Product ID: 0x10d6
  10.           Vendor ID: 0x10d6  (Actions Semiconductor Co., Ltd.)
  11.           Version: 1.00
  12.           Speed: Up to 480 Mb/sec
  13.           Location ID: 0x14200000 / 12
  14.           Current Available (mA): 500
  15.           Current Required (mA): 128
Copy the Code

So why do you list the DEVICE mode on the USB3.0 page? Since then the device acts as an USB2.0 device and you could also use any standard USB2-Micro-USB cable and insert it into the left part of the USB3-Micro-USB receptacle on the board.
And regarding the other mode: "HOST: When we use USB3.0 as host interface, we need replace USB3.0 Micro B with USB3.0 Standard A Receptacle for USB3.0 Standard A Plug to USB3.0 Micro B Cable." Does that mean we should take a solder iron and replace the Micro-B-Receptacle with an usual Type-A?

tkaiser  
Now that you released kernel sources I realize that you added the 1.3 GHz frequency with just 1250mV:
  1. static struct cpu0_opp_table cpu0_table[] = {
  2.         /*khz                uV*/
  3.         {1308000, 1250000},
  4.         {1104000, 1175000},
  5.         { 900000, 1025000},
  6.         { 720000,  975000},
  7.         { 408000,  950000},
  8. };
Copy the Code
When I made my tests I compared the increased voltage between 900 MHz and 1104 MHz (150mV) and came to the conclusion that I've to also add this additional amount of Vcore voltage if I want to clock the SoC 200 MHz higher. Therefore I ended up with 1325mV while you stay at 75mV below at 1250mV.

Is this value a recommendation from Actions Semi? Did you extensive tests with this setting and can confirm that the board runs stable?

Tido  
Edited by Tido at Fri Oct 16, 2015 19:02

In one of the posts from Tony he wrote that they did extensive testing, even with different heatsinks.
So I guess we can count on that value.

tkaiser  
Edited by tkaiser at Sat Oct 17, 2015 08:51
Tido replied at Fri Oct 16, 2015 12:00
In one of the posts from Tony he wrote that they did extensive testing, even with different heatsink ...

I remember he wrote about '24/7 server' (not CPU bound at all) and '48 hours games and video' (might not be CPU bound as well: GPU/VPU if acceleration is available). Since LeMaker has tons of S500 lying around I still wait for the answer "we immediately started to let stress tests run on all 4 CPU cores for at least three months @ 1.3GHz without any issues"

Anyway: if you need 150 mV Vcore more to increase the clock speed from 900 MHz to 1104 and then you just need an additional 75 mV to again increase the clock speed by the same amount of 204 MHz then either Actions Semi produces magic silicone or one of the two settings (for 1.1 or 1.3 GHz) is wrong. Based on the assumption that Actions Semi's SoCs follow the natural law that the more voltage you feed the less lifespan you can achieve. This is what I understood following many discussions on the linux-sunxi list: you've to adjust voltages so that they're not too high (negatively impacts consumption, temperature and also lifespan) and not too low (negatively impacts stability).

You have to log in before you can reply Login | Sign Up

Points Rules