Micro USB plz?


  • Avoid the EXT connector
  • Avoid trying to do USB host with a full-size B port
  • Avoid obscure connectors like dockport
  • Ensure that every connector is electrically and logically capable of everything the user expects it do do.
I agree with the first point and the fourth point. As for dockport, I'm not sure why it isn't worth considering. (Just not as a replacement for the dedicated power outlet) As for your second point: Could you reword that? I'm having a hard time understanding what you are meaning since USB host goes through the USB-A port, not USB-B.

The point here is that the jack is perfectly fine for anybody that's taking any kind of reasonable care when inserting and ejecting the jack.
 This is good.

It suggests that with proper care and attention, robust microUSB implementations are possible.
Yea, sounds all well and good until a lot of users are sending their Pyras back to EvilDragon for RMAs because of microUSB ports that they didn't give "proper care and attention" to when plugging and unplugging.

-God Ginrai
 
Last edited by a moderator:
We're well aware of that,  you're just completely missing the point.  We want both USB slave, and USB 3 Host to be available to the user.   Your recommended configuration doesn't allow for that.


- Neelix
You might be, I wasn't commenting on what you were aware of. _wb_'s posts, however, have continuously suggested that I wanted USB 3.0 speeds for USB slave and that I should just settle for USB 2.0 speed for USB slave, which is not possible with the SoC.
 
No, your position is that you want a full-size USB slave port at whatever speed and only USB 2.0 speed for USB host. My position is to have USB 3.0 OTG so we can have both slave and host at that speed.


And no, the board space needed for micro-USB 3.0 OTG and full-size USB-B is not the same. Micro-USB 3.0 takes 13mm x 5mm = 65mm^2. Full-size USB-B takes 12mm x 18mm = 216mm^2, which is more than 3 times the board area. In terms of volume it is even worse.
Actually, I wanted eSATAp for the USB 2.0 host. And are you suggesting we chuck USB 2.0 host altogether? Everyone else in the topic has been suggesting this micro port in addition to a full size USB-A 2.0 host port.


Also, isn't that 18mm upwards? If so, that's not equivalent to boardspace.

-God Ginrai
No, it's "only" 13mm in height. That still twice as high as USB-A and four times as high as micro-USB.

Where did I say we have to get rid of the USB 2.0 Type A port? My proposal was, and still is:

- USB 2.0 host (of course with backwards compatibility to 1.x) + eSATAp

- micro-USB 3.0 OTG (perhaps with SlimPort), of course with backwards compatibility to 1.x and 2.0

In this proposal, you can pretty much connect anything without needing adapters. You only need an adapter if you want to use a gadget with a full-size USB 3.0 connector (e.g. a fast memory stick); in that case you'll need an adapter if you want the super speed, but if you're fine with high speed, you can just put it directly in the 2.0 port.

If there would be no SoC limitations, that USB 2.0 host would be replaced by USB 3.0 (so there would be both a USB 3.0 host (+eSATAp) and a micro-USB 3.0 OTG port). But if that is not possible I could live with the above.

In your proposal, we would have:

- USB 2.0 host (of course with backwards compatibility to 1.x) + eSATAp

- USB 3.0 (or 2.0, whatever) type B (slave/gadget/device) port

In that proposal, there is no way at all, not even with adapters, to get USB 3.0 speed in host mode. You would also need adapters in case you want to plug in something with a micro-USB 2.0 or micro-USB 3.0 plug, or a charger like those of many smartphones, with a micro-USB plug.
 
In the post I quoted, you said that my proposal was to have USB 3.0 slave and USB 2.0 host, whereas your position was to have USB 3.0 OTG. Whether you meant it or not, that statement implies that the USB 3.0 OTG you are suggesting replaces both of those.

And as I said, I'm fine with losing the faster speeds of USB 3.0 for USB host if it means that my Pyra is less likely to have hardware problems. You want faster speeds, I want durability. I don't see how wanting more durable hardware is not considered a valid argument.

-God Ginrai
 
Avoid the EXT connector
Unfortunately there is no standard for what the EXT port includes. It's not just the TV/audio out, it's also audio in, UART, IR, and a bunch of GPIO lines (that, sadly, no one ever took advantage of).If there isn't enough room for it and something needs to go I'll regrettably agree it should probably be the first on the chopping block but I don't think we're at that point yet where features need to arbitrarily be cut.
 
As for your second point: Could you reword that? I'm having a hard time understanding what you are meaning since USB host goes through the USB-A port, not USB-B.
#

Electrically, either could work.

There have been various suggestions of wiring the USB otg port to a non USB-AB socket in an attempt to avoid the fragility of microUSB.

I think these are a bad idea, because everyone expects USB-B sockets to be for slave mode and USB-A for host mode
 
As for your second point: Could you reword that? I'm having a hard time understanding what you are meaning since USB host goes through the USB-A port, not USB-B.
#
Electrically, either could work.

There have been various suggestions of wiring the USB otg port to a non USB-AB socket in an attempt to avoid the fragility of microUSB.

I think these are a bad idea, because everyone expects USB-B sockets to be for slave mode and USB-A for host mode
My suggestion to wire the OTG port to USB-B was to ditch the host and only be slave mode. This would fall within expectations. I wasn't suggesting we allow it to do both, and I haven't seen anyone else suggest it either.

EDIT: If a full-size USB-AB 3.0 port existed, I'd be all over it. But it does not.

-God Ginrai
 
Last edited by a moderator:
Avoid the EXT connector
Unfortunately there is no standard for what the EXT port includes. It's not just the TV/audio out, it's also audio in, UART, IR, and a bunch of GPIO lines (that, sadly, no one ever took advantage of).
If there isn't enough room for it and something needs to go I'll regrettably agree it should probably be the first on the chopping block but I don't think we're at that point yet where features need to arbitrarily be cut.
miniDP + 3.5mm line-in + headset jack + some 0.1" though hole pads on the PCB for hackers covers most of it

I would much prefer that to a completely non-standard connector, even if it makes the device 1cm wider as a result.

Regarding board perimeter space, we can gain about 6cm by moving SD cards to the side
 
Best solution:

1.  Power barrel - although I think we -could- get rid of it, if it's presence makes people feel better, fine.

2.  microUSB 3.0 OTG (wide) port.

3.  eSATAp combined eSATA and USB 2.0 host.

4.  DP++ (Display port dual mode).  Passive adapters allow connecting to anything.

Absolutely NO on the USB 3.0 type B slave only port.  That's simply a really bad idea.
 
Absolutely NO on the USB 3.0 type B slave only port.  That's simply a really bad idea.
Funny, I feel the same way about the microUSB port. That it's simply a really bad idea.

I do, however, agree about points 1 and 3, tho'. I don't know enough about point 4 to know if I should support this over HDMI or not.

-God Ginrai
 
Absolutely NO on the USB 3.0 type B slave only port.  That's simply a really bad idea.
Funny, I feel the same way about the microUSB port. That it's simply a really bad idea.


I do, however, agree about points 1 and 3, tho'. I don't know enough about point 4 to know if I should support this over HDMI or not.

-God Ginrai
Buy a device or cable with a microUSB 3.0 connector on it.  It's really far more robust than microUSB.  I'm rough on devices and don't like microUSB (normal) because it feels fragile.  MicroUSB 3.0 (the wide one) - that one doesn't bother me at all.  It is -far- more robust.

You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.

For DisplayPort++, Scroll down half-way on: http://en.wikipedia.org/wiki/DisplayPort to read up on DP dual mode.  It allows for passive adapters to be used to get to DVI, HDMI, VGA while still allowing full DP functionality.  The only drawback is I don't know if our SOC will have the right leads to run it.
 
You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.
You're making it sound like you're losing 2 features. OTG is just the features of host and slave in one connector so you can switch between both instead of having 2 connectors.

For DisplayPort++, Scroll down half-way on: http://en.wikipedia.org/wiki/DisplayPort to read up on DP dual mode.  It allows for passive adapters to be used to get to DVI, HDMI, VGA while still allowing full DP functionality.  The only drawback is I don't know if our SOC will have the right leads to run it.
Sounds interesting. I'll read more into it. Thanks for pointing me towards it.

-God Ginrai
 
You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.
You're making it sound like you're losing 2 features. OTG is just the features of host and slave in one connector so you can switch between both instead of having 2 connectors.
You can not implement OTG with a full sized B port (usb 2.0, or USB 3.0) it does not have enough wires.

Having both a full sized "A" and "B" ports connected to this port would be a really bad idea, confusing to users (only one can be used at a time) and adding extra circuitry which is very likely to cause problems to the extremely high speed superspeed wires.

Connecting just the full sized "B" connector eliminates the USB 3 host mode. This is unacceptable.

The full sized connectors are HUGE. Space limitations on the board may dictate micro SD cards, and micro A ports for the USB 2.0 hosts.

The micro AB USB 3 port is the most reasonable choice.
 
You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.
You're making it sound like you're losing 2 features. OTG is just the features of host and slave in one connector so you can switch between both instead of having 2 connectors.
What you're failing to understand is that a USB OTG port really IS three features.  It is having a host mode port, a device mode port AND having them exist as a single compact port on the device that is able to switch between the two.

1.  USB 3.0 Host port

2.  USB 3.0 Device port

3.  Compact & switchable between the two = OTG

Combining the two functions into one port IS a feature.

Think of it this way.

1.  USB 2.0 host

2.  eSATA port

3.  Compact & switchable between the two = eSATAp

Does that help?
 
You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.
You're making it sound like you're losing 2 features. OTG is just the features of host and slave in one connector so you can switch between both instead of having 2 connectors.
 What you're failing to understand is that a USB OTG port really IS three features.  It is having a host mode port, a device mode port AND having them exist as a single compact port on the device that is able to switch between the two.

1.  USB 3.0 Host port

2.  USB 3.0 Device port

3.  Compact & switchable between the two = OTG

Combining the two functions into one port IS a feature.

Think of it this way.

1.  USB 2.0 host

2.  eSATA port

3.  Compact & switchable between the two = eSATAp

Does that help?
It's not a feature worth noting when you make it redundant. (We already have a USB host port)

-God Ginrai
 
You're simply not going to win on the argument for full sized 13mm tall USB 3.0 type B tower that disables OTG and 3.0 host modes.  It sacrifices WAY too much for a single perceived gain in one dimension/factor.
You're making it sound like you're losing 2 features. OTG is just the features of host and slave in one connector so you can switch between both instead of having 2 connectors.
 
What you're failing to understand is that a USB OTG port really IS three features.  It is having a host mode port, a device mode port AND having them exist as a single compact port on the device that is able to switch between the two.


1.  USB 3.0 Host port


2.  USB 3.0 Device port


3.  Compact & switchable between the two = OTG


Combining the two functions into one port IS a feature.


Think of it this way.


1.  USB 2.0 host


2.  eSATA port


3.  Compact & switchable between the two = eSATAp


Does that help?
It's not a feature worth noting when you make it redundant. (We already have a USB host port)

-God Ginrai
We have one USB 2 host port.  We do not have 10.  

We do not have a USB 3 host port if you take away the microUSB 3.0 OTG port.

The difference in speed is more than a factor of 10.

USB 3.0 top transfer speed:  5 Gbit/sec.

USB 2.0 top transfer speed:  480 Mbit/sec = 0.480 Gbit/sec

USB 1.1 top transfer speed:  12 Mbit/sec

USB 1.0 top transfer speed:  1.5 Mbit/sec

In data transfer times it's needing 5 minutes on USB 3.0 host to do what could take an hour on USB 2.0, two days on USB 1.1 or two weeks on USB 1.0.

We're not talking apples to apples when you say that it's redundant.  USB 3.0 host <> USB 2.0 host.
 
We have one USB 2 host port. We do not have 10.
We could have two, I believe.

Still, I agree that it would be a bad idea to discount the value of a 10x speedup

16MB/s SDIO must have seemed like enough for anyone back when the pandora/OMAP3 was being designed.

I would say though that its a little misleading to refer to the USB 1.x specs there.

Very few devices indeed use low-speed or full-speed  for bulk data transfer these days.
 
In data transfer times it's needing 5 minutes on USB 3.0 host to do what could take an hour on USB 2.0, two days on USB 1.1 or two weeks on USB 1.0.
Theoretically. But that situation doesn't happen often in actuality. What do the large majority of Pandora users do that takes them an hour on USB 2.0?

-God Ginrai
 
In data transfer times it's needing 5 minutes on USB 3.0 host to do what could take an hour on USB 2.0, two days on USB 1.1 or two weeks on USB 1.0.
Theoretically. But that situation doesn't happen often in actuality. What do the large majority of Pandora users do that takes them an hour on USB 2.0?

-God Ginrai
Copy ~30 gigs of data to a 32+ gig SD card.. I don't have a card reader on my desktop it's easier to go through USB cables in host mode.
 
In data transfer times it's needing 5 minutes on USB 3.0 host to do what could take an hour on USB 2.0, two days on USB 1.1 or two weeks on USB 1.0.
Theoretically. But that situation doesn't happen often in actuality. What do the large majority of Pandora users do that takes them an hour on USB 2.0?

-God Ginrai
Copy ~30 gigs of data to a 32+ gig SD card.. I don't have a card reader on my desktop it's easier to go through USB cables in host mode.
How often do you find yourself repeating this procedure with this amount of data?

-God Ginrai
 
Last edited by a moderator:
Back
Top