PDAGOLD Takes on USB Host [Archive] - Thoughts Media Forums

View Full Version : PDAGOLD Takes on USB Host

Jason Dunn

12-08-2004, 10:00 PM

<div class='os_post_top_link'><a href='http://www.pdagold.com/articles/detail.asp?a=220' target='_blank'>http://www.pdagold.com/articles/detail.asp?a=220</a><br /><br /></div><i>"Independent observes may be puzzled a little regarding USB Host. Some users (mainly owners of new LOOXes) glorify it uncritically, whereas others tend to play down its importance (the rest:). This article does not aim to decide the dispute but show what this functionality can be used for by ordinary users. In the first place, the USB Host gives you a POSSIBILITY to enhance your device. It is up to you to use it or not."</i><br /><br />Confused about exactly what USB host really is, and how to use it? Pavel Koza from PDAGOLD has written an excellent article that covers some real-world testing of the USB host capabilities of the Loox 720.

rob_ocelot

12-09-2004, 12:17 AM

Great article.

I don't seem to have the same USB host problems on tHe Toshiba e800 as Pavel is having with the FS720. I would be the first one complaining about having to do so many soft resets, especially with WM2003 and it's slow startup times. :-)

Now that I do have USB host functionality I don't know how I lived without it. I wish HP or Dell would integrate this functionality into their smaller form factor units, especially those that are not dual slot. An HP4150 form factor with a VGA screen and USB host (and maybe a jog dial) would be just about perfect. Once USB host is more common I'd expect more drivers or products that are designed with pocketpc in mind.

USB host works great with the iPod. The only real downside is that it's USB 1.0, which is painfully slow compared to 2.0. I have no problems playing full sized Divx files off the iPod however.

Judgeless

12-09-2004, 12:34 AM

The only real downside is that it's USB 1.0, which is painfully slow compared to 2.0.
What devices do you use a USB host for on your e800? Most of the efforts for new peripherals are for Bluetooth not USB. USB 2.0 does not make things faster then 1.1. There are two speeds of USB 1.1, 1.5Mb and 12Mb. There are three speeds of USB 2.0, 1.5Mb, 12Mb, and 480Mb. Every PDA based on Intel XScale CPU’s (SA1110, PXA250, PXA270) only supports 12Mb modes of USB 1.1 and 2.0. That accounts for more then 98% of the PDA’s that have shipped in the last four years.

johncruise

12-09-2004, 02:30 AM

Most of the efforts for new peripherals are for Bluetooth not USB.

Depends on anybody's point of view. I still see alot of USB devices being developed (especially flash disks -- don't get me wrong... I love love LOVE BT). The way I am seeing it, USB peripherals are here to stay. I've heard of the same argument when Firewire step into picture 4 years ago, but USB is still here. It's like what other people keep on saying about CF. Even though we keep on seeing SD and other small form factor flash cards are popping up, there is no denying that CF will stay. It might slow down but it will never go away for some time. (wooookey... I'm way off topic. sorry :oops: )

By the way, if a particular version of USB supports multiple speeds, how do I know which one mine supports? (can't find any good documentation about that for my good 'ol E-200).

Thanks in advance.

Judgeless

12-09-2004, 03:20 AM

Most of the efforts for new peripherals are for Bluetooth not USB.
Depends on anybody's point of view. I still see alot of USB devices being developed (especially flash disks -- don't get me wrong... I love love LOVE BT). The way I am seeing it, USB peripherals are here to stay
I agree USB devices will be here for years to come. There are a lot of new USB devices coming out all the time. I was referring to USB devices that support being hosted on a PPC. As mentioned in the article above the USB Flash drives work fine, a couple of keyboards work but that’s about it. I would predict a lot of companies that are creating peripherals targeting PPC will not use USB. Bluetooth makes a lot more sense.

By the way, if a particular version of USB supports multiple speeds, how do I know which one mine supports? (can't find any good documentation about that for my good 'ol E-200).
The E-200 uses a Intel SA1110 CPU that has a built in USB 1.0 controller that supports 1.5Mb and 12Mb. The same speed as the newest devices shipping today like the HP 4700 and the Dell x50v. They only support USB 2.0 1.5Mb and 12Mb not the 480Mb spec.

alizhan

12-09-2004, 03:42 AM

What devices do you use a USB host for on your e800? Most of the efforts for new peripherals are for Bluetooth not USB.
I can tell you one I'd love to be able to use: a Garmin eTrex Vista C GPS. USB-only, and all the better for not requiring BT (longer battery life for both devices, more stable connection, less GPS interference, etc.). The iPod also ain't a bad idea, and it doesn't do BT either. Or, really thinking outside the box, how about USB speakers? Or MIDI control of a synthesizer?

BT is not the universal answer the industry wants us to think it is. I understand why the industry wants to cram BT down our collective throats, and I even agree that some of the arguments are compelling. Wireless certainly has its advantages, but so does wired. I am greatly disturbed by the recent industry trend of making users take a side: "Either you support BT, and want every single stinking thing you own to be BT, or you are a hard-connector, and want nothing to do with BT."

Why does it have to be either/or? Why not support both? Isn't that sort of flexibility one of PPC's most touted features?

USB 2.0 does not make things faster then 1.1. There are two speeds of USB 1.1, 1.5Mb and 12Mb. There are three speeds of USB 2.0, 1.5Mb, 12Mb, and 480Mb.
Technically true, but most industry wags promote "USB 1.x" as synonymous with "12Mbps," and "USB 2.0" as synonymous with "480Mbps" (in fact, they usually don't even give speeds, just labeling them as "slow" and "fast"). Inaccurate, annoying, and potentially misleading, yes, but that's the idiom, and I doubt it's going to go away any time soon.

-- Mark

Judgeless

12-09-2004, 04:42 AM

SassKwatch

12-09-2004, 02:24 PM

Most PDA’s are shipping with Bluetooth now and there are lot of peripherals that support it today. A USB host on PPC is not a standard and has very little support. Why create a new standard when one is working fine.
I want the USB host option simply to have the ability to connect a portable hard drive to the ppc. Even though SD/CF prices have come down significantly, they are still *obscenely* priced in comparison to a good old fashioned spin up hd, and it simply makes no sense to pay the price differential when all we need is the same option on a ppc that's standard on every desktop machine in the world these days (at least on the Wintel side of things).

IF BT can be made to provide that functionality, that would suit me equally well. Or if someone can tell me that I'll be able to purchase a 20gb CF card for $100 (US) any time soon, I'd be even happier. But I'm not holding my breath on either.

Menneisyys

12-09-2004, 02:26 PM

The only real downside is that it's USB 1.0, which is painfully slow compared to 2.0.

The recent PDA hardvare can't even utilize the full 12 Mbps bandwidth of USB 1.1, so I don't think the lack of USB2 support is that bad. In the other direction (USB slave), the inclusion of the USB2 client on the iPAQ 4700 didn't deliver any performance boost in ActiveSync-based communication.

The average readng/writing speed between a USB hard disk and the F-S PL 720 I've measured is around 300-400 kbytes/s in both directions, which is about the half as the real speed of an average 1.1 unit. (See http://www.firstloox.org/forums/showthread.php?t=2379 on this subject.)

Menneisyys

12-09-2004, 02:45 PM

Judgeless

12-09-2004, 04:05 PM

Most of the efforts for new peripherals are for Bluetooth not USB.
Well, very few (if any) image tanks (the perfect candidates for really portable and in-the-field usable HDD's; see their list at http://forums.dpreview.com/forums/r...ssage=8951076.) support BT... Image tanks and other, USB-only gadgets are pretty common at tech geeks so USB host remains the one and only way to communicate with them for 1-2 years at least.
I agree that USB only gadgets are very common with tech geeks. The article that started this topic states that very few are supported because the manufactures of these USB gadgets do not write drivers for Pocket PC. There has been CF USB host cards out for Pocket PC for many years. The industry has created a lot of Bluetooth peripherals for the mobile wireless Pocket PC market because they fit well. I have never heard the term image tanks, does this mean HD to store images? I have seen ads for devices that will move the contents of a storage card to a HD and remove the files at the push of a button but they are not called storage tanks. If your goal is to use a PPC to take images from a CF card and put them on a USB 2.0 12Mb drive its going to take a very long time.

bcries

12-09-2004, 05:55 PM

The thing is, if you look at desktop PC perhipherals, USB is the norm and Bluetooth is rare. I think Bluetooth is a great niche for primarily portable-use peripherals, but for that which desktops already enjoy (full sized keyboards, VGA output, portable hard drives, etc.), which do you think is easier?
a) create new bluetooth or CF equivalent products for the PPC,
b) simply write new PPC drivers for existing USB devices

I think the argument stands that if we want greater and more affordable access to desktop-ish peripherals (I'm thinking particularily of video output stuff and mass storage), USB host would be a nice compliment to bluetooth.

Judgeless

12-09-2004, 09:09 PM

The thing is, if you look at desktop PC perhipherals, USB is the norm and Bluetooth is rare.
The majority of wireless keyboards and mice shipped over the last two year for the desktop have been using Bluetooth. All the wireless keyboards and mice from Logitech and Microsoft use Bluetooth.

which do you think is easier?
a) create new bluetooth or CF equivalent products for the PPC,
b) simply write new PPC drivers for existing USB devices
I am a hardware Engineer and strongly believe that the statistics you read about were it takes 5 software Engineers for every Hardware Engineer. So I would say it would take 5 times the effort to create Pocket PC USB drivers.

I think the argument stands that if we want greater and more affordable access to desktop-ish peripherals (I'm thinking particularily of video output stuff and mass storage), USB host would be a nice compliment to bluetooth.
I do not think a lot of the desktop peripherals have a lot of use on a portable wireless device. USB printers, USB scanners, USB web cameras, USB label printers, USB card readers. You will never see a USB video adapter to drive a display. USB at 12Mb is to slow, USB at 480Mb is way to slow. You will find USB video capture devices but that is completely different. Those devices heavily compress the video in mpeg or some other format before it is sent over USB. That is not an option when driving a LCD or RGB monitor.

alizhan

12-10-2004, 05:04 AM

I can tell you one I'd love to be able to use: a Garmin eTrex Vista C GPS. USB-only, and all the better for not requiring BT (longer battery life for both devices, more stable connection, less GPS interference, etc.).
GPS units run at 4800 baud and the USB units do USB to serial conversion (COM2:) and provide power to the GPS. If the PDA is powering your GPS your battery life will be a lot less for both devices.
In the world of slaved GPSs, like most of the PPC world seems to use, I agree. The Vista C (http://www.garmin.com/products/etrexVistac/), however, is a standalone consumer unit with native USB support. Unless one is trying to recharge the GPS batteries from the PDA (which would be misguided, at best :)), the only power the GPS requires from the PDA is the power to send the signal down the cable.

Bluetooth, serial, and USB GPS all offer a stable connection.
That's debatable. Search these very forums to find scads of posts about unstable connections (BT, USB, serial, WiFi, IR, yodeling, etc.). In my experience, BT is more flakey than any two of the others combined. YMMV.

A GPS radio operates at 15GHz and a Bluetooth device operates at 2.4GHz there is no interference.
The signal bands don't overlap, no, and yet they apparently interfere quite nicely. Just like signal leakage from the PDA often affects the CF or SD based GPS units. Look around on GPS Passion (or here, for that matter). This situation is slowly improving as manufacturers figure out how to shield things better, though.

Most PDA’s are shipping with Bluetooth now and there are lot of peripherals that support it today. A USB host on PPC is not a standard and has very little support. Why create a new standard when one is working fine.
Three reasons: because it is obviously not working fine for many people, because not everyone has a BT PDA or wants to go purchase one, and because USB actually has a larger and more varied taxonomy of toys than BT is likely to have for some time. USB isn't the new standard; BT is. It just isn't being leveraged on the PPC (IMO, mostly due to lack of vision and/or the perception of the market being infeasibily small).

Question: how many BT-enabled devices (other than phones and PDAs) are useful in their own right? Most BT GPS units, just as an example, are expensive doorstops without a BT host. While the same can be said of many USB GPS units, units like the eTrex are designed to be useful on their own. USB on such devices is not required for the device to be useful; it simply makes the device more useful.

Not allowing PPC users to interact with all the wonderful toys available via USB just because "PPC uses BT" is frustrating and short-sighted.

I really think based on the industry, people want wireless devices. Wireless cell phones continue to explode. More people are buying laptops then ever because they are wireless device that have full internet access. Most PDA’s include WiFi and Bluetooth because a PDA is mobile wireless device.
I really think that most people end up with wireless because the industry keeps telling them how great it is, and glossing over all the problems that come with it. The cell phone argument is apples-to-oranges: cell phones are not PDAs (convergence devices aside); cell phones are defined in terms of being wireless. PDAs are defined by being glorified memory aids; connectivity may help them accomplish that goal, but is not their whole reason for existing (that's more the Blackberry's wonk). Wireless on PDAs was, and still is, secondary.

I've seen far too many people buy wireless PDAs, only never to use it for anything beyond what a Palm III or Gameboy could do. They "know" that the wireless capability is really cool and neat and powerful and they just had to have it (gasp!), but they never seem to find any real use for it beyond occasionally squinting at pixelated pornography from the comfort of their couch. They really wanted it only because marketing sold it to them as a "must have" item.

There's also the buy-in required. Going BT ain't cheap. Not only do you pay extra for the BT PDA, you then have to pay extra for the "convenience" on every other device you use. Between access points, dongles, converters, and all the surcharges on the peripherals themselves, one can easily end up spending more than a hundred dollars to replace a $15 cable. This sort of self-propagating scheme is great for the aftermarket industry, but just hurts consumers.

Outside of business, wireless is often a solution in search of a problem. Our wireless access point broke a few months back, and I haven't bothered to replace it. I haven't missed it. Hard connections are way faster, the Internet looks much nicer on my 19" monitor than my 4" PDA, and I've learned that the world won't end if I can't answer e-mail while I sit on the pot. What, exactly, am I gaining here?

Oh, and don't get me started on WiFi vs. BT. :roll:

Anyhoo, that rant went on a bit longer than expected (and FWIW, don't take any of this personally; I'm just frustrated at what I see as either a seriously wasted opportunity, or a calculated attempt at up-selling the general public). The PDA industry is yet young. Who knows what wonders tomorrow will bring? ;)

-- Mark

alizhan

12-10-2004, 05:18 AM

which do you think is easier?
a) create new bluetooth or CF equivalent products for the PPC,
b) simply write new PPC drivers for existing USB devices I am a hardware Engineer and strongly believe that the statistics you read about were it takes 5 software Engineers for every Hardware Engineer. So I would say it would take 5 times the effort to create Pocket PC USB drivers.
And as a software enginner, it is my duty to contend that this is true, but only because we have to deal with the mess hardware engineers leave behind. ;)

Seriously, though, it does tend to take more SEs to get stuff done. Interacting with the hardware is usually easy enough, but interacting with the rest of the system is often an exercise in strained patience. Even simple operating systems like CE are still horribly complicated, containing execution paths and composite effects the original designers never would have dreamed of. Sort of like a "butterfly effect" for software.

To whit: much of the problem early BT had was because the software just wasn't up to snuff. Much of the software wasn't up to snuff because the specifications were either too vague, or were outright contradictory. The hardware was occasionally to blame, but most of the problems were software-based. Now that the software has finally settled out a bit, BT is more usable.

I think the argument stands that if we want greater and more affordable access to desktop-ish peripherals (I'm thinking particularily of video output stuff and mass storage), USB host would be a nice compliment to bluetooth.
Hear, hear! Both have their place.

Judgeless

12-10-2004, 04:21 PM

Unless one is trying to recharge the GPS batteries from the PDA (which would be misguided, at best ), the only power the GPS requires from the PDA is the power to send the signal down the cable.That is not true. A serial GPS needs external power for the device to operate. USB uses 4 pins 2 for data 2 for power. The USB to serial chip in a USB GPS needs power (100mA) and the GPS receiver needs power (145mA). If the PDA was connected over USB powering the GPS the overall battery life of the PDA would be a lot less.
Search these very forums to find scads of posts about unstable connections (BT, USB, serial, WiFi, IR, yodeling, etc.). In my experience, BT is more flakey than any two of the others combined. YMMV. That’s true consumers have trouble with any interface “(BT, USB, serial, WiFi, IR, yodeling, etc.)” BT is not more flakey then the other combined. The worse GPS solution by far is a CF GPS. Its loaded with issues, yet people still buy them.
Just like signal leakage from the PDA often affects the CF or SD based GPS units. Look around on GPS Passion (or here, for that matter). This situation is slowly improving as manufacturers figure out how to shield things better, though. This is 100% not true. There is a group called the FCC that heavily heavily regulates EMI and RF interference and RF emissions. Device can not be sold with out FCC approval.
because it is obviously not working fine for many people, because not everyone has a BT PDA or wants to go purchase one, and because USB actually has a larger and more varied taxonomy of toys than BT is likely to have for some time. USB isn't the new standard; BT is. A lot less people have a PDA with USB Host capabilities then people with Bluetooth. There are very few PPCs PDAs that can host USB. I agree there are a lot more USB peripherals out there then Bluetooth but only a couple of keyboards and USB Flash drives work with USB Hosted based PPCs. That’s what the whole article was about.
how many BT-enabled devices (other than phones and PDAs) are useful in their own right? Most BT GPS units, just as an example, are expensive doorstops without a BT host. While the same can be said of many USB GPS units, units like the eTrex are designed to be useful on their own. USB on such devices is not required for the device to be useful; it simply makes the device more useful. A BT GPS unit is door stop with out a BT host is correct. But so is a CF GPS with out a CF slot to use it in. A serial GPS is a door stop with out a PDA or laptop to plug it into. A USB GPS is a door stop with out a host to plug it into. I do not understand the point. It is true you can pay a premium for a stand alone GPS that also can be used with a PDA or a Laptop.
Not allowing PPC users to interact with all the wonderful toys available via USB just because "PPC uses BT" is frustrating and short-sighted. I suggest you go on a mission calling all the toy companies that have USB in there product demanding they write drives for PPC.
I really think that most people end up with wireless because the industry keeps telling them how great it is. I do not think you understand what makes a device wireless. Its not just wireless communication. It’s the entire solution. A laptop is a wireless device that is very portable. It does not need a wired power connection. It does not need a wired monitor. It does not need a wired keyboard and mouse. It does not need wired speakers and a mic. With WiFi it does not need wired Ethernet. Its truly a wireless device as a PDA is. The industry is feed on demand. People want wireless devices like Laptops, PDA’s and cell phones. Using wired USB is taking a step backwards.
Oh, and don't get me started on WiFi vs. BT. I hate when people compare WiFi to BT. The two standards do not compete with each other. That’s why you find both in PDA’s and Laptops. WiFi is like the wired Ethernet on all desktops with out the wires. BT is like the USB port on every desktop with out the wires. There will never be a day when the desktop gets rid of USB in place of wired Ethernet. The same is true with WiFi and BT. The both have a purpose.

Talon

12-10-2004, 06:33 PM

A GPS radio operates at 15GHz and a Bluetooth device operates at 2.4GHz there is no interference.
GPS L1 is 1.575GHz not 15, GPS L2 is ~1.2GHz, Bluetooth is 2.4-2.48GHz.
So some Bluetooth systems do have a significant impact on GPS L2 but that is only used for high accuracy systems, it's not an issue for consumer level equipment.

A USB host on PPC is not a standard and has very little support. Why create a new standard when one is working fine.
While no one has said it anywhere I would expect this to be a USB On The Go host rather than a full host. One of the big problems with USB is that you can't just connect two USB devices together. USB OTG (an add on to the 2.0 spec) is an attempt to fix this situation. As with any new spec it will take time to reach critical mass. The USB standard was around for years before USB ports became common on new PC's.
Over time I expect to see more and more USB OTG devices. All the new generation of PDA processors support it (although the bulverde's implimentation leaves a lot to be desired, Intel behind the rest of the competition again).

Judgeless

12-10-2004, 08:11 PM

I did leave out a decimal point. I meant 1.5GHz. Still they to not interfere with each other as Talon posted

USB on the go does have potential. It is used by some camera manufactures so the camera can be directly connected to a printer that supports USB OTG. I am sure over time we will see more peripherals that support OTG. OTG uses generic profiles like Bluetooth so drivers are not as important.

Intel behind the rest of the competition again).
The problem with PPC’s is it relies so heavily on the Intel ARM based CPU’s. I would guess over the last four years 98% of PPC’s shipped have a Intel ARM CPU. (SA1110, PXA25X, or PXA270). One or Two have a TI OMAP. I think one has a Sharp based ARM. It would be nice to have other options.

Talon

12-10-2004, 11:44 PM

The problem with PPC’s is it relies so heavily on the Intel ARM based CPU’s. I would guess over the last four years 98% of PPC’s shipped have a Intel ARM CPU. (SA1110, PXA25X, or PXA270). One or Two have a TI OMAP. I think one has a Sharp based ARM. It would be nice to have other options.
Well MS don't want to have to support multiple instruction sets but in theory PPC will run on any ARM9T CPU or later. The OMAP2's are actually ARM11 cores.
There are a few other people producing suitable parts but they aren't major manufacturers.
Of the three you mentioned the OMAP is by far the lowest power and the most efficent/highest performance per clock design currently around. It's internal design is great, TI stuck to what they know well, the thing has DMA and parallel busses all over the place. The downside being that getting the most out of it is not a trivial thing.
Intel have the raw number crunching lead and higher clock rate but only by adding pipeline stages (it's not a true ARM core) and buring power, clock for clock the xscale is the worst part out there. But this makes xscale simple to use and simple to port to since most people started with a SA1110, a lot of things haven't changed since then. The Intel part is also the most expensive by a fair margin despite the OMAP having two CPUs.
The samsung sits in between, not as elegant as the OMAP but it makes up for it with a brute force approach. It's cheap and it's reasonably fast, in other words Samsung saw a hole in the market for a low end CPU and filled it.

I'm not sure where there is a hole in the market for any other company to get a foot in the door. Hopefully the other two will eat into Intels market sufficently quickly to prevent the intel non-standard instructions from locking users to Intel parts in the future but I can't see a 4th manufacturer showing up any time soon.

Judgeless

12-11-2004, 12:14 AM

The problem is there are few ARM based CPU that have a built in LCD controller and all the peripherals a PDA needs. Intel has XScale, Atmel has a couple of ARM 9 devices with a LCD controller but they run at 200MHz. Sharp has a few with LCD but they are slow. Mot has the DragonARM AKA MX1. The Dragonball was the CPU used in all the Palm Pilots over the years because it had a built in LCD controller. Mot pulled 68K core and replaced it with an ARM9 core but it runs at 200MHz.

Intel always wins because they are shipping the fastest device with the most advanced peripherals. Intel was lucky when they acquired Digital semiconductor. Digital took the highest speed process used for the Alpha CPU back then and used it to create a ARM based device (SA110). The SA1110 used that core. Intel acquired Digital and grew that process. Alpha was supported by Windows NT for a while but it died when MS dropped it and MIPS.