Compound USB device design

T

Terry Dawson

Guest
Greetings,
I'm designing a compound USB peripheral that will combine a USB sound
card with two serial ports. My design objectives are in order:

- low component count
- small physical size
- low power consumption
- ideally it'd be powered from the upstream USB 2.0 interface.

Ideally there'd be minimal programming required although it would be
nice to have the device readily and uniquely identifiable when they're
plugged in.

At this stage I'm thinking of basing the design on:

- Future Technology FT2232H dual USB UART, and a
- Burr Brown/TI PCM2901 Audio Codec with USB interface

with an

- Alcor Micro AU9254

to bind them together.

I've had a bit of difficulty finding candidate USB hub ICs, I'm guessing
I'm just not using the right search terms or something. I ended up with
a couple, the Alcor seemed ok.

I've no previous USB design experience and I'd appreciate your input
regarding:

- any potential traps or pit-falls to watch out for
- suggestions for other USB controller ICs to use
- any reference designs that I might plunder

regards
Terry
 
Terry Dawson wrote:
Greetings,
I'm designing a compound USB peripheral that will combine a USB sound
card with two serial ports. My design objectives are in order:

- low component count
- small physical size
- low power consumption
- ideally it'd be powered from the upstream USB 2.0 interface.

Ideally there'd be minimal programming required although it would be
nice to have the device readily and uniquely identifiable when they're
plugged in.
The drivers for the individual audio and UART chips would handle that.
They just get treated as two seperate USB devices.
Audio/UART device selection would in this case be based on the ease of use
and availability of suitable drivers and example apps in your chosen
language.

At this stage I'm thinking of basing the design on:

- Future Technology FT2232H dual USB UART, and a
- Burr Brown/TI PCM2901 Audio Codec with USB interface

with an

- Alcor Micro AU9254

to bind them together.

I've had a bit of difficulty finding candidate USB hub ICs, I'm
guessing I'm just not using the right search terms or something. I
ended up with a couple, the Alcor seemed ok.
Digikey have many usb hub chips:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Cat=2556697&k=usb%20hub
as do Mouser:
http://mouser.com/Semiconductors/Integrated-Circuits-ICs/_/N-6j73k?Keyword=usb+hub&Ns=Pricing%7c0&FS=True

Not many in nice easy SO packages though, but if you want small go for QFN.

You only need USB 1.1 for your intended devices.

I've no previous USB design experience and I'd appreciate your input
regarding:

- any potential traps or pit-falls to watch out for
- suggestions for other USB controller ICs to use
- any reference designs that I might plunder
The datasheets for the chips usually have all you need hardware wise in
their example apps.

If this for a potential commercial product?

Dave.

--
---------------------------------------------
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
 
David L. Jones wrote:
Terry Dawson wrote:
I've had a bit of difficulty finding candidate USB hub ICs, I'm
guessing I'm just not using the right search terms or something. I
ended up with a couple, the Alcor seemed ok.

Digikey have many usb hub chips:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Cat=2556697&k=usb%20hub
as do Mouser:
http://mouser.com/Semiconductors/Integrated-Circuits-ICs/_/N-6j73k?Keyword=usb+hub&Ns=Pricing%7c0&FS=True
Thanks for those, I'll take a close look at them. I've not done business
with either digikey or mouser before.

Not many in nice easy SO packages though, but if you want small go for QFN.
Size isn't my number 1 priority, so I'm prepared to compromise.

You only need USB 1.1 for your intended devices.
correct.

I've no previous USB design experience and I'd appreciate your input
regarding:

- any potential traps or pit-falls to watch out for
- suggestions for other USB controller ICs to use
- any reference designs that I might plunder

The datasheets for the chips usually have all you need hardware wise in
their example apps.
The datasheets have lots of good information, but many of them refer to
"standard USB design principles" without providing a reference to them.

Maybe I'm imagining things are there that aren't. It all seems fairly
straight-forward.

If this for a potential commercial product?
Did you mean "Is this" ? No, strictly hobby/personal.

If I was wildly optimistic I could imagine enough other people being
interested that I might consider making a batch and selling them on a
cost-recovery basis but I'm intending to give the completed design away
(Open Hardware). I'm basing other aspects of the design on other peoples
work so I can't claim any sort of real ownership in any case.

thanks!

Terry
 
Terry Dawson wrote:
David L. Jones wrote:
Terry Dawson wrote:
I've had a bit of difficulty finding candidate USB hub ICs, I'm
guessing I'm just not using the right search terms or something. I
ended up with a couple, the Alcor seemed ok.

Digikey have many usb hub chips:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Cat=2556697&k=usb%20hub
as do Mouser:
http://mouser.com/Semiconductors/Integrated-Circuits-ICs/_/N-6j73k?Keyword=usb+hub&Ns=Pricing%7c0&FS=True

Thanks for those, I'll take a close look at them. I've not done
business with either digikey or mouser before.
Delivery is the only killer for low value orders, Mouser are a fixed US$30.

Farnell have some too:
http://au.farnell.com/jsp/search/browse.jsp?N=500001+1003198+511&Ntk=gensearch_001&Ntt=usb+hub&Ntx=&No=0&getResults=true&appliedparametrics=true&locale=en_AU&catalogId=&prevNValues=500001+1003198&originalQueryURL=http%3A%2F%2Fau.farnell.com%2Fjsp%2Fsearch%2Fbrowse.jsp%3FN%3D500001%2B1003198%26Ntk%3Dgensearch_001%26Ntt%3Dusb%2Bhub%26Ntx%3D%26No%3D0%26getResults%3Dtrue%26appliedparametrics%3Dtrue%26locale%3Den_AU%26catalogId%3D

Some are stocked in Oz, and remember that Farnell now have free delivery and
no minimum order, so likely a much better option than Digikey or Mouser for
small orders.

The datasheets have lots of good information, but many of them refer
to "standard USB design principles" without providing a reference to
them.
They are probably just referring to generic controlled impedance lines and
power supply current limitations etc.

Maybe I'm imagining things are there that aren't. It all seems fairly
straight-forward.
It should be.
USB hubs are complete drop-in solutions, as are the UART/Audio chips. Your
only hurdle should be suitable drivers.

If this for a potential commercial product?

Did you mean "Is this" ? No, strictly hobby/personal.

If I was wildly optimistic I could imagine enough other people being
interested that I might consider making a batch and selling them on a
cost-recovery basis but I'm intending to give the completed design
away (Open Hardware). I'm basing other aspects of the design on other
peoples work so I can't claim any sort of real ownership in any case.
Open Hardware is cool.
Hope it goes well.

Dave.

--
================================================
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
 
David L. Jones wrote:

Some are stocked in Oz, and remember that Farnell now have free delivery and
no minimum order, so likely a much better option than Digikey or Mouser for
small orders.
I've already availed myself of their free delivery, it's certainly got
me dealing with them more seriously than I ever have before.

The datasheets have lots of good information, but many of them refer
to "standard USB design principles" without providing a reference to
them.

They are probably just referring to generic controlled impedance lines and
power supply current limitations etc.
ok, that could be right.

Maybe I'm imagining things are there that aren't. It all seems fairly
straight-forward.

It should be.
USB hubs are complete drop-in solutions, as are the UART/Audio chips. Your
only hurdle should be suitable drivers.
I'm hoping drivers won't be too much of an issue either, as the vendors
seem to supply them for their chipsets. I'll be using Linux myself but
any broader audience might not, so I'll need to give that some thought.

Do I need to worry about anything in PCB layout? I've seen some mention
of avoiding running signalling traces over power planes in some
datasheets for example. I'm assuming this is generally good practice but
it leaves doubt in my mind that I'm missing something.

Open Hardware is cool.
Hope it goes well.
thanks :)

Terry
 
Terry Dawson wrote:
David L. Jones wrote:

Some are stocked in Oz, and remember that Farnell now have free
delivery and no minimum order, so likely a much better option than
Digikey or Mouser for small orders.

I've already availed myself of their free delivery, it's certainly got
me dealing with them more seriously than I ever have before.

The datasheets have lots of good information, but many of them refer
to "standard USB design principles" without providing a reference to
them.

They are probably just referring to generic controlled impedance
lines and power supply current limitations etc.

ok, that could be right.

Maybe I'm imagining things are there that aren't. It all seems
fairly straight-forward.

It should be.
USB hubs are complete drop-in solutions, as are the UART/Audio
chips. Your only hurdle should be suitable drivers.

I'm hoping drivers won't be too much of an issue either, as the
vendors seem to supply them for their chipsets. I'll be using Linux
myself but any broader audience might not, so I'll need to give that
some thought.
Do I need to worry about anything in PCB layout? I've seen some
mention of avoiding running signalling traces over power planes in
some datasheets for example. I'm assuming this is generally good
practice but it leaves doubt in my mind that I'm missing something.
There would only be two things you need to be wary of.
First is the controlled impedance lines for the USB. Plenty of info on this
is available, Google will no doubt give countless references to what's
required. If you keep your lines short enough the actual impedance ain't
going to matter too much, so no need for proper controlled impedance PCB
manufacture, do some basic calcs to get close enough to the required
impedance and that will do. It's important to keep the USB trace lengths
matched. But you are only running USB 1.1, so no biggie, the preverbial
coathanger would probably work.
You don't run controlled impedance traces over *splits* in your ground
plane, that's a big generic no-no.

Second would be your audio circuit. I don't kow what performance you are
after, but keeping your analog and digital grounds seperate is important.
They should meet only at the audio chip. Keep all your digital stuff on one
half of the board ,analog on the other, that kind of thing.

Dave.

--
================================================
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
 
David L. Jones wrote:

There would only be two things you need to be wary of.

First is the controlled impedance lines for the USB. Plenty of info on this
is available, Google will no doubt give countless references to what's
required. If you keep your lines short enough the actual impedance ain't

Second would be your audio circuit. I don't kow what performance you are
after, but keeping your analog and digital grounds seperate is important.
They should meet only at the audio chip. Keep all your digital stuff on one
half of the board ,analog on the other, that kind of thing.
That's precisely the sort of stuff I'm after, I'd not have given thought
to either of these.

Thank you for the tips.

Terry
 
Terry Dawson wrote:
Do I need to worry about anything in PCB layout? I've seen some mention
of avoiding running signalling traces over power planes in some
datasheets for example. I'm assuming this is generally good practice but
it leaves doubt in my mind that I'm missing something.
For USB 1.1? - I wouldn't worry about it. Just use the usual commonsense
rules for MHz signals (lots of decoupling caps, fast signal traces as
straight as possible, etc) & you should be fine.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------
 
"Terry Dawson" <vk2ktj@gmail.com-fu> wrote in message
news:hbpg8f$lci$1@news.eternal-september.org...
Second would be your audio circuit. I don't kow what performance you are
after, but keeping your analog and digital grounds seperate is
important.
They should meet only at the audio chip. Keep all your digital stuff on
one
half of the board ,analog on the other, that kind of thing.

That's precisely the sort of stuff I'm after, I'd not have given thought
to either of these.

Thank you for the tips.

My opinion is why would you want to build such a device when separate USB
audio devices and USB-Serial adapters are cheap?
(and small, so size is not a real issue)
Secondly, you will not match the audio performance of the higher quality
audio devices given what you say above. Even the cheap ones are pretty good
these days!
Thirdly I wouldn't even consider USB 1.1 if you want to keep the audio
latency at all low, especially considering the minimal savings.
(if you don't know what that is, it probably won't affect you)

Of course if you are just looking for an interesting project to keep you
busy while learning something, then go for it. Personally I don't waste my
time when cheap off the shelf solutions are already available.

MrT.
 
Mr.T wrote:
My opinion is why would you want to build such a device when separate USB
audio devices and USB-Serial adapters are cheap?
(and small, so size is not a real issue)
heh, my initial response is "hobbies don't have to be economically
rational".

In truth I have a small pile of $6 USB hubs and $5 USB sound cards
sitting on the desk in front of me. I already have the thing
"prototyped" by cobbling together off-the-shelf units. But it's not
robust and it isn't small. USB connectors are bulky. I forgot to mention
robust in my list of requirements, anyway, it is one :)

I was considering at one point just ripping the guts out of these,
desoldering the connectors, remounting the pcbs and hardwiring them
together. That's my plan B if I run out of enthusiasm for doing it myself.

Secondly, you will not match the audio performance of the higher quality
audio devices given what you say above. Even the cheap ones are pretty good
these days!
There is no requirement for especially high-fidelity.

Thirdly I wouldn't even consider USB 1.1 if you want to keep the audio
latency at all low, especially considering the minimal savings.
(if you don't know what that is, it probably won't affect you)
Low latency isn't a requirement either.

Of course if you are just looking for an interesting project to keep you
busy while learning something, then go for it. Personally I don't waste my
time when cheap off the shelf solutions are already available.
This isn't a commercial exercise, I'm just indulging in a hobby. In the
absence of an original project idea I'm opting for an original
implementation instead.

Having fun and learning stuff is the ultimate objective.

Terry
 
Bob Larter wrote:
Terry Dawson wrote:
Do I need to worry about anything in PCB layout? I've seen some

For USB 1.1? - I wouldn't worry about it. Just use the usual commonsense
rules for MHz signals (lots of decoupling caps, fast signal traces as
straight as possible, etc) & you should be fine.
cool. thanks! That's reassuring.

Terry
 
"Terry Dawson" <vk2ktj@gmail.com-fu> wrote in message
news:hbtbf7$s01$1@news.eternal-september.org...
My opinion is why would you want to build such a device when separate
USB
audio devices and USB-Serial adapters are cheap?
(and small, so size is not a real issue)

heh, my initial response is "hobbies don't have to be economically
rational".

In truth I have a small pile of $6 USB hubs and $5 USB sound cards
sitting on the desk in front of me. I already have the thing
"prototyped" by cobbling together off-the-shelf units. But it's not
robust and it isn't small. USB connectors are bulky. I forgot to mention
robust in my list of requirements, anyway, it is one :)

I was considering at one point just ripping the guts out of these,
desoldering the connectors, remounting the pcbs and hardwiring them
together. That's my plan B if I run out of enthusiasm for doing it myself.
That would probably be my plan A. I'm betting you do run out of enthusiasm
too.



Secondly, you will not match the audio performance of the higher quality
audio devices given what you say above. Even the cheap ones are pretty
good
these days!

There is no requirement for especially high-fidelity.

Thirdly I wouldn't even consider USB 1.1 if you want to keep the audio
latency at all low, especially considering the minimal savings.
(if you don't know what that is, it probably won't affect you)

Low latency isn't a requirement either.

Of course if you are just looking for an interesting project to keep you
busy while learning something, then go for it. Personally I don't waste
my
time when cheap off the shelf solutions are already available.

This isn't a commercial exercise, I'm just indulging in a hobby. In the
absence of an original project idea I'm opting for an original
implementation instead.

Having fun and learning stuff is the ultimate objective.

Fair enough, if you really couldn't find a more worthwhile project to waste
your time on I guess :)

MrT.
 
Terry Dawson wrote:
Bob Larter wrote:
Terry Dawson wrote:
Do I need to worry about anything in PCB layout? I've seen some
For USB 1.1? - I wouldn't worry about it. Just use the usual commonsense
rules for MHz signals (lots of decoupling caps, fast signal traces as
straight as possible, etc) & you should be fine.

cool. thanks! That's reassuring.
No worries. Glad I could help. Also, as someone else said: keep your
analog power & ground lines as far apart as possible.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------
 

Welcome to EDABoard.com

Sponsor

Back
Top