The SAM7 has 5-volt tolerant I/Os, but it's the host you're signalling with the pullup and I don't think that the spec lets you pull the data bus any higher than 3.3v. Scott ----- Original Message ----- From: "Lars Andersson" <laran@lara...> To: <AT91SAM7@AT91...> Sent: Monday, May 09, 2005 2:13 PM Subject: RE: [AT91SAM7] Re: UDP pull-up control What about pulling the signal up to VBUS instead of the 3.3v rail. I assume this will give an even better pull up and make the system more robust or is there something that I am missing here. Correct me if I am wrong, but isn't the IO-pins 5v tolerant.

> The main problem I see with the npn circuit in the datasheetis that it
> does not pull up as hard as a pulldown driving pullup circuit. The npn
> will only be able to pull up to 3V3-0.6 = 2.7V. The down-up circuit
> (which you could do with an npn driving a pnp) will pull to the rail.

> This won't consume appreiably more current (perhaps an extra 100uA),
> but does drive the pull-up better. This will make the USB more robust
> on longer wires and worse noise conditions etc.

What about pulling the signal up to VBUS instead of the 3.3v rail. I assume this will give an even better pull up and make the system more robust or is there something that I am missing here. Correct me if I am wrong, but isn't the IO-pins 5v tolerant.

--- In AT91SAM7@AT91..., "microbit" <microbit@c...> wrote: > Hi Scott, > > I'm laying out the USB interface portion of my board right now, and I've got > > a question on the pull-up control. The SAM7 datasheet shows a simple BJT > > controlling the pull-up, but the AT91SAM7S-EK board appears to use an > > n-channel FET driving a p-channel FET. Why the added complexity? I've used > > that type of circuit before, but only for switching a lot of current. This > > only needs to handle like 2 ma. Is there any advantage to using this scheme > > versus a BJT or logic level FET? > > I don't know enough about USB, so leave that for the buffs. > But AFAIK the enumeration process includes an initial limit of current that the hub is able > to give, and then it tells how much current the USB slave device can use. > If you don't power anything major through the USB, I don't think you need major external > current switching at enumeration. > > USB buffs, if I'm wrong, please educate us :-) The main problem I see with the npn circuit in the datasheetis that it does not pull up as hard as a pulldown driving pullup circuit. The npn will only be able to pull up to 3V3-0.6 = 2.7V. The down-up circuit (which you could do with an npn driving a pnp) will pull to the rail. This won't consume appreiably more current (perhaps an extra 100uA), but does drive the pull-up better. This will make the USB more robust on longer wires and worse noise conditions etc.

I forget what the limits are (the shutdown limits are really a bigger problem), but this doesn't affect that. At least, not as far as I can see. The +5V bus voltage goes straight into the regulator, though a diode. The dual FET circuit only controls insertion of the 1.5k pullup that's used to signal device connection. See page 19 of the EK manual for the relevant schematic. Scott ----- Original Message ----- From: "microbit" <microbit@micr...> To: <AT91SAM7@AT91...> Sent: Monday, May 09, 2005 10:50 AM Subject: Re: [AT91SAM7] UDP pull-up control I don't know enough about USB, so leave that for the buffs. But AFAIK the enumeration process includes an initial limit of current that the hub is able to give, and then it tells how much current the USB slave device can use. If you don't power anything major through the USB, I don't think you need major external current switching at enumeration. USB buffs, if I'm wrong, please educate us :-) B rgds Kris

Hi Scott,

> I'm laying out the USB interface portion of my board right now, and I've got
> a question on the pull-up control. The SAM7 datasheet shows a simple BJT
> controlling the pull-up, but the AT91SAM7S-EK board appears to use an
> n-channel FET driving a p-channel FET. Why the added complexity? I've used
> that type of circuit before, but only for switching a lot of current. This
> only needs to handle like 2 ma. Is there any advantage to using this scheme
> versus a BJT or logic level FET?

I don't know enough about USB, so leave that for the buffs.

But AFAIK the enumeration process includes an initial limit of current that the hub is able

to give, and then it tells how much current the USB slave device can use.

If you don't power anything major through the USB, I don't think you need major external

current switching at enumeration.

USB buffs, if I'm wrong, please educate us :-)

B rgds

Kris

I'm laying out the USB interface portion of my board right now, and I've got a question on the pull-up control. The SAM7 datasheet shows a simple BJT controlling the pull-up, but the AT91SAM7S-EK board appears to use an n-channel FET driving a p-channel FET. Why the added complexity? I've used that type of circuit before, but only for switching a lot of current. This only needs to handle like 2 ma. Is there any advantage to using this scheme versus a BJT or logic level FET? Thanks, Scott