Question - USB with internal RC Oscillator

[lpcware]

Content originally posted in LPCWare by Clovis on Tue Jul 24 08:43:32 MST 2012
Hi everyone. I'm introducing myself in the world of CortexM3 and I would really appreciate your help in this matter.
I've been succesfully testing some example codes for USB HID with an  LPC1769 on a LPCXpresso Embedded Artist's Base board that I've found on  the Internet.
I would like to modify this examples to operate the USB HID using the  internal RC oscillator in order to simplify a future PCB Design.
I've been reading LPC17xx's User Manual, Clocking and Power Control Section.
The internal RC Oscillator runs at a 12MHz frequency, and the system PLL  is capable of taking it as input and multipliyng this frequency four  times, so I could generate a 48MHz main clock. This clock would allow  USB HID to work fine.
HOWEVER, I've seen SEVERAL times in the user manual that it's  strongly suggested (almost as if it were an obbligation) to use the  system oscillator (external) and not the internal RC oscillator when  using the PLL0 output for USB clocking.
Is it impossible to use the IRC?
Has anyone faced this issue before?
Thanks in advance!

[lpcware]

Content originally posted in LPCWare by Clovis on Wed Jul 25 10:23:40 MST 2012
Thanks everyone, I think the answer is quite clear.

[lpcware]

Content originally posted in LPCWare by NXP_Europe on Wed Jul 25 03:37:26 MST 2012
Hi Clovis,

in the testprogram the IRC-oscillator is being set as accurate as possible.
But we have to deal with the full range of voltage and temperature, so we have to be sure that we stay within the specification.

By coincidence it may work on an IRC, but we advice to use an external Xtal.

[lpcware]

Content originally posted in LPCWare by frame on Wed Jul 25 00:02:36 MST 2012

---

Quote:
With 1% IRC accuracy that's hopeless :)

---

And if I don't mess up vendors, that 1% are not satisfied over the temperature range. It is actually 3%.

The internal RC oscillator has the advantage of being simpler to use and much more robust mechanically.

However, all professional boards come with an external quartz...

[lpcware]

Content originally posted in LPCWare by Tsuneo on Tue Jul 24 20:12:10 MST 2012
If you were a hobbyist who could take the risk that your board occasionally fail on USB, you might go with the internal RC OSC. But for product design, which should always work, a crystal (or a high-accuracy resonator) is required, to satisfy +/-0.25% tolerance for Full-Speed on the USB spec.

Thanking to 4x oversampling DPLL, USB engine can decode bus signal up to around +/-25% deviation. Therefore, deviated clock works in most cases. But when host schedules a transaction at the end of frame, and when the device is too slow, babble error (transaction overlaps on SOF timing) occurs.

Tsuneo

[lpcware]

Content originally posted in LPCWare by ex-kayoda on Tue Jul 24 09:07:40 MST 2012
With 1% IRC accuracy that's hopeless :)

[lpcware]

Content originally posted in LPCWare by Clovis on Tue Jul 24 08:53:04 MST 2012
thanks for the quick reply!

I kind of thought so, but I was wondering if anyone really tried if a HID can handle with the IRC precission...

[lpcware]

Content originally posted in LPCWare by Ex-Zero on Tue Jul 24 08:48:51 MST 2012
That's an accuracy problem :mad:

See: http://knowledgebase.nxp.com/showthread.php?t=28