FE1,1S USB Hub Oscillator Question

[MotoDan]

I'm working with a new USB Hub design that uses the somewhat popular FE1.1S 4 port hub controller chip. I'm not seeing any activity on either pin of the 12 MHz crystal. I've checked for oscillations with and without a device connected to the upstream port. The crystal is a YSX321SL with no external load caps added. The FE1.1S datasheet mentions that it has integrated crystal load caps.

[ataradov]

How are you observing that activity? Look at the behavior after power cycle. Are there any oscillations that die after a while? Or there is nothing at all?

[MotoDan]

No oscillations at power cycle or after. Both crystal leads are at approximately 2 vdc.

[ataradov]

How are you observing it? Can you capture the power on transition to that 2 V?

It is hard to tell what may be wrong without seeing what is happening exactly. Also, show your schematic.

[MotoDan]

It tuns out that the oscillator needed some external capacitance across the crystal. Added 12pF and the oscillator came alive.

[ataradov]

This is not a correct approach. It may be just marginally stable. And if this is the case, there should have been some oscillations at the begging.

[MotoDan]

Alex, I think you are correct. The oscillator no longer starts regardless of what I try. I was told by the chip maker that the FE1.1 USB hub should not be run off 5 as the datasheet would imply. Instead, they suggest not using the 5V input and instead power via the 3.3V input. I found that by powering the device from 3.3V does not cure the oscillator start up problem - unless I reduce VDD to approx 1-2.5V. Also tried a different crystal type and experienced the same results.

[ataradov]

Asa I said, show your schematic.

It should work fine when powered by VBUS (+5V). In that case you should use output of the 3.3V regulator and connect it to the 3.3 V input. You will need recommended decoupling capacitors on the outputs off all the voltage regulators.

[MotoDan]

This version is powered via 5V. The 5V decoupling caps are on another sheet. The 3.3V is output from the internal regulator and only connects as shown.

[ataradov]

Is 3.3 V even present when you power it iup? If you are using internal regulator, then VBUSM should be connected to the +5V. Otherwise you have a weird dependency where VBUSM==VDD33F, but VDD33F is disabled when there is no VBUS voltage.


TESTJ should be left unconnected. It has internal pull-up. Given that it also serves as I2C pin, device may try to drive it on power on.

And capture the XIN/XOUT pins at power up. If there are any oscillations at all, you likely need to populate C5/C6.