Can the LPC546xx's XTAL pins be run from an external oscillator rather than a crystal?

[redough]

Most of the documentation in the data sheet seems to point to using a two port crystal. Is it possible to use TCXO's or any other external oscillator instead of a crystal for the XTAL?

[ianbenton]

As far as I am aware, every microcontroller under the sun that has two pins for the crystal has them connected to the input and output of a 74HCU04 to make a Pierce oscillator - see figure 14. The reason for the lowered supply voltage is to reduce the standing current. If the supply is greater than 2x Vgs(th) for the MOSFETs then they start to draw lots of current when biassed into linear mode to make an oscillator (see figures 8, 9 and 10). I bet that the crystal oscillator in the  LPC546xx  is copied from the LPC15xx which is copied from the LPC11xx, so they will probably all work the same. The LPC11xx datasheet has even more detail on connecting to an external oscillator (see page 95).

If you attenuate the signal from your TCXO so that it doesn't exceed 1.9V then it can't do any harm. It's worth a try.

You could even try a simple resistive divider, or a series RC on the input and an extra resistor between XTLOUT and XTLIN (looks like an AC coupled inverting amplifier using an op-amp)

[redough]

Ian, thanks for the information! I will try placing the TCXO and and limit below the maximum voltage, and I'll let you know the outcome. Do you have a link for the LPC11xx datasheet you are referring to? I only found this one for the LPC111x, which appears to have the same documentation as the LPC15xx. Thanks for your help!

[ianbenton]

Erm - sorry about that - should've read that LPC15 datasheet more carefully. You're right - they are the same.

So, if you try the circuit in the LPC15xx datasheet, figure 46.  Assuming the TCXO runs on a 3.3V supply then you need  an attenuation of a bit more than 2.35:1, and they suggest a coupling capacitor of 100pF, so that requires >135pF to ground (Cg), say 150pF. Worth a try.

However, the TCXO is now driving a pure capacitive load of 60pF, and a frequency in the MHz or tens of MHz region. That's quite a lot of capacitance at that frequency, and the pulse currents will put your clock signal all over your power supply! I think I'd try a series RC, with 100pF for Cg. The C can be quite large, 1nF or more, as all it does is remove the DC; and the R needs to be about 2.2x the reactance of the 100pF cap - 350Ω at 10MHz, say 390Ω as precision isn't required. It will make the signal on XTALIN rather more sinusoidal, but that's what it is expecting from the crystal.

It would probably work with 3.9kΩ and 10pF to ground, so it needs even less output current from the TCXO.

Let me know which works best - I've been looking for a 13.11MHz* crystal, and I can only find that frequency in an oscillator module!

*so I can run the processor at 52.4MHz, and have a 10 bit PWM counter, and a further 10 stage divider to get 50Hz.

[soledad · Accepted solution]

Hi,

The oscillator input pin (XTALIN) cannot be driven by an external clock. Must connect a crystal between XTALIN and XTALOUT.

Have a great day,
Sol

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[b49]

The last two lines on page 3 of the data sheet for this MCU say:

• External clock input for clock frequencies of up to 25 MHz.
• Crystal oscillator with an operating range of 1 MHz to 25 MHz.

I recommend that NXP remove the first of these two bullets in the next revision of this data sheet because it is misleading to engineers who are engaged in product selection.  A better bullet would say:

• External crystal oscillator with an operating range of 1 MHz to 25 MHz.

[ianbenton]

The crystal pins have the same spec as the LPC15xx, which is a maximum of 1.9V. The LPC15xx datasheet mentions "slave mode" where  the XTALin pin is driven from an external oscillator, but the LPC546xx datasheet doesn't.

My guess is that it would work if you follow the instructions in the LPC15xx datasheet (couple into the XTALin pin using a capacitive divider)

[redough]

Ian, thanks for the reply and pointer! In terms of the LPC15xx will it only operate in "slave mode" with an external oscillator attached to the XTALin pin or can it operate in "master mode" with this arrangement too?