LPC55S04: Can device run with no external crystals or clocks

[cpm72]

Hello,

We are using an LPC55S04 microcontroller in a new low cost, space-constrained design. Can you confirm that the device can start up, run an application and be used with a debugger when there is no external crystal or clock provided on the XTAL_32M pins?

Various sections in the user manual (1.2, 4.2.3, 4.5.32, 4.5.33 and 4.6.3) indicate that the device starts up from the internal 12 MHz trimmed FRO. I just wanted to double-check that I don't need to leave room for external components.

Regards

[Julián_AragónM · Accepted solution]

Hi @cpm72,

There is no need for an external oscillator as long as you dont need a higher frequency or accuracy for your application; I suggest also looking at chapter 7.20.1 & 7.20.2 (Clock sources & PLL), as it has extensive information about the external and internal clock sources this board supports.

Best regards, Julian

[cpm72]

Hi Julian,

Thanks for your reply. We don't need accuracy for our application, but is there any reason why higher frequencies can't be used?

We would want to switch to the internal 96 MHz FRO after startup and, if necessary, use one of the PLLs to boost the speed further. I can't see anything in section 7.20 that would prevent that when we don't have an external crystal.

Regards,

Chris

[frank_m]

Tolerances will multiply when using the PLL, and USB or ethernet would definitely not work (no connection due to clock deviation).

I suppose you might even get trouble with RS232 connections. Longterm timers might also deviate noticeably, depending on temperature.

[cpm72]

Ok thanks, good points. If no crystal is fitted, we can probably avoid having to use the PLL and we don't need to worry about USB or Ethernet for this application anyway.

The main peripheral interfaces are SPI and I2C, where frequency accuracy doesn't matter. Also, I'm hoping the 1% trimmed FRO over 0-85C should be good enough for a UART interface running ISP or debug. Timers should all be short term.

[frank_m]

I was just pointing out some interfaces and peripherals that depend on clock accuracy in some way. According to my experiences (with other devices), issues tend to occur at extreme temperatures for most devices, and at moderate temperature for a smaller percentage of devices.

Not sure if you get along without PLL, performance-wise.

In regard to UART/debug, very much depends on the "other" device. While the RS232 standard defines maximal bit time tolerances, some devices can digest larger deviations without problems. 

UART is relatively prone to signal corruption anyway, so a robust protocol with consistency checks and possibly repetitions could alleviate those problems.