Very little is written in the MCP2211 datasheet regarding the oscillator input configurations. The microcontrollers I have been using lately allow you to use the self-powered, single output oscillators rather than the integrated amplifier feedback mechanism on-chip.

In general, can one do this sort of thing for all cases when working with an oscillator that is fully integrated with the load capacitances and amplifier/feedback loops?

i.e. rather than doing this:

enter image description here

Can we always do something similar to this?:

enter image description here


Typically, when external clock source is accepted in place of OSC it is also so stated in the datasheet, and even then there could be caveats. For example, TI TUSBxxx full-speed USB hub controllers would accept 12 MHz quartz or 48 MHz clock.

You can test it with not much difficulty yourself. Route a board for the large crystal, like HC49 can, then just leave cap footprints empty and mount an oscillator dead bug connecting its output to OSC1.

There is another thing to try. I'm not sure about this part but Microchip USB to Serial converter is actually one of their micros ( PIC18 variety if I'm not mistaken) with some of the peripherals non-functional. After erasing MC code from the flash the part can be programmed/debugged as usual. "MCP2211" part can very well be the same micro; in this case to switch to external clock source you'd have to access conf.bits somehow which is not trivial if code protection is turned on inside the chip.

  • \$\begingroup\$ I would guess that a lot of parts that say nothing about injecting an externally-supplied signal will work just fine when used in such fashion. On the other hand, some parts may make use of the fact that when connected to a crystal the signals on the oscillator pins won't be perfectly anti-phase, and may use that fact to e.g. generate multiple non-overlapping clock signals. I wish data sheets were more specific about what aspects of device behavior one may safely rely upon, and which ones the manufacturer explicitly reserves the right to change. \$\endgroup\$ – supercat Jan 20 '15 at 19:03
  • \$\begingroup\$ The devil is in the detail. Typically, you want to know whether your part is present in the set of "a lot of parts", not that this set exists. \$\endgroup\$ – Oleg Mazurov Jan 20 '15 at 20:42
  • \$\begingroup\$ My point was that parts where injecting a clock would work just fine are hardly "atypical". Parts which can accept an externally-supplied clock probably should typically list that information in the data sheet, but I'm not sure that even 50% of such parts actually do include that information. \$\endgroup\$ – supercat Jan 20 '15 at 20:45

It will work in this case, however in the case mentioned in your text (integrated load caps) it might not be a good idea without a series resistor- the external oscillator will have sharp edges that could cause current to flow in unexpectedly large quantities.

Another caveat is that the internal reset circuitry may assume a certain clock start-up time, and if the external oscillator is especially slow spooling up the chip may not properly reset. Probably not a strong likelihood, but worth mentioning.

  • \$\begingroup\$ I have wondered what this series resistance was used for. You mentioned that it limited the current flow due to spikes on the oscillator, but the schematic shows that it is in series with the output of the amplifier. Is it good practice to include these resistances and is it necessary for both types of oscillators (integrated load caps/amplifier and also devices with only the crystal)? \$\endgroup\$ – sherrellbc Jan 20 '15 at 18:22
  • \$\begingroup\$ The purpose of the resistor in the datasheet schematic is to limit the drive power to the crystal. Excessive drive power can cause the crystal to drift in frequency or even fail. It should be included if it is necessary ;-). \$\endgroup\$ – Spehro Pefhany Jan 20 '15 at 18:23
  • \$\begingroup\$ I read that section of the datasheet some time ago and it mentioned it is necessary for high-drive oscillators, but I was unsure what that actually meant. The oscillator I planned on using draws ~30mA nominally, but nothing is said about it being a high-drive device. And if I may without sounding too ignorant, want do you mean exactly about drive power? \$\endgroup\$ – sherrellbc Jan 20 '15 at 18:26
  • \$\begingroup\$ Crystals are rated for maximum drive power- from perhaps 1 mW down to maybe 1uW for a tuning fork 32.768kHz type. Too much drive power and the crystal can be damaged (too little and it might not start under some conditions). That work of figuring out the proper drive level been done for you in the packaged oscillator of course. \$\endgroup\$ – Spehro Pefhany Jan 20 '15 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.