MSp430G2553 freq calibration

The datasheet tells that the lowest tap of the DCO has a frequency of 60 to 140kHz. So worst case, 100kHz are impossible.
However, if you have a 32768Hz crystal attached, there is a software that calculates the required DCO settings for a wanted frequency. Search the forum for it. It gives you the values to write into the DCo registers, so you come closest possible to the desired frequency.
This must be done for each and every MSP individually.
Still, worst case, it is impossible,

The other solution is to attach an external 100kHz signal to XIN. While the datasheet limits the frequency on XIN in bypass mode to 50kHz, experiments have shown no such limitation. (well, maybe the current consumption will leave the specified range if >50kHz are applied, so this was the reason for the limit on the datasheet)

You can calibrate DCO to 200kHz and divide that by 2 for MCLK and SMCLK. Or 400/4=100. Or 800/8=100. Doing it this way will consume slightly more power, but the clocks will be slightly smother (smaller jitters) and well within the DCO capability.

For production, Elprotronic may be able to do it one by one. Ask them.

If you look into the MSP430G2xx3 code examples, there's a file msp430g2xx3_dco_flashcal.c which shows you how to calibrate the DCO to a desired frequency.

Note that running the code as is will override the existing calibration data in INFO_A.

With the VLO, while it is wildly inaccurate and unstable, you could measure its frequency using an accurate clock source (like the 32.768khz crystal Jens-Michael suggested), so that you know roughly where the VLO operates within the 4khz-20khz range. You then only have to deal with frequency drift and instability instead of outright inaccuracy.

If you can tolerate more inaccuracy then you can use one of the pre-calibrated DCO frequencies to measure the VLO.

Of course if you can use a watch crystal you should be using *that* instead of the VLO or the DCO for low frequencies (<1MHz).

Tony