| ??? 10/19/06 01:03 Modified: 10/19/06 01:09 Read: times |
#126719 - Difference of both half periods is to be measured! Responding to: ???'s previous message |
Ertan said:
Due to the high costs of VFCs and ADCs in one of our products we are planning to use a NE555 in astable mode as a VFC to convert the analog signal into digital and measure the frequency with a 52's timer2 in capture mode.
A similar application is described in one of st's application note: AN1755 Take note, that the application note does not suggest to measure the frequency, but the difference of the both half periods! Only by this the inaccuracy and drift of involved parts do show a first order vanish. Ertan said:
Altough the application note is titled as 'PRECISION THERMOMETER...' they do not mention how precise it is and they do not say anything abut the accurcy of the overall system. Yes, because they do not know which NTC you plan to use. But they demonstrate, that many of the inaccuracies and drifts vanish. If high precision is needed, only a state-of-the-art NTC should be used, like these ones, for instance: http://www.eik.ktu.lt/personala...temp_4.pdf For applications at moderate temperatures long term drift of these parts is less than about 0.1K in 10 years! Ertan said:
For timing error at page 3 it states that for astable operation the initial accuracy is 1.5% typically.
Is this the accuracy of the DUTY CYCLE or FREQUENCY of the generated output signal? This error simply reflects the fact, that the internal voltage divider doesn't exactly divide Vcc into 1/3 and 2/3 at the comparator inputs but shows some manufacturing tolerances. A deviation which can easily be corrected by a suited calibration. Take note: Not the error itself is interesting, but only its long term drift and this vanishes in first order with the methode explained in the application note. Ertan said:
Is such kind of analog to digital conversion reliable for a 0.1% accuracy? With a bit luck. But you must follow the methode explained in the application note and must calibrate the curve at several temperatures (look-up table / piece-wise linear interpolation). By the way, for highest precision you should choose the CMOS555. Kai |
| Topic | Author | Date |
| NE555 as a VFC (to Kai) | 01/01/70 00:00 | |
| No way. | 01/01/70 00:00 | |
| Difference of both half periods is to be measured! | 01/01/70 00:00 | |
| Thanks... | 01/01/70 00:00 | |
just a note | 01/01/70 00:00 |