Working with LSBs

[Markdem]

Hello, I have been tring to work this out for hours now, but can't get my head around it. From the data sheet we read the following;

"The resolution of this register is 2.2 ms/LSB. When the line frequency is 60 Hz, the value of the period register is approximately 7576d"

So I know that at 60Hz, the period is 1/60 = 16ms. What I don't know is how you get 16ms out of the 7576 that the register is holding? How do I work this out.

Thanks

Mark

[Ttelmah]

What data sheet?.
It is hard to 'guess', which register you are talking about (except that a timer is involved). Given there are several hundred PIC models, it'd be nice to have a 'hint'.....

Best Wishes

[Markdem]

Sorry, the data sheet is not for a PIC, it is for a AD7753.
I did not think that it would matter, as a LSB is a LSB, regadless of which device it is on, or is there somthing I am not understanding.

Thanks

Mark

[Wayne_]

the value 7576d = 0x1D98 or 0b0001110110011000

So the LSB (Byte) = 0x98 or 0b10011000
= 152d

The resolution is 2.2ms, with a value of 152 = (152 * 2.2) = 334.4ms

Something not quite right, hold on a min

[Markdem]

Thats what I mean. I just can't work out what they are saying. Even if the datasheet has a typo in it, and it is ment to be 2.2ns, instead of 2.2ms, it would still be wrong.

The link to the product and datasheet is below. Section in question is on page 18 in acrobat reader, under "Period Measurment"

http://www.analog.com/en/analog-to-digital-converters/energy-measurement/ADE7753/products/product.html

Thanks

Mark

[Wayne_]

OK

LSB refers to Least Sig bit (LSb)

60Hz = 1/60 = 16ms

Calculation 38 shows Line period(s) = PERIOD * (8 / clckIn)

Their example shows clkIN as 3.579545 MHz so
Line period(s) = PERIOD * (8 / 3579545) = PERIOD * .0000022...

This shows the resolution to be 2.2us NOT 2.2ms as stated

7576 * 0.0000022 = 0.016... = 16ms

[Ttelmah]

Yes, an LSB is an LSB, but how the calculation 'gets' to 60Hz (which is what you asked), cannot be explained, without knowing what the numbers are referring to. It is also not plain, whether this number is actually 7576 decimal (which is what you seem to have assumed), or 7576D Hex.

I'm going to 'guess' that you mean the ADE7753 metering IC?.
Finding the actual paragraph, then tells us that this is clocked off 3.579545MHz. Then,we see that this is a zero crossing detector, which means it will trigger _twice_ in each mains cycle, making the actual time involved, 1/120th second, not 1/60th second. Looking further through the sheet, we find that the master clock, is divided by four (page 30), so we have the count in this being:
(3579545/4)/120 = 7457.

A slight, and 'interesting' difference.

The actual formula for the 'lline period', is given as equation 38.

They also seem to be doing quite a bit of 'rounding', using 1.1uSec as the /4 clock period, when it should be 1.11746uSec for the quoted clock...

Even more oddly, the data sheet says just in front of the bit you have quoted, that the 'MSB of this register is always 0'. Makes quoting a 16bit value rather 'weird'....

Now, they are talking about the 'resolution', which suggests that this should be the step size for a single bit charge. A quick 'thought', tells us that the resolution, for the frequency (given the /8, rather than /4 that this involves), will be in 2.2uSec steps. Suggesting that this is a simple factor of 1000 misprint, and should be in uSec...

There have also been erratas in this part of the sheet already listed. I'm afraid their accuracy is not good...

The number they are quoting, seems to be based on a master clock at 3.63648MHz, which is actually a quite likely, and common value, but it does make working out what they mean 'interesting'.

Best Wishes

[Markdem]

Thank you all for the help. I guess I was on the right track, but did not know that the calculation involves the clock.

Anyway, it is working.

Thanks again

Mark