SLC 32b math

I need to multiply a float (0 - 36) by 1,000,000 and then totalize this. What I am thinking of doing is using the 32b math operations, but I'm not sure exactly how it will work.

Let's say that I take this float and multiply it by 1,000,000.0 it will of course come back as a float and not a integer. That's bad. How do I fix that>?

Second, how do I utilize S:13 and S:14 for 32b add operations?

Thanks,
Jason

Why do you have to multiply it by 1,000,000?
If that's a constant scaling factor, there is no reason I can imagine to include it.

That's a good point ... but I would like to know hown many total gallons. ... or well, maybe at least how many thousand gallons. That would give me X.XXX million gallons. Maybe I'll just multiply by 1000 ... but 36 x 1000 = 36,000 which is above 32,767.

You are simply running up against the encoding limits on your system. Scaling by a factor, any factor, won't help; especially if using floating point numbers.

Eventually, even if you maintain a count as a real, you will not be able to add a new value to it as it will fall outside of the machine epsilon value of the floating point representation. AFAIK, the SLC deals only with 32 bit floating point values, which at most can contain 7 digits of precision. You are already at that limit by multiplying 36 * 1,000,000.

Break it down into two (or more) separate integer counters, one for units (0-999), one for thousands (0,000 - 999,000) (continue as necessary).

Re-aggregate the values in a higher end system if necessary (HMI, Database, etc).

You will have to convert the float to a 16 bit integer ranged 0 to 3600, and then do a 16 bit multiply by 10,000. You'll get an integer overflow in the destination register and the 32 bit result will be in the math registers. After the multiply make sure you clear the overflow minor error bit or the processor will fault at the end of the scan.

If you are just wanting to display the data then go with Ron's suggestion or use a large volume unit (kgal?) in the PLC and then multiply that by some conversion factor at the HMI instead of the PLC.

Thanks guys ... sounds plausible ... I'll write up some code and see what happens.

Thanks

Questions ...

When I have an OVERFLOW and the result (of an ADD or MUL) is moved into S:13 (lo) and S:14 (hi), what happens to the sign-bit of S:13 and in my destination?

When I use an STI, I'm assuming that the subroutine that I designate will only be executed once. I'm assuming one shots won't be needed on (certain) math functions. Righto?

Thanks

The sign bit of the result is preserved in the math register. The destination register contains the unsigned integer overflow, however since the SLC always interprets bit 15 as a sign bit then you will see a negative number in the data monitor if bit 15 is set.

An STI performs a single scan for each configured time period, independent of the main scan. So a 500mSEC STI runs twice a second. As for whether one shots are needed I can't say without seeing your code.

Please correct if I'm mistaken, but on the SLC, an overflow of an integer field will also result in setting the minor error bit, which must be cleared before the end of the scan, or the processor hard faults.

That is correct Ron. If a programmer wants to use the math register for 32 bit math he must correct any minor error bits before the end of the scan. IMO, its good practice to watch the minor error bits anyways.

Alaric said:

That is correct Bob. If a programmer wants to use the math register for 32 bit math he must correct any minor error bits before the end of the scan. IMO, its good practice to watch the minor error bits anyways.

Click to expand...

Just a little correction