Reverse Decode - is there a better way to do this.

I have a DINT tag named ToolSelect (ControlLogix) in which only one bit will ever be set at a time. Each bit corresponds to a location in a tool holder. The set bit is the tool that is currently selected.

To determine the number of the bit which is set (for display on HMI as the tool number) I am using the following equation.

ToolNumber = LN(ToolSelect)/LN(2)

or

CPT ToolNumber LN(ToolSelect)/.69314718

Quick and simple. It also flies right over the head of Bubba on the maintenance staff.

I've tried to come up with an easier to understand method that can be programmed in one or two rungs.

I know I could do a repetitive divide by 2 and count the number of iterations, but Bubba doesn't get loops either.

Short of just commenting the rung and telling Bubba that is just how it is, does anyone have any ideas.

I'd stick with what you have. Just comment the snot out of it. In the greater scheme of things this will always work assuming only one bit is set. As long as you comment what it is and why its there no one will get too worked up about it.

Keith

Well..... I'd comment it, and hope that Bubba will just accept that it works...

Or, you can try using the FBC (File Bit Comparison) instruction.. that might work.

i didn't find ENC instruction in control logix so i was always using XPY, it should be a hell lot faster than two logs and division.

XPY works building a DCD equivalant. Log and LN are the inverse functions of XPY. How do you use it for creating an ENC equivalent?

Don't use floating point functions!!!!

Why not use the FBC function that Gerry uses?

The FBC instruction would do it. I seriously thought about using it when I originally worte the routine. I didnt use it because of the extra tags it would require. I think FBC would confuse Bubba even more - the set up of the instruction is more complex. At least with the log computation I can show Bubba it works on a calculator.

Alaric,

Check your PMs.

Marc

OK, then it should take only 5 tests to find the bit.

By clever use of anding and testing for 0 you can find the bit index by doing only 5 tests to find the bit number in a 32 bit DINT. A INT would required 4 tests.

I don't have time to expalin now because I am 'working'.

I am not familiar with ControlLogix but I am presuming a DINT is a double integer. Why can't you declare an array or structure of bits for your tool holder and then refer to the bits as required in the program ? DINT variables should be used to store DINTs.

Aaahh, strongly datatyped Mr Goldsworthy!
I agree, but I'm now waiting to hear all the reasons why Alaric shouldn't do it this way.

Ken

The tool position is modified by a mathematical operation. A simple math operation also returns a number for the position. Its easier to do math on DINTs than BOOL arrays.

The program is object oriented. Thus the tool holder/changer is a single object tag defined by a UDT containing all information pertintent to the control of it. Becuase its a UDT a BOOL array vs DINT does not affect memory size.

Addressing single bits in a DINT is the same as addressing bits in a bool array and creates no difficulty.
DINTTAG.1
BOOLARRAY.1

Finding bit index

Now one has a good algoritm for finding the index of a set bit?
There are better ways than testing all 32 bits one at a time.
Let see some creativity, Terry?

Has anyone tried searching the internet? This stuff has been beat to death as it is a very common problem.