Radius Not Enough % of a Circle

I was taught something called the T-value method, and I'm curious to see what others think. If you have the model (or can correctly program the points using calculations) you take individual points on the radius instead of a circle. You then report the T-value for each point and use a variable to calculate the average. You then take the nominal radius size and add or subtract the average T-value deviation depending on whether your value is positive or negative and if your radius is an inside radius or outside. Of course if you have both positive AND negative deviation this doesn't work, but I generally don't run into that. Thoughts?

What is T value anyway? Isn't it essentially the same as a polar radius? My interpretation if the help file is that T value is polar radius except it's reporting deviation from nominal instead of a radius value.

What is T value anyway? Isn't it essentially the same as a polar radius? My interpretation if the help file is that T value is polar radius except it's reporting deviation from nominal instead of a radius value.

Deviation from nominal. 

Right. Relative to a coordinate system (datum system locking 6DOF) right? Cookea, I believe you're doing it the hard way. Look into PRAD (Polar Radius option)

T-Value is the amount of deviation from nominal along the vector of the feature I believe. I saw a post from Neil about it.

Quick Edit: Found post here at the bottom

https://nexus.hexagon.com/community/public/pc-dmis/f/pc-dmis-for-cmms/138169/convert-tp-to-profile

With a vector point, it's deviation along nominal vector (machine can't do anything else).  Not sure if it's nominal or actual vector with surface points, if I ever need to use T-value, I'll test first.

I wonder how a CMM measures the actual vector of a point. Like what is happening mechanically...somehow, the machine knows the direction of the surface that the probe contacts?

(Technically, that's not completely true, depending on probe head, but software would get a little complex if vector points were treated differently depending on head)

Yes, I think PRAD would do the job, but I'd still have to assign a variable to calculate the average. I'd also have to have an alignment that puts my origin at the nominal center of the radius, right? T-value just gives me a way to do it minus the alignment part, since a lot of the radius measurements I'm reporting are at various locations/angles on the part. I was just curious if anyone knew of anything wrong with this method. Hopefully not, because I find myself using it quite a bit lately.

Calculating the average seems like a pretty good idea. Yes, you have to translate your origin to the center of the radius to do P-Rad. Using T-value seems totally fine to me. The reason I don't use it is because I don't really know how to convert that to a radius number on the FAI report. Would I do Nominal Radius Size plus T-Value Result? Just curious.

It depends on if it's an outside radius or an inside. If it's an inside radius and the t-value is positive, that means the hit occurred too early, which means your radius is too small, so you would do nominal radius minus the t-value average. This way if the t-value is negative it gets added, which is correct since your radius would be too big. If it's an outside radius and the hit occurs too early your radius is big, so you would do nominal radius size plus the average.