Is this before, or after, doing a full process characterization? I suppose some sort of genetic algorithm could possibly hunt and peck its way to an optimized process, but even then, a mold is essentially cast in concrete, so process can only go so far.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm

There is a classic example about why process control cannot compensate for random process variation.

Hold a funnel two feet above a target on the floor. Drop a marble into the funnel 100 times and you will note the pattern of hits is a Gaussian distribution around the bullseye. Then repeat the experiment but after each marble, move the funnel to over the spot where the previous marble hit (feedback control). You will find that the accuracy of hitting the target has gotten worse, not better.

You have to fix the cause of random process variation in temperature, pressure, speed, feed materials, etc. Once you fix that you will not need automatic dimensional feedback controls.

There are machines integrated with cavity pressue monitoring, which involves expensive tooling.
Personally, I think it's marketing. I have moulded over 1.0m parts not-stop and the size difference between first off and last off (and all in between) was immeasurable in any reasonable sense. Modern im machines are all closed loop for all the parameters anyway.

Why do you ask?

Edit: Compositepro got it spot on.


Politicians like to panic, they need activity. It is their substitute for achievement.

Hi all,

Thanks for your responses. The context here - I’m doing some commercial feasibility work with a company that is developing a new component measurement technology designed to overcome the limitations of metrology. Their system will increase the measurability of plastic component features at high speed, in-line or near-line, repeatably and to high tolerance accuracy. Key features:

• Can hold any component on a single fixture without having to re-fixture and without compromising its form or geometry
• Can orient the component in any direction so that its high tolerance measurability is significantly increased.
• Can measure components at speed, in real-time, repeatably, on the shop floor.
• Can measure components to high tolerance accuracy (comparable with CMM).

I’m out there speaking with a few world class manufacturers to get the full picture on the problems this type of technology might solve if any, and why. So, we are in listening mode as we seek to ensure that the technology addresses a problem worth solving. A closed loop aspect is an obvious consideration where the dimensional data streams and the other factors could perhaps be used to predict dimensional issues and make corrections before they arise. I'd be interested in your views on any perceived benefits of the core aspect to this... and then on the forward view prediction aspect. It would also be useful to know if there are any existing systems out there that would take/absorb measured dimensional data streams from such a system and feed it back to enable process optimization and valdiation to better control the process.

Thanks in advance

Best Regards

Kieran