Uniformity of thickness of molded plate

[LSasso]

I'm having an issue getting uniform thickness on a flat part that I am molding. The part basically amounts to a 3" square plate, 0.1" thick, gated radially near the center. I don't have a choice with the gate position. It's vented over about 35% of the perimeter, at the corners, at 0.0015" depth. The material is cyclic olefin copolymer. The entire mold is aluminum.

The problem is that there is a thickness gradient from the center to the corners of about 0.003". I will need less than 0.001" variation for this part to work properly. I've tried longer hold times, shorter hold times, high initial pressure with a long follow up at lower pressure, larger gate (as large as it can be), higher injection pressure, lower injection pressure, warming up the cavity to around 130F. The only thing I can affect is the overall thickness which varies with the injection pressure and somewhat with the hold time, but the thickness gradient remains the same with every parameter I've changed.

Any idea on what might cause this and what I might try adjusting? I realize that there is a pressure drop away from the gate, and perhaps it's causing it to shrink more at the outer edges, but I have no idea how to address this.

[pjhall]

Do you have support pillars behind the core side? Note where they are and where your thickness variation is and see if there is a correlation!!!

[Suhas]

Hi Lasso,
Adding to PJ above
1. The part is thicker at the center - correct?
2. How many cavities?
3. Hot or Cold runner?
4. Have you checked (you must have but just confirming) the steel and is there any variation there?
Let us know.
Suhas

[rickbatey]

You may be fighting the tendency to act like a potato chip and the outside edges roll up. That is differential shrinkage from the centered gate and pressure gradient across the cavity. Since you state you can't move the gate, then let's talk about very short hold times with a downward ramped pressure profile. Higher mold temps on both sides might help...way hotter than you have tried.
Another option is to add injection/compression to the press and use it. Many parts are produced in this manner.
You may even be experiencing mold plate deflection. You should see how much preload is on the support pillars on the mold: pay special attention to the center on the moving half.
Another possible solution is to use a CFA or blowing agent. I don't know if these work on your resin.
Rick.

[LSasso]

Thanks everyone for the very helpful responses. I've been investigating these ideas and have come up with some possible solutions. It's a single cavity, cold runner. I had previously checked the flatness of the cavity, but I hadn't looked into the possibility of deflection since the tool is very small. After running a stress analysis it seems this was a poor assumption. The deflection predicted by the simulation agrees with the deflection I'm seeing in the part. I'm designing some additional cavity supports which I will test in the next day or two. I will also add precise mold temperature control, which I should have regardless for dimensional stability. Hopefully these changes will bring the part into the acceptable tolerance range, I will post again once I've tested this.

Thanks!

[LSasso]

Just an update on this topic. According to the FEA model, most of the movement of the mold cavity is due to compression of the plates from the injection pressure, and less from deflection of the plates, which was quite surprising. With this in mind I decided to tackle the mold heating issue first so that injection pressure can be backed off before the part freezes. By running the mold at 95C, the maximum recommended for this polymer, I was able to get the gradient to drop from 0.003" to 0.002". It always seems like the last 10% of the job takes 90% of the time. Next I will build a cavity support system that adds the most support I can possibly fit and still have a working ejector system. Hopefully the model isn't completely accurate and this will get me that last 0.001" or better. I'll update again when I get this tested.