Material cushion variation

[nemanja]

Hi,

On one of products i have problem that not in each shot is same material cushion. Anyhow i have like 100 shots OK, than few with less or more cushion than should be. On this parts i have shout thru material (is overmold), or not fulfilled part. And that again all is OK for some time.
Now i need to fix this problem, but i cant relay find solution.
I check injection pressure, switch over volume, switch over pressure, cycle, dosage and injection time, and all values are within tolerances.
First what i looked was screw coupling but since i dont have problems on other products (different material) i excluded that as problem, also we use only virgin material.
Wet material test didn't give us wanted result, so im wondering what else i could try to prevent this from happening.

Best regards,
Nemanja.

[nemanja]

Forgot to mention that material is PA 6.6 GF30.

[brentb]

Nylon should be consistently dried.

KOM

brent

[chrisprocess]

How big of cushion difference are we talking about? Electric or hydraulic?
If those data points are clearly 'outliers', you could have a sticky valve somewhere on your injection, or maybe during dosage.

[chrisprocess]

May be worth while to clean the linear transducer or replace it if you're getting odd readings

[MTUHusky]

It would be helpful to know what % of the barrel you are using for each shot and the amount of cushion variation you are seeing. When you say "Wet material test didn't give us wanted result"; what was the result? With regards to the moisture, you are using PA 66 GF30, which neat (unfilled) PA has a max moisture of 0.20%. However, you have a glass filled PA and glass absorbs zero water so the max moisture for PA66 GF30 should be 0.14% (approximately). I have seen data sheets from the material manufacturer for PA66 GF50 that states the max moisture is 0.20% which is absolutely wrong. PA66 GF50 should have a max moisture of 0.10% because 50% of it is glass and glass absorbs no water.

You also mentioned that it is an over-molded process. Is the mold loaded by automation or a human? If it is loaded by a human, is their cycle time consistent every shot? when the cushion goes out of tolerance did they have a long cycle before that? If your operator runs a long cycle time then your residence time will increase causing the material to drool out of the barrel and fluctuations in your process. You can combat this by adding time to your cooling time and extruder delay (if you raise your cooling time by 5 seconds then raise your extruder delay 5 seconds), this will buy you more time without adding residence time. Be careful with this as it might change dimensions on your part so if you go this route, make sure you have your QA department do a dimensional check on the parts after. Its best just to get an operator that can keep up with the machine if possible.

Chris has an excellent suggestion, before you go messing with your process, check your linear transducer to make sure it is not a mechanical/electrical issue.

Hope this helps,

Husky

[rickbatey]

What type of mold are you using for this over-mold? I’ve seen rotation over-mold where the cavity volumes are close but not the same!

[nemanja]

Hi,

Brant, material is 100% of time dried on same conditions what manufacturer recommended.
Chris, difference in material cushion is about 0,2-0,3cm3, i was excluding machine as problem, since i have only on this one product issue. other products that are 10x smaller dont have problem, and is almost same material. Only difference is that problematic material have some additives for self-extinguish. All linear transducers are clean regularly during weekly maintenance of machine.
Husky, im using around 35% of barrel per shot. Wet material test give us just more often bad shots because of parameters out of tolerances.
It is overmolded process, but cycle time is always 32,3+/-0,1s.
Rick, we use slider table, 2x2 cavities. veight of molded parts in all cavities is 22+/-0,1g.

Now i will check linear transducer, and i will try to reduce residence time of material, and lest see what results will be.
Thanks for answers and tips. :D

[Brisli]

Are you running a standard heat profile on the barrel? or reverse profile? I have had instances where I wasn't getting it melted soon enough and causing blockages in the compression zone resulting in cushion variation without significant dose time changes.

[nemanja]

Are you running a standard heat profile on the barrel? or reverse profile? I have had instances where I wasn't getting it melted soon enough and causing blockages in the compression zone resulting in cushion variation without significant dose time changes.

For now we running standard heat profile. But I will try reverse profile, and see what will happen.

[Jendalf]

Hi,
is this problem still remaining?
Can You share some data output from the machine shot by shot incl. with the cushion drop?

If its solved, what was the root cause?

thx
j

[nemanja]

Hi,
is this problem still remaining?
Can You share some data output from the machine shot by shot incl. with the cushion drop?

If its solved, what was the root cause?

thx
j

Hi,
We didn't solve anything. so far tried reverse, flat and hump profile no results on different temperature range. Here are pictures of problem and some of process data. It is always or to much or to less material. I can send more data, but i cant upload here excel file.
254
255
256

[brentb]

What's the problem?

Are you making a mountain out of a mole hill?

KOM

brent

[JayDub]

For the page of data you show, there is no cushion variation at all, which is a bit weird. If you are seeing part to part variation, I'd be concerned that your zero position is off and the screw is actually bottoming out.

[brentb]

If I ever saw a process with ZERO cushion variation out to 3 decimal places, I'd suspect alien (ET Type) intervention?

KOM

brent

[chrisprocess]

I'd be more concerned about your cycle time varying 6/10 of a second every other shot.
How much time is there between recovery/decompression and mold open?

Use the arburg time-sequence viewer, and look to see what exactly is changing every other shot.

Your issue could be related to this change in cycle time - maybe after a number of shots the press is almost "resetting" this condition changing your end cushion
EDIT: Evidence of the above can be seen in your recovery time (t4015) before the incident is > 5.0s and after the incident is < 5.0s

I'd suspect something is incorrect with your sequencing or your machine has sticky valves.

[chrisprocess]

I would suggest adding a delay before and after screw recovery and also ramping down on pack pressure before recovery.

ex: If you're packing at say, 1800bar and after pack-time you abruptly change to 0bar, the screw can have an aggressive bounce-back affect (with inertia) Sometimes the machine won't always pick up where the end-location is, or get false readings, and do wonky things. I experienced this very issue on a sodick machine

[nemanja]

Criss, time in process is different because of sliding table. Movement of table make this process not equal. I didnt think that this can be problem, but i will try to fix that, hold is always on ramping in my programs, ending with 25bars. Nice and smooth. :cool:
And since i started this topic i tried few things:
-Change temperature profiles, no result; more or less it was same, except when i tried reverse, i had problem with starting machine (overburned at nozzle, and unable to purge).
-Delay before dosage, no result;
- Wet material test, no result; unable to inject two same shots with wet material plus some stuck material in feeding zone;
- Play with back pressure, no result
Also i tried combinations of this. Since is new machine (from February 2017), i already sent in ARBURG questions about parts that you mentioned.
Brent, i really wish that i dont need to lose hours on this matter but company owner want to reduce scrap rate from 3-5% to 0,....%. And deadline is till end of July.
Only left thing for me to test is material from different bags. Still cant find someone willing to test this for me in my country.

[MTUHusky]

That first picture looks like you have a terminal there that is now filled with plastic, I would check your terminal shutoff to make sure it's not worn out but I would expect that you would see a lot of that blow through if it was. If you turn pack pressure off, how full are your parts?

The parts look grainy so I would still think that it is contaminant or moisture related. What is your material dryer rated for as far as mass? We had a process that was having similar issues with cushion and we found that we were running 65 lbs per hour through a dryer that was rated at 50 lbs per hour.

[chrisprocess]

It would be good to understand if those cushions (V4062) are truly outliers or if there's a predictable behavior. How much data do you have on this? A few hours or days would be best.
Changing cushion to linear inches is also preferred because volume is a calculated value.

Send me xcel spreadhseet if you can offer, I'd like to take a gander :) chrispchicken11Hotmail.com
As much data as Arburg machine will give, but importantly: Fill Time, Cushion, Maximum Inj. Pres, Transfer Pressure, Recovery Time, Cycle time, End dosage position, Start Inject position (End dosage + decompression).

[nemanja]

Hi,

We have terminals but for now this is not problem. That was my first thinking but i checked tool and is OK. I had same idea about moisture, and now, just in case, im running two dryers. So material is dry. Look of material is because it have some additives for self-extinguish, UL compatibility etc.
If i turn off pack pressure, part will be around 98% full. Basically im using pack here just to "flat out" surface.
Criss, i send you data that i collected so far and convert to excel. Im having tons data more. :)

[chrisprocess]

Criss, i send you data that i collected so far and convert to excel. Im having tons data more. :)

nemanja - had a little time this am so here's what i got!

Received 6 data sets, 2 of which appear to be duplicates, and the last 2 appear to be engineering trial runs and fall outside normal run conditions.
The remaining 3 data sets are shown in sequence in the photo.
Data sets are separated by shot# (X-axis) where I drew red lines to indicate new data set.

263


Based on this information I didn't note any repeating peaks or cushion 'anomalies' that fall outside of the overall process variation itself.
Arguably some drops in or around process start-up, but this may be expected.
Maybe some type of incident noted near shot 485 in data set 3 - machine shutdown maybe or alarm?
In terms overall Material Cushion Variation, in my mind, would be difficult to pin-point it as being the root-cause of your issue.



However I did see some switch-over position changes (probably a process change) which I will email back to you, along with some other graphs you can look at.
Also if you want keep collecting data on that cushion over time and see how your overall process varies.

[nemanja]

Switch over point change was caused by shooting throu terminals. Operator need to change volume of injected material for little and after is OK. Again in some number of cycles again switch over need to change because part is unfulfilled.
Only thing what is left and i didn't tested so far is material. And i finally found institute willing to do tests. As soon as i get results, ill notify you.

[nemanja]

Hello,

We did materials testing on University for chemistry and results show 1% difference in glass fibers from loot to loot.
Did anyone else did similar tests? I also wonder if this can be problem for material cushion to change. Since i have problem only on this material, I was testing only that one, so I dont really know for sure what is "allowed" deviation of glass fibers in materials.

BR.