Unofficial OSHPark Turnaround Times for 4 layer boards

[ShaunAndFRED]

After listening to Amp Hour #149 Laen from OSHpark.com, it sounded like the delivery time was starting to speed up as more and more people used the service. I was just wondering what people were seeing for delivery times.

[marshallh]

About 2 weeks usually, generally he does 2 batches per month.
The problem as it stands is he rotates between 3-4 different fabs, and I along with some others keep getting plating problems, drill mishaps, etc...
Laen is pretty good at responding but nonetheless we keep having problems. I'm trialling another fab because of it.
This is all for 4layer of course.

[Laen]

I have actual data on this!

The tempo I'm going for right now is two panels a month, with a 2 week turn time on them.  That should mean that if you order at any random moment, you should expect your boards to ship in about 21 days.  If you time it really well, it can be as low as 8 days, but that's difficult.  Especially since panels can fill up and go to the fab early.
 
Anyway, the stats!

Over the past 9 months:

Minimum time from ordering to shipping:  8.00 days.  That's pretty much the minimum possible time.  Only 2% of orders got this, and only 5% hit 9 days.

25% of orders were shipped within 14.00 days of ordering.

Median time from ordering to shipping: 17 days.
Mean time from ordering to shipping: 18.14 days.

75% of orders shipped by 21.75 days.  95th percentile was 29 days.

Not great, but as it grows in popularity, I'll be able to order panels more often, which will drive that time down. 

In fact... If I only run the query on the last 4 four layer panels, the mean drops to 16.86 days.   So it's getting there slowly.

As marshall said, there were definitely some quality issues on the 4 layer panels when I first switched to FR408.  Mostly things like poor drill registration, and hole sizes being slightly out of spec.  I worked with the fab to make sure our expectations matched and found a couple of mismatches (they thought they were allowed a 3.5 mil hole size variance instead of 2.5 mils, and a 3 mil registration tolerance instead of a 2 mil) which have been taken care of.  My test boards from the past 3 panels has been clean so I _think_ we're now pretty good.

-Laen