SMT Line Performance Measurement

Hi Processman,

Where I worked before, as an SMT Department Leader, there were two more departments who ran SMT lines, and we had a task from the CEO to decide to how we measure our performances. The result was like this:

- first of all, you will have a changeover time for every product. Depending on your business, machines and complexity of the machines, you should define a 'legal' average changeover time which is not considered as downtime. We all worked in the 3 SMT departments with ASM Siplace machines, so we decided this legal changeover time to be 30 mins (interval of measurement of this time was last board at entrance of oven from one job, to the first board at entrance of the oven of the next job);

- second, there will always occur small/big problems on the line, so we said that only those downtimes will be investigated first, which are bigger then 10 minutes (same measurement point, oven entrance; time difference between 2 boards);

- third, we defined together that time spent on new products (sample production), which run on the line for the very first time, not to be considered downtimes, as we carried out process optimizations, case studies of new components/tooling and preparing production procedures for serial production;

- fourth, planned maintenances are not downtimes (our decision) because they are planned, and if carried out in the given timeframe (our case, 1 shift) then it is legal. All the time which was more then 480 minutes over maintenance, was considered as downtime.

Add all the illegal downtimes together, and take it out of your available production time (between 15-19 shifts) and you can calculate a percentage.

We used traceability system, so behind all this, there was a massive amount of automatically gathered data, but it worth it. It made us more competitive, and more attentive to details.

If you want more information regarding this, you can contact me.

BR, Tom

reply »

Words of advice:

There needs to be an audit of downtime data being logged by operators. Everybody has to be on the same page and understand how and when to log specific reasons.

It makes the boss feel better that you are measuring something BUT you have to be realistic..... Taking the resource to collect massive amounts of data that you really don't do anything with is of no value and needlessly saps resource.

reply »

Back in the old days, based on "The Goal" by Eliyahu Goldratt, we wanted to optimize throughput. In this ...

* THE GOAL: Make more money. Increase throughput. Make more money.

* Throughput is gated by the bottleneck.

* Bottleneck is where the material is converted more slowly than other parts of the process. Bottlenecks have buffers in front of them.

* Spending money on improving things in areas other than the bottleneck will not improve throughput.

More recently, Ron Lasky [ronald.c.lasky@dartmouth.edu 603-646-9197] wrote "The Adventures of Patty and the Professor." It is similar to "The Goal." Where "The Goal" has no industry focus, "The Adventures of Patty ..." focuses on specific electronics assembly process improvement tools and is excerpted from Dr. Lasky's blog. Both are good books.

reply »

The best way to measure is by money made(quality product and minimal rework). I worked for a company that asked me to achieve building 100 boards/week for an assembly. A year later they were unhappy that we don't hit 150 boards/day of the same board consistently everyday. Most of the monitoring nowadays is useless. Managers place goals on engineers without understanding the goals. Then goals are not met and requirement will be shifted to meet the goal. It is all made up charts and graph, but this is all corporate America understands nowadays. "Ohhh the DPMO graph goes up and this is great". My point is figure out what you want to measure and what is important to measure. Your machine might be moving 80% of the time and you can still build crap.

reply »