Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

SMT Production Performance Metrics

Views: 10920


SMT Production Performance Metrics | 24 February, 2010

I am looking for some suitable SMT production performance metrics for a high mix, low volume setup. Currently we use CPH and Machine Utilization. However, in the process of catering to the HMLV setup, we have more changeovers and so the utilization and CPH is really getting compromised. Are there other metrics that we can consider? Please advise.

reply »


SMT Production Performance Metrics | 24 February, 2010

You're correct, CPH doesn't mean dip. You want: * Perfect parts. * On-time delivery. * Cpk ~2 each quarter * Zero lost machine time as a result of product change-over.

... Right?

reply »


SMT Production Performance Metrics | 25 February, 2010

Why don't you use OEE (or OLE for the whole line) where you will involve downtime (availability), performance and quality rate.

We're currently using this metric for our SMT lines, if you need info just let me know.

reply »


SMT Production Performance Metrics | 26 February, 2010


Thank you for your reply. I would greatly appreciate if you can provide me with some more details pertaining to OLE or OEE.

Thanks, Aajmera

reply »


SMT Production Performance Metrics | 26 February, 2010

you can find a very helpful guide here...

We basically have a database where operator input the boards produced per hour and if he can't reach the "standard" he can input a downtime issue and assign to a specific category and subcategory (no OTHERS!! in any category as this will contaminate your database).

this will create a matrix log that we download in excel, then I count each hour that they logged in database and based on that I made the calculation...

total time = sum of hours (in minutes) planned downtime (as a category in database)

Net time = total time - planned downtime

Total Available time = Net time - Unplanned downtime

Availability = (Total available time)/Net time *100%

lets say that you have 150 min of Net time and you had 30 minutes because of a feeder damaged and 20 minutes of setup between model A and B so your total Available time will be = 150 - (30+20) = 100 min

so Availability = 100 / 150 * 100% = 66.66%

Performance, for this you need to consider the total available time since you can't count time that line was down... it is basically how good (or bad) the line ran when it was available for production. the formula can be confusing when you use "high mix", but basically is how many minutes was running compared with available time:

Total produced time = Sum(Pc/Rate)

example: rateA = 100 pc/hr rateB = 140 pc/hr

Pc produced productA = 150 pc ProductB = 15 pc

Total produced time = (150/100 + 15/140)pc/(pc/hr)*60 min/hr = 96.43 min

Total available time = 100 min

Performance = 96.43 / 115 * 100 = 83.85 %

Quality rate is basically good pieces against total produced.

lets say you have 165 boards produced and 3 went to repair so you have a Quality rate of = 163/165 * 100% = 98.78%

based on this your OEE (or OLE if you consider a SMT line complete with your product rate as the bottleneck process)

OEE = (0.6666*0.8385*0.9878) * 100% = 55.21%

if you graph the tree indexes (managers love that) you will see that you have to improve your availability... and if you do a breakdown you could see that you need to improve your machinery (feeders) and changeover strategy.

as for the goal... there's a lot of debate but for our "old lines" we have a 60% goal and 80% for the new line.

hope you will find helpful this info.

reply »


SMT Production Performance Metrics | 3 March, 2010

Thanks Manuel

reply »


SMT Production Performance Metrics | 3 March, 2010


How do you calculate the rate? Is it the cycle time for the bottleneck process on the line?

Please advise.

Thanks, Abhinav

reply »


SMT Production Performance Metrics | 3 March, 2010

Yes, it is basically:

Rate = Time interval/Cycle time I put time interval since you can measure however you want (pc per hour, per day, per quarter hour... it is relative) the most used is Pc/Hour, so the time interval is 1 hour (3600 sec). Lets say your bottleneck is the pick and place machine and the cycle time is 25 sec per pz, then:

Rate = (3600 sec/hr)/(25 sec/pz) = 144 pz / hr

So the line "ideally" should be getting 144 pz each hour.

reply »


SMT Production Performance Metrics | 9 March, 2010

Thanks Manuel.

In an OEM facility, do we know what could be a good way to keep a track of how well SMT lines are catering to the requirements of the Final Assembly lines?

Any thoughts?

reply »


SMT Production Performance Metrics | 11 March, 2010

Currently we measure only in schedule attainment... based on Final Assembly requirements, how many boards were produced during the week. We're working on link Final assembly requirements to be able to produce "leaner" in smaller batches, but is a pain in the lower-lower back. According with management JDE (Enterprise Resource Planning system) will be solve all of our broken links, But I will belive until I see.

reply »


SMT Production Performance Metrics | 13 March, 2010

Manuel: What is the objective of OEE?

It seems to say it's OK to give-up 4% of your quality, if you can increase your machine speed by 5%.

Does it work on machines that are not capacity constrained?

reply »


SMT Production Performance Metrics | 14 March, 2010

Well, not really... the index does not say much by itself, as you say, you can improve your cycle time by 5% but if you don't update the product rate in your calculation, then you're not providing accurate data so your performance is going to be higher (sometimes higher than 100%, there's when the math screams: you're doing something wrong!!!). If you improve your machine speed, you're increasing troughput but you're not improving OEE. In few words, OEE the metric of how good you produce what you say your line is capable to produce, in the time that is stated, with the highest quality.

The best visual tool is to graph (as attached) the three factors involving OEE. let's say you improve your performance but decrease your quality so your OEE index is the same, you can notice what the difference is if you track it. On the chart Attached, if you can see the first week (WK 5) is different than the previous month trend and also different than the next Weeks results, the root cause was that even if we improve on our availability (not so much machine issues neither material shortages) we decreased performance because the lead operator left on vacation, and the backup operator hasn't the same skills than the every-day op.

if you mean a not capacity constrained machine as a low volume / high mix machine, then yes.. it is the same, the only difference is that your OEE is expected to be lower, and your main focus is availability (changeover time) and quality (of course). As you have all the time of the world to complete your requirement, performance tend to be lower.


reply »


SMT Production Performance Metrics | 19 March, 2010

Just a comment, we have been using the OEE as metric in our plant, what I've seen is that can be misunderstood because a good OEE level does not mean a good machine utilization, let me try to explain this on my own words. I have a HMLV line, so when I'm running product "A" my capacity is 50000 CPH, my cycle time is 30 sec, and those are my considerations for Rate, but when I'm running Product "Z" in the same line my capacity is 15000 CPH and my cycle time is 60 sec. (product B is the less common product and set up is not optimized). I can even have a better OEE whith product Z, but I will run the line @ less CPH than with product A. So the OEE number by Itself does not reflect the line performance. Although the managers just want to see a high number of OEE. So you have to be careful of how to present the numbers.

reply »

Reflow Oven

Used SMT