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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

Downtime Calculation



Downtime Calculation | 6 August, 2002

How do you calculate downtime of an SMT Line with 5 machines (1 Printer, 3 Pick and Place and 1 Oven)?

Let's say that my stencil printer is down, since it is the beginning of my process my whole line is down. But if my 1st pick and place machine is down for my 15 mins and the rests are running, how will I calculate my line downtime for that machine only? What if the next machines stopped due to PCB waiting because of the downtime, will it be considered downtime?


reply »

Ken Bliss


Downtime Calculation | 6 August, 2002

Hi Brandon

There are at least two reasons to calculate down time on each machine. One is to know if the individual machine is having ongoing problems and either needs to be serviced or replaced. I doubt this is much of a problem for most machines. The key area of downtime on a line is based on the bottleneck of the specific line. I have always found it to be the pick and place machine. if you have more than one machine on the line, then it is a matter of timing the actual placement time for each machine and the one that runs the longest cycle is the bottleneck, a cycle being the actual time the head is placing components. With that information measure the down time of that machine (the non placing time) and that will give you a true answer of down time. The non bottleneck machines are really not an issue on the line from a downtime perspective unless another machine is having a specific problem Defining down time is another issue, some cannot be avoided. Scheduling maintenance during non production times is best. Feeder reload and changeovers need to be staged offline and ready to swap out feeders to shorten the changeover time to the minimum typically less than 20 seconds to swap out a reload and 15-20 minutes to do a complete changeover. A simple overall measurement is to track how many components per shift and per week are placed and track that average by the hour, you will see if improvements are actually improving anything, the more placements the more uptime or less downtime. You can find a Calculator online to see how much uptime can be recaptured and its impact at then click on PDM calculator.

reply »


Downtime Calculation | 17 August, 2002

Hi Ken

First of all, I think you are right to point out that the uptime should be calculated on the SMD line bottleneck.

But I have some comments to your posting in general:

The SMD line bottleneck, you say, is always the palcement machine. This is not always the case. It depends very much on the cycle time of the PCB produced. I find that the screen printer (cycle time 15 - 60 sec.) and reflow oven (cycle time 20 - 40 sec) often are the actual line bottleneck.

Also you must differentiate between the SMD line bottleneck and the company bottleneck (constraint).

Company constraint = a process or resource limiting the company�s throughput.

In you calculations on your web-site, it seems to me that, you assume that all SMD lines / areas are the company's constraint. Which is not true in all cases. A lot of companies do not run 3 - 5 shifts to cope with the market demands. So it is therefor not a company constraint. And you can therefor not assume that increasing efficiency on the SMD line will increase sales. The only costwise improvement you can get on a Non-constraint SMD line is a decrease in labor cost (Operational expense). And this will give you a total different set of numbers.

But I find it very positive that you are pointing out that a lot of things can be done to increase the SMD area output without actually investing in another expensive line.


reply »


Downtime Calculation | 17 August, 2002

Hi Brandon.

You should always try to optimize / "balance" your line in a way so the machine with the shortest time per placement is your bottleneck. The other machines in the line should then have a 10 - 20 sec sorter cycle time, to be able to cope with reel change etc. This way you will get the most out of the uptime you have !

This is the uptimum situation but in real life this may not be possible in all cases. So the downtime / uptime measurement should be measured on the actual line bottleneck for the particular PCB.


reply »

Ken Bliss


Downtime Calculation | 17 August, 2002

Hi Brian

Thank you very much for you comments, Points well taken, forgive the length of my response here, but I wanted to respond to each of your points.

The Pick and Place machine �should� be the bottleneck. The reason is in my 20 years in the industry we have been able to relieve every bottleneck in the plant except the pick and place machine, meaning that once the bottleneck has been chased to that machine we have not been able to relieve it as a stand alone machine. This is after we have recovered virtually all down time caused by feeder exhaust and reloads down to seconds and minutes. Maintenance is a separate not easily controllable issue, and tweaking certain jobs is also difficult to control.

Our online calculator that I am pleased you have tried out, is intended to be used in conjunction with our white paper �Profit-Driven Manufacturing�. Actually the process of finding the the first real bottleneck takes some expertise, which is where consultant companies like mine at Bliss industries comes into the picture to help find it and relieve it. You mention about the �company� bottleneck, that is also correct. As you relieve the factory bottlenecks you will create others including a �market constraint� meaning you do not have enough business to support the factory. In this economy I already hear everyone yelling, my bottleneck is the market constraint since most are slow or slower than they would like to be. The key here is for sales management to be involved with understanding that as bottlenecks are relieved in the factory more finished product can be produced by the same staff, which means if you where producing say 100 assemblies per day your true cost is the entire cost of running the plant for the day including labor and materials divided by 100. If you increase output 30 percent in my argument that would increase your output to 130. You then can divide your total cost by 130. Your cost per unit comes down dramatically.

Now take that reduced cost to sales and split the savings with them in the way of allowing sales to increase discounts to win new business that they could not get before as they were not competitive enough. This will relieve the market constraint and make you more profitable at the same time.

Clearly this is an ongoing process, our methods and processes are bases on �TOC� Theory of constraints as written by Eli Goldratt and his book �The Goal� and �It�s not Luck�. Clearly proven in industry to work by Dr. Goldratt.

The key focus must be to continually identify the true bottleneck and relieve it and then do it again, efforts on throughput anywhere else has little or no true impact on the companies profits. Only machines should be allowed to be a bottleneck. If people are, that process needs to be corrected. Again my experience is once you push the bottleneck to the pick and place machine you have all that accomplished. Then if you have the business to support it you put an additional machine on the same line to relieve it until each machine is about the same amount of time. Then its time for a new additional line.

For those that already have a second pick and place machine on the same line will require focus somewhere else.

On your cycle time for reflow and screen printing, perhaps I am not understanding something. The time you quote are extremely long from my experience with virtually all major machine manufactures. The screen printer cycle time should only be the time it takes for the first conveyor staged board that is waiting to move forward and start the print cycle go through its process until the next staged board begins to move. I would not count any other time as cycle time of the printer as when the line is up and running the initial full cycle only occurs once on the first board going down the line.

The same should be true on the reflow oven, one board after another as fast as they come out of the pick and place machine. The conveyor moves at a certain speed through the reflow oven, the actual time from one end to the other may be 30-60 seconds but if everything is running right once the first board has entered the reflow oven the next board should be right behind it at the exact time it takes for the pick and place machine to complete its work.

I look forward to your response to my comments. I really enjoy this forum.

Ken Bliss

reply »


Downtime Calculation | 18 August, 2002

Hi Ken.

It's always fun and a good exercise to debat on a interesting subject like this.

I can see at your response that I have to clarify some of the issues.

Yes, the placement machine should be the line bottleneck. And in a line with more placement machines, the machine with the lowest tact time should be the line bottleneck. But only enough to take up the downtime due to reel exhaust on the other line non-bottleneck placement machines. How many sec. the cycletime should be lower than the line bottleneck depends on the buffer (conveyor) between the machines and the time it takes to change a reel. The important thing is to keep the machine with the lowest tact time running as much as possible.

But again this really depends of the company's ability to sell everything produced.

This leads us to your calculations again. You assume in your reply that if you can produce 130 PCB�s in the same time as you did 100 not using PDM or TOC methods, then you can go to sales and convince them to lower the sales price and get more sales. Starting a snowball effect ! But you overlook that not all companies make money, unfortunately. So, in these situations an uptimization on the SMD line will not turn the company around entirely unless the extra time is used in a productive way.

Here the increased performance should really be used to lower batch sizes and reducing of work in process inventory. This will then lead to shorter lead time and lower inventory. And when this turn the company towards making money then your sales plan should be initiated.

Now, cycle times. Not many screen printers has a complete cycle time (load + print) less than 15 - 20 sec. And if the company has a offline screen printer, a cycle time around 40 sec. is absolutely not unrealistic. On the reflow process, we also have to clear up a thing or two. A complete reflow profile don't take 30 - 60 sec. but 4 - 5 minutes. And when I state that the cycle time could be 20 - 40 sec. I maybe forgot to stress that this was on a nitrogen oven. These type of ovens often has gates opening and closing. So the PCBs are not let in as quickly as on a completely open oven.

You say that you enjoy the forum. So check out the forum on SMT in FOCUS as well !!

I hope you will contribute on my forum as well, Thanks !!

Regards Brian

reply »

pcb components vacuum pick up

Voidless Reflow Soldering