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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

Typical placement speeds - please advise

Views: 5793


Typical placement speeds - please advise | 28 June, 2014

Based on your knowledge and experience, could you tell me about speed differences when placing a mix of SMD at mass production? We could consider packages like the small R & C chips, then SOD323, SOT-23, SOT-23-5, SOT23-6, SON6, QFN8, SOIC-8, SOIC-16, TSOP-28, TSSOP-28 QFN-40, HTQFP-64,TQFP-100, 456FBGA also 6- to 20-way connectors with 1.27mm pitch, tantalum elcos with 7.3x4.3x2.8mm, etc. In short, how do you would group them for establishing placement speed classes? My final goal is to estimate total placement time departing from the parts list.

reply »


Typical placement speeds - please advise | 29 June, 2014


The issue is very complex and difficult to arbitrarily divide the types of components. From research we perform today together with our customers to design a work-at-SMT accurate as much as possible, I can summarize that vast time differences depending on the type of feeder (like a tray, stick, etc.), and it depends on the size of the component. For example if you are talking about diodes, resistors (and the like) so time is very short (eg types 0201, 0603, etc.) that usually comes in the most simple and some of the heads can pull some components of this type. Nation so if it comes in QFP , BGA (and more) so time is much longer that the complex has a lift and can take only one time each.

From the experience of the design work we have seen the average number of times is a huge deflection. But if you look at all the works then it is offset in part so the deviation is negligible for most.

Can you say why you need the time of placement of the components?

Best Regards, Alexei

reply »


Typical placement speeds - please advise | 30 June, 2014

Thanks Alexei for your advice. I consider mainly high-volume production with tape/reel feeders for most components and tray for big ICs. My ultimate goal is a cost estimation of the production activities. So departing from a parts-list, I need to estimate the placement time. This will be a cycle-time in mass-production. Multiplying this time by the machine and labor hourly rates (USD/hour), we get the cost of assembly. Of course, the model is more complex than this but I am explaining just the core.

reply »


Typical placement speeds - please advise | 3 July, 2014

If you are developing a cost model for quotation/estimation purposes, we would recommend time studying a handful of existing jobs/set up's that you have, and averaging your actual CPM rate. Then determining your cost to operate from there. This will give you the baseline to which you can add your indirect labor costs, any mark-up's or margins, etc, to determine an SMT placement rate cost to use in estimation.

If you are time-studying a particular job to insure that cost estimation was accurate, we would suggest doing the same thing, only on a specific basis. We would recommend averaging the run time over multiple runs to get a more accurate picture of the placement times.

Cheers, ..rob

reply »


Typical placement speeds - please advise | 4 July, 2014

Thanks Rob for your meaningful advice. Instead of just a general average time per component, I think if we have 3 averages, 1 per "component group", we will gain significant accuracy in the cost model. For instance, a motherboard may have 2500+ R and C components placed at, say, 0,045 sec/component, and 12 big ICs at 0,9 sec/component. As to small and intermediate size Ics, this is where I have most doubts. Will e.g. the optical alignment or inertia considerations slow down also their placement to an intermediate value? Could "fine pitch" together with package size play a role in the criterion (component lead/terminal spacing less than 1.27 mm)? Actually I have little direct contact with the factory floors nowadays, that is a reason for me posting my question here, in case some practicioner could share his/her data or expertise. Alternatively, I am considering contacting machine suppliers.

reply »


Typical placement speeds - please advise | 12 July, 2014

If you're getting quoted .045 seconds per chip, or .9 seconds per IC, one of two things is happening. The most likely is the supplier has two machines in the line, one placing small parts quickly, and one placing large parts slowly. The part will run on one or the other, with the associated speed. Minimal middle ground.

You have a machine like a Siplace that can pick up one big part or a bunch of little parts per beam movement. Still, it'll be doing one or the other.

On one of my lines, the parts will be placed by either an Advantis AC30S (realistic 23,000 parts per hour, .15 seconds per chip) for anything shorter than 6mm tall and less than 10mm square, or else it will be placed by a GSM (about 1 second per part) for anything bigger or taller. If its small enough to fit on the fast machine, it places fast. If it is not, it places slow. Doesn't matter if that part is a resistor, a capacitor, a small transistor, a 64 pin QFN IC or whatever.

Alignment time is usually pretty trivial these days, cameras are fast. Inertia is only a factor with truly huge odd-form, and those would be on a slow machine anyway.

reply »


Typical placement speeds - please advise | 14 July, 2014

I've primarily used MyData equipment, personally, and, optically centered devices definitely take longer than mechanically centered parts on a MyData, owing to the construction of the machine. With a fixed upward looking camera, the head has to pick the part, move to the camera, and move to placement...where as, with the mechanically centered devices, the head picks the part, and centers it while moving to the placement location.

However, I viewed a number of other machines at IPC/APEX this year, and I've found that optical centering is getting better and better, and faster and faster. A number of machines have lasers mounted on the head, and do optical centering in line with pick/place; rarely moving to the upward looking camera. Other machines have the upward looking camera located between the pick area and placement area, so that the head loses minimal time passing over the camera on it's way to placement. Image processing speeds have increased dramatically over the years as well, so that the centering process can be conducted nearly in line with the pick/place operation.

Higher mass components can often have slower pick/place times, as adjusted in line with manufacturing, if they're causing issues on the machine, as well; adding another dimension into establishing a time model.

With all that said, narrowing an exact time/placement for each different component group could be one of those things that becomes more difficult than it's worth. A 9-pin BGA (or LGA) should take slightly less time in image processing than, say, a 1000 pin BGA...or, a 1000p BGA with non-symetrical ball patterns. Do you have one price for small, medium, and large BGA's? Another price for heavy IC's, another price for chip components, etc, etc? Do you account for tool changeover (if applicable on your equipment) between the size groups? Do you account for slower placements/picks as components require it?

In my mind, the solution to the above is to determine a base cost/component for running your machines, based upon the average efficiency of the machines, and your operating costs for the machines (direct and indirect labor, estimates of MRO purchases for maintenance, cost of the machines, etc). If you determine that your average placement rate is, say, 15,000 cph; and you calculate the cost to run the machine at that rate, you'll have a pretty good benchmark. Then, you just need to review your cph rates at whatever interval is desired (per job, per week, per month, per quarter, per half, per year), and keep an eye on your fixed costs.

About the only time I'd think you'd need to look at it in more detail is if you're trying to shave every penny off of a job for a huge win, or for a continuous cost savings policy with a customer.

cheers, ..rob

reply »

ii-feed SMD Intelligent Feeder

Used SMT machines