What are the steps involved in bringing a product to production phase

[Sai teja]

Hello,
         I am working in a startup and our electronics product is close to final stage. We are in a phase to start mass production soon. I would like to know standard process involved in developing multiple products(as per electronics is concerned). We are new to this area. Please someone provide information regarding this.

[Richard Crowley]

We don't know what you are asking about here? 
Are you asking about getting PC boards made?
Are you asking about having PC boards assembled (components, soldering, etc)?
Are you asking about having wiring harnesses created?
Are you asking about final assembly of the complete product?
Is this a specialty product for a particular kind of customer?
Is this a general consumer product for millions of units?
The answer to your question depends on many details you have not revealed here.

[Sai teja]

Specifically, I would like to know about assembly and testing process.
1) What is the procedure for testing individual electronic components and ICs before assembly?
2) What is the the procedure for testing the functionality of the assembled PC board ?

Thank you for time.

[AndyC_772]

1) Don't. Buy them only from authorised distributors and insist they're delivered in sealed anti-static packaging. The chances of components getting damaged in between the factory and your assembly plant are minimal.

2) This is entirely up to you. You know your design, the quantity it'll be built in, the target market, and the economics.

It could be:

- a functional test which you carry out on each unit
- a JTAG boundary scan test
- built-in self test
- electrical testing using a bed-of-nails
- some combination of the above

It's completely your choice as to what you believe is necessary or desirable. Nobody else can tell you what is best for your particular product.

As a general point, leaving the question of "how will we test this?" until right before the product goes into volume production, is a bad idea.

[anishkgt]

i would suggest you mention your product so that others here would have an idea to suggest the fab house anything that maybe required and check with manufactures. Here is one such who does testing as well and famous too https://www.eevblog.com/forum/manufacture/what-actually-happens-at-the-fab-house-and-some-facts/msg1282113/#msg1282113.

The solution to the question is in the part you haven't revealed.

[Howardlong]

Specifically, I would like to know about assembly and testing process.
1) What is the procedure for testing individual electronic components and ICs before assembly?

You assume that the parts supplied work. If you are using a reputable CM, they will use parts supplied from reputable sources. If you risk going to China and don't have someone you inherently trust working with them on the ground, all bets are off, they will substitute and use the cheapest parts they can find with questionable provenance, irrespective of what they promise.

2) What is the the procedure for testing the functionality of the assembled PC board ?

You must test every signal path end to end, over a reasonable set of use cases. To dramatically speed this up, if it's at all possible I recommend an integrated self-test to test as many signal paths as possible. Note that an integrated self-test in itself is not a substitute for physical testing on a test jig. For example, you can't test a connector with an integrated self-test!

One further thing, I would go out of your way during design to avoid any need for manual adjustment like trimmers, this is hugely time inefficient, and trimmers, having moving parts, are unreliable. Spend time working out how to make your product automatically self-calibrate at the design stage instead.

[zeqing]

Specifically, I would like to know about assembly and testing process.
1) What is the procedure for testing individual electronic components and ICs before assembly?

you can not test every single components, huge work;

2) What is the the procedure for testing the functionality of the assembled PC board ?

for bare PCB, normally the PCB factory will test them all ; for assembled PCBA, test them as your method. normally you can ask the vendor to make a testing jigs to accelerate the testing, such as https://www.makerfabs.com/index.php?route=product/product&path=93_141&product_id=516

[coppice]

Specifically, I would like to know about assembly and testing process.
1) What is the procedure for testing individual electronic components and ICs before assembly?

It is usually a bad idea to test components before assembly. It involves additional handling steps, and each handling step adds risk of mechanical distortion or damage. This is especially true for things like QFPs, which have rather soft metal legs, which are very easily bent. Why would you want to test incoming components, apart from sample testing for quality control? If you can't trust your supplier to have a very low defect rate you really need to change suppliers.

A lot of companies who subcontract their PCB assembly want to program parts, like flash MCUs, before the devices go to the assembly shop. I have never found one who managed to do this without a measurable increase in the boards which fail test due to a bent leg, or other physical damage. The best way to program these devices is usually to do it at the board test stage, after PCB assembly. However, if the subcontractor is running the board tests, and they aren't to be trusted with the code, lots of companies prefer to accept the higher defect rate. The devices are programmed and the security fuse blown. Then they go off to the assembler.

[Corporate666]

Everyone so far is saying that you don't or shouldn't test components before assembly.  But that's not true - it depends on what you are making and your tolerances and necessary paper trail and such.

My pick and place came with an on-the-fly component testing system consisting of, essentially, V-shaped probes and an extremely fast RCL meter tied into the controller.  It could pick components and press them on these probes (using vision to accurately locate them) and verify the component value/specifications prior to placement. 

It certainly slows down the placement quite a bit but depending on your application, it can be useful or even necessary.  If you are making medical devices, military/space and such, then these processes might be required.  But for consumer electronics, as said above, you just buy from a reputable manufacturer/distributor, ensure the parts are sealed and you can trust that the parts will meet their specifications as long as you use them within their specifications.  Make sure you trust your assembler too!  It's easy to the wrong kind of paste, the wrong soldering profile, bad handling techniques or do numerous other things which can cause high failure rates either during the assembly phase or early failures in the field.

[frozenfrogz]

Also: Building stuff and making sure it works as intended is just part of the whole mission. Sadly a lot of start-ups and small companies totally forget about: Certification for safety and compliance (not always mandatory, depends heavily on the product and intended audience), paper work to be included with the sold item - without that you will either: Not find re-sellers willing to stock your product or: Face legal issues you did not consider beforehand.
There are more issues you can run into, but in my opinion these are the most severe you can easily cover by getting your research straight.

[HalFET]

Don't forget to do a simple monte-carlo style analysis on the circuit if you have any critical analog part, wouldn't be the first person who always uses nicely binned parts from Farnell and gets slaughtered by that 5 or 10%!

[zeqing]

Everyone so far is saying that you don't or shouldn't test components before assembly.  But that's not true - it depends on what you are making and your tolerances and necessary paper trail and such.

My pick and place came with an on-the-fly component testing system consisting of, essentially, V-shaped probes and an extremely fast RCL meter tied into the controller.  It could pick components and press them on these probes (using vision to accurately locate them) and verify the component value/specifications prior to placement. 

do not understand much, so the RCL meter only measures the Resistor and capacitor? but how about the ICs/connectors? besides, the resistor and capacitor should be the least risk in all the components, they are counted as Millions even for a middle-level production, or thousands in a small batch/prototyping projects. 

[coppice]

Everyone so far is saying that you don't or shouldn't test components before assembly.  But that's not true - it depends on what you are making and your tolerances and necessary paper trail and such.

My pick and place came with an on-the-fly component testing system consisting of, essentially, V-shaped probes and an extremely fast RCL meter tied into the controller.  It could pick components and press them on these probes (using vision to accurately locate them) and verify the component value/specifications prior to placement. 

do not understand much, so the RCL meter only measures the Resistor and capacitor? but how about the ICs/connectors? besides, the resistor and capacitor should be the least risk in all the components, they are counted as Millions even for a middle-level production, or thousands in a small batch/prototyping projects.

I didn't understand that comment about the PnP machine doing testing, either. There is no way a PnP machine is going to be able to meaningfully test anything but the most trivial components on the fly, or to be able to test them in better than the most trivial way (e.g. it might test continuity in a transformer, but a HiPot test is more significant for such a component). These simple tests of simple parts are not going to significantly affect the quality of the assembled boards.

[AndyC_772]

I've had production batches of PCBs delivered where someone has put the reels on the P&P machine in the wrong order. A simple test, at least on the first few parts from each reel, would have saved a great deal of rework.

[coppice]

I've had production batches of PCBs delivered where someone has put the reels on the P&P machine in the wrong order. A simple test, at least on the first few parts from each reel, would have saved a great deal of rework.

That would help with the simple passives, but it won't help with anything more complex. Even with passives it would be inefficient to test every part. You only need to check, as a new reel is loaded, that it is the correct reel. It rings a slight bell than the PnP machines we used in late 80s did some kind of simple reel load check like that. We still had problems with more complex components, though. I remember a batch of mispackaged chips that ended up in the feeders backwards. Because testing was done in batches in that place, many thousands of boards had been assembled before testing started and the 100% failure rate was noticed.

[Corporate666]

Everyone so far is saying that you don't or shouldn't test components before assembly.  But that's not true - it depends on what you are making and your tolerances and necessary paper trail and such.

My pick and place came with an on-the-fly component testing system consisting of, essentially, V-shaped probes and an extremely fast RCL meter tied into the controller.  It could pick components and press them on these probes (using vision to accurately locate them) and verify the component value/specifications prior to placement. 

do not understand much, so the RCL meter only measures the Resistor and capacitor? but how about the ICs/connectors? besides, the resistor and capacitor should be the least risk in all the components, they are counted as Millions even for a middle-level production, or thousands in a small batch/prototyping projects.

It doesn't test every part in every way - it offers the capability to do a some types of testing.  There are various shapes/sizes of probes (some custom made) to allow testing things like connectors. 

[Corporate666]

I didn't understand that comment about the PnP machine doing testing, either. There is no way a PnP machine is going to be able to meaningfully test anything but the most trivial components on the fly, or to be able to test them in better than the most trivial way (e.g. it might test continuity in a transformer, but a HiPot test is more significant for such a component). These simple tests of simple parts are not going to significantly affect the quality of the assembled boards.

That's not true - it depends what it's set up for.  It can verify that the parts being placed match what is on the BOM (vision obviously ensures size and shape and orientation match, electrical testing verifies value matches).  It can do spot checking or 100% part checking if desired. 

Yes, my experience is that I get very few failures from bad components in general when they are purchased in sealed packages from a reputable manufacturer - and I echoed that above to the OP - I am simply saying that a blanket statement of "you don't test parts before" isn't accurate, because it is something that is done in some circumstances.  We don't know what the OP is doing - I imagine it's military/space/medical or possibly complicated boards that take a long time to place that do spot checking or first article checking.  The PnP that came equipped with this system was bought from a medical device manufacturer.  I pulled it off the machine and sold the testing system for almost as much as I had paid for the PnP machine. 

[zeqing]

Everyone so far is saying that you don't or shouldn't test components before assembly.  But that's not true - it depends on what you are making and your tolerances and necessary paper trail and such.

My pick and place came with an on-the-fly component testing system consisting of, essentially, V-shaped probes and an extremely fast RCL meter tied into the controller.  It could pick components and press them on these probes (using vision to accurately locate them) and verify the component value/specifications prior to placement. 

do not understand much, so the RCL meter only measures the Resistor and capacitor? but how about the ICs/connectors? besides, the resistor and capacitor should be the least risk in all the components, they are counted as Millions even for a middle-level production, or thousands in a small batch/prototyping projects.

It doesn't test every part in every way - it offers the capability to do a some types of testing.  There are various shapes/sizes of probes (some custom made) to allow testing things like connectors.

i understand you but this is not what i found in my visiting of factory.  i have visit some very big assembly factories in shenzhen and Germany,  the standard procedure for the PNP is making the first PCBA sample firstly and a special engineer checking with the BOM , for components / directions/resistance/capacitance.etc.  and if all OK, the mass production go on. but not before the soldering.

[mrpackethead]

I dont' have it on my PNP machines, ( the ability to meausre passives ), but many reels ( particaully samsung ceraminc caps ) will have a 1 or 2 parts loaded on teh tape at the start of the reel, ( then 50cm of empty tape to thread it through ).   This is so you can test them to double check it if you want.   

Loading reels onto Machines is an abolustley critical job to get it right!  you can place a ot of parts in a big hurry and get it very wrong very fast!

[mrpackethead]

Hello,
         I am working in a startup and our electronics product is close to final stage. We are in a phase to start mass production soon. I would like to know standard process involved in developing multiple products(as per electronics is concerned). We are new to this area. Please someone provide information regarding this.

you've done this arse about face.    To make production work well, you need to know how that all works when your designing it.   I see this all the time with people bringing me things to make.    ( even though we are not a CM, and just do inhouse they come. )..    You'd be amazed at how much bbetter your product will be if you'ved designed it with produciton in mind as well as fucntion and form.

[coppice]

Hello,
         I am working in a startup and our electronics product is close to final stage. We are in a phase to start mass production soon. I would like to know standard process involved in developing multiple products(as per electronics is concerned). We are new to this area. Please someone provide information regarding this.

you've done this arse about face.    To make production work well, you need to know how that all works when your designing it.   I see this all the time with people bringing me things to make.    ( even though we are not a CM, and just do inhouse they come. )..    You'd be amazed at how much bbetter your product will be if you'ved designed it with produciton in mind as well as fucntion and form.

Its pretty much standard in large organisations for an industrial engineer to be present for all design reviews. A good one will usually pick up a few physical issues, that if changed will make production smoother and more predictable.

[Gyro]

1) Don't. Buy them only from authorised distributors and insist they're delivered in sealed anti-static packaging. The chances of components getting damaged in between the factory and your assembly plant are minimal.
...

One exception here... specify Bare Board Test on the PCBs, they're a component too.

[AndyC_772]

True story that seems relevant

https://www.eevblog.com/forum/projects/i-am-furious-ordered-pcbs-are-defected/msg1074215/#msg1074215

[Kean]

Actually, even a bare board test can give incorrect results...

I had a batch of PCBs with 33% failure rate.  The programming header for the mcu on one PCB in each panel was mostly shorted to an internal ground plane.  Because this was caused by a "bug" in the software used during panelisation by the board house, it completely passed e-test as the gerbers used for creating the e-test program had this exact short.  Thankfully it was only a small sample batch.

[Kean]

Here is a snip of the Gerbers where I've highlighted the error.  No idea how the board house managed this, but at least they were open about it and shared their panel gerbers with me so I could see the rest of the panel wasn't affected.
I was new to Altium at the time and had recently taken on a client using a bunch of old Altium designs.  The board house said they could do the panel layout (3 designs) for me at no charge, which was good because I was on a tight deadline to get the new revisions I'd just done verified before we went into mass production.