EEVblog #939 - How Is A PCB Manufactured

[EEVblog]

How is a double sided plated through hole solder masked and silk screened component olverlay PCB manufactured?
Take a ste-by-step tour of a PCB manufactuing plant.




http://www.pcbzone.net

Desing for manufacture tutorial:
DFM Automated PCB Panel Testing:

Alba PCB Group Video:
Eurocircuits Video:

[ludzinc]

If you're ever in Canberra you should visit Lintek.

http://www.lintek.com.au/products-and-capabilities/#process

They clad their substrates using vacuum deposition so can deal with exotic materials.

Nice guys too!

[trophosphere]

Another awesome video Dave. I have actually never looked deeply into how PCBs get made (just knew the general process) even though I got my undergraduate degree in EE back in 2009. Now I am on a PCB manufacturing video binge on Youtube.

[julian1]

Interesting video. Especially the negative photo-resist process, where they tin the copper in the areas of no resist. And then remove the resist, and etch the copper.

Does tin resist ferric chloride, or any of the other common etchants?

[Brumby]

Interesting video. Especially the negative photo-resist process, where they tin the copper in the areas of no resist. And then remove the resist, and etch the copper.

Does tin resist ferric chloride, or any of the other common etchants?

I thought about that for a moment - and the reasoning is quite simple.

The negative resist process allows for the retention of the plated through holes.  Otherwise, a positive resist process would need to find its way into the holes or they would get etched away.

[julian1]

Tin may also be superior in resisting the harsh etching conditions, compared with the thin layer of UV cured polymer. Meaning less undercut, and/or stronger etchant/faster process.

[EEVblog]

If you're ever in Canberra you should visit Lintek.
http://www.lintek.com.au/products-and-capabilities/#process
They clad their substrates using vacuum deposition so can deal with exotic materials.

Yep, used them back in the military days for exotic boards. That's all they do now, there is no general mass production PCB manufacturer left in this country.

[Moshly]

Here is a video on how tek did it in 1969 ->



I like the way they use gold as the enchant resist.

[sleemanj]

Tin may also be superior in resisting the harsh etching conditions, compared with the thin layer of UV cured polymer. Meaning less undercut, and/or stronger etchant/faster process.

I expect economics plays a role.

By using the film resist to mask the bits you don't want, you can then plate on the copper to thicken out the bits you want to keep, plate tin on top of those bits to mask it, strip the film which is non-recoverable, etch away the copper you didn't want, which is recoverable, and then strip the tin, which is also no doubt recoverable, so the non-recoverable materials in the etch process is limited to one application of film.

Doing it the other way (using the film to mask the bits you don't want) would mean that you'd either have to apply and strip film twice (once to copper plate to thickness, and then again to etch away the bits you don't want), or start with PCB which is totally copper clad to finished thickness increasing the time to etch significantly, and I expect while the copper etched is recoverable there would be losses so if you are etching more copper, there'd be more losses.

Applying film twice would also mean you increase the chance of registration errors between applications of film, where as because you are tinning through a single film application you don't have that problem.  And you'd be doing the second film application over a non-flat surface which would be fault prone I expect.

[timb]

Here is a video on how tek did it in 1969 ->



I like the way they use gold as the enchant resist.

That was great! Good ole' Tek, doing it all in house.

I swear, that narrator sounds *a lot* like Dr. McCoy from Star Trek...

[free_electron]

the double plating step is simple : save copper AND create a more uniform copper

reason 1 :  anything you plate on you need to etch off.

They plate a thin layer of copper first on the entire board.
then apply resist in negative.
then grow copper again . the copper only grows on the exposed area's ( which are what you want to keep anyway ... )

tinplate, strip resist . etch.

reason 2 : uniformity.  the layer thickness in the holes is much more controllable if the entire board is pure copper. imagine this: a thin small sliver of a trace with a tiny 0.1mm via. 1 cm away from it is a big fat trace with a 2cm hole. if you try to plate that the big hole will grow copper faster than the tiny one. simply because the current in the hole is given by the contact surface area ...
the current is pulled out of the copper plate by the plating electrode , but the current goes in through the plating liquid. so that tiny trace gets less current than the thick one.. as the current determines the plating speed ...  by not having photo resist all holes benefit of the entire contact surface to the liquid. so they all grow uniformly.

once there is copper in the hole it doesn't matter anymore. this effect only happens to get the copper initially in the hole.

[zapta]

Dave, why do you need non plated holes? Can't you just isolate them on one or more layer?

[Len]

This is not a silkscreening process

... while a silkscreening process is being shown. 

[G7PSK]

When plating through hole what makes the copper stick to the bare pcb edge of the drilled hole, is it pre treated in some way. If you want plastic plated you have to have a vapour or chemical surface applied first.

[timb]

When plating through hole what makes the copper stick to the bare pcb edge of the drilled hole, is it pre treated in some way. If you want plastic plated you have to have a vapour or chemical surface applied first.

Old Style Method: Basically, they immerse it in a bath that contains a copper solution. It creates a very light plating inside the holes, which is then thickened during the electroplating step.

They show this in the Tektronix video above, if you want to see it in action.

Not sure if it's any different with a modern process.

[ion]

When plating through hole what makes the copper stick to the bare pcb edge of the drilled hole, is it pre treated in some way. If you want plastic plated you have to have a vapour or chemical surface applied first.

That's what I was thinking - I guess there's no footage of that step.
A few years ago I came across a chemical kit form MG Chemicals specifically to make bare PCB material conductive so it can be plated.  Seems to be discontinued now and I'm not surprised - it was an expensive, multiple step process and you'd be left with several containers of toxic waste once the chemical baths reached the end of their (very short) shelf life.  I'd imagine most poeple wouldn't go through all that for a prototype or DIY project.

[rrinker]

 Now that I've had a chance to watch Dave's video as well as the Tek one - it's kind of amazing how much HASN'T changed. Still a significant amount of manual operation.
 And that NC drill in the Tek video - now that is an old bit. GE C100 maybe, even older than the GE 550 we had at my first job - like the 550, the one in the Tek video does no storage of the program, it simply runs step by step reading the paper tape. After one run, the tape is rewound for the next set of boards. You can control stuff like feed rates, but you can't edit the program on the machine. The 550 dates from the mid 60's and lasted until GE released their first true CNC control in the 1050 in 1974 (we had one of those, too). The one in Tek's plant is much older, early 60's or maybe even late 50's vintage. But in our shop, that 550 was just as reliable as newer fancier controls and put in a full day's work every day, so no reason to replace it.
 Kind of funny watching the worker silkscreen on the photosensitive solder mask and labelling materials while Dave is saying "in the past this was a silkscreen process but now it's photosensitive" Yes, not really silkscreening when you have no pattern in the screen, but still funny.

[WattSekunde]

I remember inspecting a pcb production because of a quality problem with the via hole positions in some boards. And yes we used teardrop vias. We looked through the whole process and found the problem at the drilling stage. It was caused by drilling too much boards at once in a stack. The drill bend sometimes sidewards on the lowest boards in the stack. The manufacturer had to reduce to 4 boards in a stack to fulfill our quality needs.

[vzoole]

When plating through hole what makes the copper stick to the bare pcb edge of the drilled hole, is it pre treated in some way. If you want plastic plated you have to have a vapour or chemical surface applied first.

The answer is here:
Black Hole
This machine fill the whole board and holes with graphite then remove it from the surface, so the holes still conductive too.

https://youtu.be/T7S40GYESbY?t=3m57s

[ludzinc]

If you're ever in Canberra you should visit Lintek.
http://www.lintek.com.au/products-and-capabilities/#process
They clad their substrates using vacuum deposition so can deal with exotic materials.

Yep, used them back in the military days for exotic boards. That's all they do now, there is no general mass production PCB manufacturer left in this country.

They still do boring 2 and 4 layer FR4 stuff.  But they now have a cool laser drill too!

[ali_asadzadeh]

Maybe they could use TI DLP for direct exposing to the PCB, or maybe they could use a 4K LCD panel for direct exposing. they are just plain awesome Ideas.

[timb]

Now that I've had a chance to watch Dave's video as well as the Tek one - it's kind of amazing how much HASN'T changed. Still a significant amount of manual operation.
 And that NC drill in the Tek video - now that is an old bit. GE C100 maybe, even older than the GE 550 we had at my first job - like the 550, the one in the Tek video does no storage of the program, it simply runs step by step reading the paper tape. After one run, the tape is rewound for the next set of boards. You can control stuff like feed rates, but you can't edit the program on the machine. The 550 dates from the mid 60's and lasted until GE released their first true CNC control in the 1050 in 1974 (we had one of those, too). The one in Tek's plant is much older, early 60's or maybe even late 50's vintage. But in our shop, that 550 was just as reliable as newer fancier controls and put in a full day's work every day, so no reason to replace it.
 Kind of funny watching the worker silkscreen on the photosensitive solder mask and labelling materials while Dave is saying "in the past this was a silkscreen process but now it's photosensitive" Yes, not really silkscreening when you have no pattern in the screen, but still funny.

Yeah, I was wondering about the drill in the Tek video. So, it's purely a mechanical solution then? No electronics? (Aside from the motors, naturally.)

I assume it works a bit like those old player pianos? I always wondered how they automated this sort of thing before CNC came about. The aforementioned pianos have been around since at least the mid to late 1800's, so I figured it would be something like that.

[vzoole]

Maybe they could use TI DLP for direct exposing to the PCB, or maybe they could use a 4K LCD panel for direct exposing. they are just plain awesome Ideas.

This direct imaging technology is already exist (at 5:03):


https://www.youtube.com/watch?v=T7S40GYESbY&feature=youtu.be&t=5m3s

[peterjohnston]

Dave, you need to retool your silkscreen so you've got all the letters on your OSHW logo!

[free_electron]

Maybe they could use TI DLP for direct exposing to the PCB, or maybe they could use a 4K LCD panel for direct exposing. they are just plain awesome Ideas.

This was invented By Maskless Imaging here in San Jose. they use 4 DLP engines to direct expose the board. there is a dithering process so they can create perfect rounded corners that are not possible with classic scanned laser direct imaging. they could make like 30 micron traces .. ( on 1/3 ounce copper) that is 1.1 mils trace and 1.1 mils gap ....

I saw these machines in action about 4 years ago at APCT here in San Jose. they had 3 of them. very impressive.