Chip On Board

[NivagSwerdna]

Most people will be familiar with the epoxy blobs you see on PCBs.... but as I brushed my teeth this morning.... I wondered...

how do the economics of COB work?  I can see you save on the packaging but doesn't the cost of attaching the device to the board and putting resin on top offset that?

How are these things actually attached?  Doesn't it disrupt the manufacturing process since I presume this isn't a job for a pick and place?

Also... is it possible to remove the epoxy to such an extent you can see the bonding or does that imply dangerous chemicals?  (I assume heat won't work).

Just interested.... thx.

I found...   So the bonding is automated... I presume there is a previous step to glue the die to the board (and possibly a cleaning step)... and a subsequent step to put on the epoxy.

And I've learnt a new phrase... "Purple Plague".  There's so much I don't know!

[NivagSwerdna]

Isn't google wonderful.

https://learn.sparkfun.com/tutorials/how-chip-on-boards-are-made

I guess the economics work out as long as you either automate the workflow on a line or pay the humans very little.

[Berni]

I do wonder how much they actually save with all this wire bonding.

I understand packaging a chip costs money but the wire bondering process is also a lot of extra trouble. For some of the common old jellybean ICs you can buy then for 10 cents each on digikey in reasonable a quantity. So the IC manufacturer has to be making them for like 1/4 that cost. This means it costs less to make the chip than it is to pay someone to work for 10 seconds in the developed world.

[Jeroen3]

They put the black glob on by hand 

[NivagSwerdna]

I do wonder how much they actually save with all this wire bonding.

I guess the bonding cost has to occur at some point anyway.

[NivagSwerdna]

is it possible to remove the epoxy to such an extent you can see the bonding or does that imply dangerous chemicals?  (I assume heat won't work).

No seems to be the answer. I won't be hot using concentrated Nitric acid anytime soon.  http://www.epoxyworktops.com/epoxy-resin/chem-resistance.html

[Ian.M]

I do wonder how much they actually save with all this wire bonding.

I guess the bonding cost has to occur at some point anyway.

No leadframe, package moulds, or package marking and you also save all the handling and bulk packaging steps from when the die would have been wire-bonded to the leadframe up to placing a conventional chip on the board.

is it possible to remove the epoxy to such an extent you can see the bonding or does that imply dangerous chemicals?  (I assume heat won't work).

No seems to be the answer. I won't be hot using concentrated Nitric acid anytime soon.  http://www.epoxyworktops.com/epoxy-resin/chem-resistance.html

Decapping COB without hot fuming Nitric acid + careful masking is going to be difficult, because anything that attacks the epoxy will also attack the board.  Hot Abietic acid does attack epoxy and can be used for decapping, (simply boil a chip in rosin for about 20 minutes), so I suppose you could boil a blob of rosin on top of the COB encapsulant using a soldering iron, changing the rosin frequently, but unlike the lower temperature Nitric acid process, the >300 deg C temperatures required are likely to degrade the chip and probably the board under it.

[NivagSwerdna]

No leadframe, package moulds, or package marking and you also save all the handling and bulk packaging steps from when the die would have been wire-bonded to the leadframe up to placing a conventional chip on the board.

Ah. OK. That makes sense.  (I once visited a silicon fab in Plymouth)

Decapping COB without hot fuming Nitric acid + careful masking is going to be difficult, because anything that attacks the epoxy will also attack the board.  Hot Abietic acid does attack epoxy and can be used for decapping, (simply boil a chip in rosin for about 20 minutes), so I suppose you could boil a blob of rosin on top of the COB encapsulant using a soldering iron, changing the rosin frequently, but unlike the lower temperature Nitric acid process, the >300 deg C temperatures required are likely to degrade the chip and probably the board under it.

I'll leave that to others. 

[Zero999]

Cost isn't the only factor. Chip on board is often associated with cheap electronics but there are other advantages too: it saves space and the parasitic inductance and capacitance is reduced too.

[ajb]

COBs may also allow the rest of the board to be assembled using a cheaper process than a conventional package really permits.  In cheap electronics you'll occasionally see COBs mounted on little two-layer carrier PCBs that slot into the main PCB.  Now there's no need for careful alignment of a many-legged IC package, the main board can be much coarser pitch and the whole thing can be wave soldered instead of stenciled and reflowed.

[tszaboo]

I do wonder how much they actually save with all this wire bonding.

If you dont know, it means that it is not relevant to you. I imagine, below 10K pieces it is not really worth it. Also, it will not be worth it, if your system consist of more than 2 ICs on the board. I mean, COB for sure will have lower yield than SMD, and if you have, say 4 chips on the board, the economics are probably not working anymore.
So if you are not designing single use singing greeting cards, dont even bother.

[macboy]

I do wonder how much they actually save with all this wire bonding.

If you dont know, it means that it is not relevant to you....

Thanks for clearing that up. I guess I can stop learning now since everything I don't already know is irrelevant. What an attitude to have toward someone who is curious and wants to learn. 

[Alex Eisenhut]

Well COB's been around for decades, I remember as a kid tearing a VIC-20 game cartridge apart and being befuddled by the black blob. I had only ever seen them in Japanese calculators.

[tszaboo]

I do wonder how much they actually save with all this wire bonding.

If you dont know, it means that it is not relevant to you....

Thanks for clearing that up. I guess I can stop learning now since everything I don't already know is irrelevant. What an attitude to have toward someone who is curious and wants to learn. 

Its the same thing Dave is always saying: "If you have to ask, you cannot afford it."

[Zero999]

Well COB's been around for decades, I remember as a kid tearing a VIC-20 game cartridge apart and being befuddled by the black blob. I had only ever seen them in Japanese calculators.

I wonder when chip on board was first invented?

My guess is it was pretty early on in the history of integrated circuits, when manufacturers realised the package was much larger than the die and they could make the product much smaller by putting the die directly on the board. The first device with a chip on board was probably a quartz watch, in the 1970s.

[ejeffrey]

Another way I have seen COB used is for small volume custom ASICs.  In this case, I assume that a small custom packaging run is more expensive than a small run of COB.

A fair number of RF / microwave devices are available as bare dies.  I have never used them, but I would guess that you can do clever stuff like place two dies next to each other and wire bond die to die without ever hitting the PCB, which would minimize inductance and losses.  Basically miniature dead bug assembly.

[amyk]

Well COB's been around for decades, I remember as a kid tearing a VIC-20 game cartridge apart and being befuddled by the black blob. I had only ever seen them in Japanese calculators.

I wonder when chip on board was first invented?

My guess is it was pretty early on in the history of integrated circuits, when manufacturers realised the package was much larger than the die and they could make the product much smaller by putting the die directly on the board. The first device with a chip on board was probably a quartz watch, in the 1970s.

Hybrid circuits were already available in the 60s. I guess COB could be considered a variant of that.

[Retep]

No leadframe, package moulds, or package marking and you also save all the handling and bulk packaging steps from when the die would have been wire-bonded to the leadframe up to placing a conventional chip on the board.

And for dodgy Chinese manufacturers concerned about protecting "their" IP: no need to scratch the number of the package

COB's may be packaged in a tray, in which case you still have the packaging steps.

The wire bonding machines I have seen for standard packages did not require any operator intervention and are much faster. Also mounting standard packages on the PCB can be fully automated with run-of-the-mill pick-and-place machines. 

I guess whether a COB makes sense depends on the local labor cost, production volume, electrical- and physical requirements.

[MK14]

The following link, has a nice explanation and brief (approx 30 sec) video, showing how they are made.

https://learn.sparkfun.com/tutorials/how-chip-on-boards-are-made

Basically, it is just a normal electronics production factory WITHOUT using any clean room or anything. The "chips", are like tiny surface mount capacitors/resistors, and are picked and placed, by hand, just like a normal surface mount component would be. After five minutes for the glue to dry, they are put into an automatic wire bonding machine. Which is about the size of a giant inspection Microscope.
It takes only about ten seconds for the automatic machine to wire it up. One blob of black epoxy and it's done.

It looks really quick, cheap and simple!

[NivagSwerdna]

The following link, has a nice explanation and brief (approx 30 sec) video, showing how they are made....

See earlier in the thread.    https://www.eevblog.com/forum/chat/chip-on-board/msg1259420/#msg1259420

[MK14]

The following link, has a nice explanation and brief (approx 30 sec) video, showing how they are made....

See earlier in the thread.    https://www.eevblog.com/forum/chat/chip-on-board/msg1259420/#msg1259420

I've read/followed the thread a number of times. I totally missed that link, sorry.