>>Could be, this is not an area I am all that familiar with. Still, having custom chips manufactured involves considerable up front costs so it would require either high volume low cost or low volume high priced products to justify it.
Indeed up front costs would be high, but when you have reusable components, it's cheaper than making all variant designs on one die.
>>True, but why limit yourself to just ram and flash if you can build your product with off the shelf chips. Having custom chips made is expensive. Why do it unless there is a cost saving or technical/maketing reason to do so.
Sure, it might be possible to get bare dies from different manufacturers, but it's better to design components that have been built from the ground up to be as compatible with each other as possible, requiring minimal passive components off-die as possible. That's not to say you should always be reinventing the wheel, when a sufficient (or better) solution is available.
>>Very good point. They may have enough clout with suppliers to make getting parts like bare chips possible. The only reason I could see for doing this in house would be if you could do it for a lower cost or if quality control was an issue.
Only deal with reputable companies Might cost more, but it would be worth it, IMO.
As far as packaging is concerned, the options are virtually limitless, which can be much more convenient than common chip packages.
Sure, it might be possible to get bare dies from different manufacturers, but it's better to design components that have been built from the ground up to be as compatible with each other as possible, requiring minimal passive components off-die as possible. That's not to say you should always be reinventing the wheel, when a sufficient (or better) solution is available.
I agree it is better to have components that have been designed to work together and to have the smallest number of components possible. Great for manufacturing large numbers of low cost devices.
On the other hand, I would be quite happy to build small modules in small quantities at a reasonable cost using off the shelf components. If I could do that with CoB's and bare dies or flip chips that would be great. If I could print my own ceramic substrates and do the same that would be absolutely fantastic.
Only deal with reputable companies**Might cost more, but it would be worth it, IMO.
Definitely a good thing, but not always easy to do. When the components you need for your design are only available from a company you have not dealt with before, and whose reputation has not been established yet, you may have to take a leap of faith. Thankfully the Internet makes that somewhat easier.
As far as packaging is concerned, the options are virtually limitless, which can be much more convenient than common chip packages.
Very true. Although I try to stick with established standards as much as possible lack of a standard or being unable to meet a specific packaging standard should not be a road block.
>> If I could print my own ceramic substrates and do the same that would be absolutely fantastic.
I wonder how different it is from etching fiberglass PCBs, or are you talking about something different than just sticking the die on a substrate? What does ceramic offer over say a mil-spec PCB?
>>Actually there are a few wire bonding systems out on ebay in the $3-7000.00 range. Conditions and options vary.
Just doing a quick eBay search pulls up plenty of wire/ball/die bonder machines for a lot cheaper than I expected. If you can purchase various bare-die ICs, this might not be such a high price endeavour.
>> If I could print my own ceramic substrates and do the same that would be absolutely fantastic.
I wonder how different it is from etching fiberglass PCBs, or are you talking about something different than just sticking the die on a substrate? What does ceramic offer over say a mil-spec PCB?
Ceramic has much better heat conduction, higher temperature/pressure tolerance, and can be hermetically sealed. Potting compounds (including those used on CoB's) allow moisture to migrate through them which eventually damages the chips. High temperatures makes the problem worse.
I guess the learning curve is quite steep..
Massimo
I definitely don't have the expertise to design a chip or the resources to pay to have it done. If it were possible to print 3d ceramic substrates it would not be much different from laying out a multilayer PCB, and that I can do. If 3d ceramic printers were available at reasonable cost it would not be too long before chip makers would have to make bare dies available. For sure a win/win situation.
I'm not sure what the cost difference is now but back in the mid 70's when I did come in contact with them a small (1"x1") frit sealed ceramic module could cost as much as 100 times what an equivalent potted PCB would cost.
I would think that for your application PCB's with SMT components are your best bet. Potting compounds have improved quite a bit over the years so potted or sealed in some way for harsh environments should be adequate.
There appears to be a lot of companies that offer services for CoB/MCM (Multi-Chip Module) assembly in various packages including ceramic. From the information I've come across, MCM is likely to really take off with the push for smaller and smaller devices, so these assembly services will drop in price proportionally. It looks as if ASIC providers are already starting to go this route. Another interesting aspect of MCM is the ability to offer high-quality analog interfaces by using good ADCs/DACs and OPAMPs.
You have to show in concrete terms, that your MCM solutions are cheaper, more reliable, easier to build than current SMT designs. If the MCM people had such advantages, they'd be the dominate format today. They haven't so far.
Comments
Indeed up front costs would be high, but when you have reusable components, it's cheaper than making all variant designs on one die.
>>True, but why limit yourself to just ram and flash if you can build your product with off the shelf chips. Having custom chips made is expensive. Why do it unless there is a cost saving or technical/maketing reason to do so.
Sure, it might be possible to get bare dies from different manufacturers, but it's better to design components that have been built from the ground up to be as compatible with each other as possible, requiring minimal passive components off-die as possible. That's not to say you should always be reinventing the wheel, when a sufficient (or better) solution is available.
>>Very good point. They may have enough clout with suppliers to make getting parts like bare chips possible. The only reason I could see for doing this in house would be if you could do it for a lower cost or if quality control was an issue.
Only deal with reputable companies
As far as packaging is concerned, the options are virtually limitless, which can be much more convenient than common chip packages.
I agree it is better to have components that have been designed to work together and to have the smallest number of components possible. Great for manufacturing large numbers of low cost devices.
On the other hand, I would be quite happy to build small modules in small quantities at a reasonable cost using off the shelf components. If I could do that with CoB's and bare dies or flip chips that would be great. If I could print my own ceramic substrates and do the same that would be absolutely fantastic.
Definitely a good thing, but not always easy to do. When the components you need for your design are only available from a company you have not dealt with before, and whose reputation has not been established yet, you may have to take a leap of faith. Thankfully the Internet makes that somewhat easier.
Very true. Although I try to stick with established standards as much as possible lack of a standard or being unable to meet a specific packaging standard should not be a road block.
I wonder how different it is from etching fiberglass PCBs, or are you talking about something different than just sticking the die on a substrate? What does ceramic offer over say a mil-spec PCB?
>>Actually there are a few wire bonding systems out on ebay in the $3-7000.00 range. Conditions and options vary.
Just doing a quick eBay search pulls up plenty of wire/ball/die bonder machines for a lot cheaper than I expected. If you can purchase various bare-die ICs, this might not be such a high price endeavour.
http://www.mosis.com/
The group chip designs on a wafer. Not sure about the cost.
for the design, gnu electric is compatible http://www.gnu.org/software/electric/ and free
I guess the learning curve is quite steep..
Massimo
Ceramic has much better heat conduction, higher temperature/pressure tolerance, and can be hermetically sealed. Potting compounds (including those used on CoB's) allow moisture to migrate through them which eventually damages the chips. High temperatures makes the problem worse.
I definitely don't have the expertise to design a chip or the resources to pay to have it done. If it were possible to print 3d ceramic substrates it would not be much different from laying out a multilayer PCB, and that I can do. If 3d ceramic printers were available at reasonable cost it would not be too long before chip makers would have to make bare dies available. For sure a win/win situation.
My experiments general don't need to last long, low cost and lot of 'em is my want. Shorter life might be a fine trade off
I'm not sure what the cost difference is now but back in the mid 70's when I did come in contact with them a small (1"x1") frit sealed ceramic module could cost as much as 100 times what an equivalent potted PCB would cost.
I would think that for your application PCB's with SMT components are your best bet. Potting compounds have improved quite a bit over the years so potted or sealed in some way for harsh environments should be adequate.
http://www.ami.ac.uk/courses/topics/0134_sp/index.html
techdigest.jhuapl.edu/TD/td2001/Blum.pdf
www.bourns.com/pdfs/bourns_mm_packaging.pdf
www.fuse-network.com/fuse/training/mcm/mcm_v5_3.pdf
http://en.wikipedia.org/wiki/Multi-chip_module
There appears to be a lot of companies that offer services for CoB/MCM (Multi-Chip Module) assembly in various packages including ceramic. From the information I've come across, MCM is likely to really take off with the push for smaller and smaller devices, so these assembly services will drop in price proportionally. It looks as if ASIC providers are already starting to go this route. Another interesting aspect of MCM is the ability to offer high-quality analog interfaces by using good ADCs/DACs and OPAMPs.
You have to show in concrete terms, that your MCM solutions are cheaper, more reliable, easier to build than current SMT designs. If the MCM people had such advantages, they'd be the dominate format today. They haven't so far.