ASIC Foundry with JFET Process?

Or better, post it to the NG for the benefit of everybody. JT's asking for a commercial offering, not some hush-hush trade secret.

Cheers

Phil Hobbs

Yep. I'm sure you could make use of that same process ;-) ...Jim Thompson

ing

Probably not, since my electronics work is divided between board-level and device-level stuff. I've never done an ASIC, though it would be fun one of these days.

I just think it's a bit unseemly to ask for help without being willing to s hare.

Cheers

Phil Hobbs

I didn't mean it that way... sometimes people don't want to post publicly. ...Jim Thompson

What about DI? Perfect isolation..

Company name that provides what I asked for? ...Jim Thompson

Analog Devices? Their process seems to include quiet FETs - but it does seem to be unusually elaborate.

If memory serves, the fancier processes that produced good FETs included ion-beam implantation, which is pretty exotic.

To the best of my knowledge ADI does not offer foundry services. I suppose anything is possible if the volume is big enough, but it'd have to be pretty big (many many many millions of dollars, probably) for that to happen. I can't speak for the other biggies in the analog space like LTC, TI, Maxim, but I'd be surprised if they offered foundry access. That's a very different business model and requires a very different customer support structure.

I have heard of them desiging ASICs for people in return for NRE, which could achieve a similar outcome under the right circumstances.

Most CMOS is ion implanted IIRC, so not exactly exotic.

Yes, in fact my product group at ADI frequently does ASICs. But it usually takes more than simply NRE, there has to be a pretty big purchase commitment behind it as well. An ASIC has to compete against internally-generated products for development resources in a non-foundry company so has to stand on its own legs as a business proposition. Opportunity costs in any large non-foundry semiconductor company set a pretty high bar for ASIC business.

Jim's been around a while and understands this, but people not directly in the biz often don't. No insults intended.

It was more exotic back in the 1980's when I was working on electron-beam microfabricators - for Cambridge Instrument - and had occasional contact with people in the integrated circuit processing business.

Most modern CMOS processes might use ion-implantation at some stage in the process, but the original RCA and Motorola processes didn't seem to have done so.

They don't do litho with crayons anymore, either. ;)

Cheers

Phil Hobbs

Yep. The only "big name OEM" that I know of that does foundry work is ON-Semi, but just their AMIS facility in Pocatello, doesn't have an appropriate process.

My usual fab source is X-Fab. Their JFET has the back/body at substrate... making it pretty worthless for low-noise amplifier use :-( ...Jim Thompson

That had its advantages ;-)

My Master's project had a JFET. I just drew what I wanted on a quadrille pad and they (Motorola's layout guys) cut the Rubylith to match... those were the good-ol-days... no grand authorization-by-committee required to conduct an experiment... parts out in three days ;-) ...Jim Thompson

I used to do a lot of my own processing too, back in my silicon photonics days, but it was with a Leica e-beam writer with 30 nm minimum feature size. Fun stuff, except that some of my run sheets had over 600 steps by the time I was done. (That included rework and so on, which I had to do a fair amount of since I wasn't using what everybody else was, and had to optimize it on my own.) One learns patience that way, for sure.

Cheers

Phil Hobbs

Precisely when? I worked on the Cambridge Instruments EBMF 10.5, which did offer than kind of resolution. That had been incrementally developed over a bout a decade. Sometime around 1990, Philips gave up on their electron-beam microfabricator - which was a good machine, but needed more supported than they were set up to give - and sold it to Cambridge Instruments-Leica, who were happy to get a machine with fewer obsolete part problems and a non-wi re-wrapped back-plane.