Universal Transistor

December 21, 2011 8:47 PM

Are you tired of having to pick between N-type and P-type transistors when designing electronics circuits. Well, help is on the way with the new Nanowire Universal Transistor.
posted by Confess, Fletch (16 comments total)

This may be a very important breakthrough in solutions to a problem I never was aware existed. Cool!
posted by meinvt at 8:59 PM on December 21, 2011

This is NOT cool OR uncool OR NOT cool AND uncool OR cool AND NOT uncool.
posted by twoleftfeet at 9:02 PM on December 21, 2011 [16 favorites]

Is this real, or is it more hand-wavey cold-fusionesq grant-money grubbing?
posted by lekvar at 9:06 PM on December 21, 2011

It's cool, but it's highly unlikely that this will see large-scale production in the next 15 years, unless it's got better radiation hardness (and from my rather limited viewpoint, I can't see how it would be better). Transistor density just isn't an issue anymore, scaling is.
posted by spiderskull at 9:23 PM on December 21, 2011

How do they sound when you run a guitar signal through them?
posted by drezdn at 9:26 PM on December 21, 2011

Might lead to slightly smaller transistors, but from a technical standpoint, couldn't this be done with four regular transistors? An n-type, a p-type and another two to act as switches to connect those two to the source depending on the input? Maybe it's a little more complicated, I'm not an EE. But those four transistors might have a much simpler structure.

How do they sound when you run a guitar signal through them?

Digital circuits aren't usually designed with transistors that do well in terms of amplifying things linearly, so probably not very good.
posted by delmoi at 9:29 PM on December 21, 2011 [1 favorite]

Digital circuits aren't usually designed with transistors that do well in terms of amplifying things linearly

Then again, neither are the best guitar amps.
posted by flabdablet at 9:55 PM on December 21, 2011

Finally NP-complete.
posted by benzenedream at 9:59 PM on December 21, 2011 [5 favorites]

How do they sound when you run a guitar signal through them?

Getting an imaginary single universal transistor would be interesting to play with. If it could be biased like a dual gate FET, it might actually function reasonably well in a more or less linear mode.

But no matter what, it won't have that special mojo unless it's made from germanium in a metal TO-18 can.
posted by 2N2222 at 11:03 PM on December 21, 2011

I trust 2N2222 on this matter.
posted by 7segment at 12:05 AM on December 22, 2011 [6 favorites]

Feh. Give me four of these and four of those and ±110V power rails and I'll show you special mojo.
posted by flabdablet at 1:18 AM on December 22, 2011

spideskull wrote:
Transistor density just isn't an issue anymore, scaling is.

Any chance you can elaborate a little please? Are these two things not related?
posted by kcds at 4:25 AM on December 22, 2011

The article says n-type transistors or p-type transistors, but don't they mean n- or p-type semiconductors? (Like the links above)
In my experience, transistors have both n-type and p-type semiconductor materials. That's why they get called NPN or PNP. Do they mean the choice between n-MOS and p-MOS? I thought CMOS was plenty dense enough. Why do we care?

(note: I'm sure transistors have changed a lot since I went to school)
posted by MtDewd at 6:40 AM on December 22, 2011 [1 favorite]

I trust 2N2222 on this matter.

I think he's biased.
posted by Crabby Appleton at 8:28 AM on December 22, 2011 [7 favorites]

I trust 2N2222 on this matter.

Me too. 2N2222 is a transistor.
posted by exphysicist345 at 10:55 AM on December 22, 2011

MtDewd: Yep, NPN and PNP. P-type and N-type are diodes/semiconductors.

There are multiple problems in the next level of circuit density, spiderskull's claim aside.

The next-smaller printing wavelength, "Extreme UV" - really x-ray spectrum - requires vacuums to work in, where materials can behave bizarrely (titanium can absorb enough elemental hydrogen to actually crumble like dried bread!), and the lasers involved as sources are incredibly reactive (spewing tin ions like a flamethrower, for instance).

ASML is working on technologies that effectively double the resolution limit at a given wavelength. But it has limits in when it can be used; nothing comes free. Basically, it's useful for memory, but not logic. Logic is the bigger driver, economically, however.

Not sure what spiderskull's comment about "radiation hardening" is about. I've never heard of this requirement for circuits at all.

As for this not seeing large-scale production in the next 15 years... new technologies take 5-10 to develop anyway, even in this field. Next-gen machines are bought & paid for 2 years before the parts are made. 15 years isn't a problem; it's just a timeline.
posted by IAmBroom at 7:09 PM on December 23, 2011