MAT 04

AD doesn't mention that on their web site (yet). Arrow has 90 on the shelf according to theirs. Not a whole lot, says 16wks if you want more. Maybe time to call your AD rep and if not happy with the answer stock up ...

The part is apparently going away in all packages.

Everybody email Analog or a distributor and demand that the MAT-04 be saved!

John

Farnell carries the THAT340N which looks a little better

-- Bill Sloman, Nijmegen

"Demand"? Buy some. It's the American way of doing things. ...Jim Thompson

They'll go by the sales numbers. If too low it'll be snuffed out. Problem is the price tag, it's a boutique part that is most likely used only by scientists. Us regular people usually only need temp tracking and then 50 cents usually do the trick:

Are those separate, isolated die? Amazingly, they don't say.

If isolated die, thermal tracking will be mediocre.

I recently did some static and transient thermal tests on the UPA800 dual RF transistor. That's two chips, and thermal coupling was shockingly bad.

John

It isn't great. Best is to put a blob of goo over it and place them away from any moving air. For monolithic you are pretty much stuck with FETs (CD4007) if it has to be cheap. The rest is all either sliding towards lalaland or carries boutique pricing.

The only truly matched ones in current production I know of is the THAT300 series that Bill mentioned. Out of Taxachusetts, of all places, Four bucks a pop, I guess they'll make you support their state-run health care if you buy ;-)

Hans Camenzind probably isn't too far away from you guys and he used to offer some sort of bone-simple custom chips. You were very limited in resistors and such so I never liked the concept but he could probably give you an array in a jiffy by leaving everyhting out and piping as many transistors as there are pin for:

I wish the ECL guys would offer a chip with uncommitted devices in there. That would really rock.

They do say "Multi-Chip".

Best regards, Spehro Pefhany

True. So it must be separate silicon. There's no spec for inter-transistor voltage, which should be an obvious thing to spec.

But I'm getting used to not seeing capacitances and voltages and all sorts of stuff on data sheets.

John

They are separate silicon.

The upside is that only a few select people will then be able to do unorthodox stuff with such devices. The downside is that often it won't be allowed because non-spec :-(

The real problem is differences in power dissipation driving differences in Vbe. I think the MAT-04 was an interleaved structure to minimize that. The separate-chip things are terrible that way.

For monolithic you are pretty much stuck with FETs

It's DI so I'm guessing the transistors aren't interleaved. No die photo.

There are some insanely fast transistors available in arrays, but by the time they wirebond the bits out to pins, they tend to turn into oscillators.

John

True. Problem is, most likely only one out of 100,000 engineers ever needed that. And only for one unit.

At their prices they should be able to furnish that information. Worst case you'd have to pop one open.

I've had Japanese RF transistors that came as arrays in SIP packages and they were very behaved.

Hans' stuff wasn't very fast, those chips should behave. But I am not sure how old he is now and whether he still offers the service. Maybe he could be bribed with some Swiss chocolate :-)

The problem is that (like most multiple transistor packages) that's a multi-chip part, and doesn't temperature track like a monolithic--it's about 100 times worse.

Cheers

Phil Hobbs

Been doing some digging--the BCM847 is not a bad part if it's monolithic, which it may or may not be, and costs 24 cents in onesies.

Then there's the THAT340 (already mentioned) but its beta is the pits.

Linear Systems makes a number of very good ones, if one can really get them--e.g. the LS313.

NXP only makes current mirrors, e.g. BCV61D, but they're pretty good if you're making mirrors. Not a lot of use for other stuff.

Intersil makes monolithic arrays, e.g. the HFA3046. None of these is as good as the MAT04, although most are faster.

Cheers

Phil Hobbs

In this case, you're not far wrong. There are things you can do easily with MAT04s that are extremely difficult without them--such as effectively reduce the noise of a $5k laser by 70 dB for measurement purposes. $50 is pretty cheap at that point.

Cheers

Phil Hobbs

So really the monolithic would have to be more than $50 to be less bang-for-the-buck. No sense being penny wise and pound foolish. ;-)

Best regards, Spehro Pefhany

I was about to use that as a replacement for a Zetec dual Darlington, until I saw its price. Yeouch! May it rest in peace.

That is a common dilemma we analog guys face. There's a challenging problem and a really cool solution. Then, poof, the solution goes away because we are too small a market. Happened to me with the SD5400, did a lot of nice things with it but it has already gone from mainstream to boutique. From there it's often just a matter of time :-(