We're looking for a PNP transistor capable of dissipating about a watt (in some air flow) and that has low output capacitance, where Co includes the mounting as well as the part itself. This is for use as a current source in a 1 GHz bandwidth driver circuit.
The 2nd best choice seems to be 2N2907. The best may be 2N1119, but they look hard to get.
-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at
formatting link
| 1962 | | | | Vote Barack... Help Make America an Obama-nation | | | | Due to excessive spam, googlegroups, UAR & AIOE are blocked! |
More details, please...How many volts do you have available for the current source? What's the current? What will be the voltage swing at the output of the current source?
There used to be some good high-freq uniwatt PNPs (MPSU56, MPSU60 iirc) but I can't find even a datasheet. The Cobo for most available parts is really high (>10pF @ 10V), so if your GHz driver is inflicting a large voltage swing then you're probably SOL. If you had a bunch of volts available you may be able to add a series R to help decouple the effect of the Cobo.
You're right -- that 2907 still looks like a crumulent choice.
Bob
--
== NOTE: I automatically delete all Google Group posts due to uncontrolled
SPAM ==
120 mA max, 1 watt max. The 2907 is borderline on power, maybe OK with a clip-on heatsink. Only about 8 pF. If we go with a surface-mount part, SOT-223 or such, we add too much pcb capacitance trying to get the heat out. [1]
There were lots of 2N5000-series fast PNP's once, but they've become hard to find and very expensive. Central still makes some interesting parts, and I'm trying to get specs and pricing.
It's amazing how good some of those ancient discretes were.
And it's amazing how many transistors nowadays don't even spec capacitances.
John
[1] we're considering using something like a 2-pin faston tab, sticking up in the air, as a cheap heatsink for a SOT-223. We'd just add a couple of thru holes right next to the transistor tab and solder the sucker in. Less puffs than a big copper pour.
Your U60 comment is interesting. Central makes a CEN-U60, in a nice TO-202 package (good cooling!) but has no proper datasheet. May be the sameish thing as MPSU60.
I only need 12 volts Vce, 120 mA max, 1 watt max, 5 volts max p-p swing, but low Cc.
We do have a ferrite bead in series with the collector, for the really fast stuff, but lots of capacitance causes slow whoopies (like, 50 ns) on output edges. No headroom for additional R.
Have you considered spreading the heat into two die and eliminating the heat sink and letting you go with a much smaller (and faster) die size? Perhaps a pair of BFG31 in parallel.
Can you still get the 2N2907 in TO-5? That would allow one of those circular finned sinks.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
| |
| Vote Barack... Help Make America an Obama-nation |
| |
| Due to excessive spam, googlegroups, UAR & AIOE are blocked! |
"John Larkin" wrote in = message news: snipped-for-privacy@4ax.com...
watt
a
How about the 2N5583?
30 V, plenty of current, 5 pF max, 1.3 GHz Ft, TO-39
35 C/W theta jc The amazing thing about this one, as I recall, was that S21 was almost flat from 50 to 150 mA. And it had 1.3 GHz Ft in a PNP in 1969.
Don't know about status and availability, though.
For less C, there used to be a PNP complement to the BFR-96. BFQ-32? But that was in a "pill" package and I doubt you could put 1 W in it.
Of course, if you don't need to go down to DC you can make an "artificial inductor" with an NPN and a big capacitor from base to emitter. Still limited by the Ccb of the transistor, though.
--=20 Regards, Howard snipped-for-privacy@ix.netcom.com
That's pretty radical. But I still have to get rid of the heat. The prototype board uses a SOT-223 version of the 2907, and there's not quite enough pcb copper to keep the case temp below 100C, and there's already 10 pF of copper pour on the bottom side. I had to add a copper flag to get it down to 80C
ftp://66.117.156.8/Copper.jpg
So if I use two transistors, I still have to get rid of the same amount of heat. The nice thing about a leaded part is that it sticks up, into the air flow - we're adding a servoed fan to the 8-channel production unit - and air has an Er = 1!
Fast stuff is nasty: it gets real hot, but everything has to be really close together, and stray C must be kept down, so cooling is usually a pain. It would be nice to make boards out of BeO or AlN: both have incredible ratios of thermal conductivity to dielectric constant. Or diamond.
which is 8 pF max and has the virtue of costing 10 cents. But as I mentioned to JP, adding enough pcb copper to cool off a surfmount part adds at least 10 pF, likely more. So I'm back to old-fashioned leaded parts sticking up in the air.
Yup; Digikey has them from several vendors, and the Lady Liberty clip-on heat sinks are available.
The original TO-5 was rotten thermally (the belly side was epoxy) but I think we can also get them in the package with the solid header, TO39 or something.
How about the 2N2905 that Rich suggested? Ok, it's around 50c which kind of crimps the fun but comes in TO-39. It has a lot better thermal conductivity than the TO-18 package. Digikey has them:
formatting link
The moped cylinder type heat sinks are IMHO a bit flimsy nowadays. Had to use some at a client after they requested a surprise post-layout change of about 1000% in the output power spec. Those things slip off rather easily and according to Murphy they fall into some place where they short 12V and 3.3V ... bzzzt ... POOF.
--
Regards, Joerg
http://www.analogconsultants.com/
"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.