Transistor arrays?

Does Intersil still make CA3086? Five per package, but it's an OLD part, probably not cheap. RCA design, then GE, Harris, Intersil. My datasheet still says RCA.

Somehow, Jameco seems to have them at $1.95 each:

And better yet, JDR has them at $0.85 each:

Way too expensive for my taste. And maybe bigger than 5 discretes, though monolilthic.

Jon

John

My PN2222A's are less than 1/2 cent each. :)

Jon

Not from Rohm, but those were the RETs I was talking about. It looks like they're indeed the best solution.

Nope. Gotta be able to see them in full sunlight. The reds are driven at 33mA and the blues at 22mA, though both PWMed.

At least it can be seen. ;-)

Nope. PWM, but no multiplexing. I'll probably get fancy with the PWM though. There is an FPGA on the other end of the wire.

Yep, my first thought (the current design uses IRLML2504s at 17 cents each) but SOT-23s are big. ;-) The RETs are about 1/4 the size but still easy for manufacturing.

Resistor Equipped Transistor. A transistor with resistors from input to base and base to emitter. Real cheap (moderately) high current drivers for UCs and FPGAs, but they only come one or two to a package.

Too big. Too expensive, but I never would have thought to use analog switches. They are cheap ones, but not cheaper or smaller than BJTs.

On a sunny day (Thu, 03 Sep 2009 19:48:24 -0500) it happened krw wrote in :

I use clear LEDs (clear plastic). That way if not on, they look well: clear. .5mA is clearly (red) visible in direct sunlight. I never ever use the ones made of colored plastic, with the sun on those they seem lit when they are not.

Just try a few of the high efficiencey ones :-)

I can read a book with 20 mA red in the dark.

Yes, it's an unorthodox selection, but analog switches sometimes can be used nicely for high side switching. Also input levels of some 5V capable devices are TTL like, not CMOS in contrast to tiny logic.

MIKE

I think someone brought these up, they get down to the range slightly above $0.01 if you buy serious quantities:

Comes in tiny SC-75. Of course if you guys have it assembled in the US then SMT placement costs per part are high. In that case a ULN2003 in TSSOP might be more economical:

Gaak!

Darlington. I think the OP said something like "No" or close to it.

I've used those. They are relatively lower cost. 15 cents and maybe down to a dime each in some qty. That makes them reasonable.

Also, the TSSOP is 108 C/W, thetaJA. The DIP is 67, just for comparison. Worst case Vcesat at Ic=100mA is 1.1V. Keeping that (not strictly correct) and at 30mA, that's 33mW per gate. Assuming 100% setting on his PWM, that's times 7 for all or 230mW. About 25C rise in the TSSOP, 15C in the DIP, not counting other measures added to the picture or counting power dissipated during transition between ON and OFF, which can amount to something. There's also 13k ohm to ground on the input, besides the drive to the Darlington. The 1.1V mentioned above also is given with an input drive current of 250uA. At 30mA on Ic, this is a lowsy equivalent beta of 120. 100uA of that or more may be due to that 13k.

Why nothing with just a simple BJT arrangement? (Offered both with and without protection diodes, which I decidely may or may not want. Oh and might as well as for PNP both ways, as well.) Is there no market?

Jon

I should have added that also by comparison, the PN2222A in TO-92 is

....

By the way, a problem I sometimes face is a low Vcc -- say a +5V rail; barely enough to drive a 400nm LED at pulse currents up to 80mA. And I need to control the current (PWM via a filter or else a micro DAC output.) I'm using feedback with software for closed loop control and do not need linearity (in fact, some non-linearity actually helps), but I do need monotonicity for the PID to work well. The application has physics associated with fluorescence that does NOT permit on/off PWM to achieve an equivalent, as this isn't about human eyes.

I can pay for an expensive device (multiple dollars) that provides crafted, linear control -- though finding something to do that with low overhead voltages is an issue. But I don't need that and I certainly don't want to pay that much for what I don't care about having. I can use a BJT, but there's no headroom to speak of for control. I currently use a current mirror with parts like the BCV62 (infineon, nxp, philips) plus a separate BJT. The whole thing is just a few cents in cost, uses all multi-source parts, but takes space and includes more than half a dozen parts -- some of which are there to protect against the wayward screwdriver on an LED connection.

It is here where simple BJTs (not Darlingtons) in a package of 8 would be nice. I'd form up current mirrors, nicely, and get 4 such in a package. On the other polarity side, I'd get 8 BJTs for one leg of the formed-up mirrors. For things I do with multiple such LEDs, that would be nice even if it doesn't include the resistors I add. Am I in some small minority on this? Why isn't there a non-darlington package of BJTs?

Jon

There are, e.g. the MPQ2222. Thing is, they're usually quads--there are quints, e.g. the CA3046, but with three leads per transistor you start running out of leads at that point (the 3046 is actually a hex die, but only 5 sections have metal).

Cheers

Phil Hobbs

T'is how it is. Que sera, sera. I have to switch design philosophy depending on where my clients says production will occur.

There are, but you can almost only get them in Asia. I think the last one I used was from Korea Electronics (KEC). They looked similar to the old Motorola parts:

Seen this stuff while in Korea but not anywhere in the West lately. It also is not always on the web site of manufacturers.

Then there is:

But again, you may even need to source the Mits arrays in Asia. Over here they probably only sell them complete with a car around them :-)

I can with these blues too. That's why we have a brightness control.