EmbeddedRelated.com
Forums
The 2024 Embedded Online Conference

Simple "LED driver" chip?

Started by Richard H. April 6, 2006
On Fri, 07 Apr 2006 07:39:59 -0700, Richard H. wrote:
...
> That brings me back to the question about the generic part numbers - in > this case, a general-purpose transistor. Is there a book (website?)
MMBT2222: http://www.fairchildsemi.com/ds/MM/MMBT2222.pdf Cheers! Rich
Richard H. wrote:
> That brings me back to the question about the generic part numbers - in > this case, a general-purpose transistor. Is there a book (website?) > that offers a good reference for these part numbers and their > characteristics? (E.g., Lancaster's TTL and CMOS cookbooks have about > 100 pages of reference for logic chips, but it seems there's a lot more, > and there are newer processes.)
Resistor Equipped Transistors are now quite common, and cheap. So they are a good digital interface device, where you go outside the uC pins abilities. See the Philips web site, and they also have dual RETs with PNP/NPN designed for level shifting, giving (eg) 12V hi side drive from a uC pin, in a single SOT23-6 style package. -jg
Richard H. wrote:

> Spehro Pefhany wrote: >> If it's several, as in 2 or 3, just use transistors, BJT or MOSFET. >> Darlington arrays as others have suggested would be better if you have >> more than four... keep in mind that they have significant foward drop >> on a 3-5V circuit (which may be a factor if you're using blue or white >> LEDs). You might also be able to use a HCMOS shift register. > > > Thanks to everyone for the quick comments! > > The need here is very simple, so I'm reluctant to use a feature-rich / > pricey part (we'll be making a lot of these; the forward voltage drop > would also be an issue). We'll probably do 4-5 LEDs per MCU. > > The flexibility of discrete transistors sounds good (and possibly very > cheap); since I'm doing SMT, we could get them pretty small too. Hmmm. > > I'll do some AoE reading and look at the discrete designs offered here. > > > That brings me back to the question about the generic part numbers - in > this case, a general-purpose transistor. Is there a book (website?) > that offers a good reference for these part numbers and their > characteristics? (E.g., Lancaster's TTL and CMOS cookbooks have about > 100 pages of reference for logic chips, but it seems there's a lot more, > and there are newer processes.) > > >> If you're running small LEDs at that high a current you >> may well have reliability and lifetime problems, particularly when the >> ambient temperature gets high. > > No problem here. They're not normal LEDs; they're high intensity > (several hundred mcd, but not the multi-candela ones), and I'll be > running them well below rating. > > Thanks! > Richard
Please do not rule out transistor arrays in sot23 packages. Half a dozen transistors in a 14 pin package (emitters tied to pins 7/14).
On Fri, 07 Apr 2006 07:39:59 -0700, the renowned "Richard H."
<rh86@no.spam> wrote:

>That brings me back to the question about the generic part numbers - in >this case, a general-purpose transistor. Is there a book (website?) >that offers a good reference for these part numbers and their >characteristics?
No, AFAIK. Look at distribution quantities-in-stock and prices or ask. There are thousands of part numbers which will work, and dozens which are close to optimal depending where the product is being built and how many. For low to moderate quantities in North America you can't go far wrong using MMBT4401/4403 for NPN/PNP. Google for data sheets, there are many, many manufacturers. 'Digital' transistors (transistors with base and maybe E-B resistors built-in) and digital transistor arrays were developed originally in Japan by Rohm and are now multiple sourced & are very popular in consumer goods, but generally not quite as cheap and easily available as resistors or resistor networks and discrete parts in low quantities. Generally the transistors 'inside' are referenced to popular Japanese discrete transistor part numbers. Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Spehro Pefhany wrote:
> On Fri, 07 Apr 2006 07:39:59 -0700, the renowned "Richard H." > <rh86@no.spam> wrote: > > >That brings me back to the question about the generic part numbers - in > >this case, a general-purpose transistor. Is there a book (website?) > >that offers a good reference for these part numbers and their > >characteristics? > > No, AFAIK. Look at distribution quantities-in-stock and prices or ask. > > There are thousands of part numbers which will work, and dozens which
The 2N7002 may be a good fit. FET with low threshold voltage and low on resistance. Regards Robert
Rocky wrote:
> The 2N7002 may be a good fit. FET with low threshold voltage and low on > resistance.
Indeed, this looks very promising. In particular, ON Semi's version in a tiny 1.6mm SC-75 package for 4 cents. After wallowing a while in Digikey's search engine, I tried a different approach... I jumped over to ON Semi's site and sorted by price (something Digikey won't do). Thinking I'd work my way up the price list, I stopped after the first entry, which was a 2.7 cent 2N7002. Good enough for my purposes, but the smaller version was worth the extra penny. :-) Cheers! Richard
"Richard H." <rh86@no.spam> wrote in message news:31m%f.426$AB3.11@fed1read02...
> Rocky wrote: >> The 2N7002 may be a good fit. FET with low threshold voltage and low on >> resistance. > > Indeed, this looks very promising. In particular, ON Semi's version in a tiny 1.6mm > SC-75 package for 4 cents. > > After wallowing a while in Digikey's search engine, I tried a different approach... > I jumped over to ON Semi's site and sorted by price (something Digikey won't do). > Thinking I'd work my way up the price list, I stopped after the first entry, which > was a 2.7 cent 2N7002. Good enough for my purposes, but the smaller version was > worth the extra penny. :-) > > Cheers! > Richard
Depending where you have stuff manufactured, the LED-specific constant current drivers are likely to be cost-effective against the loading and procurement cost of all those discretes, and the reliability issues surrounding so many parts. Part prices are only part of the picture. They're damn easy to use too. -Andrew M
Richard H. wrote:
> Spehro Pefhany wrote: > > If it's several, as in 2 or 3, just use transistors, BJT or MOSFET. > > Darlington arrays as others have suggested would be better if you have > > more than four... keep in mind that they have significant foward drop > > on a 3-5V circuit (which may be a factor if you're using blue or white > > LEDs). You might also be able to use a HCMOS shift register. > > > Thanks to everyone for the quick comments! > > The need here is very simple, so I'm reluctant to use a feature-rich / > pricey part (we'll be making a lot of these; the forward voltage drop > would also be an issue). We'll probably do 4-5 LEDs per MCU. > > The flexibility of discrete transistors sounds good (and possibly very > cheap); since I'm doing SMT, we could get them pretty small too. Hmmm. > > I'll do some AoE reading and look at the discrete designs offered here. > > > That brings me back to the question about the generic part numbers - in > this case, a general-purpose transistor. Is there a book (website?) > that offers a good reference for these part numbers and their > characteristics? (E.g., Lancaster's TTL and CMOS cookbooks have about > 100 pages of reference for logic chips, but it seems there's a lot more, > and there are newer processes.) > > > > If you're running small LEDs at that high a current you > > may well have reliability and lifetime problems, particularly when the > > ambient temperature gets high. > > No problem here. They're not normal LEDs; they're high intensity > (several hundred mcd, but not the multi-candela ones), and I'll be > running them well below rating.
Interesting that everyone focuses on transistors rather than more integrated solutions. An interface chip I use a lot is the NC7WZ17P6X dual buffer in a 6 pin SC70 package, about 2 mm square. It will drive 24 mA at LVCMOS levels and will handle up to +-7 volts on the input. With 5 volt Vcc it will drive up and down at 32 mA. I don't think you can touch the density with transistors and we are only paying $0.06 each or $0.03 per driver. I recently removed a bunch of SOT23 transistors from my design and replaced them with these, some to drive LEDs.
rickman wrote:
> Interesting that everyone focuses on transistors rather than more > integrated solutions. An interface chip I use a lot is the NC7WZ17P6X > dual buffer in a 6 pin SC70 package, about 2 mm square. It will drive > 24 mA at LVCMOS levels and will handle up to +-7 volts on the input. > With 5 volt Vcc it will drive up and down at 32 mA. I don't think you > can touch the density with transistors and we are only paying $0.06 > each or $0.03 per driver. > > I recently removed a bunch of SOT23 transistors from my design and > replaced them with these, some to drive LEDs.
Thanks! Half the battle seems to be finding these bargains when you need them. The other half seems to be getting distributors to talk with a small guy (even if you're willing to buy a reel). I agree with you on using arrays where possible, and this package certainly seems small. I'll give it a look. Small PCB size is important to this project, so discretes are attractive because they can be squeezed into small spaces (even though the overall density is worse). So far, I've found a discrete 2N7002 MOSFET with great specs (overkill, actually) at $0.04 in a similar tiny package; at this price, I could afford to buy a reel if I can't find a disti that'll do cut tape.
> ... > So far, I've found a discrete 2N7002 MOSFET with great specs (overkill, > actually) at $0.04 in a similar tiny package; at this price, I could > afford to buy a reel if I can't find a disti that'll do cut tape.
We got a few hundreds spare 2N7002 (100mA). If you can't find what you need, we can send you some. Shipping is probably more than the cost anyway.

The 2024 Embedded Online Conference