> > > I'm driving 12V LED load =3D LED strip - 3 LEDs in series + internal resistors. > using Arduino PWM. While looking for transistor > suggestions and before reading your post, I used MPS2222A (Motorola TO92 > 2222) from by parts bin. > > The current draw from the LEDs (with PWM) was 135mA @12v. I determined th= e > wattage to be 1.62W. Ta < 25C. > This is total LED + transistor dissipation. When on transistor dissipation =3D Von x Ion Von for transistor driven as an on/off switch is saturated voltage Using BC337-40 Minimuim Beta =3D 250. See data sheet at http://www.onsemi.com/pub_link/Collateral/BC337-D.PDF Say we use Ibase =3D 3 mA See fig 4, page 3. At Ic =3D 300 mA , Ib=3D 3 mA, Vsat =3D 0.4V At Ic =3D 100 mA, Vsat =3D 0.08V At 300 mA, Vsat =3D 0.4V, Pd_transistor =3D V x I =3D 0.4 x 0.3 =3D 120 mW Page 1 , thermal characteristics. Rja =3D 200 C/W That is rise in degrees C (or K) per Watt in free air with no assistance from heat sinking viua leads etc. 200 C/W x 0.120W =3D 24 degrees C temperature rise. Tmax operating (page 1) =3D 150C (NOT recommended. At tambient of say 50C (> Death Valley on most days) Tj =3D50 + 24 =3D 74C Max allowed tambient =3D 150C tmax - 25C =3D 125 C. This would perhaps survive in a car engine compartment on a hot day. At 135 mA and say Vsat =3D 0.1V you get Pd =3D 0.1 x 0.135 =3D~ 15 mW. Temperature rise =3D200 C/W x 0.015 =3D~ 3C :-) !!! Driving the transistor with a "forced beta" well abiove nominal miunimum for the transistor gives low Vsat and low Pd and low temperature rise. > I used a 1K base resistor and at 100% duty cycle, the base current was > 3.8mA (hFE ~100) > At 100% duty cycle the collector current was 136mA > At 50% duty cycle the collector current was 67mA > http://i.imgur.com/E8IjHhq.jpg > > Although MPS2222A in a TO92 package can only sustain 625mW dissipation, it > was not running hot. I ran it >24 hours and I saw no heat issues. So you > are right that small transistors (TO92) like BC337 can be used. > > As above. > 1. In a similar case (load, transistor, supply, etc) how much can I expec= t > from a TO92 case? > See data sheet and example above. Ingredients: 200 C/W Rja Decreasing Vsat with increasing Ib/Ic Tambient Tj max. Roast until tender. 2. How to calculate the tipping point to go to TO126 or TO220 package? > As above.. Depends on how hard you can drive it to get Vsat down and how close to wind you wish to sail. If you limit rise to say 50C then you can get 250 mW dissipation. That may be getting adventuresome. Imax for this version is 800 mA ! Say at 500 mA and 250 mW you need Vsat <=3D 0.5V Fig 4 says 8 mA drive gives 0.5V at 500 mA. That's 'typical' and getting decidedly risky. More drive a good idea. IF you can heatsink the TO92 or blow some air over it you can look at the Tjc (rather than Tja) of ~=3D 85C/2. That's for perfect case to heatsink rtransfer, which you will not get. but shows that you can get a bit ore out of TO92 if you must. Somewhere around here you can use D-PAK / TO252 SMD which are easy to use in a hobbyist environment and are closer to TO220. They are SMD but can be used in a through hole system without much problem. They can get appreciableheat sinking from PCB copper and some from strip board copper and can be het sunk. TO220 is substantially bettwer than TO92, even with no heat sink. TO126 is usually indicative of a steam driven age component and I'd not expect to see much of these now. ie liable to be low Beta etc. [There MAY be exceptions - eg darlingtons]. > 3. Is it safe to assume that using a MOSFET can even be thermally more > effective due to low Rds(on) > > MOSFETs are usually much easier to drive hard (Vgs min factor, current only an issue for switching speed). Select Rdson rated to suit.. Uusally much lower than can be easily got with bipolar. eg the BC337 at 500 mA and say 0.5V requires 8mA drive typical. Rdson equiv =3D V/I =3D 0.5/0.5 =3D 1 Ohm !!! <=3D 0.1 Ohm Rdson FETS are usual, much lower is common and < 0.01 Ohm available. MOSFETS start to need appreciable gate current at switching edges a speed rises. Under 10 kHz PWM may be OK with processor pin drive. Edges startto get noticed on short pulses. If yuou want switching times around 1 uS or less you will need a gate driver. At not too fast PWM the driver can be as simple as 2 x jellybean bipolars + no R's etc. (1 x npn, 1 x pnp, join bases, join emitters, drive bases, output from emitters, collectors to supply (NPN) and ground (PNP). Vout swing is two Vbe less than Vin swing. This has no shoot-through protection but has a dead spot as Vin transits area withing +/- Vbe of Vout which is enough in most cases to work "OK". Russell --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .