Shop OBEX P1 Docs P2 Docs Learn Events
Is there a 'more' reliable transistor? — Parallax Forums

Is there a 'more' reliable transistor?

T ChapT Chap Posts: 4,223
edited 2012-04-09 01:52 in General Discussion
In an application where a Prop will drive IR LEDs at 38K for infinity, is there a transistor that could be deemed more reliable than others? Reliable meaning, under normal temperatures(70-80F) last a long time will the least failures. I am testing 15 IR LEDs(TSAL6200 Vishay). The transistor will be an NPN. Optionally, I do not mind the expense of a mosfet.

In example circuits online, I see 2n2222, BC337.

Comments

  • jmgjmg Posts: 15,185
    edited 2012-04-05 23:16
    T Chap wrote: »
    is there a transistor that could be deemed more reliable than others?

    Temperature rise (and thermal cycling) are a known failure drivers, so a cooler transistor will be deemed more reliable than a hot one. I think oft quoted is a 10'C junction rise, doubles failure rates.

    There are also failures due to electron migration so pushing close to a rated current is also not good for reliability.

    So choose a power package that is easy to cool, and use a fraction of the rating.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2012-04-05 23:26
    Take a look at the LP395 or LM395, depending upon your current requirements.

    -Phil
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2012-04-06 01:13
    The 2N2222a in a metal cap is very rugged for small applications. It is rated at around 800ma and 75volts. Sure, you could go to something with a 5amp rating and 120volts - but do you really need to do so.

    Keep your transistors strongly derated for durability - use the 2N2222a for 100ma at 12 volts or less is likely to last indefinetly.

    http://www.eng.yale.edu/ee-labs/morse/compo/datasheets/2n2222.pdf

    Heat is the great destroyer. Individual metal packages are better than multiple transistors in epoxy packages. So be wary of SMDs.

    MOSfets generally run cooler, but have two disadvantages - they often do not do well when switching is driven by 3.3VDC or 5VDC; and they can be easily damaged by static electricity. So my 'hobby choice' is to keep some 2n2222a on hand. There usually isn't much heat, but if you must have a low-power consumption design, the MOSfets are worth the additional trouble.
  • LeonLeon Posts: 7,620
    edited 2012-04-06 03:11
    Metal can devices tend to be more reliable than ones in plastic packages, and dissipate heat better.
  • Tracy AllenTracy Allen Posts: 6,666
    edited 2012-04-06 11:05
    The super e-line series of bipolar transistors from Zetex would qualify. Here is an app-note in relation to automotive temperatures and hazards. They use an advanced planar process that cuts down on hot spots within the structure, and add passivation layers to protect surface, and encapsulate in a high temperature silicone plastic. They have high gain, low saturation voltage, and very high DC and pulse current handling capabilities, even in a TO92 type package.

    I keep a small stock of ZTX1049A (npn) and ZTX1149A (pnp) on hand. They are superbeta, with gain typically 800 or 1000, so I often use them with op-amps in current sources, as well as for switching where that high gain eases the drive requirements. They do seem to be bulletproof. Its TO92 type package is rated for 1W, 4A continuous at saturation, 20A pulse. So it could coast along driving a few leds, and do so with a very low base current.

    Whether they would be worth the price premium for you application is quite a different question.

    Zetex was a very innovative company, and I was kind of sad to see their identity swallowed in acquisition by Diodes Incorporated.
  • T ChapT Chap Posts: 4,223
    edited 2012-04-06 11:17
    Thanks for all the info. Tracy, digi has that part for .74(npn). They have the part Phil mentioned for 3.75(1amp version). The whole point is overkill, the 2n2222 is the defacto for most of the circuits online, but then again the application requires the least potential for breakdown, a few bucks extra is insignificant on a one-off.
  • BigFootBigFoot Posts: 259
    edited 2012-04-06 15:27
    T Chap wrote: »
    Thanks for all the info. Tracy, digi has that part for .74(npn). They have the part Phil mentioned for 3.75(1amp version). The whole point is overkill, the 2n2222 is the defacto for most of the circuits online, but then again the application requires the least potential for breakdown, a few bucks extra is insignificant on a one-off.

    I use the smt versions of the 2N3904 & 3906 in all of our products and have never seen one blow.
  • xanaduxanadu Posts: 3,347
    edited 2012-04-06 20:36
    I like the 2N2222 TO-18. I have used much more expensive solutions that didn't go so well.

    [video=youtube_share;yNDO04g9Ugc]
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2012-04-07 03:15
    Regarding the 2N3904 & 3906,
    They are certainly good transistors, but one might as well buy the 2N2222a and the appropriate counterpart 2N2907 for usually similar cost. There are occasions that the 2N2222 and 2N2907 will drive a relay or an H-bridge that is marginal with the 2n3904 and 3906.

    Handier, and more economical

    Smt is a separate issue as the package is a significant thermal issue. It really helps to just have the transistors off the board and individually cooling. But if 5VDC and 30ma are your usuall parameters, heat is not much of a problem.
  • jmgjmg Posts: 15,185
    edited 2012-04-07 20:27
    T Chap wrote: »
    digi has that part for .74(npn).

    I like the SOT223 packages for moderate drive applications.
    There, in MOSFETS, Digikey lists
    100V 1.7A 350mOhms 56c 1+
    30V 5A 60mOhms 68c 1+
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2012-04-09 01:52
    MOSfet are optimal for SMTs - but different engineering and assembly issues. Unlike the bipolar transistor that seems to go to saturation at 3.3vdc; a MOSfet must be carefully selected or specifically adapted to do so.
Sign In or Register to comment.