Small-signal transistors(bjt) SOA question

[RW988]

Hi,
I was looking for a safe operating area diagram of the BC550/60 but none of the official specifications(fairchild, on-semi, cdil) contained it. There are absolute maximum ratings, but conditions for Ic or Vce were not included. Most of the bipolar junction transistors operating area is limited by secondary breakdown, but this information is not included in BC550/60(and similar small-signal bjts) datasheets.

My question is:
Is SOA for these bjts limited only by Ic, Vce and power dissipation line(Ta)? Or maybe I missed something? Where can I find this information?

I mainly care about project reliability. If you know load parameters(of these bjts or similar) from your own experience, please share this knowledge. Thanks

 

[Mark Tillotson]

SOA is simple for a small transistor dies as there is no secondary breakdown with the die being smaller than hot-spot dimensions. Hot spot dimension reduces with power density. On a device without a built-in metal heatsink the power density is much more limited anyway, the whole die overheats before power fluxes can create significant temperature differences across the die.

 

[RW988]

Ok. Thanks for the explanation. So the only limitations are the maximum power dissipation, Ic and Vce? Maybe I should take some safety margin for these parameters for reliability?

 

[indianajo]

Maximum power limit is at 25 deg C case which usually can't be maintained. People used to put metal clips around TO92 cases but I doubt that it helped much. If I'm nearing 1/4 watt on TO92 I install a 1 cm long case version (MPSW06/56) 1 watt rated or go on to TO 126.

 

[GoatGuy]

… and of course, it should be said, that if one is getting anywhere near 1 W, it does beg “but why?”. After all, the best performance of small-signal BJTs is no-where near their theoretical safe-operating-area free-air power rating at room temperature!

⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅

 

[RW988]

So for a 1/2W device 1/8W is a reasonable limit? In the worst case scenario, of course.

What about parallel connection of small signal transistors? I found two topics, but they did not give me a clear answer as to what values of emitter resistors and possibly additional in the base circuit should be used for effective current distribution. I think they should be many times larger than in a typical medium or high power amplifier but here there is a complete lack of thermal coupling by the heat sink. Anyone tried? Are there any thumb rules for these configurations?

 

[indianajo]

There have been amps discussed on this forum where the speaker is driven by 50 pn2222 TO92 transistors or something. I remember the words "massively parallel" but the only link I come up with is op amps: Building a massively parallel op amp power amplifier
With emitter resistors you would be trying to cover up variations of .2 v in the Vce of different parts, so if each transistor provided 200 ma at full power, you would want the emitter resistor to drop about .4 volts, so V/I=R, 0.5 ohms? Your design current would be to make the transistor dissipate no more than .25 watts, so pick a collector resistor that makes that happen when the transistor has half the power supply voltage as Vce. If your parts vary more than .2 volts at a reasonable current by test, then use that number as the voltage to calculate the emitter resistors.
With circuit boards costing by the square foot, and even worse for point to point builders like me, each wire taking about 2 minutes to add, I don't see the point of using TO92 transistors for output unless you are taking a college electronics lab and the BC550 are free. Besides 2 of the 3 distributors I use (digikey, mouser) charge about $.12-18 for each 1/2 watt resistor. (I do better at farnell that has closeouts on end of bin #, but you can't get $3 a hundred on every value every day).
I've used $1 30 Mhz Ft TO220 transistors (8 amp rating) as VAS, the transistor after the input and before the drivers, which shocks the greybeards around here. But it doesn't sound bad IMHO. The AX6 has 6 transistors/channel. Most distortion is in the speaker anyway, .3% HD is negligible IMHO. That number won't sell amps on the consumer market where bragging with numbers is the game. I find high freq IM distortion at low volume more annoying, and no sales brochure quotes that number. I find hiss & hum between program items annoying too, which is why I buy metal film resistors.

 

[john_ellis]

An ON Semi datasheet from 2007 shows no SB limit for device types BC556-558. Full 625mW dissipation at Ta=25C and just over double for Tc=25C at up to 65V (BC556). I would imagine that as the BC560 is from the same series and as only 45V that would be equally OK.
The problem with these devices would be the high thermal resistance. At about 250K/W the junction is going to get pretty warm for more than 100mW dissipation, which may need some compensation.
It is also worthwhile remembering that high reliability circuits need to keep junction temperatures low, because thermal cycling can induce damage. That is probably a greater concern than SB.

 

[RW988]

but the only link I come up with is op amps: Building a massively parallel op amp power amplifier

Wow, extreme. But very interesting. Finally only 4.7ohms as current distribution resistors and it works!

With emitter resistors you would be trying to cover up variations of .2 v in the Vce of different parts, so if each transistor provided 200 ma at full power, you would want the emitter resistor to drop about .4 volts, so V/I=R, 0.5 ohms?

So twice the expected voltage difference divided by the maximum expected current as a rule? 0.4/0.2 should be 2ohms

I don't see the point of using TO92 transistors for output unless you are taking a college electronics lab and the BC550 are free

hehe, you're right. I don't want build power amplifier to drive 4ohms speakers. Inductive load, yes, but ~44-50ohms lowest impedance. So I think about connecting in parallel... maybe 5-10 pairs of transistors? Supply voltage +/- 17.5 maybe 18V and an unpredictable input signal with slightly lower peaks.

Your design current would be to make the transistor dissipate no more than .25 watts, so pick a collector resistor that makes that happen when the transistor has half the power supply voltage as Vce.

So 250mW(worst case) is for 500mW TO92 device safe limit? Half power as a reasonable margin?

An ON Semi datasheet from 2007 shows no SB limit for device types BC556-558. Full 625mW dissipation at Ta=25C and just over double for Tc=25C at up to 65V (BC556).

Yes, 625mW for full device or 500mW collector power dissipation. It depends on the datasheet version, but both for Ta=25*C. I really don't know how reliable this data is. Thermal resistance(junction−to−ambient)=200-250K/W, maximal Tj=150*C, mathematically, everything is correct.

The problem with these devices would be the high thermal resistance. At about 250K/W the junction is going to get pretty warm for more than 100mW dissipation, which may need some compensation.
It is also worthwhile remembering that high reliability circuits need to keep junction temperatures low, because thermal cycling can induce damage. That is probably a greater concern than SB.

Thanks, this can be very important. 100mW for 250K/W gives for junction 25K over the ambient temperature. It's quite low compared to the graphics card temperatures. It really should be so cold?