Disclaimer: This post has nothing whatsoever to do with trans-sexuals or transvestites.
I bought two batches of medium power BJTs off Ebay (I know, I'll never learn). They are all PNP BD140s. Obvs the first thing I wanted to do was check they were the real deal. First test didn't look good at all. Beta for both batches was way too high: between 150 and 200. Shouldn't be more than about 80 according to the datasheet! Next test, stick 70V across Vce (the datasheet stated abs. max is 80V). Result: pass. Nothing went phut. Next test (not yet implemented owing to the limitations of breadboard to pass high currents) is going to be Ic and power dissipation. I plan to run an amp Ie and get a Vce of 5 to get
5W. The DS says these devices - if kosher - should be good for 12W so that's not asking too much. I will be mounting them on a suitable heatsink. Any other test anyone can think of? Or should I just have binned 'em all straight after the Beta figures were too high?
Nope. I already have a batch of genuine BD139s (the NPN compliment to the 140) which I had planned to use in a class B audio amp output stage together with the suspect ones. The gains of the two types need to be as close as possible for minimum distortion whereas as things stand, the PNPs are double that of the NPNs.
Modern day semiconductor processes tend to produce new transistors with better gain and closer tolerances than the original devices. They usually specify a minimum current gain and if you are lucky you might get a bit more than that now. Any half decent design the beta shouldn't matter provided that it is enough to make the feedback loop work.
Fakes tend to have either less gain than the real part or even a dead short between one or more of the pins.
Why wouldn't you buy a matched set of NPN and PNP from the one manufacturer ? If they've fiddled the Hfe on one of them, they've probably fiddled it on its mate as well.
This is what datasheets and trace-ability are for.
The last parts I bought, were from Element14. which is now owned by Farnell as far as I know. Not all parts are available as "cut tape", but some are.
Yeah, I didn't bother going for a matched pair in this instance since it's a very 'lo-fi' signal source (and old AM radio tuner intended solely for sports commentary). But I'm thinking of betas of 78 vs. 178 might be too far apart even for lo-fi.
If they are producing new devices with "better gain" than NOS then that's a big deal-breaker for me I'm afraid. I don't think that's acceptable as one runs into issues such as I'm having here, plus it means the new ones don't fully conform to the original datasheets. If they're going to do that, they should call them something other than the original part number.
Not necessarily less gain. Fake power trannies could have higher betas if they'd re-potted small signal devices just to pass a simple transistor tester check and that was what I suspected had been done in this case. I'll soak test 'em at 1A Ic and set them aside for future use in something else if they pass. If they fail, then the bin beckons.
The transistors will absolutely have emitter resistors in to nullify gain issues. In practice the driver transistors develop a voltage across the base emitter resistors, and the emitter resistors turn that into essentially a transconductance.
They are genuine. The specs are very loose on beta. These are not valves....
In a production design with less than 0.05% total distortion at all power levels I *never**ever* selected transistors to be of balanced gain.
Instead one uses localised negative feedback to linearise them. The spec sheet I am looking at has a beta that varies wildly with emitter current anyway.
formatting link
At 100mA collector current it is rated at 60-160 At less than that its rated up to 250 at 0.5A its rated at 'better than 25'
It is impossible to match an NPN to a PNP. The dopants are different the Vbe curves are different and the beta versus Ic curve is generally different. So called 'matched pairs' are merely a manufactures best efforts attempt.
My very first hard lesson in transistor design was when I left a small RC receiver I had built in the sun. It stopped working. The change of gain with temperature was of the order of two to one.
As a professional designer of power audio kit I learnt never to expect too much gain from a single transistor - always use undecoupled emitter resistors and quite low base emitter resistors to produce a combination with a gain that was predictable over wide ranges of power and temperature.
I.e. a transistor with a raw beta of 100-200 becomes a combination with a feedback beta of 20-22.
All this knowledge will be built in to any design of any quality these day.
A typical 5W audio amp will have a pair of temperature compensating diodes and a resistor across the complementary bases, and a one ohm emitter resistor for each transistor.
Transistor design is not valve design. A valve is typically a much more linear device with a predictable transconductance and it drives a loudspeaker via a sub optimal transformer that limits the amount of negative feedback that can be applied overall. Transistor amplifiers have far more gain available, but are far less linear, but use of massive amounts of negative feedback both locally and overall is possible to get the performance.
The only time I have ever seen mismatched output transistors cause perceptible distortion was in a Sinclair 12W unit supplied in 1967. We tested the gain of the output transistors (both NPN)... one was 5, and the other was 20. The usual Sinclair trick of using reject transistors bought as scrap..so far below spec as to be unusable...the circuit design was OK, but not with out of spec transistors.
Oh and by the way the power dissipation of a 5W audio amp of a push pull nature is something like 2W works case split between the output transistors - so 1W apiece.
Which manufactures data sheets? Some manufacturers may have tighter parameters and de-rating.
Also after manufacture devices could have been selected for better performance and although given the same generic device number they would have additional prexix numbers/letters.
Fairchild BD140-16S hfe = 63 to 160 BD140-16STU hfe = 100 to 250
Phillips have two designations BD140-10 and BD140-16 which match the two hfe ranges as the Fairchild.
mailto : news {at} admac {dot} myzen {dot} co {dot} uk
That's like an Angry emoji on Facebook isn;t it ;-)
BTW totaly agree on the Hfe of transistors and other ratings are usually minium specs. I have some demo metal clad 100W resistors , but they are only rated at 100W provided they are mounted on 995cm^2 3mm thick piece of aluminium , I assume that's in free air with no fan etc..
As NT has suggested, beta is a secondary consideration, As a general rule higher the better. Your amplifier is designed to have voltage gain, where if anything the input resistance will depend on the overall current gain.
Perhaps share the circuit? There are also some very simple single package amplifiers that can drive a few watts.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.