Jellybean hobbyist general purpose transistors..

[floobydust]

1970's Tektronix tested and ranked all transistors and diodes; ranked by hFE, leakage current, avalanche voltage etc.

Some manufacturers have crap datasheets and I buy from those that have good information, like On-Semi and NXP.
Diodes Inc datasheets are terrible, missing tons of the usual specs. If a spec is absent, it's not tested so buyer beware.

Example BC846A datasheet:
Infineon 13 pages, 17 graphs
Fairchild 7 pages, 17 graphs
Taiwan Semi 6 pages, 12 graphs
On-Semi 13 pages, 10 graphs ea.
NXP 15 pages, 8 graphs
Diodes Inc 7 pages, 3 graphs

[Vtile]

Is there a general rule of thumb for high(ish) voltage (ie. 100V to 140V range), but low leakage (Vce and maybe Vbe) BJT transistor type. ...Or is the higher voltage the rule of thumb to find lower leakage type.

** Low leakage, <15nA **
**When searching that transistor model***

PS. If we still try to stay on "general purpose" category.

[David Hess]

Is there a general rule of thumb for high(ish) voltage (ie. 100V to 140V range), but low leakage (Vce and maybe Vbe) BJT transistor type. ...Or is the higher voltage the rule of thumb to find lower leakage type.

** Low leakage, <15nA **
**When searching that transistor model***

PS. If we still try to stay on "general purpose" category.

Leakage depends more on area, construction, processing, and temperature than voltage rating.  For small signal transistors, it is only tested down to a level of 10s of nanoamps because that covers almost all applications and a more sensitive test would take longer.

Voltage rating has more of an effect on recovery time.

A thorough search will turn up transistors tested down to 10 nanoamps but I think of low leakage as 10 picoamps and lower which I have had no trouble finding in common parts like the 2N3904.

[exe]

If it meets the datasheet, and your application works based on datasheet information, then you're fine.

The problem is many parameters are not specified, or only "typical" values are specified. Or how this parameters change with temperature, supply voltage, etc.

[Vtile]

Is there a general rule of thumb for high(ish) voltage (ie. 100V to 140V range), but low leakage (Vce and maybe Vbe) BJT transistor type. ...Or is the higher voltage the rule of thumb to find lower leakage type.

** Low leakage, <15nA **
**When searching that transistor model***

PS. If we still try to stay on "general purpose" category.

Leakage depends more on area, construction, processing, and temperature than voltage rating.  For small signal transistors, it is only tested down to a level of 10s of nanoamps because that covers almost all applications and a more sensitive test would take longer.

Voltage rating has more of an effect on recovery time.

A thorough search will turn up transistors tested down to 10 nanoamps but I think of low leakage as 10 picoamps and lower which I have had no trouble finding in common parts like the 2N3904.

Makes sense, thx. PS. I just noticed that at least some types of PN2222 are "rated" to 10nA Icbo max. at 25 deg.C. I'm more and more confident the bulk box of cheap as dirt (they are sand after all) parts (from reputable source) is right way to go for most all non-production hobbyist needs, most critical feature seems to be that you have the stock pile to pick from. 

It would be interesting to know if the garden variety actually do have different internals between models or are they actually just from one outcome from a machine and then sorted to meet different key specifications of garden variety part numbers. I wouldn't be too surprised (without having even general knowledge of semiconductor production) if this would be the case looking at how close the specks are in many parts.

[David Hess]

It would be interesting to know if the garden variety actually do have different internals between models or are they actually just from one outcome from a machine and then sorted to meet different key specifications of garden variety part numbers. I wouldn't be too surprised (without having even general knowledge of semiconductor production) if this would be the case looking at how close the specks are in many parts.

It is common for one design to result in multiple part numbers and it happens with some ICs as well.  Bipolar transistors are often graded for Vce and of course hfe to be divided into different part numbers.  Some power MOSFETs are graded for Vds like Vce in bipolars.  Besides things like offset voltage, bias current, and offset voltage drift, many old bipolar operational amplifiers were also graded into 28, 36 and 44 volt versions.

And as mentioned earlier, some part numbers like the 2N3055 become a dumping ground for other parts which did not meet their intended specifications.

[T3sl4co1l]

I recall measuring a random 2N4401 in the low nA.  It was a cold day, YMMV.

If it's not on the datasheet, you have to test it yourself.  Fine for breadboarding, or small run test equipment, say.  Not so great for production.

Again, jellybeans aren't tested for leakage, because it's slow -- expensive to test!  There may be special-purpose designations which are, but they'll be expensive for that reason (and also the much smaller quantity of product being moved).

Tim

[Cerebus]

If it's not on the datasheet, you have to test it yourself.  Fine for breadboarding, or small run test equipment, say.  Not so great for production.

But doable nevertheless. The HP/Agilent/Keysight 34401A had a selected JFET* in the input amplifier and they made 25,000 a year of the suckers.

*Q104, selected for a given range of V_GS at an I_DS of 1.36 mA.

[SeanB]

Simple reason they do not change the spec is because the specification is for a MINIMUM requirement, that the transistor is guaranteed to meet when manufactured. That your device you bought is so much better is moot, you are buying one that gets that minimum spec exceeded.

As to different devices getting different part numbers, randomly opened the NS small signal transistor book, and looked at process 63, PNP medium power device. There the same die can be put into 5 different packages, and be variously a 2N2905A, 2N2907A, 2N4403, 2N3702, 2N3645 or a 2N4143.

Then you get the interesting ones, like process 59, a dual small signal MOSFET in TO99 package that is a whole series of FM3954 or FM1100 series.

Then process 1, which is graded into various power parts, including the venerable 2N3055.

Fun one is a MPQ6502, which is not a processor.

[Cerebus]

Wouldn't life be easier if, instead of the JEDEC/whatever part numbering scheme evolving, we'd had something like the National book. Here's the set of dies, here's the parameters we actually test and here's the bins we put them into and the part numbers for those bins.

I still find myself going to the (defunct) National databook to find a suitable part from time to time just because the layout by process makes it easier to navigate.

[Lassivv]

Try to buy some good "middle-way" mosfets on my stock.

Normally i drive something 12-24vdc (leds, motor etc) with mosfets (90% different leds).

Now i have stock
- ULN2003
- IRFZ44N
- Here maybe need one more cheap option (trough hole) model.

Then i need some SMD models. (use many hours on Mouser find and Nexperia chips seems interesting)
cheaper, not that good RDS
https://www.mouser.fi/ProductDetail/Nexperia/PSMN7R5-30YLDX/

little pricier, better RDS
https://www.mouser.fi/ProductDetail/Nexperia/PSMN2R4-30MLDX/

Find so many different voltage/Ampere/RDS models and same footprint. Quite cheap parts and because Mouser/Farnell i think they are good quality and not any fakes or something like that.

Every hints/tips are welcome. Need one more good/cheap throught hole model (like mosfets more than transistors). And what you think Nexperia chips.

[David Hess]

Wouldn't life be easier if, instead of the JEDEC/whatever part numbering scheme evolving, we'd had something like the National book. Here's the set of dies, here's the parameters we actually test and here's the bins we put them into and the part numbers for those bins.

I still find myself going to the (defunct) National databook to find a suitable part from time to time just because the layout by process makes it easier to navigate.

That would be nice and I use the old National Discrete databook for the same thing.  For instance it is especially annoying with On's (and now Fairchild's since they bought them) product portfolio because there are parts I know exist, like fast saturated switches, which cannot be searched for.  Transistors built using different processes like perforated and ring emitter structures should be labeled.  This would have helped distinguish all the variations of the 2N3055.

The same thing occurs with Texas Instruments and their various linear parts.  They often do not include even simplified or equivalent schematics so it is not always apparent how the input and output circuits work which is important in some applications.

I wonder what advantage the various manufacturer's think is provided by withholding this information.

[Wolfgang]

the 150 volt 2N5401 and 2N5551 would be more suitable.

Or the similar MPSA46, or whatever they are, from that family.  MJE350 and complement are also quite popular for audio (driver stage), though the datasheet is sorely wanting.  I think On Semi makes a detailed datasheet?  Or, there are equivalent parts with good data out there, shop around.

Did you mean the MPSA43/MPSA93?  They are a little slower.

The Motorola/On MJE371/MJE521 are the same way.  They are recommended for specific applications but lack a full set of specifications.  I ran across them when doing a search for fast TO-126, TO-225, and TO-220 parts to use in high performance regulators.

The BD135 through BD140 are like that also.  The ST and Fairchild datasheets say nothing about dynamic performance but there are SavantIC Semiconductor (who?) datasheets which say 190 MHz for the NPNs and 160 MHz for the PNPs.  What?

For HV work if often use MPSA44 and MPSA94 (400V) or MJE340 and MJE350. Medium speed, but reliable.

[vk6zgo]

How disturbing it is that VHF means very high frequenzy in a 30 MHz to 300 MHz range while every dog collar today transmit in Gigahertz range Wifi. Anyone else annoyed. 

No, as the distinctions are well known, & you  can get charts showing the division of the spectrum.
The actual terms are as they are for historical reasons, & changing them now would be even more confusing.

[vk6zgo]

Back in the day, a supplier in my home city used to have a "lolly jar" ( candy jar to NA folk) containing a huge quantity of "cleanskin" unmarked NPN silicon transistors, which they sold for a few cents each

They had the same package as the BC107, BC108 series, & would work in most of the same circuits which used those.
We called them "BC10?" Transistors.

In a repair situation, you will often find manufacturers will use the same device in both a critical & non-critical  position.
If the one in the critical spot dies, steal the non-critical one, & replace it in the non-critical spot with a generic transistor.

Another trick is to have a look at what device some other company uses in a similar position.
Sometimes, it is cheaper & has better specs.

I replaced the horizontal output transistors in a lot of Sony 27" picture monitors with Philips BU208D devices with universally good results.
Although Sony were good with parts supply, BU208D transistors were cheaper & available "anywhere"..

[Brumby]

Let's get in early for the future....!

How about:

Band name	Abbreviation		ITU band number	Frequency	Wavelength
Extremely low frequency	ELF		1	3–30 Hz	100,000–10,000 km
Super low frequency	SLF		2	30–300 Hz	10,000–1,000 km
Ultra low frequency	ULF		3	300–3,000 Hz	1,000–100 km
Very low frequency	VLF		4	3–30 kHz	100–10 km
Low frequency	LF		5	30–300 kHz	10–1 km
Medium frequency	MF		6	300–3,000 kHz	1,000–100 m
High frequency	HF		7	3–30 MHz	100–10 m
Very high frequency	VHF		8	30–300 MHz	10–1 m
Ultra high frequency	UHF		9	300–3,000 MHz	1–0.1 m
Super high frequency	SHF		10	3–30 GHz	100–10 mm
Extremely high frequency	EHF		11	30–300 GHz	10–1 mm
Tremendously high frequency	THF		12	300–3,000 GHz	1–0.1 mm
Ridiculously high frequency	RHF			3–30 THz	100–10 μm
Fantastically high frequency F***ing high frequency	FHF			30–300 THz	10–1 μm
Obscenely high frequency ** Optically high frequency	OHF			300–3,000 THz	1–0.1 μm
Preposterously high frequency	PHF			3–30 PHz	100–10 nm

** Covers visible light

[spec]

I have started to think that I stock a few hundred
 general purpose through hole transistors (an investment of a few tens of bucks). To be sure I will have them to tinker around as to me it seems that the jellybean variety is vanishing or vanished already from THTs.

I'm mostly thinking of small signal, low freq (<200MHz), throw in buffer etc. general purpose uses of tinkering and maybe slobby repair or hack here and there.

BJTs:
BC547B(npn) & BC557B(pnp) comes to my mind.
Is there other worth to look that are still readily available.
Darlington models to maybe look at?
.
JFET
Is there cheapish (<$0.50 ) replacement for 2n4117 other than mbf4117(smd). It seems to be pretty alone in fart detector range.

MOSFET
For logic level switching use. Any ideas, in leds/signals on/off range of power. Should be also as robust as possible for ESD and ideally should have the protective diodes. 2N7000 ?? or were it JFET&BJT hybrid, hmm.

Known matched pairs (for that random differential stage) that wouldn't cost arms and legs and could still be found as THTs.    ..none or should I look from IC listings..

Any other than TO-92 cases to look at for jellybean THT parts. Metal cans seems to be premium only now.

Hi Vtile

This is my list of complementary goto jellybean BJTs:
BC546C/BC556C       : 65V,  100mA, 500mW,                                TO92    (plus clip-on heat sinks)
BC337-40/BC327-40 : 45V,  500mA, 625mW,                                TO92    (plus clip-on heat sinks)
TIP41C/TIP42C         : 100V, 6A,       65W,    2degCW,     150degC  TO220  (plus clip-on heat sinks and heatsink mounting kits)
TIP35C/TIP36C         : 100V, 25A,     125W,  1degCW,     150degC   TO247  (plus clip-on heat sinks and heatsink mounting kits)
2N3055/MJ2955        : 70V,  15A,     115W,  1.52degCW, 200degC,  TO3     (You just have to have these old work-horses, but only the revised HF type [epitaxial base] )

The clip-on heat sinks are handy, to get you out of the situations where you want the performance characteristics of a smaller transistor, but slightly exceed the bare case dissipation (BC337/BC327 especially). For the bigger transistors the heatsink clips often eliminate the need for a full-blown heatsink.