Looking for Transistor vendors, low and medium power

[JensRasmussen]

Hello All,

I have restarted a project that has been shelved for some (many) years, and I'm now moving from concept to implementation.

I'm looking for help finding suitable BJT transistors for Input, VAS, Cascode and Pre-drivers.

So far I have some candidates

Input:
NXP BCM846DS/BCM857DS matched pair.

I'm happy with these, but they need to be cascoded to work in higher voltage applications.

VAS, Cascode, Pre-drivers:
NXP BF820/BF821
NXP BF722/BC723
NXP MMBTA42/MMBTA92
NXP PZTA42/PZTA92

I'm not 100% happy with any of these, and would like advise on something more linear.

I have looked at the Onsemi KSC3503/KSA1381 pair, but I have not been able to find these with matching HFE for the NPN and PNP versions. (I know they are in the datasheet, but I have not found them available anywhere)

Output stage:
MOS based, please post ideas! 🙂

Please share your favourites and don't hold back on vendors, I'm somewhat limited by what I know from European and US vendors, and have very limited knowledge about the Japanese vendors.

 

[RM]

Without a schematic this is almost a plea for tomorrows menu! Fleskesteg! Mmmmm

R

 

[bucks bunny]

Any good design must not rely on exotic parts.
Just my 2c

 

[JensRasmussen]

Hey,

I'm after transistors that have these graphs in their datasheet:

The flatter the curves, the more linear the part is going to be. My problem is that most datasheets these days don't have this kind of information.

Another question could simply be what vendors have datasheets where this information is given?

\\\Jens

 

[rikkitikkitavi]

NXP have Spice models for their BF parts, most of them atleast. That would make simulation feasible.
Sorry for the bump

 

[Citizen124032]

The flatter the curves, the more linear the part is going to be.

Really? They tend to tie together at the Early voltage, so not one single bjt will be linear.
The left characteristic has a strange tilted line at 100µA: the bloke who copied it from the scope with a transparent sheet shifted it, and back in position @ 120µA. Or even worse, if the bjt behaves like that, there's a hidden nasty nonlinearity around that base current value.
The right characteristic does not look 'right' to me: 40µA is flatter then 30µA and 50µA - rarely seen or another quirk.
Is your intended design depending on the supposed linearity of the involved semi's? You aim for a non-feedback design?
Anyhow (10 cents with duckduckgo):
link 1
link 2

 

[stigigemla]

Generally bipolar transistor have flatter ic/vce the lower the Hfe is. But if you are using them in common base configuration as for example a cascode the the smaller base current seems to equalise the greater procentual shift in it. With very high impedances the capacitances influences more and more.
To see what is best for your project we need to see the schematic.
The 42/92 are the same transistor, just in different housing. You can have them in SOT23, SOT89 and SOT223 just as BFx21/22.
The high voltage transistors often have a high lowest voltage in high speed switching. You have to test them to see how Hfe are sinking with low voltages and fast pulses. When they were used as video output for cathode ray tubes in tvs we had to leave about 10 volts minimum or the picture diffused in the high light parts.
To see what is best for your project we need to see the schematic.

 

[Alexandre]

To see what is best for your project we need to see the schematic.

Yes, for example if the input differential has R as collector load driving BJT VAS, then the Ic x Vbe curve is more important for the VAS (and I would use a cascoded VAS, if possible Baxandall/Hawksford cascode).

Agree on everything you wrote, and I appreciate these insights coming from your practical experience (with video circuitry).

Thanks,
Alex

 

[Alexandre]

@Citizen124032 I need to finish my design of a curve tracer to use with analog scope. I need it to have a return trace, unlike the digital tracers.
Other than selecting linear parts, with adequate Early voltage, it will also allow to see the quirks you mentioned (I have also noticed this in many datasheets).
It's also interesting to get an idea of transistor thermal behaviour / "memory" (return trace should not be distinguishable, the traces should be sharp).
Spot checking of the Hfe can also be helpful, it would be interesting to check at both the highest and lowest Ic that will be solicited from the transistor, at the specific operating voltages.

Does anyone check the capacitances, to make sure they conform to specs?

Thanks,
-Alex

 

[Citizen124032]

I need it to have a return trace...

Acquired a Leader LTC-905 curve tracer when it became available as 2hand, low priced.
But as with TV's, the backward swing is suppressed & very swift, hard to see.
Better estimate the Vearly based on what's available.