It's 2010 and what's your fav VAS transistor?

[prairiemystic]

It's 2010 and what's your fav VAS transistor?

The requirements might be:
high Vce (rail-to-rail) say 120V-200V, Ic ~ 1A
high linearity
medium power dissipation, say <25W
medium Ft say 15Mhz up

Higher Ic and PDiss mean larger (slower) junctions, so the TO-220 packaged parts are out. I usually see the TO-126 (TO-225) package for these parts. TO-92 is too tiny.

Not many North American offerings. On-Semi offers the venerable MJE340/350, ST the 2N3439/5415 in the old TO-5 can. The BD139/140 are low in Vce like the 2N4921/23. The TIP series are dogs, IMHO.

I haven't looked at the Japanese (Sanyo/Toshiba) offerings, which are surely superior but tough to procure as originals.

[Workhorse]

Better to use cascoded VAS, Fast CE like BC546 and Slow CB like MJE340 makes great combination.

[infinia]

Fairchild has many offerings instead of 2Sxxxx they use KS
For Vas I look for low Cob and high linear gain > rather than higher Pd > 5-7 Watt is plenty
KSC3503ESTU http://www.fairchildsemi.com/ds/KS%2FKSC3503.pdf

[CBS240]

Originally Posted by Workhorse

Better to use cascoded VAS, Fast CE like BC546 and Slow CB like MJE340 makes great combination.

But.... the VAS should not be required to supply much current. To add an extra current gain stage in return for using small signal devices for the VAS is a good trade off, IMHO.

For VAS, I use THAT matched arrays for the current amps and cascode SOT-563 duel mosfets as the voltage amps (common gate comp. bridge); bias of each element = 2mA @ 25V. 22V swing, but being full bridge, 44V. In my case it is followed by HEC OPS, or basically a triple, so plenty of current gain. Since the 'OPS circuit' only requires ~8uA of current it doesn't bog down the VAS.

Cascode allows the exploitation of characteristics from two different types of transistors. Of course it's a bit more in depth than that.... but unless this is your first project or a learning exercise I believe the small addition of this 'complexity' should be regarded as nothing considering to the results.

[lineup]

Originally Posted by Workhorse

Better to use cascoded VAS, Fast CE like BC546 and Slow CB like MJE340 makes great combination.

Good idea, solution, Workhorse.

Cascoding will add some components making circuit less simple.
But performance there are several good things to gain.
Not only possibility of higher voltage.

Questions.
BC546 you take for example.
Why not lownoise BC550?
I could guess, that for VAS the low noise is less important.

In my diagram I use 2N5551/2N5401 + MJE340/MJE350.
Say I use something like 10-20 mA in VAS stage.
Would those transistors be better or not, compared to BC546/BC556?

[Aldotin]

Originally Posted by Workhorse

Better to use cascoded VAS, Fast CE like BC546 and Slow CB like MJE340 makes great combination.

Someone told me that cacoded VAS only works fine with triple output stage. For the simple emmiter follower driver+output it's not recommended.

Your opinions?

[Workhorse]

Cascoding offers many advantages.

1. CE transistor gives you stable gain because of near constant VCE
2. Miller effect is eliminated, more gain at HF and better linearity.
3. CE transistor also enjoys stable thermal equilibrium and hence current drift is reduced in VAS.
4. CB or CG stage gives you one advantage of using mute or shutdown function if symmetrical complementary VAS is used.
5. CE transistor is easy to drive by previous stage due to its high HFE, less loading .


Lineup,

i agree with you, but BC550/BC560 are no longer in production by NXP. Other sources i dont know, but we have BC549/559C having low noise performance as replacements.


CBS,
VAS current normally is between 5mA to 20mA, depends on driver stage input current requirement, if you have Triple EF output stage, 5mA for VAS is enough.


regards,
Kanwar

[Workhorse]

Originally Posted by Aldotin

Someone told me that cacoded VAS only works fine with triple output stage. For the simple emmiter follower driver+output it's not recommended.

Your opinions?

Not really,

Run the Cascode pair with 20mA idle current

[prairiemystic]

The MJE340/350, an old part with many changes. 2003 die design change (to perforated emitter) and then a die shrink, like most power transistors at On-Semi. Assembly in China, fab moved to Malaysia.
ON Semiconductor
So it's different part, too bad they didn't make something new.

[lineup]

These DPAK package MJD340/MJD350
is a new variant similar to MJE340/MJE350.
They might be easier to find as they probably are replacing MJE340.

ON Semiconductor MJD340: High Voltage Power Transistor
ON Semiconductor MJD350: Power .5A 300V Discrete PNP

Designed for line operated audio output amplifier, switchmode power supply drivers and other switching applications.

* Lead Formed for Surface Mount Applications in Plastic Sleeves (No Suffix)
* Straight Lead Version in Plastic Sleeves (-1 Suffix)
* Lead Formed Version in 16 mm Tape and Reel (T4 Suffix)
* Electrically Similar to Popular MJE340 and MJE350
* 300 V (Min) VCEO(sus)
* 0.5 A Rated Collector Current
* Pb-Free Packages are Available

Regarding MJE340/MJE350 they are not as good as e.g. BD139/BD140 as drivers.
But just like Workhorse tells, I have seen MJE340/MJE350
several times used for Cascoding in very good amplifiers VAS.
So for drivers there are better choices, but for Cascoding these are very alright.