VAS and driver Transistor?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Generally speaking for VAS I'd recommend the 2SC3505/2SA1381 or the 2SC3423/2SA1360 pair.
Search for devices with high Vce0 (150V min.), high beta (100 min) and good beta linearity, low Cob (<10pF), and high Ft (min. 100MHz).
As driver my favourite is the 2SC4793/2SA1837 pair.
For low-power stages, say LTP I prefer the 2SC2240/2SA970 devices, and for any other low-power stage (reference, protection, etc...) I use the 2SC1815/2SA1015 pair.
maxpou said:
Thank you for reply Andy,
but i have a other question. If i change the VAS and driver transistor for more speed transistor is it possible to burn my amp? Thank You! Maxpou

Of course! You'll need to redesign your compensation circuit (oops I cannot find any compensation capacitor in your schematic).
Have you got any pic related to the square-wave response of your amplifier? I'm interested how does your amplifier works with 10...20kHz squarewave input without any lag- or lead compenastion.

Is your design stable now? Any oscillation(s)?
Just bump the Vas current to 40mA and put a clip-on heatsink on the 2N3440/5415 and skip the driver stage.

The Ft on the Vas transistors is over 30Mhz at 40mA, and 40mA is enough to drive those outputs. You can do the resistor pair for feedback right off the bias network (like Nelson Pass did on the Nakamichi stasis receivers).
For a VAS: 2sc2705, 2sc3423 and 2sc3953 (and siblings) are hard to beat. I
also like obsolete NEC 2sc639 (TO-18) with fT=750MHz and Cob=1.5pF - but no Spice models or datasheets are readily available, and Vceo=40V only.

Also repeating my request for a decent 2sc3953/ksc3953 Spice model. It sounds great in actual listening, but I want to get some simulated numbers for H2/H3 etc.
Depending on the stability of your feedback loop, it is conceivable that you can cause it to become unstable after substituting a different VAS transistor. Usually this happens because the new transistor has higher *gain*, not higher fT.

If the instability shows up as a high-amplitude self-oscillation, it is possible that you can burn up the Zobel resistor, output transistors and/or emitter ballast resistor(s). As always, it pays to have a quick-blow fuse protecting the output stage if you want to try these kinds of experiments.
To test for stability,

Simply power up the amp through a 60W light bulb on the mains side, or better yet, a variac.

Amp stability is intimately affected by rail voltage. Reduce rail voltage to about half (making any adjustments to ensure output offset is within a volt or two) and keep an eye/finger on the Zobel resistor (typically 10R 1W) as you raise the rail voltage.

With an 8R//100nF load connected to roughly simulate a speaker, this test is never destructive, and will point you in the right direction as far as compensation is concerned.



Joined 2004
Paid Member
fab said:


With 8R and 2.2uf at 1 KHz, it should be damped a lot more quickly...

Anyway, you can test only at 10 KHz for that purpose. Your waveform at 10 KHz could be more damped opinion...

Good luck

Also, if you re-adjust compensation caps do not forget to verify other caps value like explained in

Maybe you could consider a Zobel...

Good luck
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.