What are the pros and the cons of those 3 vas ?

[tessier]

Hi

Between the beta enhance vas, the buffered vas and the cascoded vas, what are the pros and the cons of those 3 vas ?

Thanx

Paul

[CBS240]

Hi

Some of my ........................

The beta enhanced has a much higher input Z and your input stage likes this very much but it adds another stage within the closed loop. The buffered VAS allows you to use a small signal transistor for the VAS which tends to have better gain linearity and higher GBW. The cascode VAS requires extra voltage for the supply rails in order to get around the lost cascode voltage. However cascode has wonderful benefits. For starters it allows you to use a lower Vce small signal device for the current amplifier and this brings out all kinds of possible components. Lower Vce transistors tend to have higher Hfe, much faster GBW and lower noise, and it does not add an extra stage in the CL. Also having a constant Vce not only will linearize the AC load line, it inhibits the Early effect. There are also benefits to having the voltage amplifier being common base instead of common emitter. Certainly as far a quality is concerned, cascode is the best way to go for sure, but like I said, it needs a boosted supply voltage to get rail to rail swing at the output stage and this adds complexity to the overall circuit. There are always trade-offs.

As for my amp, I use a voltage doubling circuit that then looses several volts in a series-shunt voltage regulator for the low current VAS and input stage supply placing it about 10V greater than the output stage rail. It's not a terrible addition to complexity but is well worth the benefits.

[Bigun]

the middle option looks a bit like a Sziklai - offers some promise of good linearity.

[CBS240]

....or a common emitter driving a common collector (emitter follower) judging from where the Miller cap is placed.

[Bigun]

yup, but perhaps all we need is to add some emitter degeneration to the first device and fold the collector of the 2nd device back above the resistor to create that local feedback loop. Cdom offers some scope for experimentation, perhaps 2-pole ?

[AndrewT]

If Cascode is adopted, it must drive a high impedance load, either a buffer or a 3stage output.

[balaboo]

Andrew is absolutely correct - a small common-emitter stage coupled to CFP emitter-follower is the way to go (think Baxandall super-pair).

[CBS240]

If the amplifying transistor has Vce of 5V and a med pwr device is used for the cascode transistor, 40-50mA bias could be used. Do you need a VAS with more current than that? A small signal TO-92, even with Vceo less than 25V could easily be the current amplifying transistor, perhaps with Hfe greater than 300. This would negate the need for the extra current gain stage and still only require ~150uA of DC bias from the input stage. You could even use J-fets. The biggest issue is the extra boosted voltage supply required to make up the cascode voltage difference. This may be the reason to go with one of the other solutions but cascode really is hard to beat in terms of performance.

[gaetan8888]

Hello

There is also the VAS with a collector buffering, like in the image.

I never try this one, do we connect the collector of the buffer transistor to the upper rail ?

Anyone did try this one, any amps who use this VAS ?

Thank

Bye

Gaetan

[AndrewT]

D. Self recommends it for good performance.
He chooses not to use it in his published designs because it uses more components and takes up more PCB space than the simple EF before the VAS.