VSSA Lateral MosFet Amplifier

[Lazy Cat]

two CRDs instead the 470ohm resistors (input pair's collector to rail), improve THD by 30%, mostly 2nd order.

Not the right place for CRD. 15k injection resistor can be replaced with CCS. 470 ohm determine input stage's voltage gain, so by replacing it with CRD would make extremely unstable and nonlinear transfer function. Sims swallow many unrealistic solutions.

 

[Lazy Cat]

Not fully satisfied with data i found about small heatsink for VAS i decided ti swap to ones well known and that i have in stock. Miller caps has swap to RF SMD type the more close i could to Vas transistors legs.

Good to use four 1mF as in sch. Nice PCB.

 

[sonnya]

What do you mean? To combine two amps? Please elaborate Sonny.

At VSSA we are fixed to six transistors only, as promised.

Sorry i forgot about the thread saying only siz transistors.

The only reason to use this output section, is that the current is much lower in every part.

I think with your setting you are at about 20watt per channel instead of 4 - 8 watt per channel!?

But forget about what i have wrote.

 

[Lazy Cat]

I think with your setting you are at about 20watt per channel instead of 4 - 8 watt per channel!?

Two 50 kHz graphs clearly shows 50 Wrms/8 ohm at +/-32 V rails. Settings are: 1,8 mA input stage bias, 15 mA VAS bias, 100 mA output bias and all resistors values as in the VSSA 1.2 schematic. VAS transistors on the main heatsink, TLVH431 as bias spreader set to 1,565 V, Miller caps 12 pF.

 

[Samuel Jayaraj]

LC, can you please point to the schematic of the bias spreader using TL431. Thanks.

 

[tinitus]

regarding onboards caps

new develloped small lytics can have very low ESR and high current handling

I would say no poly caps needed, only one lytic cap per rail, thats all
and the VSSA board will look neat and simple

get rid of all those stupid caps
just my personal opinion on it, ofcourse

 

[Lazy Cat]

LC, can you please point to the schematic of the bias spreader using TL431. Thanks.

VSSA having VAS transistors on the main heatsink, because of the thermal considerations, require TLVH431 as bias spreader. Once bias is trimmed to proper value, 1 k trimmer can be measured and replaced with two fixed resistors. Vgg fixed to 1,565 V in my case, enabling 100 mA output bias current.

 

[Idefixes]

Latest version with VAs on their own heatsink. 1mF package swap from 12.5mm/5mm legs spacing to 16mm/7.5mm legs spacing because of bigger and easier purchasing possibility. I swap zobel network under board with SMD parts.

 

[Samuel Jayaraj]

LC, thank you for the schematic. Would you like to give a brief comparison of sound quality between VSSA, TSSA and SSA(IGBT)?

 

[tinitus]

neat looking board
looking at small changes
if you move one lytic up, and make the other smaller, you might be able to shrink the board maybe 20%
maybe it would help to make it just 10mm wider

your Vas heatsink design appear somewhat 'space wasting' ?
maybe consider making 'optioanl' mountinghole, somehow
would make optional mounting on main heatsink possible

 

[tinitus]

but actually, I would suggest to design a universal driver board instead
to hook up with all the different output stages

 

[tinitus]

well, since I have my own funny ideas...more to come

 

[Lazy Cat]

well, since I have my own funny ideas...more to come

I'm planning to design specific VSSA PCB, maybe we are on the same track ...

 

[tinitus]

adding a small PCB...just to illustrate the basic idea

 

[Lazy Cat]

two CRDs instead the 470ohm resistors (input pair's collector to rail), improve THD by 30%, mostly 2nd order.

Did you try to sim with 1N5297 or similar?

Question is thermal stability and VAS Iqc stability if 470 ohms are replaced.

 

[Lazy Cat]

adding a small PCB...just to illustrate the basic idea

... so still small separate on-board heatsink for VAS. I'm planning to go to the main heatsink with VAS, just like in my test setup.

 

[pergo]

current in the 470ohm resistors is quasi-costant. Very low variation, sine shape (distorted sine).
With CRD or another CCS there's no variation at all.
Sims are pretty good but i will try a temperature sweep up 70°C.

 

[tinitus]

... so still small separate on-board heatsink for VAS. )

output source pins will be hardwired ... look at it 3 dimensional, instead of one plane

 

[Lazy Cat]

output source pins will be hardwired ... look at it 3 dimensional, instead of one plane

ALF pins connected to PCB with wires? Not very good idea.

 

[tinitus]

I have Renessas, not ALF

anyway...gate and drain pins are connected directly to board