SX-Amp and NX-Amp

www.hifisonix.com
Joined 2003
Paid Member
I know that 3503/1831 are very good for VAS but for drivers can I use 2sb1086a-Q/2sd1563a-Q with matched HFE of 205 instead (to spare my 3502/1380)?

Thanks

Fab

I am not familiar with these fab - I would need to look at the data sheet.

What is important:

Very low Cob
Vce(max) >100V
hFE reasonably high, and within 10% or so match on complementary pairs
Good Ft

Lower Vce(max) devices generally show a lower Early voltage and higher distortion. However, their hFE tends to be higher, and in the big scheme of things this offsets the Early voltage issue somewhat.

I'll take a look at the data sheets in a few hours time and give you my thoughts
 
www.hifisonix.com
Joined 2003
Paid Member
I know that 3503/1831 are very good for VAS but for drivers can I use 2sb1086a-Q/2sd1563a-Q with matched HFE of 205 instead (to spare my 3502/1380)?

Thanks

Fab

Fab,

I took a qucik look. These will make excellent drivers or pre-drivers, but they are not suitable for serious VAS work. The Cob is very high at 30pF and the Ft is also low at 50 MHz. 30pF is a typical miller comp cap value in a VFA amp, or a big portion of the the Alex comp cap value in a CFA. They will result in high distortion compared to the types I spec'd.

My suggestion is that you do not use these for VAS/TIS.

The ideal transistor for this type of application is

  • Cob <3pF (the good ones are at <1.8pF)
  • Vce(max) 200V with >300 better still (high early voltage
  • hFE - per earlier comments, need to be matched to within 10% on complementart pairs.
  • Ft 100 MHz or better
 
From the njw1302 datasheet the HFE is a flat line from 10ma to several amps...

Fab

Yes, it is, you are correct.
You should be measuring Vbe at your intended operating current for matching paralleled transistors.

Single pair drivers do not need to be Vbe matched. They do benefit from a close hFE, 15% is probably adequate. I would try for 5%. It affects the current coming from/to the VAS output.
 

fab

Member
Joined 2004
Paid Member
Fab,

I took a qucik look. These will make excellent drivers or pre-drivers, but they are not suitable for serious VAS work. The Cob is very high at 30pF and the Ft is also low at 50 MHz. 30pF is a typical miller comp cap value in a VFA amp, or a big portion of the the Alex comp cap value in a CFA. They will result in high distortion compared to the types I spec'd.

My suggestion is that you do not use these for VAS/TIS.

The ideal transistor for this type of application is

  • Cob <3pF (the good ones are at <1.8pF)
  • Vce(max) 200V with >300 better still (high early voltage
  • hFE - per earlier comments, need to be matched to within 10% on complementart pairs.
  • Ft 100 MHz or better
Hi Bonsaï

I do intend to use the 350x/183y pair for VAS and my question was only for pre/drivers before the output stage. So then they are correct?

For my 3502/1830 my new arrived stock from Ampslab allows me to get a match within 3% HFE.

Thanks

Fab
 
Last edited:

fab

Member
Joined 2004
Paid Member
You should be measuring Vbe at your intended operating current for matching paralleled transistors.

Single pair drivers do not need to be Vbe matched. They do benefit from a close hFE, 15% is probably adequate. I would try for 5%. It affects the current coming from/to the VAS output.

Are you talking about the output trannies or the drivers of these trannies?

Fab
 
The PDTA115ET,215 is available from Newark on cut tape, as long as they have a partial in house, but only IF you call and ask for a quote. Very quick process, actually, once you figure out the rules.

(Not normally available except by the reel.)

I am ordering 100 pieces, and they will go out with the other transistors, to everyone who got the boards, and with the partials I am sending to those who asked. Just added them to another order, so there is no extra shipping, just a PITA to have to get these separately... :rolleyes:
 
www.hifisonix.com
Joined 2003
Paid Member
PMI,
These were base stoppers I fitted for Q12 and Q13 in order to prevent the possibility of HF parasitic distortion.

They were not needed (verified in tests) so you fit the 0 Ohm jumper. If you are handy with your soldering iron, you can also just 'flash' them over with a blob of solder.

I will add further explanation in the write up when I update it after all the feedback is in.
 
You should be measuring Vbe at your intended operating current for matching paralleled transistors.

Single pair drivers do not need to be Vbe matched. They do benefit from a close hFE, 15% is probably adequate. I would try for 5%. It affects the current coming from/to the VAS output.

But my original question related to output HFE % matching...:D
Do not worry sometimes peoples here do not understand my posts until I repeat the same thing twice...;)

Fab
Look again at my post.
You posed a question, PMI replied, I said "you should be measuring Vbe at your intended operating current for matching paralleled transistors."

Then I added that drivers are done differently, just in case you thought that my answer applied to both outputs and drivers.
 
I know I am confused, although I am slowly getting used to it... ;)

More important, someone who reads this in a few weeks might be also, and it would be nice to leave some useful tips behind for the next group of nx-Amp builders.

I assume that for the purpose of this thread, parallel transistors refers to the output pairs, NJW3281/NJW1302 (Q1,2,4,5). And, we are calling them parallel even though they have individual 0.33R emitter resistors.

I also assume that "drivers" refers to Q12, Q13, KSC3503/KSA1381, the same transistors where the base resistor is shown as 33R in the schematic and 0R in the BOM.

@Fab&Andrew: if you would care to share some of your reasoning about the transistor selection, and for matching them, it would be appreciated, by me for certain, and probably by others reading this later.
 
I think I got it!

Q6, Q7 (VAS) should be as specified KSA1381/KSC3503 and well matched if possible.

Q12, Q13 (drivers) does not have to be high spec like KSA/KSC pair and does not have to be well matched (but won't hurt if they are high spec and well matched)

Q1, Q2, Q4, Q5 (output transistors) does not have to be well matched because there are 0R33 emiter resisitors (and matching them would require huge quantity and hence a lot of money)