advertisement


Matching Driver and Output Transistors for the NAP/Avondale Designs

O

OzBrit Audio

pfm Member
What do people think about driver and output transistor matching in the Avondale NCC220/300/NAP designs. More particularly what benefit does this have on the sound quality?

I am particularly interested in the complementary designs where matching HFE is difficult regarding differences between the NPN/PNP devices re the MJ15003/4 seems nigh on impossible. My latest batch of 5 of each showed NPN average HFE 45 and PNP 175 so matching becomes rather difficult!

On the driver side the MJE15030/31 devices have similar problems although probably easier to match than for the power transistors.

My thoughts are for single output transistor designs (NCC220) match on Vbe and for dual transistor (NCC300) match each pair on Hfe. And for the driver transistors match on Vbe.

So what do people do/think on this issue. And have any of you had experiences you wish to share on how it affects sound quality?

Cheers
 
Let's put it this way - Naim were sticklers for matching everything that needed it, and never bothered with this on their regular amps. I believe the NAP500 and statement use multiple devices and I think they're matched.
 
My understanding is that HFE matching is fairly irrelevant in these designs. The output stage acts as a big unity gain buffer, and HFE hardly changes that, over fairly broad ranges.
 
MJS said:
Let's put it this way - Naim were sticklers for matching everything that needed it, and never bothered with this on their regular amps. I believe the NAP500 and statement use multiple devices and I think they're matched.
Click to expand...
Thanks for all the replies

Interesting so Naim only match the input devices TR1 and TR2 for HFE on their mainstream amps, even their quasi complimentary designs they don't bother matching the output devices?

But for paired output devices what do you think best to mach on HFE or VBE?
 
OzBrit Audio said:
Thanks for all the replies

Interesting so Naim only match the input devices TR1 and TR2 for HFE on their mainstream amps, even their quasi complimentary designs they don't bother matching the output devices?

But for paired output devices what do you think best to mach on HFE or VBE?
Click to expand...

When you say paired, you mean parallel? In that case, depends on the details of the circuit. Broadly, if the drive environment is low impedance (voltage controlled), then I think VBE matching matters most. If the drive is high impedance (current controlled), then HFE matching. This is to optimise current sharing. Doug Self argues somewhere that you can reduce crossover distortion by not having an exact match, to blur out the crossover region!
 
I have just implemented dual (read parallel) output transistors on my NCC220's along with a current mirror for TR1/TR2 bringing them closer to the NCC300. I must say I"m quite impressed with the results - I can see why people get excited about this amplifier design.
One thing I have noticed however is that the bias currents on one amplifier don't match well. What I mean is if I check the voltage across the 0.22R resistors for the original output pair I get say 12mv for the the NPN side and 16mv for the PNP side. Obviously this is offset by the currents flowing through the second set of output transistors as what goes in must come out as they say. But I am wondering if this really matters - I know that the heat distribution won't be equal across the four transistors but will this affect sound quality?

I thought I matched on Vbe best I could using a cheapo multimeter but am wondering if I am best investing a proper transistor tester. I think Peak do quite a good one thats tests for Vbe gain etc.

I will probably start a separate thread on this upgrade project posting pics etc but am wondering what people think on here re the output transistor matching especially those who have built the NCC300 boards?

Cheers
 
Cheapo multimeter will only use tiny currents (circa 1mA) which is not enough to match Vbe values on power transistors. You'd want something around 100mA collector current TBH. I think that PD said elsewhere that Doug Self indicated that splitting the bias currents unevenly in parallel output pairs distributes the switching noise slightly, which is a good thing. You will probably find that the other pair measure in reverse, i.e. 16 and 12mV or thereabouts which is about 60mA through a single 0R22 string so you're pretty much on the nose for the 120mA total.
 
Thanks again for the replys. I’ll think I’ll leave as is then and continue with my build

As I mentioned earlier quite impressed with the results so far. Well worth the effort

Cheers
 
You must log in or register to reply here.


advertisement


Back
Top