advertisement


Hackernap parallel output transistors

Pete MB&D

Pete Maddex, the one and only!
Hi Chaps

My Hackernaps are based on my 135 clones so I am running +-56V on the output stage so a but close for comfort.
I had planed to add parallel output transistors to give me some peace of mind, and after the success of the NCC300 and it spurred me on.

I thought the easiest way would be to use a piece of U section aluminum to mount all 4 transistors with a small board to connect each pair up and add the extra link to another emitter resistor.
I use a small board with 3 1R 1W resistors for the emitter resistors and a wire from the second output transistor.

Output transistors by Pete Maddex, on Flickr

Its a bit tight in there!

Hackernap by Pete Maddex, on Flickr

I got the first one done and it sounds very nice, I had to tweak the bias which drifter a bit as it warmed but it was nice and stable as confirmed by the scope and generator.

I was doing the second one when everything started to look blurred, I had a silent migraine a nice C of flashing lights that slowly gets bigger and moves out until it disappears into my perifferal vision, so I stopped.

Pete
 
The amount of power for a single pair of transistors to handle is more than recommend, you could overload blowing them and the speakers.

This way the work is shared between the pair.

Pete
 
Are you using MJL4821? from the SOAR curves you are at around 3.5A max current at DC with no temperature derating, so yes I agree it's a good idea. At the minute I am only using 2SC5200 in mine at 47V and will get round to replacing them with 4281 or 2SC2922 or similar. Mind you mine only drives full range speakers (no crossover) and never gets played loud for long.
 
Yep MJL4281s mine drive a pair of Arcs which like a bit of current, one of the reasons I decided to do the mod.

I need to get the other one finished and back into the system it running with a B&O ICE amp at the moment.

Pete
 
Current sharing between the output transistors, you don’t want one working harder than the other, the emitter resistors do the same thing. The more current through the emitter resistor increases the voltage across it making the voltage across the base-emitter junction lower effectively lowering the signal and reducing the current flow.

Pete
 
Small-value resistors in line with the base connection of power transistors is a different thing entirely: it is about avoiding HF (MHz range ) oscillations by gyration / negative input impedance. Sugden show 2R2, and usu values for such things aare indeed in the 2r2-4r7 range.

It's very hard to find a decent analysis of this, but the best found easily online is by Dennis Feucht (eg: http://audioworkshop.org/downloads/AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf) - there's a part 2 well worth reading if you look for it.

ETA: if that pdf looks scary, it boils down to this: at high-enough frequencies, an emiiter-follower can behave like a Colpitts oscillator, depending on what is on each of its leads. For useful, stable amplification - adding a bit of real resistance in the base connection can totally damp this effect out. Specific to our context - it's greatly preferable to not using such a small resistor, for example - in the hope of 'lower output impedance in the audio band' (which is nonsense for a a feedback-controlled amplifier anyway)
 
Small-value resistors in line with the base connection of power transistors is a different thing entirely: it is about avoiding HF (MHz range ) oscillations by gyration / negative input impedance. Sugden show 2R2, and usu values for such things aare indeed in the 2r2-4r7 range.

Grid stoppers.
 
Small-value resistors in line with the base connection of power transistors is a different thing entirely: it is about avoiding HF (MHz range ) oscillations by gyration / negative input impedance. Sugden show 2R2, and usu values for such things aare indeed in the 2r2-4r7 range.
Interesting (and helpful, thankyou). It happens that the monster Sanken transistors Sugden used run all the way to 50Mhz. I wouldn't want 100W of that launched at my innocent tweeters.

BugBear
 
First listen on Sunday after finishing the last one was completed, they sound very good, much better on complicated stuff and they control my arcs much better in the bass department.
They should be nice and warm when I get home.

Pete
 
First listen on Sunday after finishing the last one was completed, they sound very good, much better on complicated stuff and they control my arcs much better in the bass department.
Well, that's interesting. (*)

Your stated reason for the change was piece of mind, because the trannies were close to their limit.

Was the sound improvement a secondary goal, or is it a happy surprise?

BugBear

(*) (c) Capt Jack Sparrow
 
The overload was the main reason behind the change, but after hearing how good the NCC300 is I decided to pull my finger out and go for it, plus I has worked out how to implement it easily.


Pete
 
Sounds very good now they have warmed up, I am hearing lots more detail and clarity as well as the deeper and better bass.
Parallel your output transistors NOW!

Pete
 
Sounds very good now they have warmed up, I am hearing lots more detail and clarity as well as the deeper and better bass.
Parallel your output transistors NOW!

Pete
So the only electrical change should be that the current in (each) transistor is halved/shared. Do you think this is putting the transistor into a better (more linear?) part of its characteristic curves?

BugBear
 


advertisement


Back
Top