Thanks for bringing these to our attention. Sanyo had a plethora of awesome BJTs that were severely culled in the last few years
...
They have lower Cob and better SOA than the 2SA1930/2SC5171, but the NPN part has substantial variation of beta versus Ic. The 1930/5171 are both ruler flat in this regard (i.e., perfect).
I have searched for drivers for a few years.
Considered the 2SA1930/2SC5171. Poor SOA won't drive paralleled outputs.
Looked at CFP pre-driver+driver pairs and even FETs.
Finally settled on individual 2SA1381/2SC3503 for each NJL4302/NJL4281 output.
They are very linear, ~200MHz fast @ 20mA, excellent breakdown Vce. exceptional VAF, the SOA is decent individually and exceptional once paralleled, and the TOTAL paralleled capacitance is still less than a TO220 driver.
I think Toni's amp could be improved with this set-up.
And the total cost is reasonable.
Maybe even use in push-pull (totem pole) but haven't simulated this yet.
The F rank ~200 Hfe are now discontinued.
I plan to buy a thousand or so, quite soon.
Unless some one has a better recommendation?
Best wishes
David
Last edited:
Finally settled on individual 2SA1381/2SC3503 for each NJL4302/NJL4281 output.
…
I plan to buy a thousand or so, quite soon.
Unless some one has a better recommendation?
I thought the 2SA/2SC prefix was no longer available. Sanyo made these parts, but so did/does Fairchild. They are still manufactured and sold as KSA1381/KSC3503. I like your approach - and these are also useful in the TIS/VAS, so worthwhile having a few.
According to the Fairchild website, the only current parts are:
KSC3503DS, KSC3503DSTU and KSA1381ESTU
It is a shame the higher beta grades are no longer available.
The datasheets are exactly the same, they even seem to have used the graphs from one to the other, there are another two even smaller packages that use the same die. Changing the die brings some benefits such as lower interconnect capacitances but these are very small differences.
Nowadays different materials are used that reduce thermal resistance so one tends to find same die parts with higher ratings. Sankens are producing TO3P output transistors that have higher SOA than the bigger package TO264.
We just need to hope that the few parts still available for audio dont get discontinued alltogether.
FYI: I have already presented a discrete darlington variant (alpha version) to play with in post #278. "asc"-file can be downloaded here: http://www.diyaudio.com/forums/soli...ab-power-amp-200w8r-400w4r-6.html#post3500615
To use other power bjt's than C5200/A1943 should be not really a problem.
For mosfet fan's there is a (alpha version) in post #114: http://www.diyaudio.com/forums/soli...ab-power-amp-200w8r-400w4r-3.html#post3485544
Both versions need some rework and finetuning and of course a real life test. The story continues ...
BR, Toni
No need to worry, KEC has also been producing these parts for a while now if ever Fairchild discontinues them. Although hard to get if you are just a DIYER I believe its possible with a group buy.
As these parts are mainly used in the voltage amplification stages there is basically no penalty but benefits to the smaller hfe. As hfe lowers, the early voltage increase.
Milestone SA2013-Rev2-RC1
All pcb's are now ready for mounting and cabling.
The case needs some work (drilling and engraving) and maybe the first prototype amplifier will be ready in the next 2 - 3 weeks for extended real life audio tests.
All pcb's are now ready for mounting and cabling.
The case needs some work (drilling and engraving) and maybe the first prototype amplifier will be ready in the next 2 - 3 weeks for extended real life audio tests.
- 2 x input/vas stages
- 2 x output stages
- 2 x power suppy
- 1 x microcontroller board (controls: inrush current, temperature, remote functions, speaker DC, speaker output relais, mute)
- 1 x mains supply board
- 1 x dual mono star ground
- 1 x power supply 12V for output relais board
- 1 x dual mono speaker output relais board
Attachments
Last edited:
Hi Harry
Walkalone is correct. The F rank is discontinued so I plan to buy one or two thousand while there is still some old stock left in the warehouse.
Thanks to everyone who commented. Sorry not to answer individually, and quicker but time zone difference you know.
Yes, I have started to work on that. I prefer to think of it as distributed driver. The drivers are on a separate heat sink so the the thermal stability is better and they do not use the same rail as the outputs either.
Nice to read positive response to the idea. I will present a version for Harry and everyone when I have a bit more tested.
What do people think of totem pole drivers?
I had planned to use Thermal Trak diodes between each driver to lower the dynamic impedance but now the Thermal Traks are discontinued and this seems a bit of a waste of my stockpile. Instead just use one pair in the Vbe multiplier and use totem pole drivers perhaps. Have not really studied this yet.
Best wishes
David
Last edited:
As explained, I plan to use them as drivers so hfe does matter. More is better here.
Still undecided on the VAS.
Often lower hfe transistor types have better VAF.
Never considered if this applies within a particular type.
Why does it? Any data?
Best wishes
David
I don't see how unconditional stability is the same as "conditional" "stability", even with quotes.
No doubt. But he found a solution that was important, non-obvious and unique. Any maths book will show the importance of uniqueness as a mathematical property. And at the front of the book, directly underneath the title, in nice clear letters, it doesn't mention "practical". It reads.
Hendrik W. Bode Ph.D.
Research Mathematician
As the optimized solution to a wicked problem.
There is only a maximum feedback function. The particular network to implement that will depend on circuit values, transistor capacitances, trace inductance and capacitance and so on. If you are lucky these may result in a fairly decent feedback function with just a "Cherry" capacitor as the only added discrete component. But that would be exceptional. The extra resistor and capacitor of TMC are just a couple of extra variables to tweak the feedback function closer to optimum.
Only a PDF. I would like a hard copy of the later edition. Apparently there were some additional notes or addenda. If you ever see one buy it for me
Best wishes
David
Last edited:
Thank you!@ astx
WoW ! They all look very pro
Using asc file from post #278 and merging with current input/vas stage development shows:
THD20k@200W@8R with TMC 68p/470p/470R
~ 0.0006% ( only 1-2 ppm worse compared to current output stage)
Not so bad. It is worth a try to build this on breadboard ...
"Totem pole" confuses me here. Have you a simple example how your outputstage differs from that of post #278?
BR, Toni
Thanks too!
And yes - TMC will be realized in the final amplifier as it gives the best real life measured THD figures. But TPC or Miller can also be used if one does not like the small phase shift at high frequencies.
Worse! It is supposed to be better.
Some work required to optimise I suspect.
Perhaps my bad use of the term.
I meant push pull drivers as seen here
http://www.diyaudio.com/forums/soli...ogy-construction-troubles-17.html#post2268194
Perhaps not practical with distributed drivers. Multiple pairs of Vbe multipliers is not tidy. One pair of Vbe multipliers to drive all the distributed drivers? Just an idea.
Best wishes
David
Last edited:
It does, to obtain higher beta the base region is narrower.
http://www-physics.lbl.gov/~spieler/physics_198_notes/PDF/VIII-3-BJT.pdf
