LOL
I measured some long ago, and they really do follow the plots given by Toshiba (I was surprised too).
Peufeu,
This may be useful; 2SC3423 and the complement, 2SA1360, datasheets available here. These devices have 150V, 50mA, 1.2W, 200MHz, Cob around 2pF, flat beta from 0.3mA to 20mA, and a Y beta grading from 120 to 240. They are readily available, highly linear, very fast, and made in Thailand by Toshiba.
I do like your circuit; I designed one very similar some time ago but have not done listening tests yet. I reluctantly decided in light of the many cheap DACs in the market that it was not commercially viable.
Cheers,
Hugh
This may be useful; 2SC3423 and the complement, 2SA1360, datasheets available here. These devices have 150V, 50mA, 1.2W, 200MHz, Cob around 2pF, flat beta from 0.3mA to 20mA, and a Y beta grading from 120 to 240. They are readily available, highly linear, very fast, and made in Thailand by Toshiba.
I do like your circuit; I designed one very similar some time ago but have not done listening tests yet. I reluctantly decided in light of the many cheap DACs in the market that it was not commercially viable.
Cheers,
Hugh
Hi Aksa,
I use 1360/3423 (5W) devices as VAS and pre-driver.
So far I have never sourced Y grade for the PNP. That limits hFE<=160.
I would never have thought of using them in the front end!
I use 1360/3423 (5W) devices as VAS and pre-driver.
So far I have never sourced Y grade for the PNP. That limits hFE<=160.
I would never have thought of using them in the front end!
- KSA992 is interesting
- ksc1845 has large beta droop at 10mA
(both are hard to find apparently)
- 2SC3423, 2SA1360 : souceable at 2€/pair, nice curves, but can't find spice model...
So far I like 2SC2240 + 2SA970.
Latest circuit and sims updated. I think the design is done. Thanks to all. Final transistor choice will be determined by listening tests I think. Please comment 😉
http://audio.peufeu.com/node/57
Andrew : actually there are several arguments for the use of high-beta high-current transistors even in the differential at the front end of an amplifier :
- more linear base current (reduces errors)
- larger chip = slower thermal effects
However a diff pair is rarely biased above 10 mA unless you need funky high slew rates.
Drawbacks are increased noise and capacitance, but some of the japanese parts quoted above seem to be quite good on that.
Hugh : I'm flattered 😀 however I understand you dropped the DAC stuff since the market is saturated now (even though most products are severely compromised and do nothing about jitter).
- ksc1845 has large beta droop at 10mA
(both are hard to find apparently)
- 2SC3423, 2SA1360 : souceable at 2€/pair, nice curves, but can't find spice model...
So far I like 2SC2240 + 2SA970.
Latest circuit and sims updated. I think the design is done. Thanks to all. Final transistor choice will be determined by listening tests I think. Please comment 😉
http://audio.peufeu.com/node/57
Andrew : actually there are several arguments for the use of high-beta high-current transistors even in the differential at the front end of an amplifier :
- more linear base current (reduces errors)
- larger chip = slower thermal effects
However a diff pair is rarely biased above 10 mA unless you need funky high slew rates.
Drawbacks are increased noise and capacitance, but some of the japanese parts quoted above seem to be quite good on that.
Hugh : I'm flattered 😀 however I understand you dropped the DAC stuff since the market is saturated now (even though most products are severely compromised and do nothing about jitter).
Hi,
I would never dream of substituting a 2sa1360 O grade for a bc560 C grade in the front end LTP (even at high tail current).
Much as I said a couple of posts ago.
There are many other small signal BJTs that can do the job better than 1360 or other power BJT with their low hFE, low fT, high Cob (1360 is good here), poor linearity, high cost, etc.
I would never dream of substituting a 2sa1360 O grade for a bc560 C grade in the front end LTP (even at high tail current).
Much as I said a couple of posts ago.
There are many other small signal BJTs that can do the job better than 1360 or other power BJT with their low hFE, low fT, high Cob (1360 is good here), poor linearity, high cost, etc.
> So far I like 2SC2240 + 2SA970.
Me too. Any the ones I measured were really very linear.
But I admit that I also like 2SA1015 / 2SC1815, at least the gain curves from the datasheet. But apparently there ar a number of suppliers other than Toshiba, and their gain curve is nothing but linear. So make sure you get original Toshiba parts.
http://www.datasheetcatalog.com/datasheets_pdf/2/S/C/1/2SC1815.shtml
(Datasheet too large to attach.)
Patrick
Me too. Any the ones I measured were really very linear.
But I admit that I also like 2SA1015 / 2SC1815, at least the gain curves from the datasheet. But apparently there ar a number of suppliers other than Toshiba, and their gain curve is nothing but linear. So make sure you get original Toshiba parts.
http://www.datasheetcatalog.com/datasheets_pdf/2/S/C/1/2SC1815.shtml
(Datasheet too large to attach.)
Patrick
ROHM perhaps
ROHM has a series of high hfe transistors, only NPN though.
Apart from the COB i think they look great but then again i am not an expert.
2sd2144s and
2sd2227s
ROHM overview
In what range would the COB be acceptable?
ROHM has a series of high hfe transistors, only NPN though.
Apart from the COB i think they look great but then again i am not an expert.
2sd2144s and
2sd2227s
ROHM overview
In what range would the COB be acceptable?
Well the ksc1845 does drop pretty sharp but look where it starts and ends. It's over 600 at 5ma and 575 at 10ma and it's still 250 at 35ma .
As for availability if your in the states you can get them just about anywhere I get them from Mouser .
They are great for ltp's there gbp and there peak gain curve fall very close to currents found in ltp's.
Linear TO-92 bjt
Sorry to interrupt peufeu and I do not want to hijack your thread but just ask for advise for the same criteria as yours but for a complementary pair of bjt TO-92 pkg with c-b-e pin out for substitution (range of about 1ma to 20ma). Please other than 2N5551/5401.
Thanks
Sorry to interrupt peufeu and I do not want to hijack your thread but just ask for advise for the same criteria as yours but for a complementary pair of bjt TO-92 pkg with c-b-e pin out for substitution (range of about 1ma to 20ma). Please other than 2N5551/5401.
Thanks
Hi Fab,
bc550c/560c
or bc546b/556b if you can accept lower gain in return for the higher voltage.
bc550c/560c
or bc546b/556b if you can accept lower gain in return for the higher voltage.
I have chosen my transistors. Thanks all.
And for low currents, yes BC550C/560C is good, mister fab.
And for low currents, yes BC550C/560C is good, mister fab.
Hi Peu,
keep us updated on results from the real amplifier and how close the sym model was at predicting performance/behaviour.
keep us updated on results from the real amplifier and how close the sym model was at predicting performance/behaviour.
Bc546/550c
thanks AndrewT and peufeu
HFE for BC546/550 with "c" suffix seems high and capacitance seems low. However, for HFE linearity I am not sure. See attached extract of datasheet or Am I missing something here?
2N5551 has a more stable HFE curve function of IC for the range from let say 0.3 to 10 ma I think...
thanks AndrewT and peufeu
HFE for BC546/550 with "c" suffix seems high and capacitance seems low. However, for HFE linearity I am not sure. See attached extract of datasheet or Am I missing something here?
2N5551 has a more stable HFE curve function of IC for the range from let say 0.3 to 10 ma I think...
Attachments
Hi Fab,
which manufacturer's data did you post?
Most of the curves I have are nearly ruler flat. Only Fairchild 556 showed a significant slope (100 to 190) but not as bad as that one.
It would be worth checking on individual Qs, if absolute linearity is critical to circuit performance.
which manufacturer's data did you post?
Most of the curves I have are nearly ruler flat. Only Fairchild 556 showed a significant slope (100 to 190) but not as bad as that one.
It would be worth checking on individual Qs, if absolute linearity is critical to circuit performance.
Re: Bc546/550c
Audio designers, judging from circuits schematics I have seen,
does NOT want to use BC550C / BC560C for more than max 5-6 mA.
I have got it, from projects that i have read about,
that the best use of BC550C / BC560C is, ROUGHLY speaking:
mA
0.150 - 0.400 .... for lowest noise
0.450 - 2.000 .... a compromise, and most often used
2.000 - 6.000 .... for speed and good gain, if lowest noise is secondary issue
You almost never see any BC550C used with more than 6 mA.
This is also about the max used by Douglas Self in his Discrete Design tutorials at his website.
Because above 5 mA there are other transistors, that are more suitable.
==============================
If we think of European BCxxx series, we have
BC327 / BC337 and BC328 / BC338
and if we want higher voltage = 80 Volt:
BC639 / BC640
BC327 as well as BC639 are sometimes available
in sorted hFE gain classes:
and they are like:
BC327-16, BC327-25, BC327-40 .. which is classes like BC550A, BC550B, BC550C
-16 ~ gain class 160, -25 ~ gain class 250, -40 ~ gain class 400
If sold as only BC327 or BC639, variation of gain can be any class within the transistor range
and you have to pick out, sort and match them yourself
into different hFE gain groups.
I think is true that 2N5401 / 2N5551 will have better linearity
say at current above 5 mA.
An indication that a small signal transistor is good or intended for little bit higher currents
is that Max Current and/or Max Power DATA is a bit higher
than the standard small signal ( TO92 ) have.
===============================
And this standard is, like for BC547, BC557 TO92 package:
0.5W (some other only 0.3W) and 0.1A (= 100mA)
===============================
BC327 has got:
0.8W (some brands 0.6W) and 0.5A, which is 5 times more than 0.1A
2N5401 has got:
0.5W and 0.6A, which is 6 times more than 0.1A
BC639/640 has got:
1.0W and 1.0A, which is 10 times more than 0.1A
These more power/currents TO92 transistors
pay most often the price by have a bit lower gain.
Like max 250-400, while standard BC547C, BC550C
can have as high gain as max hFE 800.
lineup
fab said:
Audio designers, judging from circuits schematics I have seen,
does NOT want to use BC550C / BC560C for more than max 5-6 mA.
I have got it, from projects that i have read about,
that the best use of BC550C / BC560C is, ROUGHLY speaking:
mA
0.150 - 0.400 .... for lowest noise
0.450 - 2.000 .... a compromise, and most often used
2.000 - 6.000 .... for speed and good gain, if lowest noise is secondary issue
You almost never see any BC550C used with more than 6 mA.
This is also about the max used by Douglas Self in his Discrete Design tutorials at his website.
Because above 5 mA there are other transistors, that are more suitable.
==============================
If we think of European BCxxx series, we have
BC327 / BC337 and BC328 / BC338
and if we want higher voltage = 80 Volt:
BC639 / BC640
BC327 as well as BC639 are sometimes available
in sorted hFE gain classes:
and they are like:
BC327-16, BC327-25, BC327-40 .. which is classes like BC550A, BC550B, BC550C
-16 ~ gain class 160, -25 ~ gain class 250, -40 ~ gain class 400
If sold as only BC327 or BC639, variation of gain can be any class within the transistor range
and you have to pick out, sort and match them yourself
into different hFE gain groups.
I think is true that 2N5401 / 2N5551 will have better linearity
say at current above 5 mA.
An indication that a small signal transistor is good or intended for little bit higher currents
is that Max Current and/or Max Power DATA is a bit higher
than the standard small signal ( TO92 ) have.
===============================
And this standard is, like for BC547, BC557 TO92 package:
0.5W (some other only 0.3W) and 0.1A (= 100mA)
===============================
BC327 has got:
0.8W (some brands 0.6W) and 0.5A, which is 5 times more than 0.1A
2N5401 has got:
0.5W and 0.6A, which is 6 times more than 0.1A
BC639/640 has got:
1.0W and 1.0A, which is 10 times more than 0.1A
These more power/currents TO92 transistors
pay most often the price by have a bit lower gain.
Like max 250-400, while standard BC547C, BC550C
can have as high gain as max hFE 800.
lineup
thanks lineup!
This makes a lot of sense. Finally, 2N5401/5551 is not so bad after all... .Baed on Philips and On semi, the linearity is better than about 7% from 0.3 to 15 ma. The curve are different between the 2 manufacturers.
Hi AndrewT
The datasheet manufacturer is Siemens. Ihave checked another one from Motorola and it is the same curve.
This makes a lot of sense. Finally, 2N5401/5551 is not so bad after all... .Baed on Philips and On semi, the linearity is better than about 7% from 0.3 to 15 ma. The curve are different between the 2 manufacturers.
Hi AndrewT
The datasheet manufacturer is Siemens. Ihave checked another one from Motorola and it is the same curve.
