What are some good low noise bipolar transistors with good linearity?
9014's claim a 0.9db typical NF. but there's no noise chart.
they seem to be very similar to 2n2222's which also have around 0.9db NF.
2SC1815's meanwhile show a 1db NF and no chart. and they're far less linear too.
2SC1845's seem to have less noise, or are they just a HV version of the 2SC1815?
9014's claim a 0.9db typical NF. but there's no noise chart.
they seem to be very similar to 2n2222's which also have around 0.9db NF.
2SC1815's meanwhile show a 1db NF and no chart. and they're far less linear too.
2SC1845's seem to have less noise, or are they just a HV version of the 2SC1815?
We do not get genuine Japanese transistors here. Only ****ese copies linked to a Japanese datasheet.
I jacked up gain on a ON MPS8099 (npn, 80 v Vce) to exceed what was standard on an Apex AX6. The board on the other side of the amp had a genuine BC108a from 1970, possibly Siemens or Telefunken. Original PC14 design was to be driven by plate of a 12AX7, a 250kohm source, through a 6' RCA cable.
I do not hear any hiss or pops. Gain was tremendous, also. On have their process under control, IMHO. Noise specs on the datasheet were for selecting the good ones by sorting, in the day when IMHO fab was full of operators wearing hair nets and no positive pressure on the room. Now the maintenance men wear tyvek suits, booties, masks, and there is nobody else in the room.
Linearity comes from the c to e resistor ratio. bipolar junction transistors are exponential.
I jacked up gain on a ON MPS8099 (npn, 80 v Vce) to exceed what was standard on an Apex AX6. The board on the other side of the amp had a genuine BC108a from 1970, possibly Siemens or Telefunken. Original PC14 design was to be driven by plate of a 12AX7, a 250kohm source, through a 6' RCA cable.
I do not hear any hiss or pops. Gain was tremendous, also. On have their process under control, IMHO. Noise specs on the datasheet were for selecting the good ones by sorting, in the day when IMHO fab was full of operators wearing hair nets and no positive pressure on the room. Now the maintenance men wear tyvek suits, booties, masks, and there is nobody else in the room.
Linearity comes from the c to e resistor ratio. bipolar junction transistors are exponential.
The closest we get anymore are the Fairchild Korea (KSA, KSC). Are they as good? Who knows. I use them and they work, and I don’t scrutinize over .00000x % distortion since EM compatibility issues swamp that by orders of magnitude. There is Toshiba, but only for power devices.
Lots of old stock from years ago, but as it gets used up there is no more. Still have a few hundred C1815 and that might just outlast me.
Lots of old stock from years ago, but as it gets used up there is no more. Still have a few hundred C1815 and that might just outlast me.
Following. Types I have looked at/used are mostly SMD like the FJV992 and FJV1845, these in theory are very quiet, but I have not used them in an application where their noise performance was critical. BC550/BC560 are not the last word in quiet, probably not suitable for an MC amp, but probably good everywhere else
KSC1845/KSA992 (Onsemi/Fairchild) are reputedly very quiet, but I have not used them yet in a location where noise counts.
Some devices in TO-126 packages may be worth looking at.
THATCorp makes some very quiet SMD transistor arrays. https://thatcorp.com/that-300-series-low-noise-matched-transistor-arrays/
(And their data sheets specify rbe in some cases.)
Linear systems https://www.linearsystems.com/bipolartransistors (matched duals)
There are several older, now closed threads on this topic, but unfortunately some deal with mostly TH parts that are no longer readily available.
KSC1845/KSA992 (Onsemi/Fairchild) are reputedly very quiet, but I have not used them yet in a location where noise counts.
Some devices in TO-126 packages may be worth looking at.
THATCorp makes some very quiet SMD transistor arrays. https://thatcorp.com/that-300-series-low-noise-matched-transistor-arrays/
(And their data sheets specify rbe in some cases.)
Linear systems https://www.linearsystems.com/bipolartransistors (matched duals)
There are several older, now closed threads on this topic, but unfortunately some deal with mostly TH parts that are no longer readily available.
Last edited:
There are some measurements of bipolar transistor noise here : http://dicks-website.eu/low_noise_amp_part5/part5.html
ZTX951 is the winner of those tested.
ZTX951 is the winner of those tested.
Apparently that ZTX951 is very low noise.. I assume the DZT951/DTZ851 use the same die, but in an SMD part, right?; https://www.diodes.com/assets/Datasheets/ds30786.pdf
What other good low noise transistors are there? I mean, yeah there's those on that list but it's not very comprehensive.
Last edited:
Hitachi 2SA872/2SC1775 for 90V VCE and 2SA872A/2SC1775A for 120V VCE, all are 50mA/300mW, so quite a limited power but should be the lower possible noise and best devices overall, otherwise the 100mA/45V BC550C/560C are the most recent equivalent of the 70s era 20V BC109C/BC179C, BC239C/BC309C and such generic proelectron named low power BC devices wich also comprise 45V BC413C/BC414C.
EDIT : There s also the 2SA970/2SC2240 as well as 2SA992/2SC1845, these are ubiquitous in japanese amplifiers along with said Hitachi, i have a preference for the Hitachi 2SA872A/2SC1775A wich i used extensively for two decades.
EDIT : There s also the 2SA970/2SC2240 as well as 2SA992/2SC1845, these are ubiquitous in japanese amplifiers along with said Hitachi, i have a preference for the Hitachi 2SA872A/2SC1775A wich i used extensively for two decades.
Last edited:
In the case of bipolar transistors, there is no one-size-fits-all transistor that provides low noise in any circuit.
Transistors like the ZTX951/851, mentioned here for their low base spread resistance Rbb’, have low noise at low source impedances like MC carts.
However, if you use it on MM carts or a general line level, the noise will be large and the Cob will be large, so linearity may deteriorate.
In such cases, a transistor with a large hfe and low current capacity rating is suitable.
Transistors like the ZTX951/851, mentioned here for their low base spread resistance Rbb’, have low noise at low source impedances like MC carts.
However, if you use it on MM carts or a general line level, the noise will be large and the Cob will be large, so linearity may deteriorate.
In such cases, a transistor with a large hfe and low current capacity rating is suitable.
The ZTX951/851 are ultra-low voltage noise only, current noise is poor so they are mainly good for low source impedances only (MC cartridge, dynamic microphone).
General purpose low noise devices typically have very high gain to allow extremely low base currents (necessary for low current noise (shot noise)).
Noise contour graphs are not as useful as voltage- and current-noise spectra, they tell you much more. This datasheet for 2SA992 has graph of voltage- and current-noise against collector current which is also useful: http://www.ids-elektronik.de/421.pdf
And remember if you ever reverse-bias the emitter-base junction of a BJT into zener breakdown its noise performance will be crippled - so a backwards diode between emitter and base is a wise precaution against power on/off transients doing this.
Can’t get Hitachi anymore, but UTC makes clones. How good are they? The D667’s Cob reportedly isnt nearly as low, and more on par with everybody else’s so you may as well use the Toshiba TTA/C these days. And I’m not sure if Toshiba has any active small signal parts in the original TO-92 (and SMD would be different type numbers).
2N5087/5210 were “discontinued” by the big makers too - a long time ago - but of course you can get all sorts of 3rd party sources. Are they as good? Probably - it’s ancient tech and more expensive than Fairchild. Are Fairchild’s C1845/A992 better? Yep (in every way), but you’ve got to bend the leads. Not an issue for new design, just go BCE/ECB and be done with it. The 50 volt limit was a deal breaker on the 2N’s for most of us making real amps anyway so it’s no big loss.
For low impedance sources it’s hard to beat the best bipolar op amps when you consider ALL sources of noise in your application. And it solves the pair matching and reverse protection problems for you. When you get it down to about 1.6 nV per root Hz it’s a don’t care below that, assuming no significant contribution from Johnson noise and nothing comes in off the power supply (don’t underestimate the importance of these - as is your circuit layout). Op amps start to look REALLY attractive unless you’re hell bent on a discrete solution on principle. Which you may be. You may be hell bent on a glass audio with a heater solution too and that’s perfectly valid. Run of the mill input pairs (ones intended as such, not dubious one size fits all like MPSA42) are quiet ENOUGH and high ENOUGH gain for power amp input stages that special consideration isn’t even on my radar. The noise gets dominated by everything that comes before it.
2N5087/5210 were “discontinued” by the big makers too - a long time ago - but of course you can get all sorts of 3rd party sources. Are they as good? Probably - it’s ancient tech and more expensive than Fairchild. Are Fairchild’s C1845/A992 better? Yep (in every way), but you’ve got to bend the leads. Not an issue for new design, just go BCE/ECB and be done with it. The 50 volt limit was a deal breaker on the 2N’s for most of us making real amps anyway so it’s no big loss.
For low impedance sources it’s hard to beat the best bipolar op amps when you consider ALL sources of noise in your application. And it solves the pair matching and reverse protection problems for you. When you get it down to about 1.6 nV per root Hz it’s a don’t care below that, assuming no significant contribution from Johnson noise and nothing comes in off the power supply (don’t underestimate the importance of these - as is your circuit layout). Op amps start to look REALLY attractive unless you’re hell bent on a discrete solution on principle. Which you may be. You may be hell bent on a glass audio with a heater solution too and that’s perfectly valid. Run of the mill input pairs (ones intended as such, not dubious one size fits all like MPSA42) are quiet ENOUGH and high ENOUGH gain for power amp input stages that special consideration isn’t even on my radar. The noise gets dominated by everything that comes before it.