If you are looking for ultra low voltage noise (2 ohm base resistance), BFG16A (SOT223A) or BFQ17 (SOT89) might be interesting. Mind you, hFE is not very high and as far as I remember the 1/f corner is not particularly low, so these are not the right transistors if you want low current noise. See www.semiconductors.philips.com for more info.
You could also try to find out if there is an SMD-version of the 2SC2547. It has a base resistance of about 14 ohm, hFE~=600 and a low 1/f-corner.
You could also try to find out if there is an SMD-version of the 2SC2547. It has a base resistance of about 14 ohm, hFE~=600 and a low 1/f-corner.
May I ask what you are going to do with this low noise transistors?
Have you checked
http://www.thatcorp.com/
BC850 isn't so bad. I get 600 nV noise in my monster headphone amp.
Have you checked
http://www.thatcorp.com/
BC850 isn't so bad. I get 600 nV noise in my monster headphone amp.
lower noise
Hello peranders & Marcel
I'd use it for a MC RIAA preamp. So I look for very low voltage noise and also a low 1/f corner transistor.
Those wideband transistors use to have a high 1/f corner. It happend to me somethign simmilar before with an op-amp, the AD8051. Everything is indeed good in it, (Low cost, 150mHz, 150V/us, 5nV...), but its 1/f corner is at 1000Hz, not good for audio. Perhaps good for GICs active filters.
I believe that a lower voltaje noise can be obtained with BJTs (so happens in OP-AMPs), at a cost of more current noise. But... by now, the lowest noise single transistor I found is the 2SK170 and 2SJ74, reaching the 0,95nV/sqrt(Hz), the same as dual ultra low noise LM394, SSM2210... BJTs. With one LM394 I could obtain 0,65nV/sqrt(hZ), but there must be any single transistor doing the same as two. Also avaibility of SSM220 and LM394 is bad (are they discontinued? MAT03 is...)
Also there's no PNP "friend" for LM394, and I'd like to make a current feedback first stage.
2SC2547 are quite more noisy than LM394.
I also use BC850C for my power amps, with low NFB resistors you can get low noise at the output, but it's not enough for a MC preamp (I've already tried it in a second stage and I had to substitute them for 2SK107 and 2SJ74)
Hello peranders & Marcel
I'd use it for a MC RIAA preamp. So I look for very low voltage noise and also a low 1/f corner transistor.
Those wideband transistors use to have a high 1/f corner. It happend to me somethign simmilar before with an op-amp, the AD8051. Everything is indeed good in it, (Low cost, 150mHz, 150V/us, 5nV...), but its 1/f corner is at 1000Hz, not good for audio. Perhaps good for GICs active filters.
I believe that a lower voltaje noise can be obtained with BJTs (so happens in OP-AMPs), at a cost of more current noise. But... by now, the lowest noise single transistor I found is the 2SK170 and 2SJ74, reaching the 0,95nV/sqrt(Hz), the same as dual ultra low noise LM394, SSM2210... BJTs. With one LM394 I could obtain 0,65nV/sqrt(hZ), but there must be any single transistor doing the same as two. Also avaibility of SSM220 and LM394 is bad (are they discontinued? MAT03 is...)
Also there's no PNP "friend" for LM394, and I'd like to make a current feedback first stage.
2SC2547 are quite more noisy than LM394.
I also use BC850C for my power amps, with low NFB resistors you can get low noise at the output, but it's not enough for a MC preamp (I've already tried it in a second stage and I had to substitute them for 2SK107 and 2SJ74)
It depends on the application, man........
Low-noise BJTs are quiet, but only if they have a low driving impedance. JFETs don't really care about impedance, because they do not have the noise current contribution. Noise figures for JFETs with low source impedance aren't really anything to go by.
Hmmm.......current feedback front end. Guess he didn't read P-As thread on current feedback.
Rohm used to make some BJTs for these applications. I forget the number......maybe I have it somewhere. But I have no idea if they still make them, and if they come in a SOT package.
Jocko
Low-noise BJTs are quiet, but only if they have a low driving impedance. JFETs don't really care about impedance, because they do not have the noise current contribution. Noise figures for JFETs with low source impedance aren't really anything to go by.
Hmmm.......current feedback front end. Guess he didn't read P-As thread on current feedback.
Rohm used to make some BJTs for these applications. I forget the number......maybe I have it somewhere. But I have no idea if they still make them, and if they come in a SOT package.
Jocko
Ultra-low noise transistors
Hi PCP,
See f.a.:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50345&highlight=2sk389&r=&session=
and
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50205&highlight=2sk389&r=&session=
Edit: the circuit I refered to in the MAT04 datasheet is current feedback!
Hi PCP,
See f.a.:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50345&highlight=2sk389&r=&session=
and
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50205&highlight=2sk389&r=&session=
Edit: the circuit I refered to in the MAT04 datasheet is current feedback!
what's wrong with CFB front end?
I don't know what's worng with CFB front end (is not really a front end, its an small signal amplifier itself, to give variable gain without bandwith variations).
Voltaje noise can be as low as in VFB amplifiers, current noise is high, but with 100R source it wouldn't be a problem. I've already made it for MM with FETs, 2SK389 ,and the result was ok.
where is it?
I don't know what's worng with CFB front end (is not really a front end, its an small signal amplifier itself, to give variable gain without bandwith variations).
Voltaje noise can be as low as in VFB amplifiers, current noise is high, but with 100R source it wouldn't be a problem. I've already made it for MM with FETs, 2SK389 ,and the result was ok.
where is it?
Quote from PCP:
"2SC2547 are quite more noisy than LM394."
This surprises me, as their datasheet specifies a 0.5nV/root(Hz) voltage noise at 10mA collector current. I've never checked that by measurement, I've only used them in microphone amplifiers with bias currents in the order of 1mA and got noise figures quite close to the calculated values.
Anyway, the other transistor types I recommended have a highish 1/f-corner in the CURRENT noise, but for extremely low source impedances, especially if they are inductive, that doesn't matter much. I know their non-SMD predecessor BFW16 has been used successfully in an ultra low noise MC amplifier.
"2SC2547 are quite more noisy than LM394."
This surprises me, as their datasheet specifies a 0.5nV/root(Hz) voltage noise at 10mA collector current. I've never checked that by measurement, I've only used them in microphone amplifiers with bias currents in the order of 1mA and got noise figures quite close to the calculated values.
Anyway, the other transistor types I recommended have a highish 1/f-corner in the CURRENT noise, but for extremely low source impedances, especially if they are inductive, that doesn't matter much. I know their non-SMD predecessor BFW16 has been used successfully in an ultra low noise MC amplifier.
Hello
Yes, that's right at 10mA, but voltaje noise is inversely proportional to the square root of colector/drain current (also to the number of devices in parallel), at 1mA it has 1.58nV, where LM394 has stated 0,95nV. You also have to check thermal noise, if the device gets hot makes itself noisier.10V and 10mA makes 100mW, juntion is about 15-25ºC over ambient temperature.
I've even thought using Peltier heat pumps to keep the device cold...
I don't know, I downloaded BFG16A and BFQ17 datasheets from philips semiconductor, and it doen't tell anything about noise. I found a noise figure in other application, but current noise is included in it.
For ELSO.
I've used it many times, and 2SJ109, it's a cool transistor, very high permormance. In my MC preamp I use it for a CFB "pre-preamp".
Yes, that's right at 10mA, but voltaje noise is inversely proportional to the square root of colector/drain current (also to the number of devices in parallel), at 1mA it has 1.58nV, where LM394 has stated 0,95nV. You also have to check thermal noise, if the device gets hot makes itself noisier.10V and 10mA makes 100mW, juntion is about 15-25ºC over ambient temperature.
I've even thought using Peltier heat pumps to keep the device cold...
I don't know, I downloaded BFG16A and BFQ17 datasheets from philips semiconductor, and it doen't tell anything about noise. I found a noise figure in other application, but current noise is included in it.
For ELSO.
I've used it many times, and 2SJ109, it's a cool transistor, very high permormance. In my MC preamp I use it for a CFB "pre-preamp".
"voltaje noise is inversely proportional to the square root of colector/drain current (also to the number of devices in parallel), at 1mA it has 1.58nV, where LM394 has stated 0,95nV."
Not quite. Actually, BJT voltage noise is the sum of base resistance thermal noise, which is roughly independent of the bias current (at least as long as the device is not in high injection), and the collector shot noise transformed back to the input, which does decrease with bias current. The base resistance term is strongly dependent on the transistor type, the collector shot noise term is not, at least not as long as you bias the transistor below its peak fT. (Near and above peak fT, all kinds of second-order effects such as high injection come into play.)
The total voltage noise in V/root(Hz) theoretically equals:
sqrt(4kT*rb+2kT/gm),
with gm=IC/(kT/q),
where IC is the collector current, k is Boltzmann's constant (1.38065E-23 J/K), q is the absolute value of the electron charge (1.6022E-19 C), rb is the base resistance and T is the absolute temperature.
For a transistor giving 0.5nV/root(Hz) at 10mA, T=320K or so, this comes down to rb~=12.76766233 ohm. A transistor with 0.95nV/root(Hz) at 1mA, 300K has rb~=41.54731389 ohm, which is not bad, but obviously worse than 12.76... ohm.
Not quite. Actually, BJT voltage noise is the sum of base resistance thermal noise, which is roughly independent of the bias current (at least as long as the device is not in high injection), and the collector shot noise transformed back to the input, which does decrease with bias current. The base resistance term is strongly dependent on the transistor type, the collector shot noise term is not, at least not as long as you bias the transistor below its peak fT. (Near and above peak fT, all kinds of second-order effects such as high injection come into play.)
The total voltage noise in V/root(Hz) theoretically equals:
sqrt(4kT*rb+2kT/gm),
with gm=IC/(kT/q),
where IC is the collector current, k is Boltzmann's constant (1.38065E-23 J/K), q is the absolute value of the electron charge (1.6022E-19 C), rb is the base resistance and T is the absolute temperature.
For a transistor giving 0.5nV/root(Hz) at 10mA, T=320K or so, this comes down to rb~=12.76766233 ohm. A transistor with 0.95nV/root(Hz) at 1mA, 300K has rb~=41.54731389 ohm, which is not bad, but obviously worse than 12.76... ohm.
Regarding the BFW16A and its family: according to the on-line SPICE model, rb is about 2 ohm. Since its an RF transistor, it is probably safe to assume that Philips modelled it reasonably accurately. I remember that the people who wrote an article about the ultra-low noise MC amplifier measured 1 ohm, but that was for the original BFW16, not the BFW16A.
Unfortunately, Philips transistor datasheets contain very little useful information nowadays.
Unfortunately, Philips transistor datasheets contain very little useful information nowadays.
Re: It depends on the application, man........
I think Jocko refers to the (now discontinued) Rohm 2SD786/2SB737 pair. I don't know what the replacements would be, although I know that Borbely recommends 2SC3329/2SA1316
mlloyd1
hmmm ... I wonder if it's time to sell mine yet ?
I think Jocko refers to the (now discontinued) Rohm 2SD786/2SB737 pair. I don't know what the replacements would be, although I know that Borbely recommends 2SC3329/2SA1316
mlloyd1
hmmm ... I wonder if it's time to sell mine yet ?
Jocko Homo said:
2SA1015L and 2SC1815L
While in a TO92 package, not SOT23, these are about the lowest noise transistors that I've seen: 2SA1015L and 2SC1815L. The 'L' signifies even lower noise than the low noise versions.
The zip file I made contains the PDF files for both devices. but this forum won't allow me to upload it, so if anyone wants them, email me and I'll send them by email: mail@visionsurgery.net
The combined file is 264K (each one was too large to post separately).
Regards, Robert
While in a TO92 package, not SOT23, these are about the lowest noise transistors that I've seen: 2SA1015L and 2SC1815L. The 'L' signifies even lower noise than the low noise versions.
The zip file I made contains the PDF files for both devices. but this forum won't allow me to upload it, so if anyone wants them, email me and I'll send them by email: mail@visionsurgery.net
The combined file is 264K (each one was too large to post separately).
Regards, Robert
Re: 2SA1015L and 2SC1815L
and put the PDF-file there.
Sending emails back and fort between several pepole
over a considerable period,
when this thread is read over and over again,
does not appeal to me.
We are lucky who have homepages!
I should guess
2SA970 & 2SC2240 is easier to find
than
2SA1015 & 2SC1815
because I think these went out of production some years before 2SA970
In these days of Surface Mounted minitrasistors,
TO92 is becoming less and less interesting to produce
for the big semiconductor companies.
But if we buy a lot now
we can have a private supply for some decades forward.
Some of the new SMT-transistors will be very good.
The smallest ones,
are impossible for us DIYers to work with, without microscope
and special equipement.
I will buy some 100-reds of 2SA970 and 2SC2240
while it is still possible.
from
this online supplier:
(english version catalogue)
http://www.elfa.se/en/
small signal bipolar htm:
http://www.elfa.se/elfa/produkter/en/2412.htm
Prices seems to be in EURO.
2SA970BL/Tsh 71-091-35 22.80€/100 PNP 0.3 120 0.1 350/700 100 Compl to 2SC2240BL
2SC2240BL/Tsh 71-091-43 22.80€/100 NPN 0.3 120 0.1 350/700 100 Compl to 2SA970BL
halofacts - brought to you
So why not make a link at your hompage rljones?rljones said:
and put the PDF-file there.
Sending emails back and fort between several pepole
over a considerable period,
when this thread is read over and over again,
does not appeal to me.
We are lucky who have homepages!
I should guess
2SA970 & 2SC2240 is easier to find
than
2SA1015 & 2SC1815
because I think these went out of production some years before 2SA970
In these days of Surface Mounted minitrasistors,
TO92 is becoming less and less interesting to produce
for the big semiconductor companies.
But if we buy a lot now
we can have a private supply for some decades forward.
Some of the new SMT-transistors will be very good.
The smallest ones,
are impossible for us DIYers to work with, without microscope
and special equipement.
I will buy some 100-reds of 2SA970 and 2SC2240
while it is still possible.
from
this online supplier:
(english version catalogue)
http://www.elfa.se/en/
small signal bipolar htm:
http://www.elfa.se/elfa/produkter/en/2412.htm
Prices seems to be in EURO.
2SA970BL/Tsh 71-091-35 22.80€/100 PNP 0.3 120 0.1 350/700 100 Compl to 2SC2240BL
2SC2240BL/Tsh 71-091-43 22.80€/100 NPN 0.3 120 0.1 350/700 100 Compl to 2SA970BL
halofacts - brought to you
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.