Little Diode of New Malden, Surrey claim to have them in stock so, like all our dwindling through-hole parts, get a quantity while you can. Tomorrow may be too late.
Depending on the voltages and location , 2SC1815/2SA1805 or the the old faithful 2SC2240/2SA970 are good and can can still be selected for lowest noise if that is your aim.
Depending on the voltages and location , 2SC1815/2SA1805 or the the old faithful 2SC2240/2SA970 are good and can can still be selected for lowest noise if that is your aim.
I use often 2N5551/5401.
Not the lowest noise perhaps. But robust and good quality and takes high voltage and current.
Besides I use 2SC2240/SA970. Takes high voltage. Low noise and precision.
Also BC550C/560C I use.
BC550C/560C are handicapped of take max 45 Volt = 2x22.5V
2SC1815/2SA1015 are good gain, low noise and think can take some voltage.
Samuel Groner:
SG-Acoustics · Samuel Groner · Discrete OpAmps
Not the lowest noise perhaps. But robust and good quality and takes high voltage and current.
Besides I use 2SC2240/SA970. Takes high voltage. Low noise and precision.
Also BC550C/560C I use.
BC550C/560C are handicapped of take max 45 Volt = 2x22.5V
2SC1815/2SA1015 are good gain, low noise and think can take some voltage.
Samuel Groner:
SG-Acoustics · Samuel Groner · Discrete OpAmps
Here I have used BC550C
in a transistor line preamp with gain x10 and no global feedback.
Is fed by a discrete lowdrop regulator at 24.0 Volt.
Shows that we can get low THD (0.003% at 1Volt) using no global feedback.
The x10 is acheived by use of two gain stages.
The output is buffered into a load of 10k
in a transistor line preamp with gain x10 and no global feedback.
Is fed by a discrete lowdrop regulator at 24.0 Volt.
Shows that we can get low THD (0.003% at 1Volt) using no global feedback.
The x10 is acheived by use of two gain stages.
The output is buffered into a load of 10k
Attachments
Unfortunately there are (as far as I know) no easily sourced complementary parts which show, at the same time:
* low rbb' (gives low voltage noise)
* high hFE (gives low current noise)
* low excess noise (gives low low-frequency current and voltage noise)
* matching hFE for complementary pairs (gives low input bias current)
To give a specific recommendation you'll need to elaborate on your specific application. What's the source impedance, what input bias current is acceptable, is low-frequency noise a major concern?
Samuel
* low rbb' (gives low voltage noise)
* high hFE (gives low current noise)
* low excess noise (gives low low-frequency current and voltage noise)
* matching hFE for complementary pairs (gives low input bias current)
To give a specific recommendation you'll need to elaborate on your specific application. What's the source impedance, what input bias current is acceptable, is low-frequency noise a major concern?
Samuel
The world is analogue, and there will always be a need for low noise frontends for sensors. As far as I can tell, the noise performance of available devices has greatly improved over the last decades.
Is there a particular reason why the input stage needs to be complementary?
Which design are you referring to?
Samuel
Samuel
I keep reading here and on other forums that low noise devices such as the 2SC2547, 3329, LM394 are near impossible to get. I diy mic preamps for home recording use and discrete one are always MUCH better sounding but
need low noise input stages., which is why I was asking....
As far as I can tell, the noise performance of available devices has greatly improved over the last decades.
I think there are many that would disagree with this statement - judging by what is being said elsewhere
Is there a particular reason why the input stage needs to be complementary?
having looked at your own designs, you use 3329 and its compliment in one of your own low noise designs. Was there a reason you did this OR could you have achieved the same results using just a pair of PNP (or NPN) and if so why not make it simpler?
I keep reading here and on other forums that low noise devices such as the 2SC2547, 3329, LM394 are near impossible to get. I diy mic preamps for home recording use and discrete one are always MUCH better sounding but
need low noise input stages., which is why I was asking....
As far as I can tell, the noise performance of available devices has greatly improved over the last decades.
I think there are many that would disagree with this statement - judging by what is being said elsewhere
Is there a particular reason why the input stage needs to be complementary?
having looked at your own designs, you use 3329 and its compliment in one of your own low noise designs. Was there a reason you did this OR could you have achieved the same results using just a pair of PNP (or NPN) and if so why not make it simpler?
Of course there are many discontinued devices which we'd have use for. Nonetheless I don't think it is correct to state that, as a trend, there are less low noise devices available. One problem is that there are now some many different transistors available that it's very hard to get, and keep, an overview. Also many transistors may have low noise, but are not explicitly specified for such, and thus just unearthed by accident.
With 2N4401/2N4403 and BC550C/BC560C you can cover 95% of the audio duties with excellence. Add SSM2212, SSM2220, LSK389 and BF862, and we're at 99.9%. All of these are comparably cheap (what did the LM394 cost when it came out?), and reasonably easy to source.
Much more of a problem is the lack of knowledge. You can't just pick a magic transistor and think that this will provide low noise per se. Deducing the appropriate number of paralleled devices, selecting the optimum operating current and minimising secondary noise sources is very frequently done wrong. Many more dBs of noise figure have been spoilt that way than with the unavailability of a specific transistor.
The main reason was the cancellation of input bias current, at no additional input current noise. But this is not necessary in many applications, and then there are other ways to do it.
Samuel
With 2N4401/2N4403 and BC550C/BC560C you can cover 95% of the audio duties with excellence. Add SSM2212, SSM2220, LSK389 and BF862, and we're at 99.9%. All of these are comparably cheap (what did the LM394 cost when it came out?), and reasonably easy to source.
Much more of a problem is the lack of knowledge. You can't just pick a magic transistor and think that this will provide low noise per se. Deducing the appropriate number of paralleled devices, selecting the optimum operating current and minimising secondary noise sources is very frequently done wrong. Many more dBs of noise figure have been spoilt that way than with the unavailability of a specific transistor.
The main reason was the cancellation of input bias current, at no additional input current noise. But this is not necessary in many applications, and then there are other ways to do it.
Samuel
Last edited:
This thread initially was discussing parts availability for Samuel's LNA -- I had a bit of trouble with my variant until I realized that I had incorrectly placed one of the 1N4148 dioes!
BF862's are available from the usual suspects -- legitimate devices will have "62" labeled on the device.
I strongly recommend Sam's method of matching devices.
My LNA clocks in at around 390 pV/Rt Hz at 10kHz -- Mr. Groner did me a bit better but at these levels it's counting angels on the head of a pin!
More tomorrow... one suggestion --- some black electrical tape over the 4 1N4148 diodes used to set the 2.4V reference.
BF862's are available from the usual suspects -- legitimate devices will have "62" labeled on the device.
I strongly recommend Sam's method of matching devices.
My LNA clocks in at around 390 pV/Rt Hz at 10kHz -- Mr. Groner did me a bit better but at these levels it's counting angels on the head of a pin!
More tomorrow... one suggestion --- some black electrical tape over the 4 1N4148 diodes used to set the 2.4V reference.
No please it's 2Ap There is a SOT 23 code book.
The SMD Code Book
2Ap BF862 NXP ZC SOT23 n-ch jfet for am car radio front end
For everyone this is an invaluable resource for sorting or iding parts on boards.
Last edited:
Mine from Mouser / Digikey have "2AW".
I think it depends on the factory.
Gerhard
(now back to my measurement -> Simulation model program for quartz crystals.
It used to work 18 years ago, but C is now different. )
Sorry, you are correct. mine say 2AP horizontal and 32 vertical. I got mine from Mouser before they went EOL.
Apologies.
My application for low noise transistors was for a microphone preamp (~310Ω source impedance @ 1.6mV nominal), with existing PNP TO-92 topology. The best I found was a ZTX 550 (with a ZTX450 complement) which would get to ~74 db S/N. I used the chart from "Art of Electronics" by Horowitz-Hill, which has an EXCELLENT section on this very topic.
A quick check found it is China vs Taiwan, the 2Aw parts being responsible for some of the high noise corner parts. They are both genuine NXP parts. It must be impossible to keep that SMD database up to date.