Hello,
I am working on a MM phono preamp, a cascode configuration, which uses a Jfet 2sk389 as the low transistor and a NPN BJT for the upper one. For the noise issue, it's important to use an high Hfe and low rbb BJT, and I came up to these 3 options, which are easily available in my country:
2SC2240 datasheet
MPSA18 datasheet
BC550C datasheet
Now my question: which one has the lowest "base spreading resistance" (rbb)? Only the BC550C datasheet show this value.
Thanks,
Claudio
I am working on a MM phono preamp, a cascode configuration, which uses a Jfet 2sk389 as the low transistor and a NPN BJT for the upper one. For the noise issue, it's important to use an high Hfe and low rbb BJT, and I came up to these 3 options, which are easily available in my country:
2SC2240 datasheet
MPSA18 datasheet
BC550C datasheet
Now my question: which one has the lowest "base spreading resistance" (rbb)? Only the BC550C datasheet show this value.
Thanks,
Claudio
Hi Claudio,
I have used the 4401 in precisely this role; beta is lowish, around 30 minimum and typically 100, but as David remarks the noise performance is exceptional - AND it's cheap.
The gain of a cascode is given by the product of the transconductance of the lower device and the load on the collector of the upper device; in truth the beta of the upper device does not, IIRC, come into it. So low beta should not matter.
Of most concern is the PSRR of a cascode - it's poor, forcing you to use either super-regulation or a battery. But I would assume this is a multistage amp, so there will be perhaps another stage after the cascode??
Cheers,
Hugh
I have used the 4401 in precisely this role; beta is lowish, around 30 minimum and typically 100, but as David remarks the noise performance is exceptional - AND it's cheap.
The gain of a cascode is given by the product of the transconductance of the lower device and the load on the collector of the upper device; in truth the beta of the upper device does not, IIRC, come into it. So low beta should not matter.
Of most concern is the PSRR of a cascode - it's poor, forcing you to use either super-regulation or a battery. But I would assume this is a multistage amp, so there will be perhaps another stage after the cascode??
Cheers,
Hugh
Claudio, a BC550 will be fine at 100-500 uA. You will get around 1 uV noise which is quite sufficient. You can check my QSXM2 RIAA amp where I use lot's of BC550C/BC560C.
Re: Re: Best rbb transistor, for MM phono preamp
You guys need to brush up on noise theory.
Claudio has stated the IP or lower device is a dual low noise
N ch Jfet, 2sk389. This is a fine IP device for the purpose.
The N ch BJT is cascoded on top of the Jfet. As such
it will have virtually no effect on the noise of the
circuit.
You do not need an ultra low noise BJT, but more
appropriately, a BJT suitable for a cascode.
Terry
peranders said:
davidsrsb said:
amplifierguru said:
peranders said:
lineup said:
peranders said:
You guys need to brush up on noise theory.
Claudio has stated the IP or lower device is a dual low noise
N ch Jfet, 2sk389. This is a fine IP device for the purpose.
The N ch BJT is cascoded on top of the Jfet. As such
it will have virtually no effect on the noise of the
circuit.
You do not need an ultra low noise BJT, but more
appropriately, a BJT suitable for a cascode.
Terry
You mean input transistors? That's what I mean also. All other positions are pretty unimportant when it comes to noise.
Re: Re: Re: Re: Best rbb transistor, for MM phono preamp
MPSA18 has lowest OP Cap and very high beta both
good qualities for cascode tranny, it gets my vote.
BC550C has slightly lower beta and slightly higher C
but it is very close... also worth a try.
Cheers,
Terry
lineup said:
Yeah, you are right.
So what is your BJT suitable for a cascode ?
I think the man wants to know.
MPSA18 has lowest OP Cap and very high beta both
good qualities for cascode tranny, it gets my vote.
BC550C has slightly lower beta and slightly higher C
but it is very close... also worth a try.
Cheers,
Terry
Making an amplifier that is quiet enough that the thermal noise of the MM cartridge dominates is fairly easy. Cascodes have the benefit that the input capacitance is low. I don't like the usual simple longtailed pair input in feedback preamps as this has a varable miller input capacitance depending on just how much negative feedback is available at any instant. As the cartridge is resonant with this load capacitance to get its high frquency responce right, this has the potential of causing phase anomalies.
Solid state allows you to follow the input amplifier with a second pnp or pfet stage to get the signal back to ground referenced and get a decent psrr, something the tube guys cannot do.
The choice of the upper transistor in the cascode has more to do with voltage rating than anything else. There is nothing to stop you running your preamp with a 300V rail like tube amps and getting a high gain in the first stage for passive RIAA equalisation.
Solid state allows you to follow the input amplifier with a second pnp or pfet stage to get the signal back to ground referenced and get a decent psrr, something the tube guys cannot do.
The choice of the upper transistor in the cascode has more to do with voltage rating than anything else. There is nothing to stop you running your preamp with a 300V rail like tube amps and getting a high gain in the first stage for passive RIAA equalisation.
Thanks everybody for the answers.
Now back to the topic: my look for a low rbb BJT for the noise issue looks useless since the upper transistor, in a cascode configuration, since it have virtually no effect on the noise of the circuit (Terry).
Looking at Borbely MC pre, it seems to me (and maybe I am wrong) a complementary cascode topology, with parallel low Jfet and upper BJT: the upper BJT used is the 2SD786-2SB737, now substituted by the 2SC3329, that is a low rbb transistor (2 ohm), low noise and low 1/f. My question: why using such a device if noise in the upper cascode transistor is not so important?
Regards,
Claudio
Now back to the topic: my look for a low rbb BJT for the noise issue looks useless since the upper transistor, in a cascode configuration, since it have virtually no effect on the noise of the circuit (Terry).
Looking at Borbely MC pre, it seems to me (and maybe I am wrong) a complementary cascode topology, with parallel low Jfet and upper BJT: the upper BJT used is the 2SD786-2SB737, now substituted by the 2SC3329, that is a low rbb transistor (2 ohm), low noise and low 1/f. My question: why using such a device if noise in the upper cascode transistor is not so important?
Regards,
Claudio
Hmmm...no one has asked about the overall preamp topology.
I assume this is a single-ended version of the Borbely preamp. This has 2 voltage gain stages with noninverting feedback to source. with about 20dB gain in stage 1 and perhaps 60dB gain in stage 2.
With feedback and the large stage 2 gain, the ratio of JFET AC drain to gate voltage should be rather small, so the input capacitance is little more than Cgs+Cgd in the frequency range where open loop gain is high, cascoding mainly provides a low and controlled Vgd for low noise.
If this is a preamp with no voltage loop feedback then cascoding will reduce Cgd, with a source impedance of less than 100 ohms a high input capacitance (500pF) shouldn't have significant effect and cascoding controls Vgd as before, but with MM the source impedance is at least 1k, rising with cartridge electrical resonance, so input capacitance is important and its reduction by cascoding is useful.
The Borbely preamp uses 100 ohm base resistors in the cascode transistors so he isn't too concerned about Rbb. Theoretically, the cascode transistor should have low input current noise, this noise is lower with high beta. This current noise shows up in the load resistor as voltage noise, since the base current subtracts from the load resistor current, similar to partition noise in pentodes.
This noise can be calculated from load resistance and base current, current shot noise is SQRT(2*q*Ib*B) where q=1.6e-19 and B is bandwidth. For a transistor with beta=100 and Ie=10mA, Ib=100uA, with 15kHz bandwidth the current noise is 0.69nA, or 69nV at the 100 ohm load, 6.9nV referred to input of a 20dB gain stage. A 40 ohm source resistance has 40nV noise, so even for MC use this current noise contribution is small and might be neglected. For MM the source resistance is 1k and the source noise is 500nV so this noise is of no consequence.
A high beta transistor would have the lowest input current noise for a given emitter current, MPSA18 has the highest beta, however the 2N4401 with beta=100 would probably be good enough.
You can get plenty of gain at 30V or less with resistive loads since the 2SK389 has high transconductance. Overload margin is poor with passive RIAA and too much gain in the first stage.
High voltage transistors usually have low beta, a Darlington pair reduces input current and current noise. Only the first transistor in the Darlington introduces current noise, all of the second transistor's base current passes through Darlington collector and emitter.
I assume this is a single-ended version of the Borbely preamp. This has 2 voltage gain stages with noninverting feedback to source. with about 20dB gain in stage 1 and perhaps 60dB gain in stage 2.
With feedback and the large stage 2 gain, the ratio of JFET AC drain to gate voltage should be rather small, so the input capacitance is little more than Cgs+Cgd in the frequency range where open loop gain is high, cascoding mainly provides a low and controlled Vgd for low noise.
If this is a preamp with no voltage loop feedback then cascoding will reduce Cgd, with a source impedance of less than 100 ohms a high input capacitance (500pF) shouldn't have significant effect and cascoding controls Vgd as before, but with MM the source impedance is at least 1k, rising with cartridge electrical resonance, so input capacitance is important and its reduction by cascoding is useful.
The Borbely preamp uses 100 ohm base resistors in the cascode transistors so he isn't too concerned about Rbb. Theoretically, the cascode transistor should have low input current noise, this noise is lower with high beta. This current noise shows up in the load resistor as voltage noise, since the base current subtracts from the load resistor current, similar to partition noise in pentodes.
This noise can be calculated from load resistance and base current, current shot noise is SQRT(2*q*Ib*B) where q=1.6e-19 and B is bandwidth. For a transistor with beta=100 and Ie=10mA, Ib=100uA, with 15kHz bandwidth the current noise is 0.69nA, or 69nV at the 100 ohm load, 6.9nV referred to input of a 20dB gain stage. A 40 ohm source resistance has 40nV noise, so even for MC use this current noise contribution is small and might be neglected. For MM the source resistance is 1k and the source noise is 500nV so this noise is of no consequence.
A high beta transistor would have the lowest input current noise for a given emitter current, MPSA18 has the highest beta, however the 2N4401 with beta=100 would probably be good enough.
You can get plenty of gain at 30V or less with resistive loads since the 2SK389 has high transconductance. Overload margin is poor with passive RIAA and too much gain in the first stage.
High voltage transistors usually have low beta, a Darlington pair reduces input current and current noise. Only the first transistor in the Darlington introduces current noise, all of the second transistor's base current passes through Darlington collector and emitter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.