Hi guys,
I've been wondering what transistors should I be using for audio amplifiers (primarily instrument amplifiers but I'm looking for studio gear too) or op amps Including ideal transistors for compression/gating.
Usually I use bc547s, 08x/07x, ne5532 2n7000 j201 or j11x I've realised through studying various diagrams these may be sub optimal.
Can anyone offer any advice?
I've been wondering what transistors should I be using for audio amplifiers (primarily instrument amplifiers but I'm looking for studio gear too) or op amps Including ideal transistors for compression/gating.
Usually I use bc547s, 08x/07x, ne5532 2n7000 j201 or j11x I've realised through studying various diagrams these may be sub optimal.
Can anyone offer any advice?
If you need extremely low noise at low source impedances, look for bipolar transistors with very low base spreading resistance. There have been several threads about this subject.
If you need extremely low noise at high source impedances or at totally unknown source impedances, use high transconductance JFETs. The BF862 is a very good and dirt cheap one, unfortunately it is out of production. Again there are several threads about good alternatives.
When you need neither very low noise nor more than a few hundred milliwatts of power, BC547 and BC557 will do fine.
If you need extremely low noise at high source impedances or at totally unknown source impedances, use high transconductance JFETs. The BF862 is a very good and dirt cheap one, unfortunately it is out of production. Again there are several threads about good alternatives.
When you need neither very low noise nor more than a few hundred milliwatts of power, BC547 and BC557 will do fine.
For an amplifier circuit use what's specified. For opamp use NE5532 will handle most situations, although sometimes you might want lower bias current in which case a JFET input opamp is better.
But remember "best" means nothing until you know all the criteria that are important in that position in that circuit - there is no overall "best" of any device type.
But remember "best" means nothing until you know all the criteria that are important in that position in that circuit - there is no overall "best" of any device type.
I got caught out with this one.
I was using a £3 op amp but wanted something cheaper but faster.
I found a £0.50 TL974 which had a good fast spec.
Put it in the circuit and the input was at 1 volt instead of ground !
I forgot to check the input bias current which was 1uA.
1uA and 1 meg resistor gives +1 volt offset !
So I just put a 10meg resistor from the none inverting input to neg rail and that fixed it.
There is a school of thought which says if 1µA of bias current is a problem for a circuit, then the circuit can be improved (its likely either noisy (impedances too high), or too low bandwidth (too much gain in one opamp stage).
Impedances circa 2k--10k and gains of 20 or less are good targets to have. If you need x100 gain, use two x10 stages (usually gives lower distortion, much higher bandwidth).
High impedances are noisy for three reasons, inherent Johnson noise from the resistors, current noise in the active devices, and susceptibility to cross-talk due to capacitive pickup.
JFET/CMOS opamps fix the current noise issue, good layout reduces cross-talk, but usually its best to avoid high impedance when there's no requirement for it.
Impedances circa 2k--10k and gains of 20 or less are good targets to have. If you need x100 gain, use two x10 stages (usually gives lower distortion, much higher bandwidth).
High impedances are noisy for three reasons, inherent Johnson noise from the resistors, current noise in the active devices, and susceptibility to cross-talk due to capacitive pickup.
JFET/CMOS opamps fix the current noise issue, good layout reduces cross-talk, but usually its best to avoid high impedance when there's no requirement for it.
The answer to your first question has already been provided. Twice.
The answer to your second question has been implied already: it depends on the context. They're different devices.
For homework, find yourself a copy of Horowitz & Hill "The Art of Electronics" and then tell us what you're building.
More specifically, the Philips BD139 - the fairly high fT device (190 MHz) that hasn't been made in years now. (NXP still makes surface mount versions: BCP/BCX56 and complementing -53, in SOT-223 and SOT-89, respectively.) Other manufacturers' BD139s that you can still get tend to be using a larger, more robust die but only have an fT of about 30-50 MHz (though none of them are too forthcoming about this).
The utility of a 1 A transistor in the thermally awful TO-92 package is somewhat dubious, but in practice a part like this will be useful at maybe 10-100 mA. It is the higher-powered (if lower beta) alternative to the BC337 (SMD: BC817), which in turn is more robust than the BC547 small signal transistors or BC550 low-noise @ low current transistors (which manage to keep high beta into the 10s of µA but are less exceptional at higher currents).
Other notable common types include:
2N types (JEDEC) -
2N5087 - low noise, low saturation voltage pnp
2N5088/89 - less exceptional, low voltage npns
2N3904/3906 - classic generic small signal npn/pnp, the latter quite low noise too
2N4401/4403 - similar to BC337/327 but lower beta
2N2222/2N2907 - classic, similar to BC639/640 or BC337/327 but lower voltage, a very common medium power part, also available in TO-92 (PN/MPS) or SMD (PZT)
2N3055 - classic TO-3 power transistor
2N5551/2N5401 - higher voltage small npn/pnp
Japanese types -
2SC945 - npn similar to BC550
2SC1845 (KSC1845) - similar but higher-voltage
2SC1815 - medium power
More Euro types -
BC546/556 - medium voltage small npn/pnp
TIP41/42 - 6 A npn/pnp power, 3 MHz fT
TIP31/32 - 3 A npn/pnp power, 3 MHz fT
BD237/238 - 2 A npn/pnp (medium) power
Chinese transistors -
S9013 - small signal
RTFDS is very much advised.
In terms of opamps, I suggest studying Samuel Groner's measurements. (Unfortunately, measurements for any JRC types are missing entirely, so no NJM2068, 2114, 4556A, 4558 or 4565, and the RC4580 that was tested seems to be one of the lesser '4580s as well.)
TL07x are probably best used for things like inverting line-level stages of moderate gain (feedback R usually >= 22k), where their paltry output driving abilities, modest GBW, relatively high voltage noise and high common-mode distortion will not matter but their decent slew rate, good transfer linearity and (typical FET input) very low current noise will be welcome. Include at least 10 pF or so across feedback R to avoid gain peaking due to input capacitance.
NE5532 are arguably still best used inverting or at least at a decent amount of gain (common mode rejection at high frequencies isn't exactly one of their strong points), but their lower voltage noise and just about infinitely more robust output stage (which drives 600 ohms decently, the part has even been used in small headphone amplifiers) make them a better fit for lower-impedance circuits. Low input impedance distortion and reasonable current noise means they can be used after 50-100k volume pots in typical 16 dB preamps without much issue. They do require some decent supply rail bypassing though. They'll also do OK in MM phono stages although I'd prefer the lower noise single NE5534(A) there.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.