May 2015
Can one of your experienced electronic engineers recommend replacement plastic transistors (maybe a la PN2907) which are easily available, in order to substitute Q1, Q2, Q3, and Q4? I want to totally eliminate 2SB54 transistors WITHOUT any draconic and/or extreme changes to the original diagram (below) for this (push-pull) phone amplifier I'm using, by attaching a suction cup to my (landline) phone's receiver.
I'm well aware of the existence of other phone amplifiers (built around the LM386) which are easier to build with fewer parts, but I'm only interested in this diagram, as I’ve already built it in the past — not only as a phone amp with ample volume (which also works very well as a P.A. system), but as a wired intercom.
Please log in to post an answer.
Mr. Franklin desires to reproduce a known circuit design that uses four 2SB54 transistors. The 2SB54 is an obsolete part. Furthermore, it is a germanium device. Mr. Franklin asks if a 2N2907 or similar silicon PNP device could be used instead.
Substituting silicon devices for germanium devices is a non-trivial exercise:
• The no-signal emitter-base junction voltage of germanium devices is about 0.3v, while that of silicon devices is about 0.7v. In a linear circuit such as that presented here, bias voltage appropriate to the transistor material type must be applied to overcome the junction voltage offset.
• Germanium transistors exhibit current leakage. This is often sufficient to provide self-biasing. Silicon devices have negligible leakage in the circuit under consideration, so external biasing must be considered.
The simplest thing to do is to replace the 2SB54 with another germanium part. The NTE100 device, manufactured by NTE Electronics, has characteristics that are similar to those of the 2SB54. It is available from multiple distributors including Newark Electronics, Fry’s Electronics, and Online Components, or you might try eBay. Pricing runs from about four to seven dollars per device. A data sheet, which includes a basing diagram, is available from NTE Electronics.
Good luck.
Those transistors date back in time quite a ways, as does the circuit! The 2SB54 is a Germanium, PNP unit. RadioShack has a drop-in generic on their website as part #55051964.
You could also toss out everything between the microphone and the speaker and connect a LM380 integrated circuit. I have an old transistor substitution guide in storage so contact me if you need a further referral. Good luck.
Q1 and Q2: the circuit is a self-biasing design. MPSA56 should work fine, even though these are silicon transistors instead of germanium.
The output stage (Q3/Q4): 2N4918 or BD136 would be my preference. You will need to increase the value of R6 (try 330 ohms) to reduce crossover distortion due to the larger Vbe drop of the silicon transistors. Note that this output stage has no temperature compensation and is subject to thermal runaway.
Better transistor designs of this era used a thermistor for temperature compensation. See Wikipedia’s thermal runaway entry, bipolar junction transistors, for more insight. I would still LM386, which eliminates the (likely) hard to get transformers too. More power? Look at the numerous class-D single chip power amps.
The circuit is well designed with feedback in the first stage to stabilize its current and an emitter resistor to limit the current in Q2. The output transistors, Q3 and Q4 are class B so draw no current until driven by audio. I simulated the Q1, Q2 circuit and it works perfectly with no changes using 2N2907 transistors. Q2 collector current is 4.8 mA which is sufficient to drive the output transistors to watts of power.
The main problem is the transistors are germanium and not silicon. So, I doubt you could plug in silicon transistors and have it work properly [although if you’re desperate, you might give it a try anyway].
The transistors have pretty much been discontinued from what I could tell. NTE makes a sub that should work for you — NTE102. You should be able to find them at several places. I checked Allied and they had some in stock — but on the pricey side.
You should be able to find a newer circuit at some electronics site that uses newer transistors or op amps.
I would consider the 2N3906. The Hfe (100 min, 300 max) puts this device near the 2SB54 specified Hfe of 140.
The 2SB54 is a fairly ordinary Germanium PNP transistor. It could be replaced with any general purpose Silicon PNP like the PN2907, 2N3906 or BC327. You will need some changes as silicon transistors are harsher than germanium types:
(a) you will have to increase the value of R6 to 1K to bias the output stage (maybe 1K5 to get more power out) keep an eye on the quiescent current, it should be < 10mA.
(b) (optional) to improve crossover distortion , remove R5, and use 33K resistors between base and collector of each of Q3 and Q4 (just like R1)
(c) (optional) to make it less tinny, put a 22ohm in series with C3 (other values will work like a tone control)
(d) (optional) to flatten out the response , try a 47K for R1 , (values in the range 10K to 100K could be tried) this also affects the gain and interacts with the source impedance. The above values were determined by simulation to give a better result than the original values, however the simulation uses perfect transformers. If you get RF oscillations, try a 1nF from base to collector of Q2 or 470pF from base to collector of Q3, Q4, or ferrite beads on the emitters of Q1 or Q2.
The swap you suggest will require some adjustment of transistor bias. The 2SB54 are Ge. material with 0.15 to 0.2V base emitter bias. The PN2907 and similar are Si. with a 0.65 to 0.7V base emitter bias requirement. R2 and R6 will likely need to be increased in value. Quick estimate for R6 at 0.7V is around 1.2K The collector feedback biasing makes the calculation of R2 value more difficult and might not have to be changed due to the feedback.
Search for similar circuits using silicon transistors to get some better resistor values. Note that Q3 and Q4 require a little forward bias to prevent crossover distortion. I would expect considerable distortion if the biasing resistors are not modified to account for differences in GE. vs. Si material GWS
A 2N3906 should work OK. It has the same general characteristics as the 2SB54 transistors, except for being silicon instead of germanium, greater power dissipation and slightly higher hfe (100 as opposed to 80).
I would try a PN2222 which is an old standby PNP transistor. I don’t think there is anything special about the 2SB54, it is just older. Jameco sells them for ten cents a piece. Part no. 28628. Download the data sheet and watch the pin locations.
