One BJT line amp/buffer

Good, all is well that ends well.
The 317 is an idealized substitute for a follower based on a depletion transistor, with ~zero output impedance, perfect linearity and an almost absent reverse transfer.
It is not perfect though: the SR is finite, limiting the full-power bandwidth, and the noise level is rather high, but for tape playback none of them matters (it would also be the case with other sources).
That said, if you start from scratch without other constraints, a good opamp wired as a follower is probably the solution to go for
Also note that the 317 is certainly not the only one usable as a buffer: more modern, higher performance regulators exist, and some have a very low noise compared to the 317.
Any such list or your special short list recommendation without the space and formfactor constraints?
Additionally, you know that I am looking for the through the hole BJT formfactor. Space is premium here, any adapter would be not suitable.
Last edited:
The selection tools are a terrible mess, bloated with useless and annoying features. There must be one or two candidates hidden in all that crap, but you should make a preselection based on the case, cost, max current and voltage, number of operators, etc.
We can the review your choices, and possibly virtually test them in LTspice if a model is available
I tried 'experience proofed&selected low(er) noise regulators'... but this field is not present in the bloated selection tools :)
The simple application case is presented here. No other one... maybe just the exception of seriously ultra low phase noise microwave oscillators but these will have a different form factor reuqirement.
The LT1083 series might be better the 317, but there is no data regarding the noise.
More modern and evolved types like the LT3045 should be better, but an application in a simple follower is fraught with difficulties, due to their configuration and complexity
Many thanks!
For the moment I will not switch to any LT, even if I may have a couple of LT1083 somewhere in the drawers. Sunday I'll have in visit a set of good ears to evaluate the sonics of 317 in the context of R2R line amp.
I asked for 317 alternatives, just in case the impression of the current performance will not be sufficiently good.
In such case, I think it will be justified to consider a new PCB from scratch and use the better LTs or whatever higher quality followers for audio. But I sincerely hope it is not the case.
Just for info, with the 317 and with the AGFA 19H Reference Tape (old) I can hear the rumble of master Vinyl behind the voice and 0dB tone section. Or maybe the rumble of the left ball bearing tape guide in A77 (just new exchanged)?
For me, this is enough sensitivity and clarity with 317 as line amp and headphone amp. For the SA tests, the tape noise floor is minimum 10dB worse than Line amp's 317 noise floor, and its non visible IMDs in this dynamic range ... it should be sufficient as well. But sonic judgement by good ears cannot be replaced.
Is it any expected any noise improvement if I put a 1mH coil (ferrite core) on the output pin of 317?
That would be dangerous, since it could have weird and unexpected effects on stability.
In addition, to effectively remove some of the noise, the inductor will necessarily curtail the frequency response, which is not huge to begin with: a few tens of kHz, adequate for normal audio, but reducing it further could be problematic
Interestingly and confusing, with all BW and power details considered, the noise analysis simulation using Tina shows a reduced SNR (let's say at 20KHz) with adding the inductance! Maybe I did something wrong.
I will do the LT simulations, but not quite soon.
For now, the coil stays out.

More, the Tina simulations indicate a relevant and easily measurable change of performance at low bass and high treble if the Headphones (300 or even 600 Ohms) are set to Max. But in reality I cannot measure it. Pointless to say I cannot hear it (beyond my ears and my headphones too). However, I can hear it, more like distortions in the 1-10KHz. This is below the Max power of the headphones, and I know how they should behave at the SPL driven.
Dial the headphones down a bit, and the sound quality gets back to normal.


  • Screenshot_46.jpg
    512.7 KB · Views: 46
Ya, a CFP is logically the same as a "White Buffer", which uses two ~NPN or two vacuum tubes, but a CFP does not have the PSRR issue. A CFP can have stability issues and require a Zobel network. But I still think a single op-amp replacing the EF and the pull-down resistor, ie 4 wires is the best solution. For stereo, a dual op-amp replaces two transistors and two resistors.


  • buff bjt vs op.png
    buff bjt vs op.png
    6.8 KB · Views: 43
Last edited: