[Sean B]

Hello,

Do different types of (let's narrow it down to small-signal devices) BJTs sound significantly different in anyone's experience, in various types of audio circuits? They all have that exponential relationship between base-emitter voltage and collector current. I'm wondering what could account for differences in sound, assuming those differences exist.

I recently built a BJT emitter follower with a another BJT as a current source feeding it, and even though it measures very well (junk down about 105 dB from the fundamental test tone) it added a noticeable brightness to audio passing through it. It just confounds me that this could be audible.



Thanks,

Sean B

[Don Pearce[_3_]]

On Fri, 13 Oct 2017 23:37:27 -0700 (PDT), Sean B
wrote:

Hello,

Do different types of (let's narrow it down to small-signal devices) BJTs sound significantly different in anyone's experience, in various types of audio circuits? They all have that exponential relationship between base-emitter voltage and collector current. I'm wondering what could account for differences in sound, assuming those differences exist.

I recently built a BJT emitter follower with a another BJT as a current source feeding it, and even though it measures very well (junk down about 105 dB from the fundamental test tone) it added a noticeable brightness to audio passing through it. It just confounds me that this could be audible.



Thanks,

Sean B

Much more likely (if there genuinely is a brightening, which I somehow
doubt), that the reduced output impedance is flattening the top end of
a subsequent capacitive load - maybe the result of some cable.

I'd suggest you make an objective frequency response measurement to
see if there actually is an effect.

Your figure of junk at -105dB surprises me somewhat. Emitter followers
are just not that linear - the internal emitter resistance is roughly
26/Ic(mA) so a signal is always modulating the output voltage,
resulting in even + odd order harmonics.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[Scott Dorsey]

Sean B wrote:
Do different types of (let's narrow it down to small-signal devices) BJTs s=
ound significantly different in anyone's experience, in various types of au=
dio circuits? They all have that exponential relationship between base-emi=
tter voltage and collector current. I'm wondering what could account for d=
ifferences in sound, assuming those differences exist.

Yes, nonlinearities differ between transistors, but if you look at the plots
most of the real differences are at low levels. So much of the secret is in
setting the bias up high enough so that you're never working down there.

The rest of the secret is in negative feedback, which can totally eliminate
part to part differences if you're careful enough.

I recently built a BJT emitter follower with a another BJT as a current sou=
rce feeding it, and even though it measures very well (junk down about 105 =
dB from the fundamental test tone) it added a noticeable brightness to audi=
o passing through it. It just confounds me that this could be audible.

1. Likely you have second harmonic distortion at low levels. Measure it
and see.

2. Likely the distortion is much worse at low levels than at maximum level
and you may have measured it at maximum level (because that's what people
do).

3. It's possible that the lower output impedance affected whatever you were
driving, also, and it's not the circuit causing the difference but the
interface.

A simple follower has 100% feedback and should have very very low distortion
as you note. So start changing things. Replace that current source with a
resistor. Change the source impedance using a shunt resistor. Pull the
transistor out and stick a TIP50 in (a personal favorite trick since a TIP50
has just awful linearity).
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce[_3_]]

On 14 Oct 2017 08:27:00 -0400, (Scott Dorsey) wrote:



A simple follower has 100% feedback and should have very very low distortion
as you note. So start changing things. Replace that current source with a
resistor. Change the source impedance using a shunt resistor. Pull the
transistor out and stick a TIP50 in (a personal favorite trick since a TIP50
has just awful linearity).
--scott

A simple follower does have 100% feedback, but unfortunately not from
the right point. The current-dependent emitter resistance fall outside
the loop, in series with the output, so it will add distortion which
gets worse as the load resistance decreases. It is this resistance
that prevents emitter followers from achieving unity gain - they are
always a bit below.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[Scott Dorsey]

Don Pearce wrote:
On 14 Oct 2017 08:27:00 -0400, (Scott Dorsey) wrote:

A simple follower has 100% feedback and should have very very low distortion
as you note. So start changing things. Replace that current source with a
resistor. Change the source impedance using a shunt resistor. Pull the
transistor out and stick a TIP50 in (a personal favorite trick since a TIP50
has just awful linearity).

A simple follower does have 100% feedback, but unfortunately not from
the right point. The current-dependent emitter resistance fall outside
the loop, in series with the output, so it will add distortion which
gets worse as the load resistance decreases. It is this resistance
that prevents emitter followers from achieving unity gain - they are
always a bit below.

Yes, this is true. And you can reduce the emitter resistance to make
it better into a lower load impedance, but then you can't swing as much
voltage on the output.

But, if this is the issue, you can tell because the distortion character
will change with the load impedance. You can see it on a scope and you
can hear it change as you add shunt resistances to your load.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce[_3_]]

On 14 Oct 2017 09:54:25 -0400, (Scott Dorsey) wrote:

Don Pearce wrote:
On 14 Oct 2017 08:27:00 -0400, (Scott Dorsey) wrote:

A simple follower has 100% feedback and should have very very low distortion
as you note. So start changing things. Replace that current source with a
resistor. Change the source impedance using a shunt resistor. Pull the
transistor out and stick a TIP50 in (a personal favorite trick since a TIP50
has just awful linearity).

A simple follower does have 100% feedback, but unfortunately not from
the right point. The current-dependent emitter resistance fall outside
the loop, in series with the output, so it will add distortion which
gets worse as the load resistance decreases. It is this resistance
that prevents emitter followers from achieving unity gain - they are
always a bit below.

Yes, this is true. And you can reduce the emitter resistance to make
it better into a lower load impedance, but then you can't swing as much
voltage on the output.

But, if this is the issue, you can tell because the distortion character
will change with the load impedance. You can see it on a scope and you
can hear it change as you add shunt resistances to your load.
--scott

Yep. More current makes this better. But better yet is some gain in
front and an external feedback loop that comes from the emitter pin.
Op-amp anybody?

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[[email protected]]

On Saturday, October 14, 2017 at 10:33:23 AM UTC-4, Don Pearce wrote:
On 14 Oct 2017 09:54:25 -0400, (Scott Dorsey) wrote:

Don Pearce wrote:
On 14 Oct 2017 08:27:00 -0400, (Scott Dorsey) wrote:

A simple follower has 100% feedback and should have very very low distortion
as you note. So start changing things. Replace that current source with a
resistor. Change the source impedance using a shunt resistor. Pull the
transistor out and stick a TIP50 in (a personal favorite trick since a TIP50
has just awful linearity).

A simple follower does have 100% feedback, but unfortunately not from
the right point. The current-dependent emitter resistance fall outside
the loop, in series with the output, so it will add distortion which
gets worse as the load resistance decreases. It is this resistance
that prevents emitter followers from achieving unity gain - they are
always a bit below.

Yes, this is true. And you can reduce the emitter resistance to make
it better into a lower load impedance, but then you can't swing as much
voltage on the output.

But, if this is the issue, you can tell because the distortion character
will change with the load impedance. You can see it on a scope and you
can hear it change as you add shunt resistances to your load.
--scott

Yep. More current makes this better. But better yet is some gain in
front and an external feedback loop that comes from the emitter pin.
Op-amp anybody?

The phono cartridge preamp I made used (2) Op-Amps per channel, via Walter Jung's design. Even published in (defunct) High Fidelity Magazine. Impressive.

Jack


d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[Scott Dorsey]

Don Pearce wrote:
On 14 Oct 2017 09:54:25 -0400, (Scott Dorsey) wrote:

But, if this is the issue, you can tell because the distortion character
will change with the load impedance. You can see it on a scope and you
can hear it change as you add shunt resistances to your load.

Yep. More current makes this better. But better yet is some gain in
front and an external feedback loop that comes from the emitter pin.
Op-amp anybody?

No need to go that far, the simple Ring of Three circuit family can be found
in all sorts of applications like that. Gain is cheap in the solid state
world, so it's frequently to your advantage to trade it for linearity.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Sean B]

Thanks for the ideas!

The follower was biased at 9mA, driving about 7 volts peak-to-peak into a lightish 4.7k load. I'm measuring with an m-audio sound card which has second harmonic about 120 dB down at an output around 3V rms @ 1 kHz (sine).

One other thing I wonder about: a lot of components made today have iron in their lead wires, can this cause noticeable distortion compared to all copper?

S

[Scott Dorsey]

Sean B wrote:
Thanks for the ideas!

The follower was biased at 9mA, driving about 7 volts peak-to-peak into a lightish 4.7k load. I'm measuring with an m-audio sound card which has second harmonic about 120 dB down at an output around 3V rms @ 1 kHz (sine).

Do you really trust that card that much? I might, if I'd measured it with
accurate enough references, but I wouldn't until I'd done that.

One other thing I wonder about: a lot of components made today have iron in their lead wires, can this cause noticeable distortion compared to all copper?

It can cause measurable distortion if there is considerable change in current
with respect to time. This is an issue for things like power amplifier stages.
There is some discussion of it and some examples with measurements in Doug
Self's book. I am inclined to think this is mostly not an issue but you can
get yourself into situations where it might be.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Phil Allison[_4_]]

Sean B wrote:

---------------
One other thing I wonder about: a lot of components made today have
iron in their lead wires, can this cause noticeable distortion compared
to all copper?



** So you are an audiophool and a looney - right ?

Plenty of forums exist that cater for people with your mental disability.

Not here though.


..... Phil

[Phil Allison[_4_]]

Sean B wrote:

----------------


I recently built a BJT emitter follower with a another BJT as a current source feeding it, and even though it measures very well (junk down about 105 dB from the fundamental test tone) it added a noticeable brightness to audio passing through it. It just confounds me that this could be audible.


** You post explains nothing.

How about a proper description of what the set up is and how you came to your conclusion.



...... Phil

[PStamler]

Don wrote:

"Unless you use an integrated op amp. 5532 is probably my favourite do-everything op amp. The count becomes one op amp and two resistors. "

If you want maximum performance and stability from the 5532, you'll probably need a bit more than two resistors, though not a *lot* more. An additional 100R resistor in series with the output helps keep the circuit isolated from capacitative loads; rolling off the response at about 100kHz by putting a capacitor in parallel with the feedback resistor also helps stability, and decoupling the opamp's power pins with stacked film or NP0 capacitors is usually necessary for stability if you use more than one package on a board. It may be necessary to add some electrolytic caps at the input and output for coupling; the 5532 is a remarkably good opsmp for audio, but its DC performance is often not so good.

The component count can still be pretty low (say three resistors, three small caps and maybe two or three electrolytics -- maybe also electrolytics on the supply rails). That's a lot fewer parts than a discrete circuit.

And yes, you can make a quite-good phono pramp from the two halves of a 5532. Put an RIAA circuit in (either passive or active) and Bob's your uncle. Make the RIAA switchable if you want the option of flat playback (some scratch-reduction software, including iZotope's Rx, works better on flat transfers; if you have Adobe Audition you can add the RIAA compensation easily in software after descratching.)

Peace,
Paul