[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Phil Allison wrote:
"Ian Bell"

Thanks for the input Patrick. I know about Johnson noise and in plate
resistors it is not much of an issue as it is swamped by other noise
sources. What I am concerned about is excess noise which is voltage
dependent. In the few places I have seen this specified it is typically
0.1uV/V so with a plate resistor with 150V across it there is possibly
15uV of noise due to this effect - enough to warrant a change of resistor
type.
** No it is not - you hee hawwwing ASS


Simple negation is no argument. I presently have about 65uV of noise
from the circuit under test. Eliminating a 15uV component of it is
worthwhile.

Cheers

Ian
OK, you have 65uV of noise. I think you said your amp has a gain of
about 15.

Suppose the input signal was 4.3 mV, then output signal would be 65mV,
and SNR = -60dB unweighted.

Removing 15uV of the noise won't make a significant SNR improvement.
Of course not. But it is just one of two such sources and I am simply
taking your recommendation of identifying each source in turn and
addressing them in turn.

Cheers

Ian
If the output signal is to be fed into a typical power amp needing 1Vrms
for clipping at 50W/6 ohms, then the power amp gain = 17.3x, 25dB,
and 65uV of applied noise will become 1.124mV not something that would
be improved by lowering to 0.87mV if 15uV was removed.

65mV of signal at the power amp becomes 1.124Vrms, enough to make 0.21
watts into a 6 ohm speaker, and if the 6 ohm speaker sensititivity is
90dB for 1 watt with 2.45Vrms, then the 0.21 watts gives about 83dB SPL
and is loud enough for most ppl, and allows a considerable headroom.

I suggest you work out the whole basic working of your system and amps
using a block diagram with signal levels or else quit mucking around
theorizing, and just build it and see if its quiet enough. After than
you can tweak things to make things more silent.
That is EXACTLY what I have done. The thing is on my bench now and the
output noise I quoted is as measured. You recommended I identify each
noise source and tackle them in turn which is just what I am trying to
do. I have fixed the low frequency noise as described in other posts and
now I am looking at other sources. Excess noise is just one of them.
From looking at metal film resistor specs it seemed they could be a
contributory factor and that wirewound types would be significantly
quieter. The only problem is finding them in the high values I need;
hence my original question on this topic, to which you and others
replied either that metal film resistors are not noisy or that the
noise they contribute is not significant.

Cheers

Ian

If you are worried by magnetically induced noise in WW R, then use a
pair of equal value and same brand with one reverse connected and close
to the other. Thus the magnetic noise of one cancels that in the other.


Interesting idea. Unfortunately it means finding WW resistors of double
the value and it's hard enough finding ones of the original value.


The excess noise isn't something so easily measured. I've never worried
about it because other sources of noise would always be greater. for
input stages of high gain amps I do like the cascode circuit using a
high gm j-fet like 2sk369 to drive a cathode of the triode with its grid
bypassed well to 0V. The triode load is usually a 22k R, and I will
typically use a pair of metal film 47k in parallel, and each 1W rated.
If excess noise exists, I ain't aware of it.


I have come to the conclusion that higher wattage metal film resistors
have significantly less excess noise and I have found one by Multicomp
that has nearly 10 times less excess noise in its 0.5W version than the
0.1uV/V typical value for other makes.

In the j-fet + triode cascode circuit made so famous by Allen Wright in
1988 in his FVP for MC, the gain at the anode of the triode is simply
fet gm x anode RL, and as the 2SK369 gm at 5mA = 40mA/V, if RL at the
anode = 20k, the gain = 800.
I have a source resistance fro local current FB to lower the gain to
about 250, and as this source R value is low, there is little added
noise from such an R.


Indeed.


So 0.4mV at the gate becomes 100mV at the anode. The first resistor to
shunt the amplified signal and able to add noise is the anode load
resistance, and afaik, the 20k does not significantly add any noise. In
the cascode case, the anode Ra is high because the effective Ra with a
an unbypassed cathode R = µ x Rk, and in the cascode with j-fet, Rk is
the very high Rd of the fet. So at the triode anode noise will be
determined by triode load value, and the j-fet gate input noise x the
cascode voltage gain. If gate ipnut noise was 0.2uV, then gain of 250
will increase it to 50uV. The reference signal will be 100mV, and SNR is
-66dB, unweighted, 20Hz to 20kHz.

If the gate noise was 0.1uV, SNR would be -72dB, and if the input signal
was 4mV instead of 0.4mV, then SNR would improve by 20dB.

In the case of my phono preamp, the RIAA filter improves the SNR.



Cheers

Ian

Patrick Turner.