[[email protected]]

I am making some noise measurements of tubes (valves) and I need some help
in turning observations into measured values. I do not have a true rms
audio millivoltmeter, just a Ferrograph RTS2 test set which has a
millivoltmeter calibrated for sine waves. From tests I have done it clearly
under reads white noise significantly. So I need help inconverting what I
see on the scope (connected to the Ferrograph's scope output) into rms
noise voltage.

For these measurement the Ferrograph is on its 1mV range. On any range, full
scale gives 1V rms out of the scope output. Initially the Ferrograph reads
at one tenth full scale (100uV) but it is clear from the scope trace that
this is mainly hum. So I engage the LF filter and the hum disappears from
the scope leaving what looks to me like regular white noise. The Ferrograph
meter now reads *zero* at this point. I increase the scope resolution so I
can see the noise clearly and it looks like it is about 400mV peak to peak.
If this were a sine wave I would say it was about 141mV rms which
translates to 141uV measured and should have been clearly visible on the
Ferrograph meter - that's why I think the Ferrograph under reads noise.

So, I have what appears to be a 400uV peak to peak noise signal on the
scope. How do I convert this to rms. I have read an article about this that
discusses the noise amplitude distribution (Gaussian) and how, as we all
know, a 2 sigma range encompasses over 68% of the signal range, 4 sigma
encompasses 95% and 6 sigma over 99&. So the question is, is what I see on
my scope likely to be 2, 4, 6 sigma or some other value.

Cheers

ian

[[email protected]]

Bret Ludwig wrote:

See if you can pick up a Ballantine thermocouple AC voltmeter. They
are big and heavy and therefore cheap.

I might if I lived in the US but they don't seem to be available here in the
UK.

Ian

[Patrick Turner]

wrote:

I am making some noise measurements of tubes (valves) and I need some help
in turning observations into measured values. I do not have a true rms
audio millivoltmeter, just a Ferrograph RTS2 test set which has a
millivoltmeter calibrated for sine waves. From tests I have done it clearly
under reads white noise significantly. So I need help inconverting what I
see on the scope (connected to the Ferrograph's scope output) into rms
noise voltage.

For these measurement the Ferrograph is on its 1mV range. On any range, full
scale gives 1V rms out of the scope output. Initially the Ferrograph reads
at one tenth full scale (100uV) but it is clear from the scope trace that
this is mainly hum. So I engage the LF filter and the hum disappears from
the scope leaving what looks to me like regular white noise. The Ferrograph
meter now reads *zero* at this point. I increase the scope resolution so I
can see the noise clearly and it looks like it is about 400mV peak to peak.
If this were a sine wave I would say it was about 141mV rms which
translates to 141uV measured and should have been clearly visible on the
Ferrograph meter - that's why I think the Ferrograph under reads noise.

So, I have what appears to be a 400uV peak to peak noise signal on the
scope. How do I convert this to rms. I have read an article about this that
discusses the noise amplitude distribution (Gaussian) and how, as we all
know, a 2 sigma range encompasses over 68% of the signal range, 4 sigma
encompasses 95% and 6 sigma over 99&. So the question is, is what I see on
my scope likely to be 2, 4, 6 sigma or some other value.

Cheers

ian

To measure 1mV of noise, you need to amplify it by at least 1,000 times
( +60dB ) by using a preamp with response tailored for 20Hz to 20kHz.

So 1mV will appear as 1V to measure.

The preamp you uses need only have an SNR of -60dB or even less and its
contribution to the noise shouldn't contribute substantially to noise.

So a good opamp with NFB will do. I use a 3 tube preamp for my noise
tests.

When you want to know what noise you have, display the noise on the
screen so the
maximum voltage you see ever just fills up the whole screen.
Then discinnect the noise signal and replace the signal with a sine wave
from a sig gene which just fills up the screen.
Say this measures 0.5Vrms. The the pk V = 0.7V, and P-P - 1.4V.

The meter you have to read Vrms will tell you lies and BS unless
it is a genuine Vrms meter and has at least 20Hz to 20kHz of bw.

So most DVM are in error with full bw of noise.

White noise rises in amplitude with rising F, so this compounds the
error, so the P-P
reading is probably better.

Usually if you have a good 6SN7 there will be say 3uVrms at its grid
input circuit if the grid is grounded and
the heaters are dc.

If its gain = 16, then there will be 48Vrms of noise at its anode,
and after amplifying 1,000 times, you get 48mVrms at the output of your
preamp, enough to easily
measure accurately.

If you measure 135mV P-P, then there would have been 8.46uV P-P at the
tube grid input.

Its an average reading you should expect with a 6SN7.

But some will measure 5 times more noise, so they are usually worn out
and gassy.

The really quiet ones might get down to 1/3 of the average, but they are
rare.

When noise voltages are added, the sum = sq.rt of sum of each noise
voltage squared.

So 1mV + 5mV gives a total of 5.099mV, so the noise of a following
preamp
won't much affect what you are trying to measure as long as its
input noise is say 5 times less than the noise you want to measure,
so if the preamp has 3 uV of noise, you can measure 15 uV OK.

Tubes have all sorts of noise which is disconcerting, and
some low frequency sputtery kinds of noise will send meter needles
swinging wildly and confuse DVMs.

Listen to the noise via amp and speaker.
Tap the tube for microphony.

Soon after you have tested 20 random samples you will know which ones
you will reserve for input tube duty in a preamp, because its usually
always the input tube which determines the SNR of the amp you're
building,
especially when its a phono or microphone amp.

Patrick Turner.

[[email protected]]

Bret Ludwig wrote:

On Jul 9, 1:24 am, wrote:
Bret Ludwig wrote:

See if you can pick up a Ballantine thermocouple AC voltmeter. They
are big and heavy and therefore cheap.

I might if I lived in the US but they don't seem to be available here in
the UK.

Ian

The shipping would exceed the value of the meter, but that isn't very
much.

Most likely. I have searched eBay but there does not seem to be any on there
right now.

Ian