Sign in with
Sign up | Sign in
Your question

how do I measure S/N?

Last response: in Home Audio
Share
May 8, 2004 9:00:39 AM

Archived from groups: rec.audio.tech (More info?)

I just finished building an audio pre-amp, and it is quiet, quieter than the
background noise on some CDs, but I'd like to measure the signal to noise
ratio. Do I measure the voltage at max signal, then measure the noise with no
signal, but same input pot position? Short the input RCA connections? Also,
another question, would caps shunting the power supply diodes reduce hum, or
RF type noise? There is some hum at max volume, (120 or 60 hz., I dunno)but
that's with max rotation of the input pot, whereas, 12:00 position is really
loud music. I want to be able to quantify the SNR so tweaking can be done
productively.
Thanks,
Don

More about : measure

Anonymous
a b ) Power supply
May 8, 2004 9:00:40 AM

Archived from groups: rec.audio.tech (More info?)

"Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
news:XPZmc.2660$z06.681183@attbi_s01...
> I just finished building an audio pre-amp, and it is quiet, quieter than
the
> background noise on some CDs, but I'd like to measure the signal to noise
> ratio. Do I measure the voltage at max signal, then measure the noise with
no
> signal, but same input pot position? Short the input RCA connections?
Also,
> another question, would caps shunting the power supply diodes reduce hum,
or
> RF type noise? There is some hum at max volume, (120 or 60 hz., I
dunno)but
> that's with max rotation of the input pot, whereas, 12:00 position is
really
> loud music. I want to be able to quantify the SNR so tweaking can be done
> productively.
> Thanks,
> Don
>
You could buy an oscilloscope to measure the noise, or you could use the
money to buy some music instead. Your choice.
Anonymous
a b ) Power supply
May 8, 2004 11:00:31 AM

Archived from groups: rec.audio.tech (More info?)

Don wrote:

> I just finished building an audio pre-amp, and it is quiet, quieter
> than the background noise on some CDs, but I'd like to measure the
> signal to noise ratio.


> Do I measure the voltage at max signal, then
> measure the noise with no signal, but same input pot position?

Pretty much.

> Short the input RCA connections?

No, use either the actual source, or a resistor that has the same resistance
as the impedance of the source, or a standard impedance for the kind of
preamp are testing. For example, the standard terminating resistor for
measuring mic preamps is usually 300 ohms though 150 ohms is used sometimes.

A good terminating resistor for measuring a phono preamp for a standard
moving-magnet cartridge would be 1000 ohms.

> Also, another question, would caps shunting the power supply diodes
reduce hum, or RF type noise?

Either depending on the kind of cap.

>There is some hum at max volume, (120 or 60 hz., I dunno)but that's with
> max rotation of the input pot, whereas, 12:00 position is really loud
> music.

A little hum is not a serious problem, particularly if the basic hiss level
is higher.

> I want to be able to quantify the SNR so tweaking can be done
productively. Thanks,

Objective quantification is good because it is stable, reliable, and it can
be quicker.

The suggestion that you use a sound card is a good one.

Signal to noise ratio is the ratio of the amplitude largest undistorted
signal, to the amplitude of the background noise. There are a number of
different but related ways to measure it. Measuring the signal voltage is
the easier part, because it is big and a pure tone in the midrange - usually
1000 Hz, though 400 Hz and other frequencies are sometimes used.

Measuring noise is a bit more tricky. Noise in the measuring device can be a
problem because the noise level of the unit under test (UUT) may be a low
voltage. Furthermore, noise measurements are only relevant if taken over a
standard bandwidth, such as 20 to 20 KHz at the -3 dB points, with 6
dB/octave slopes.

Finally, noise of a given amplitude doesn't always sound the same. Noise
generated by a circuit can have its frequency response shaped by that
circuit, and noise with different spectral shaping sounds different. 1
millivolt of noise at mostly low frequencies sounds different than 1
millivolt of noise concentrated in the midrange. If you change the circuit
to change the spectral shaping of the noise, you might not change the
amplitude of the noise, but you will change how much it bothers you.

The suggestion that you use a sound card and software for your measurement
is a good one.

You can use some inexpensive audio test software to turn your sound card
into test equipment.

Here's the free download:

http://audio.rightmark.org/download.shtml

This software will do a quick automated analysis of any piece of audio gear.
It was designed to test sound cards by looping the output of the sound card
to its input. You should do that to learn how the software works and find
out how good your soundcard is.

However, any piece of gear you hook between the output and input of your
sound card can be tested with this software.

When you test a preamp, you would probably want to build a simple
attenuator composed of couple of resistors to knock the 1 volt or so output
of your sound card down so that it would be suitable for the input of your
preamp.

If your preamp is a phono preamp, circuits like this one are specially made
for the purpose:

http://www.hagtech.com/pdf/riaa.pdf

If your preamp is really good and quiet, it might take a little trickery to
do the job. For example, the maximum output of your preamp might be 10
volts, but your sound card might clip if the input voltage is over 1 volt.

Here's what you do. You must have some kind of signal generator and
voltmeter. You use the voltmeter to adjust the output of your signal
generator to say a tenth of a volt. You measure this with your sound card
and then compare that to the output of your preamp and use the ratio to get
an accurate measurement of voltage.
Related resources
Anonymous
a b ) Power supply
May 8, 2004 1:57:03 PM

Archived from groups: rec.audio.tech (More info?)

Linear Technology has an excellent article on Noise Measurement -- it pops
up as an addendum to a couple of their Application Notes -- take a look at
Application Note 83 -- Performance Verification of Low Noise Regulators --
the http://www.linear.com/pdf/an83f.pdf
the information at the end of the note describes the steps you have to take
to acurately measure "noise". (The same general methods are also described
in other apnotes from other IC vendors.)

"Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
news:XPZmc.2660$z06.681183@attbi_s01...
> I just finished building an audio pre-amp, and it is quiet, quieter than
the
> background noise on some CDs, but I'd like to measure the signal to noise
> ratio. Do I measure the voltage at max signal, then measure the noise with
no
> signal, but same input pot position? Short the input RCA connections?
Also,
> another question, would caps shunting the power supply diodes reduce hum,
or
> RF type noise? There is some hum at max volume, (120 or 60 hz., I
dunno)but
> that's with max rotation of the input pot, whereas, 12:00 position is
really
> loud music. I want to be able to quantify the SNR so tweaking can be done
> productively.
> Thanks,
> Don
>
>
Anonymous
a b ) Power supply
May 8, 2004 10:43:54 PM

Archived from groups: rec.audio.tech (More info?)

"Mark A" <ma@switchboard.net> wrote in message
news:1O_mc.1459$hy5.39138@news.uswest.net...
> You could buy an oscilloscope to measure the noise, or you could use the
> money to buy some music instead. Your choice.

Most CRO's aren't that useful for measuring noise at CD levels. Better to
use a good soundcard and software. Even a cheap SB Audigy will give better
results than a cheap CRO.

TonyP.
Anonymous
a b ) Power supply
May 8, 2004 10:43:55 PM

Archived from groups: rec.audio.tech (More info?)

Well, I haven't had to do this very often, so I may be a bit rusty, also
there are engineers lurking here so anyway...

You need to decide what is your reference. For example, a 1 volt sine wave
output (most preamps will go to at least 10 volts maximum, but 1 volt output
is a pretty standard level).

Connect the output of the preamp to a VTVM. Set the meter to read full-scale
at your reference level.

Reduce your input (or short the input).

Turn the sensitivity knob of the VTVM (counter-clockwise) to increase the
sensitivity. Each click of the knob will be a certain number of dB, probably
10 or 20 dB for each click. Keep increasing the sensitivity until the noise
deflects better than 1/2 scale. Add the number of dB x clicks, (4 clicks =
40 db, for example) plus the difference of where the needle is now to full
scale, so in this example you'd get 46 dB.

Another way is to measure using a true-RMS voltmeter, measure the output at
reference, remove or short the input, measure the residual, then apply the
following formula:

dB = 20 x LOG (Vout / Vin)

That is 20 times the LOG of the one voltage divided by the other. A
scientific calculator has these functions.



Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"TonyP" <TonyP@optus.net.com.au> wrote in message
news:409c9dde$0$442$afc38c87@news.optusnet.com.au...
>
> "Mark A" <ma@switchboard.net> wrote in message
> news:1O_mc.1459$hy5.39138@news.uswest.net...
> > You could buy an oscilloscope to measure the noise, or you could use the
> > money to buy some music instead. Your choice.
>
> Most CRO's aren't that useful for measuring noise at CD levels. Better to
> use a good soundcard and software. Even a cheap SB Audigy will give better
> results than a cheap CRO.
>
> TonyP.
>
>
May 9, 2004 8:11:17 AM

Archived from groups: rec.audio.tech (More info?)

I'm the original poster. Thanks to every responder for all your help. I don't
have a VTVM, but I do have a Fluke 8600a DVM with an input impedance of >1000
megohms. : ) in the 200 millivolt range.
-Don


In article <XPZmc.2660$z06.681183@attbi_s01>, TAKE-OUTdobrien25@comcast.net
says...
>
>I just finished building an audio pre-amp, and it is quiet, quieter than the
>background noise on some CDs, but I'd like to measure the signal to noise
>ratio. Do I measure the voltage at max signal, then measure the noise with no
>signal, but same input pot position? Short the input RCA connections? Also,
>another question, would caps shunting the power supply diodes reduce hum, or
>RF type noise? There is some hum at max volume, (120 or 60 hz., I dunno)but
>that's with max rotation of the input pot, whereas, 12:00 position is really
>loud music. I want to be able to quantify the SNR so tweaking can be done
>productively.
>Thanks,
>Don
>
>
Anonymous
a b ) Power supply
May 9, 2004 8:18:10 PM

Archived from groups: rec.audio.tech (More info?)

"Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
news:Fbinc.8757$iF6.981550@attbi_s02...
> I'm the original poster. Thanks to every responder for all your
help. I don't
> have a VTVM, but I do have a Fluke 8600a DVM with an input impedance
of >1000
> megohms. : ) in the 200 millivolt range.
> -Don

That's on the DC ranges. On the AC ranges the input impedance is 2
Megohms.

Norm Strong
Anonymous
a b ) Power supply
May 9, 2004 8:18:11 PM

Archived from groups: rec.audio.tech (More info?)

normanstrong wrote:
> "Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
> news:Fbinc.8757$iF6.981550@attbi_s02...
>> I'm the original poster. Thanks to every responder for all your
>> help. I don't have a VTVM, but I do have a Fluke 8600a DVM with an
>> input impedance of >1000 megohms. : ) in the 200 millivolt range.
>> -Don
>
> That's on the DC ranges. On the AC ranges the input impedance is 2
> Megohms.

Is it one of your babies, Norm?

;-)
May 10, 2004 8:06:56 AM

Archived from groups: rec.audio.tech (More info?)

In article <6Rsnc.57945$Ik.4388638@attbi_s53>, normanstrong@comcast.net
says...
>
>
>"Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
>news:Fbinc.8757$iF6.981550@attbi_s02...
>> I'm the original poster. Thanks to every responder for all your
>help. I don't
>> have a VTVM, but I do have a Fluke 8600a DVM with an input impedance
>of >1000
>> megohms. : ) in the 200 millivolt range.
>> -Don
>
>That's on the DC ranges. On the AC ranges the input impedance is 2
>Megohms.
>
>Norm Strong
>
>
See, I really am a newbie. Or sleep deprived. Anyways, I assume the Fluke is
ok to measure these voltages, lacking a VTVM. At least I knew the issue was
one of impedance.
Anonymous
a b ) Power supply
May 10, 2004 12:42:53 PM

Archived from groups: rec.audio.tech (More info?)

Don wrote:
> In article <6Rsnc.57945$Ik.4388638@attbi_s53>,
> normanstrong@comcast.net says...
>>
>>
>> "Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
>> news:Fbinc.8757$iF6.981550@attbi_s02...
>>> I'm the original poster. Thanks to every responder for all your
>> help. I don't
>>> have a VTVM, but I do have a Fluke 8600a DVM with an input impedance
>> of >1000
>>> megohms. : ) in the 200 millivolt range.
>>> -Don
>>
>> That's on the DC ranges. On the AC ranges the input impedance is 2
>> Megohms.
>>
>> Norm Strong
>>
>>
> See, I really am a newbie. Or sleep deprived. Anyways, I assume the
> Fluke is ok to measure these voltages, lacking a VTVM.

YES!

VTVMs were the logical predecessors of the DVM.

> At least I knew the issue was one of impedance.

Actually, not. It's a matter of voltage.
Anonymous
a b ) Power supply
May 10, 2004 11:57:53 PM

Archived from groups: rec.audio.tech (More info?)

Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter, was it?

Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"Arny Krueger" <arnyk@hotpop.com> wrote in message
news:I96dnd0CSd0t5QLd4p2dnA@comcast.com...
> Don wrote:
> > In article <6Rsnc.57945$Ik.4388638@attbi_s53>,
> > normanstrong@comcast.net says...
> >>
> >>
> >> "Don" <TAKE-OUTdobrien25@comcast.net> wrote in message
> >> news:Fbinc.8757$iF6.981550@attbi_s02...
> >>> I'm the original poster. Thanks to every responder for all your
> >> help. I don't
> >>> have a VTVM, but I do have a Fluke 8600a DVM with an input impedance
> >> of >1000
> >>> megohms. : ) in the 200 millivolt range.
> >>> -Don
> >>
> >> That's on the DC ranges. On the AC ranges the input impedance is 2
> >> Megohms.
> >>
> >> Norm Strong
> >>
> >>
> > See, I really am a newbie. Or sleep deprived. Anyways, I assume the
> > Fluke is ok to measure these voltages, lacking a VTVM.
>
> YES!
>
> VTVMs were the logical predecessors of the DVM.
>
> > At least I knew the issue was one of impedance.
>
> Actually, not. It's a matter of voltage.
>
>
Anonymous
a b ) Power supply
May 11, 2004 10:58:20 AM

Archived from groups: rec.audio.tech (More info?)

Mark D. Zacharias wrote:

> Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter,
> was it?


No, the Fluke 77 series was/is not a true RMS meter.

I think that you have to go up into Fluke's 80 series to get that feature.
Anonymous
a b ) Power supply
May 12, 2004 3:09:50 AM

Archived from groups: rec.audio.tech (More info?)

"Arny Krueger" <arnyk@hotpop.com> wrote in message
news:QKSdnUOC4sJQLD3dRVn-gw@comcast.com...
> Mark D. Zacharias wrote:
> > Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter,
> > was it?
> No, the Fluke 77 series was/is not a true RMS meter.
> I think that you have to go up into Fluke's 80 series to get that feature.

It really doesn't matter whether it's true RMS or not considering few DVM's,
including Fluke 70 or 80 series, can measure AC signals below 1 mV anyway
(or a *long* way above 1mV with any precision for that matter).
Just *HOW* noisy is that pre-amp?????????

TonyP.
Anonymous
a b ) Power supply
May 12, 2004 3:09:51 AM

Archived from groups: rec.audio.tech (More info?)

TonyP wrote:
> "Arny Krueger" <arnyk@hotpop.com> wrote in message
> news:QKSdnUOC4sJQLD3dRVn-gw@comcast.com...
>> Mark D. Zacharias wrote:
>>> Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter,
>>> was it?
>> No, the Fluke 77 series was/is not a true RMS meter.
>> I think that you have to go up into Fluke's 80 series to get that
>> feature.
>
> It really doesn't matter whether it's true RMS or not considering few
> DVM's, including Fluke 70 or 80 series, can measure AC signals below
> 1 mV anyway (or a *long* way above 1mV with any precision for that
> matter). Just *HOW* noisy is that pre-amp?????????


Right. Most audio gear has maximum useful output between 1 and 2 volts. The
OPs Fluke 8600 spec says that it resolves 10uV on the 200 mv range. I
imagine that is the least signficiant digit on the most sensitive range.
It's a 4 1/2 digit meter.

http://www.atecorp.com/Equipment/Fluke/8600A.htm

I suppose that is something like the residual it reads when driven with a
low impedance source. This seems sufficient to make SNR measurements that
would show progress made on a phono preamp design.

The fact that it is not true rms says that any measurements made are going
to be nonstandard, but it seems like they could be helpful to a home
constructor.
May 12, 2004 3:09:51 AM

Archived from groups: rec.audio.tech (More info?)

In article <40a0d0b1$0$13706$afc38c87@news.optusnet.com.au>,
TonyP@optus.net.com.au says...
>
>
>"Arny Krueger" <arnyk@hotpop.com> wrote in message
>news:QKSdnUOC4sJQLD3dRVn-gw@comcast.com...
>> Mark D. Zacharias wrote:
>> > Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter,
>> > was it?
>> No, the Fluke 77 series was/is not a true RMS meter.
>> I think that you have to go up into Fluke's 80 series to get that feature.
>
>It really doesn't matter whether it's true RMS or not considering few DVM's,
>including Fluke 70 or 80 series, can measure AC signals below 1 mV anyway
>(or a *long* way above 1mV with any precision for that matter).
>Just *HOW* noisy is that pre-amp?????????
>
>TonyP.

It's a line pre-amp, not a phono pre-amp. I haven't measured it yet, but the
noise level is lower than the silent portions of some well recorded cd's. I'd
guess that noise from the preamp is less than 74 db, could be less than 85 db,
I don't know yet. That's why I posted the question. By the way, I'm using
three thermionic emmission transconductance devices : ) acting as the
amplifying elements, the preamp is built on perf board, so wire placement, not
traces on a PCB, are able to be re-routed easily. But as Arny says, I just
need to know if I am lowering or increasing the noise level while tweaking the
preamp, and with no gross levels of hum or hiss, I need to quantify the noise
to really know what's going on.
-Don
Anonymous
a b ) Power supply
May 14, 2004 8:03:09 PM

Archived from groups: rec.audio.tech (More info?)

"Arny Krueger" <arnyk@hotpop.com> wrote in message
news:p IWdnRz2lOp9Yz3dRVn-sA@comcast.com...
> TonyP wrote:
> > "Arny Krueger" <arnyk@hotpop.com> wrote in message
> > news:QKSdnUOC4sJQLD3dRVn-gw@comcast.com...
> >> Mark D. Zacharias wrote:
> >>> Not an expert on Fluke's, but the 77 anyway wasn't a true RMS meter,
> >>> was it?
> >> No, the Fluke 77 series was/is not a true RMS meter.
> >> I think that you have to go up into Fluke's 80 series to get that
> >> feature.

> > It really doesn't matter whether it's true RMS or not considering few
> > DVM's, including Fluke 70 or 80 series, can measure AC signals below
> > 1 mV anyway (or a *long* way above 1mV with any precision for that
> > matter). Just *HOW* noisy is that pre-amp?????????

> Right. Most audio gear has maximum useful output between 1 and 2 volts.
The
> OPs Fluke 8600 spec says that it resolves 10uV on the 200 mv range. I
> imagine that is the least signficiant digit on the most sensitive range.
> It's a 4 1/2 digit meter.

Yes, the 8600 will be much more useful than the 70 or 80 series hand helds
mentioned by Mark.
However the resolution at higher values is not the same as the precision at
low values. Rectifier linearity is an issue.

Having said that, for the simple better or worse comparison needed, it
doesn't matter about RMS vs Average responding or the actual measurement
uncertainty, provided the input is similar in both cases.

TonyP.
Anonymous
a b ) Power supply
May 14, 2004 8:03:10 PM

Archived from groups: rec.audio.tech (More info?)

TonyP wrote:
> "Arny Krueger" <arnyk@hotpop.com> wrote in message
> news:p IWdnRz2lOp9Yz3dRVn-sA@comcast.com...
>> TonyP wrote:
>>> "Arny Krueger" <arnyk@hotpop.com> wrote in message
>>> news:QKSdnUOC4sJQLD3dRVn-gw@comcast.com...
>>>> Mark D. Zacharias wrote:

>>>>> Not an expert on Fluke's, but the 77 anyway wasn't a true RMS
>>>>> meter, was it?
>>>> No, the Fluke 77 series was/is not a true RMS meter.
>>>> I think that you have to go up into Fluke's 80 series to get that
>>>> feature.

>>> It really doesn't matter whether it's true RMS or not considering
>>> few DVM's, including Fluke 70 or 80 series, can measure AC signals
>>> below 1 mV anyway (or a *long* way above 1mV with any precision for
>>> that matter). Just *HOW* noisy is that pre-amp?????????

>> Right. Most audio gear has maximum useful output between 1 and 2
>> volts. The OPs Fluke 8600 spec says that it resolves 10uV on the
>> 200 mv range. I imagine that is the least signficiant digit on the
>> most sensitive range. It's a 4 1/2 digit meter.

> Yes, the 8600 will be much more useful than the 70 or 80 series hand
> helds mentioned by Mark.

> However the resolution at higher values is not the same as the
> precision at low values.

True in general, but modern meters use precision rectifiers that are IME
really quite good.

>Rectifier linearity is an issue.

These multimeters also have amplfiiers that front-end the AC converters.
These meters are usually quite linear, with internal noise being more of a
problem at the low end.

> Having said that, for the simple better or worse comparison needed, it
> doesn't matter about RMS vs Average responding or the actual
> measurement uncertainty, provided the input is similar in both cases.

Agreed.
!