Correct Method to Measure Mic Preamp Noise Floor?

[Paul]

One site says a 150 Ohm Metal Film resistor across
pins 2 and 3:

http://www.benchmarkmedia.com/appnotes-a/mmpnoise

From the Vnoise=(4kTBR)**0.5, 150 Ohms will give about 2.229x10**-7
Volts, plus gain of the amp, plus noise figure.

But another suggests a paper clip short across pins 2 and 3:

http://www.daking.com/?p=76

I suspect the first one is correct. Then just crank the
gain all the way, and record the noise floor, so you can
compare preamps just by ear, right?

[PStamler]

Yes, the first method is basically correct. The paperclip method only tells how the preamp performs with a mic of zero impedance, and there aren't any of those that I know of.

The thing is, just cranking the gain and recording the result, then comparing by ear won't necessarily tell you anything useful, because:

1. different preamps have different maximum gains, and a preamp with a gain of +65dB will sound noisier than one with a gain of +60dB, even if both are perfect noiseless preamps amplifying the inherent resistor noise.

2. Some preamps have different Equivalent Input Noise figures at different gain settings, which might or might not make an audible difference.

The way to get around these objectione ie to choose some standard gain settings and use those for comparisons. +60dB, +40dB and +20dB might be good. Of course, then you need a repeatable way of getting to those settings. It's doable but can be tricky.

By the way, if you're planning to use a preamp to record classical music, you should probably do tests with a 30 ohm resistor, too, simulating the impedance of a Schoeps solid-state mic.

It's worth looking at recorded noise on a spectrum analyzer as well as measuring the average level and listening to it.

Peace,
Paul S.

[PStamler]

Oh yeah, your source resistor (150 ohm or 30 ohm) should be a metal film one, to avoid adding excess noise, which carbon comp resistors can do.

And the resistor should be in an XLR shell, to avoid problems with RF pickup.

Peace,
Paul S.

[Paul]

Ok, thanks for the feedback, Paul.

How about setting up an audio sig-gen at about 1kHz, with a fixed
amplitude, and adjusting the gain of amp A, until your DAW software
reads 0dB on its meters (or even higher), and then repeating for amp B,
so you know the gains are close.

Then put the shielded metal-film 150 Ohm resistor across pins 2 and
3, and record the noise floor then. I would think that would be a
reasonable way to tell if amp A is more or less noisy than amp B.

[Mike Rivers]

On 5/3/2013 6:12 AM, Paul wrote:

>> The paperclip method only
>> tells how the preamp performs with a mic of zero impedance, and there
>> aren't any of those that I know of.

Measuring with the input short-circuited will give you a noise level at
the output that's 4 to 6 dB lower than with the input seeing the
equivalent of a microphone. I have a few XLRs with a 150 ohm resistor
between pins 2 and 3 labeled "Dummy mic."

> How about setting up an audio sig-gen at about 1kHz, with a fixed
> amplitude, and adjusting the gain of amp A, until your DAW software
> reads 0dB on its meters (or even higher), and then repeating for amp B,
> so you know the gains are close.

This involves a different can or worms, or rahter, non-standards.
There's no standard for the relationship between analog input and
digital output. And you can't express the gain of an A/D converter
because the input and output units aren't the same.

If you have a preamp that's internally connected to an A/D converter so
the only output you can see is digital, you can express the sensitivity
in terms of how many dBu in it takes to get 0 dBFS out, and use that a
way to compare preamps. You can read the noise in dBFS with a properly
terminated input, but without knowing the gain, or at least the
sensitivity, you can't use this to compare equivalent noise levels.

I discuss this in some detail in my recent review of the Focusrite Forte
(see my web page).I was surprised at the rather high noise output at
maximum gain, but since the sensitivity is considerably higher than
other mic preamp-A/D boxes I've tested, the actual noise level at an
equivalent gain was actually lower with the Forte.

The way to compare noise when you have different sensitivities is (and I
think you have the right idea) is to put in a signal which gives you 0
dBFS at maximum gain (or at a setting you choose) with the least
sensitive unit, then measure the noise out with the input terminated
with 150 ohms. Then put the same signal into the more sensitive unit and
adjust its gain for 0 dBFS, replace the signal generator with a 150 ohm
terminator, and measure the noise.

This will tell you how much noise you'll get from each interface when
recording a given source with the same mic at the same (digital)
recording level.


--
For a good time, call http://mikeriversaudio.wordpress.com

[Scott Dorsey]

Neither one is really _correct_. The 150 ohm resistor will give you a
reasonable simulation of a dynamic mike, but a typical condenser mike will
have an impedance that comes much closer to a short.

Note that if you compare preamps just by ear, you _first_ need to make sure
that the gains are all the same. And by the same, I mean within 0.5 dB at
least. Secondly you need to know that the noise spectrum will affect how
loud it seems in isolation, but once there is signal up you may find that
the 1/f noise is more easily masked than the hiss.

And of course with a typical dynamic mike, there's a lot of thermal noise
from the winding resistance....
--scott

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

[William Sommerwerck]

> Oh yeah, your source resistor (150 ohm or 30 ohm) should be
> a metal film one, to avoid adding excess noise, which carbon
> comp resistors can do.

This is incorrect.

Any resistor of a particular value, regardless of what it's made of, generates
exactly the same amount of thermal noise, according to the Johnson-Nyquist
equation.

Excess noise (note the use of the word "excess") is created when a voltage is
applied across the resistor. Carbon-composition resistors are the principal
"guilty parties" in this regard. Metal resistors generate much less excess
noise. And when there is no voltage across the resistor (as when measuring
preamp noise) there is no excess noise.

[Paul]

If I understand you correctly, you're basically pointing out that
the voltage required to hit 0 dBFS will vary from A/D to A/D converter.

I believe I've taken care of that in my experiment this morning,
because I used the same bus for both amps, so I'm pretty sure the same
converter was used.

Tascam US-1641 pres VS. Mackie XDR pres (1402VLZ Pro)

Injected 1kHz into the lower gain unit, the US-1641, and
with gain cranked all the way, adjusted amplitude until 0dB
on channel mixer (+6.02dB is the peak, which is the actual 0dBFS?).
Then hooked up the resistor and recorded the noise.

Re-assigned the same bus to the port with the Mackie line-in, and
adjusted mic gain to get 0dB with same injected 1kHz amplitude, and
recorded resistor noise.

The Mackie is absolutely quieter. It's actually more of a very soft
60Hz hum. The Tascam pres are more like a white noise type of hissing,
and therefore much more noticable, even though the amplitude
of the noise appeared to be only slightly higher.

This was reflected in microphone tests too. The Tascams also have
a bit more low end. Again, the Tascam pres aren't bad sounding,
but I'm gonna use the Mackies more often from now on.

I didn't know the XDRs were that good:

http://www.soundonsound.com/sos/sep99/articles/mackie1604.htm

Of course, that article was published 13 years ago, and the
ONYX pres are the latest version, so I still might try one of these kits:

http://www.diyrecordingequipment.com/preamp/

In particular:

http://fivefishstudios.com/diy-kits

Anyone put one of these together?

[Paul]

I've gotta give credit to the Tascam people, because this US-1641
is a great value.

You only really notice the difference in the pres when you
are wearing headphones, and REALLY cranking the volume full up,
in the quiet sections of the recording. You're never going to
have the volume that high during actual listening, or you
would damage your hearing.

Nevertheless, these tests were done with only one track, so with
multiple tracks, the cumulative noise floor will be more noticable,
of course.

But I've made some DAMN clean recordings with these 1641 pre-amps!

[eth...@ethanwiner.com]

On Friday, May 3, 2013 8:37:00 AM UTC-4, William Sommerwerck wrote:
> Any resistor of a particular value, regardless of what it's made of, generates exactly the same amount of thermal noise ... Excess noise (note the use of the word "excess") is created when a voltage is applied across the resistor.

This is correct.

Also, it's not difficult to calibrate two preamps for the same amount of gain on a DAW meter, as long as you use the same A/D and the same input channel.

--Ethan

[Mike Rivers]

On 5/3/2013 9:39 AM, Paul wrote:

> If I understand you correctly, you're basically pointing out that
> the voltage required to hit 0 dBFS will vary from A/D to A/D converter.

Exactly. Of course it will vary with gain, but there are two variables
here, the gain between where the signal comes in and when it gets to the
converter, and the converter's "conversion factor" which is usually
expressed as volts or millivolts for full scale.

> I believe I've taken care of that in my experiment this morning,
> because I used the same bus for both amps, so I'm pretty sure the same
> converter was used.
>
> Tascam US-1641 pres VS. Mackie XDR pres (1402VLZ Pro)
>
> Injected 1kHz into the lower gain unit, the US-1641, and
> with gain cranked all the way, adjusted amplitude until 0dB
> on channel mixer (+6.02dB is the peak, which is the actual 0dBFS?).
> Then hooked up the resistor and recorded the noise.
>
> Re-assigned the same bus to the port with the Mackie line-in, and
> adjusted mic gain to get 0dB with same injected 1kHz amplitude, and
> recorded resistor noise.

You've completely lost me here. What's connected to what, and where are
you measuring the levels? What "same bus" are you talking about?

[Paul]

There are 16 simultaneous recordable tracks with the US-1641,
each of which has it's own "Bus" or channel. I'm pretty sure
there is one A/D converter for each one. But each bus can be assigned
any of the "device ports", which are the various inputs. I used the
SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
pre-amp, and once from the line-in from the Mackie.

Bottom line is that I'm quite sure I used the same A/D
converter, for both amp A and B.

With gain of the Tascam's pre-amp cranked all the way up, I used
an audio signal generator and injected a 1kHz sine wave into pins 2 and
3. The sig-gen has an adjustable amplitude knob, and I adjusted the
amplitude until there was 0dB on the bus channel mixer. Then I
disconnected the sig-gen, and hooked up the resistor to pins 2 and 3 and
recorded the noise.

I then re-assigned the same bus to the line-in port which has the
Mackie hooked up to it. I then connected the sig-gen to the first
pre-amp on the Mackie mixer, KEEPING THE 1kHz SINE WAVE AMPLITUDE THE
SAME. There is a "mic gain trim" pot at the top of each channel on the
Mackie mixer. I adjusted this mic gain pot to get 0dB on the bus
channel mixer. Then I once again disconnected the sig-gen, and hooked
up the resistor to pins 2 and 3 of the Mackie pre-amp, and recorded the
noise.

Bottom line is that I used a fixed amplitude signal to adjust the
pre-amp gains to be very close, and used the same A/D converter twice,
so when I measured the noise floor, it was a fair comparison.

[Paul]

BTW, all the stuff about the buses and channels are all a
part of Cubase 5, as well as the US-1641.

[Mike Rivers]

On May 3, 1:58 pm, Paul <Quiller...@gmail.com> wrote:



> There are 16 simultaneous recordable tracks with the US-1641,
> each of which has it's own "Bus" or channel. I'm pretty sure
> there is one A/D converter for each one. But each bus can be assigned
> any of the "device ports", which are the various inputs.

That's interesting. You mean that (hardware) input 1 can look like
Input 1, or input 2, or input 3, etc. to the DAW? That seems kind of
odd.

> I used the
> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
> pre-amp, and once from the line-in from the Mackie.
>
> Bottom line is that I'm quite sure I used the same A/D converter

I doubt that it would make any more than a couple of tenths of a dB
difference no matter which A/D converter you used. I'm talking about
types here, and implementation into the product (the interface). You
could have used Mic Input 1 and Line Input 2 as long as you adjusted
the gains for the same level on the digital end. I'd connect the
generator to both the TASCAM and Mackie mic inputs using a Y cable.
Turn TASCAM up to full gain, adjusting the generator output to keep
the digital level at full scale. Then turn the Mackie up to full gain
and adjust the Tascam line input gain to get full scale. Then you can
remove the generator, terminate the inputs, and compare the noise.

Where did you take the output from the Mackie to feed to the TASCAM?
The Insert send is the closest to the preamp. If you took the Main
outputs with the channel and master faders set to the "unity" position
you're adding more noise from more mixer stages, and possibly more (or
less) gain.

[Mike Rivers]

On May 3, 3:02 pm, Paul <Quiller...@gmail.com> wrote:


> There are 16 simultaneous recordable tracks with the US-1641,
> each of which has it's own "Bus" or channel. I'm pretty sure
> there is one A/D converter for each one. But each bus can be assigned
> any of the "device ports", which are the various inputs.

That's interesting. You mean that (hardware) input 1 can look like
Input 1, or input 2, or input 3, etc. to the DAW? That seems kind of
odd.

> I used the
> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
> pre-amp, and once from the line-in from the Mackie.
>

[mako...@yahoo.com]

I would start with the 150 ohm resistor (or whatever z you want to use) and then add a couple of 10 Meg or whatever it takes to attenuate you sig gen down to maybe 1 micro volt or so. This way you can crank the gain up full to year the noise and also couple in a very low level tone to use as a reference. Use your ears and or a spec an to compare one per vs another. Just keep the tone level constant to make a fair comparison. Your are listening / looking for the signal to noise ratio. To get an actual number of noise you have to define the measurement bandwidth. Since noise audibility is a strong function of freq, an audible test makes a lot of sense compared to looking at numbers. A noise density spectrum plot is the best way to do this quantitatively.
Mark

[Paul]

On 5/3/2013 2:49 PM, Mike Rivers wrote:
> On May 3, 1:58 pm, Paul <Quiller...@gmail.com> wrote:
>
>
>
>> There are 16 simultaneous recordable tracks with the US-1641,
>> each of which has it's own "Bus" or channel. I'm pretty sure
>> there is one A/D converter for each one. But each bus can be assigned
>> any of the "device ports", which are the various inputs.
>
> That's interesting. You mean that (hardware) input 1 can look like
> Input 1, or input 2, or input 3, etc. to the DAW? That seems kind of
> odd.
>

It certainly allows flexibility, but it can get a bit confusing
when Bus 1=mic input 7, Bus 2=line input 13, Bus 3=mic input 1, etc.

So I usually try to keep Bus 1=mic input 1, etc....

>
>> I used the
>> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
>> pre-amp, and once from the line-in from the Mackie.
>>
>> Bottom line is that I'm quite sure I used the same A/D converter
>
> I doubt that it would make any more than a couple of tenths of a dB
> difference no matter which A/D converter you used. I'm talking about
> types here, and implementation into the product (the interface). You
> could have used Mic Input 1 and Line Input 2 as long as you adjusted
> the gains for the same level on the digital end. I'd connect the
> generator to both the TASCAM and Mackie mic inputs using a Y cable.
> Turn TASCAM up to full gain, adjusting the generator output to keep
> the digital level at full scale. Then turn the Mackie up to full gain
> and adjust the Tascam line input gain to get full scale. Then you can
> remove the generator, terminate the inputs, and compare the noise.
>

The 4 line inputs at the back of the US-1641 do not have
dedicated pot gain adjustments. Then only have +4dBu / -10dBV
switches, and since these are essentially 11.79 dB pads, I
kept the Mackie line at -10dBV.

The 2 guitar/line inputs on the front, however, do have
gain pots. So you bring up a good point, that by the Friis
formula for noise factor, the gain of the first device (Mackie pre)
should be maximized.

So maybe I should have set the switches on the back to
+4dBu, except I'm not sure I had 11.79 dB of gain more
in the Mackie pre. The next time I record, I'll try
maxing the mic gain of the Mackie first, and then use
the line inputs on the front, where I can adjust the
level.

> Where did you take the output from the Mackie to feed to the TASCAM?
> The Insert send is the closest to the preamp. If you took the Main
> outputs with the channel and master faders set to the "unity" position
> you're adding more noise from more mixer stages, and possibly more (or
> less) gain.
>

Yes, I used Channel Insert send on the back.

I notice Mackie has the newer XDR2 pres, as well
as the Onyx ones, which you are familiar with.

But I think the XDR pres still hold up even today.

[Mike Rivers]

On 5/3/2013 1:58 PM, Paul wrote:

> There are 16 simultaneous recordable tracks with the US-1641,
> each of which has it's own "Bus" or channel. I'm pretty sure
> there is one A/D converter for each one. But each bus can be assigned
> any of the "device ports", which are the various inputs.

That's interesting. You mean that (hardware) input 1 can look like Input
1, or input 2, or input 3, etc. to the DAW? That seems kind of odd.

> I used the
> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
> pre-amp, and once from the line-in from the Mackie.
>

> Bottom line is that I'm quite sure I used the same A/D converter

I doubt that it would make any more than a couple of tenths of a dB
difference no matter which A/D converter you used. I'm talking about
types here, and implementation into the product (the interface). You
could have used Mic Input 1 and Line Input 2 as long as you adjusted the
gains for the same level on the digital end. I'd connect the generator
to both the TASCAM and Mackie mic inputs using a Y cable. Turn TASCAM up
to full gain, adjusting the generator output to keep the digital level
at full scale. Then turn the Mackie up to full gain and adjust the
Tascam line input gain to get full scale. Then you can remove the
generator, terminate the inputs, and compare the noise.

Where did you take the output from the Mackie to feed to the TASCAM? The
Insert send is the closest to the preamp. If you took the Main outputs
with the channel and master faders set to the "unity" position you're
adding more noise from more mixer stages, and possibly more (or less) gain.

[Mike Rivers]

On 5/3/2013 3:02 PM, Paul wrote:

> BTW, all the stuff about the buses and channels are all a
> part of Cubase 5, as well as the US-1641.

I don't know anything about either, so just do it the way you want.

[Mike Rivers]

On 5/3/2013 1:58 PM, Paul wrote:

> There are 16 simultaneous recordable tracks with the US-1641,
> each of which has it's own "Bus" or channel. I'm pretty sure
> there is one A/D converter for each one. But each bus can be assigned
> any of the "device ports", which are the various inputs.

That's interesting. You mean that (hardware) input 1 can look like Input
1, or input 2, or input 3, etc. to the DAW? That seems kind of odd.

> I used the
> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
> pre-amp, and once from the line-in from the Mackie.
>

> Bottom line is that I'm quite sure I used the same A/D converter

I doubt that it would make any more than a couple of tenths of a dB
difference no matter which A/D converter you used. I'm talking about
types here, and implementation into the product (the interface). You
could have used Mic Input 1 and Line Input 2 as long as you adjusted the
gains for the same level on the digital end. I'd connect the generator
to both the TASCAM and Mackie mic inputs using a Y cable. Turn TASCAM up
to full gain, adjusting the generator output to keep the digital level
at full scale. Then turn the Mackie up to full gain and adjust the
Tascam line input gain to get full scale. Then you can remove the
generator, terminate the inputs, and compare the noise.

Where did you take the output from the Mackie to feed to the TASCAM? The
Insert send is the closest to the preamp. If you took the Main outputs
with the channel and master faders set to the "unity" position you're
adding more noise from more mixer stages, and possibly more (or less) gain.

[Mike Rivers]

On 5/3/2013 1:58 PM, Paul wrote:

> There are 16 simultaneous recordable tracks with the US-1641,
> each of which has it's own "Bus" or channel. I'm pretty sure
> there is one A/D converter for each one. But each bus can be assigned
> any of the "device ports", which are the various inputs.

That's interesting. You mean that (hardware) input 1 can look like Input
1, or input 2, or input 3, etc. to the DAW? That seems kind of odd.

> I used the
> SAME bus TWICE, just re-assigning it's input: once from the Tascam's own
> pre-amp, and once from the line-in from the Mackie.
>

> Bottom line is that I'm quite sure I used the same A/D converter

I doubt that it would make any more than a couple of tenths of a dB
difference no matter which A/D converter you used. I'm talking about
types here, and implementation into the product (the interface). You
could have used Mic Input 1 and Line Input 2 as long as you adjusted the
gains for the same level on the digital end. I'd connect the generator
to both the TASCAM and Mackie mic inputs using a Y cable. Turn TASCAM up
to full gain, adjusting the generator output to keep the digital level
at full scale. Then turn the Mackie up to full gain and adjust the
Tascam line input gain to get full scale. Then you can remove the
generator, terminate the inputs, and compare the noise.

Where did you take the output from the Mackie to feed to the TASCAM? The
Insert send is the closest to the preamp. If you took the Main outputs
with the channel and master faders set to the "unity" position you're
adding more noise from more mixer stages, and possibly more (or less) gain.