The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Misunderstandings about preamp noise
Old 24th May 2019
  #1
Gear Addict
 

Misunderstandings about preamp noise

I often hear people describe preamps as "getting noisy" when you turn them up too far, as if the preamp would behave fine at lower levels, but something goes horribly wrong in that last little turn of the knob. I've long suspected that this is a misleading way of describing things, but had never bothered to test it.

Now I've finally done some tests to back up my beliefs. As I've long suspected, the signal to noise ratio of a preamp does NOT get worse as you turn it up - in fact it gets better. Sure, turning up the gain does raise the level of the noise, but it also raises the level of the signal by the same amount (or more, since some of the noise is an unavoidable noise floor that you can raise the signal out of.)

My test was pretty simple, and anyone who's interested can repeat it easily. I ran a test tone out of one of the outputs on my interface, through a line to mic level transformer and into a mic input on my interface (DI box would do the same thing, although this happened to be an all in one, integrated cable solution.) I set the mic input to maximum gain and then adjusted the level of the test tone output down to a point where there was no clipping.

Once that test tone level was set, I never changed the output. Next, I recorded 5 short snippets at different gain settings. For each snippet, there are 5-10 seconds of test tone, and 5-10 seconds of no signal (just noise.)

Once all 5 recordings were done, I copied the files and made normalized copies of all of them, set so that the test tones were at the exact same -1dBFS level in all of them. Finally, I made notes of the signal and noise levels for all 10 samples. I've also calculated how much gain was applied (relative to the lowest gain setting) at each of the various settings, as well as what the resulting signal to noise ratio was.

You can see in the table and graphs below that, as predicted, although turning up the gain turns up the noise, it also turns up the signal even more, so the overall performance of the system only gets better at higher gain settings. As long as you avoid clipping, there's no noise penalty for turning up the gain.

Anyone interested in hearing the original sound files can check them out at SoundCloud:

https://soundcloud.com/beowulf-recor...-noise-vs-gain
Attached Thumbnails
Misunderstandings about preamp noise-preamp-sn-gain-table.png   Misunderstandings about preamp noise-preamp-sn-gain-graph-normalized.png   Misunderstandings about preamp noise-preamp-sn-gain-graph-raw.png   Misunderstandings about preamp noise-preamp-sn-gain-graph-sn-ratio.png  
Old 30th May 2019
  #2
Doesn't turning up the gain just amplify everything - signal and surrounding unwanted signal/ noise? Signal and unwanted signal, whatever that may be...noise...the car down the road...a washing machine next door. Just because the signal gain rises and is above unwanted signal/ noise, doesn't mean it's good or usable. So, I think there is a noise penalty.
Old 30th May 2019
  #3
Lives for gear
I’m not sure that you are showing anything meaningful with your test. A 60 db signal to noise ratio in the absence of any external (room) noise is fairly awful, and that’s the better end of your figures. I do understand what you mean about the benefit of providing enough gain to lift the signal out of residual noise, but otherwise I’m not seeing how your test is meaningful.
But I’m not formally educated or trained as an audio tech, so maybe I just can’t translate this to a practical level.
Old 30th May 2019
  #4
Gear Addict
 

Quote:
Originally Posted by Bushman View Post
I’m not sure that you are showing anything meaningful with your test. A 60 db signal to noise ratio in the absence of any external (room) noise is fairly awful, and that’s the better end of your figures. I do understand what you mean about the benefit of providing enough gain to lift the signal out of residual noise, but otherwise I’m not seeing how your test is meaningful.
But I’m not formally educated or trained as an audio tech, so maybe I just can’t translate this to a practical level.
Yeah, I had a feeling after I posted the results that the bad numbers would throw people off. I think what we're seeing there is mostly the noise floor of the DAC outputs when they're being run at super low levels (low signal-to-noise because the signal is so low.) I think I'll re-create the test with the test tone being generated at full resolution, maximum output (best possible s/n ratio) and then turning it down with a mixer channel or other analog means. That should provide a much, much quieter set of signals to compare.

Nevertheless, even in this imperfect test, I thought something useful was demonstrated. What I tent to hear people say is that the farther they turn their preamps up, the more noisy their signal is. What this test demonstrates is that, for any given signal, the s/n ratio gets better at higher gain settings, not worse.

If I get a chance to repeat the test with lower overall noise, I'll share the new numbers here.
Old 30th May 2019
  #5
Gear Addict
 

Quote:
Originally Posted by hello people View Post
Doesn't turning up the gain just amplify everything - signal and surrounding unwanted signal/ noise? Signal and unwanted signal, whatever that may be...noise...the car down the road...a washing machine next door. Just because the signal gain rises and is above unwanted signal/ noise, doesn't mean it's good or usable. So, I think there is a noise penalty.
Yes and no. I hear it described as a negative attribute of some preamps that they get noisy at higher gain. All I'm saying is that if they're noisy at high gain, they're noisy at any gain. The signal to noise ratio gets better, not worse, when you turn up the gain.

The reason I see value in making these distinctions is that new audio engineers get the wrong idea when they read reviews describing noise at high gains, and they start thinking that they need to avoid the last 10-25% of the gain range. I've seen countless examples on these forums, but one in particular quite recently got me thinking about it again. The poster was asking about all sorts of different equipment upgrades because a Sound-on-Sound review of an audio interface preamp made them think it wasn't ok to turn the gain up past about 75% (and they felt like they might need that extra gain.) It's that kind of thinking that I'm trying to steer people away from.

Whatever gear you've got, and whatever sound source you've got, make the best of it by using the best gain-staging you can. Don't make it worse than it needs to be by leaving the knob down for no good reason. If the sound source is quiet enough that you'd otherwise be tempted to crank the gain, and you deliberately leave it low because you're afraid to use that last portion of the knob's rotation, then you'll end up with a noisier recording than if you had turned it all the way up.
Old 30th May 2019
  #6
Lives for gear
 
bgood's Avatar
People crank gain on prosumer interface preamps usually because they’re trying to use a mic that’s gain hungry (sm7, re20, etc)...

So, I’m that scenario, you’re pushing the preamp into 10 and picking up every bit of its self noise... your test is not valid as it applies to this real world example

As far as just plugging in any old LDC mic, hitting record and then turning up the gain on your interface for no reason other than to test it as you have... well, ok... yeah, that would show, what? That it’s noisy. You don't even have to plug a mic in to do the same test

And further... your test aside, having worked for over 30 years with the best pro-line pci/PCIe stuff to the prosumer $50 a channel/usb stuff I can say without doubt the latter sounds like sh1t when you dime it...
Old 30th May 2019
  #7
Gear Addict
 

Quote:
Originally Posted by bgood View Post
As far as just plugging in any old LDC mic, hitting record and then turning up the gain on your interface for no reason other than to test it as you have... well, ok... yeah, that would show, what? That it’s noisy. You don't even have to plug a mic in to do the same test
I didn't plug in any old LDC mic. There was no mic at all in this test. I played a test tone out from my interface line output, ran it through a passive DI for better impedance matching, and connected that to a mic input.

In retrospect I should've generated and then attenuated my test tone differently so it would be cleaner, which I plan to do soon, but it doesn't change the results.

As far as flawed methodology goes, what I find more interesting is that I appear to have a 16 bit noise floor (approaching -96dBFS) despite the fact that I *thought* I did this whole test 24 bit start to finish. I know full 24 bit dynamic range (144dB) is essentially impossible to achieve, but somewhere approaching 20 bit performance (120dB) should be within reach. I'll have to double check my settings and files to see if I goofed along the way, or if my interface input noise is a LOT worse than I'd have expected!

Anyway, despite the flaws in my first attempt, the result is still clear. The signal to noise ratio gets better as you turn up the gain. So the preamp isn't performing worse at high gain than it is at low gain.

The point here isn't a simple yes/no "is it noisy?" The point is that turning up the gain on a device doesn't make it perform worse.
Quote:
And further... your test aside, having worked for over 30 years with the best pro-line pci/PCIe stuff to the prosumer $50 a channel/usb stuff I can say without doubt the latter sounds like sh1t when you dime it...
That may be true, but what I'm saying is that if it sounds that bad "when you dime it," it would've sounded that bad at any gain setting. It doesn't get bad as a function of knob turning. Whatever its crummy noise specs are, they're with you the whole time.
Old 30th May 2019
  #8
Gear Addict
 

Having found a few issues that make this first test less than ideal, I'm going to try again. I don't believe any of the extra noise sources in the first test invalidate the results, because they all stay the same throughout the test, so any s/n ratio change when adjusting gain is still meaningful. Nevertheless, I imagine the results will be more convincing with fewer distractions.

I'll repeat this test as soon as I get time, and I'll be a little more thorough on the second go around:
1) better s/n on test tone
2) make totally sure it's 24 bit i/o throughout
3) I'll make a resistor plug to do a crude version of EIN testing at each gain setting to eliminate any risk of other noise sources confusing matters.
4) I'll see if I can get meaningful results doing a mic-based version of the test with my RE20. I suspect that my noisy basement studio will have too much ambient noise for me to do this effectively (I record exclusively on location, so perfect soundproofing in a strictly post-production room hasn't been a top priority.) But, I'll give it a shot.
Old 30th May 2019
  #9
Lives for gear
 
12ax7's Avatar
 

What is getting left out of your test is the REASON one might use higher gain settings on a mic pre:

Because the signal from the mic is low!

In your test, you left the test signal at the same level for all gain settings on the pre-amp.

(This is not what's happening in the "real world".)

Try seeing what happens to the signal-to-noise ratio when you change the level of the generated signal to various arbitrary settings, and then change the gain on the pre-amp to compensate for them.

(This is what's happening to S/N in the "real world".)
.
Old 30th May 2019
  #10
Lives for gear
 
nosebleedaudio's Avatar
 

The impedance loading the mic pre has a HUGE influence on the noise level AT the same gain of the pre.
So, insert a -55dB signal crank the gain to max and then see what you have, while having a low impedance on the Pre input..150 Ohms is common, the lower the load the less the noise, higher the load the higher the noise...
Old 30th May 2019
  #11
Lives for gear
Quote:
Originally Posted by Ebeowulf17 View Post
...
I'll repeat this test as soon as I get time, and I'll be a little more thorough on the second go around: ...
I commend you for developing a testing protocol for sussing out what you want to know, and working through the collected data. That's the sort of thing that solidifies the basis for your opinions and raises the level of insight that we can all share.

I did a bunch of measurements, and generated charts and graphs when we were experiencing a rash of "bypass the preamp" postings in GS. It turned into a massive 10-article series. Take a tour through it simply because you may benefit from some of the spreadsheet data showing voltage, dbV, dBu, etc, or perhaps some of my commentary may help steer your processes.

Glad you're up to this.
Old 31st May 2019
  #12
Gear Addict
 

Quote:
Originally Posted by 12ax7 View Post
What is getting left out of your test is the REASON one might use higher gain settings on a mic pre:

Because the signal from the mic is low!

In your test, you left the test signal at the same level for all gain settings on the pre-amp.

(This is not what's happening in the "real world".)

Try seeing what happens to the signal-to-noise ratio when you change the level of the generated signal to various arbitrary settings, and then change the gain on the pre-amp to compensate for them.

(This is what's happening to S/N in the "real world".)
.
Yes, of course if you turn down your signal level (and since you have the same noise level) then you'll end up with a worse signal to noise ratio. I kind of assumed that was understood.

I understand what you're saying. Your point is that in real-world usage, you only turn the gain up when you have a low enough signal to need the high gain, and so your signal to noise ratio ends up being bad. That's certainly true, but doesn't really speak to the complaint I'm raising. The point I've been trying to make in all these posts is that many people talk about preamp performance getting *worse* as you turn the gain up, and it doesn't appear to be the case. The preamp is just as noisy regardless. You're right that the signal to noise ratio is worse with lower level signals, but that's not because of where you set your gain knob.

The reason I'm doing these tests and posting results is that many of the opinions shared here are very misleading. I've heard so many similar claims about worse performance at high gains that I started doubting my own understanding of preamp behavior and wondering if they actually did get noisier at higher gains... and I definitely had the audio and electronics knowledge to know better... but after hearing enough respected engineers talk about bad performance at high gain, I couldn't simply dismiss them anymore. I had to find out for sure.

Just to be clear here, I'm not claiming that there aren't huge differences in preamp performance, nor am I denying that some preamps are much better for quiet sources than others. I'm just frustrated by the way people describe these differences and by the effect their choice of words has on other people who are trying to learn all this stuff.

If my interface is noisy and terrible, we should say it's noisy and terrible, and that it's not suitable for quiet sources because its noise level will be unacceptable. That's not the same as saying it's noisy at high gain. It's noisy ALL THE TIME, not just at high gain! A better preamp would be quieter ALL THE TIME, not just at high gain.

Most importantly, for any given piece of gear, there's no penalty for turning it up. If you're shopping for gear, you should get the best you can afford, and noise is certainly an important factor to consider. However, once you've got a piece of gear and you're trying to get the best sound you can out of it, don't hesitate to turn it up. Leaving the gain low will only make things worse.

As for the levels in my test, I was already starting with a very low signal. How else could I have recorded that test tone without clipping when I added 50dB of extra gain?! So, the test tone I used represented a low enough level to justify maximum gain, which is exactly what you're asking for. The level was so low that I'd have to crank the gain to get good levels into my DAW. Now, if I accepted the implications of many forum comments and equipment reviews, I might think that I better leave the gain a little below max in order to avoid all the pesky noise amplification... but that would just make the signal to noise ratio even worse!
Old 31st May 2019
  #13
Lives for gear
 
nosebleedaudio's Avatar
 

Guess you missed my post..
Insert a -60dB signal, crank gain to max and you WILL hear noise riding on the signal...Has very little to do with a GOOD pre ect..ALL pres have/add noise..Some far more than others..
I do this often for different reasons..
Old 31st May 2019
  #14
Lives for gear
Quote:
Originally Posted by nosebleedaudio View Post
Insert a -60dB signal, crank gain to max and you WILL hear noise riding on the signal...
I do this often for different reasons..
Are any of those different reasons GOOD reasons?
Old 31st May 2019
  #15
Gear Addict
 

Quote:
Originally Posted by nosebleedaudio View Post
Guess you missed my post..
Insert a -60dB signal, crank gain to max and you WILL hear noise riding on the signal...Has very little to do with a GOOD pre ect..ALL pres have/add noise..Some far more than others..
I do this often for different reasons..
Oops, sorry I didn't respond to your comment earlier. I'm aware of the source impedance effect on noise levels - that's why I didn't just directly feed a line output into a mic input through a simple cable adapter, but instead through an impedance matching transformer... although I haven't actually measured the impedance yet - it's on my to-do list for the next go-around.

As for the unavoidable noise in every preamp, no surprise there. I don't have access to any especially great ones, so I haven't been able to experience the extremes, but I'm certainly aware that they all have inherent noise, just some more than others.
Old 31st May 2019
  #16
Gear Addict
 

Quote:
Originally Posted by Bushman View Post
Are any of those different reasons GOOD reasons?
Maybe it's acting like dither, but in the analog domain? Use comparatively inoffensive white noise to mask and obscure other more offensive noises that made it into a recording?

Dunno, just a wild guess.
Old 31st May 2019
  #17
Lives for gear
Quote:
Originally Posted by Ebeowulf17 View Post
Maybe it's acting like dither, but in the analog domain? Use comparatively inoffensive white noise to mask and obscure other more offensive noises that made it into a recording?

Dunno, just a wild guess.
I would chide you for answering a question that was directed to someone else if I wasn’t ever guilty of doing the same thing.
I have never thought, “I need a bit of soft white noise here to mask the drummer faintly passing gas in a quiet passage.” But I guess that could happen.
Analog dither? Yeah, we called it tape hiss. It was never very popular.
Old 31st May 2019
  #18
Lives for gear
 
esldude's Avatar
EIN Equivalent Input Noise. Test is 60 db gain at 150 ohm source impedance. If a pre-amp contributed zero noise of its own the max possible result would be -131 dbU. Plenty of preamps get a rating of say -128 dbU and don't contribute much noise. Remember however, this doesn't mean the signal is -128 dbU. It would mean the signal after 60 db of gain is -68dbU. You might be thinking reduce the gain to 0 and the noise level left is -128 dbU. Usually that isn't the case. Usually it is somewhat higher in the few preamps I've tested. Some of them drop the level 10 db if you reduce gain to 50 db and one even dropped it 20 db with gain at 40 db. Below that the drops in gain don't drop the remaining noise level on a 1 to 1 basis. Though the few I've had aren't terrible in that respect.

And yes all you need is to put a 150 ohm metal film resistor across the microphone input. It is beneficial to perhaps try it with 470 and 1k as well.

What happens is with enough gain you've pushed the resistor sourced thermal noise above the noise floor of the preamp. At lower gain levels the preamp doesn't have enough SNR to have only thermal noise. Some of it is in the preamp. So I think you are correct you actually have better preamp SNR at higher gain levels within reason. Maybe I've been lucky, but I've not seen any pres that are noisy turned all the way up beyond what the EIN rating is. If your pre has like 80 db of gain, then yeah, turned up you'll hear it.

Here is a nice online calculator for thermal noise in a resistor.
http://www.sengpielaudio.com/calculator-noise.htm
Old 31st May 2019
  #19
Gear Addict
 

Quote:
Originally Posted by esldude View Post
EIN Equivalent Input Noise. Test is 60 db gain at 150 ohm source impedance. If a pre-amp contributed zero noise of its own the max possible result would be -131 dbU. Plenty of preamps get a rating of say -128 dbU and don't contribute much noise. Remember however, this doesn't mean the signal is -128 dbU. It would mean the signal after 60 db of gain is -68dbU. You might be thinking reduce the gain to 0 and the noise level left is -128 dbU. Usually that isn't the case. Usually it is somewhat higher in the few preamps I've tested. Some of them drop the level 10 db if you reduce gain to 50 db and one even dropped it 20 db with gain at 40 db. Below that the drops in gain don't drop the remaining noise level on a 1 to 1 basis. Though the few I've had aren't terrible in that respect.

And yes all you need is to put a 150 ohm metal film resistor across the microphone input. It is beneficial to perhaps try it with 470 and 1k as well.

What happens is with enough gain you've pushed the resistor sourced thermal noise above the noise floor of the preamp. At lower gain levels the preamp doesn't have enough SNR to have only thermal noise. Some of it is in the preamp. So I think you are correct you actually have better preamp SNR at higher gain levels within reason. Maybe I've been lucky, but I've not seen any pres that are noisy turned all the way up beyond what the EIN rating is. If your pre has like 80 db of gain, then yeah, turned up you'll hear it.

Here is a nice online calculator for thermal noise in a resistor.
http://www.sengpielaudio.com/calculator-noise.htm
No major revelations for me in there, but a very concise and well articulated description of the underlying factors (along with some logical analysis.)

Thanks!
Old 31st May 2019
  #20
Lives for gear
 
nosebleedaudio's Avatar
 

Quote:
Originally Posted by Bushman View Post
Are any of those different reasons GOOD reasons?
YES...
Old 31st May 2019
  #21
Gear Addict
Thanks for taking the time to do this test, despite the detracters I do beleive this does add weight to your argument that the SNR only improves with gain. I am definately one of those that have always said pre-amps don't really matter other than at ultra high gan levels where cheap pre-amps start to fall apart. This test shows I was always misinterpreting a higher basic noise floor as poor behaviour at high gain. That said, the noise floor for even "nasty" budget pre-amps is stuff of dreams only 20 or 30 years ago.
Old 31st May 2019
  #22
Gear Nut
Hi Ebeowulf17,

Very interesting thread you started on a topic very near and dear to our hearts here at Cranborne Audio - Equivalent Input Noise of Preamps.


Quote:
Originally Posted by Ebeowulf17 View Post
I often hear people describe preamps as "getting noisy" when you turn them up too far, as if the preamp would behave fine at lower levels, but something goes horribly wrong in that last little turn of the knob. I've long suspected that this is a misleading way of describing things, but had never bothered to test it.

Now I've finally done some tests to back up my beliefs. As I've long suspected, the signal to noise ratio of a preamp does NOT get worse as you turn it up - in fact it gets better. Sure, turning up the gain does raise the level of the noise, but it also raises the level of the signal by the same amount (or more, since some of the noise is an unavoidable noise floor that you can raise the signal out of.)
Yes and no. Equivalent input noise is the important spec to understand here. Anyone who knows EIN and what it means knows you are 100% spot-on in that the every preamp will have its best EIN figures at/near Max Gain - provided it is a low-noise gain source which most preamps these days are. At min gain, the EIN of a preamp will be its worst value, which is usually ~-90dB unweighted with a 150 ohm load. Whereas at max gain, most preamps achieve at least -120dB unweighted with 150 ohm with the absolute best low-noise designs achieving -127-128dB EIN unweighted with a 150 ohm load. (Our Camden 500 is 129.5db!)

BUT - the point I would make here is if you give a preamp max gain, you have a larger dynamic range before the noise floor, however, if you are giving a signal 60dB of gain, it will be raising the noise floor of your source by 60dB as well, which could potentially become problematic.


Quote:
Originally Posted by Ebeowulf17 View Post
My test was pretty simple, and anyone who's interested can repeat it easily. I ran a test tone out of one of the outputs on my interface, through a line to mic level transformer and into a mic input on my interface (DI box would do the same thing, although this happened to be an all in one, integrated cable solution.) I set the mic input to maximum gain and then adjusted the level of the test tone output down to a point where there was no clipping.

Once that test tone level was set, I never changed the output. Next, I recorded 5 short snippets at different gain settings. For each snippet, there are 5-10 seconds of test tone, and 5-10 seconds of no signal (just noise.)

Once all 5 recordings were done, I copied the files and made normalized copies of all of them, set so that the test tones were at the exact same -1dBFS level in all of them. Finally, I made notes of the signal and noise levels for all 10 samples. I've also calculated how much gain was applied (relative to the lowest gain setting) at each of the various settings, as well as what the resulting signal to noise ratio was.

You can see in the table and graphs below that, as predicted, although turning up the gain turns up the noise, it also turns up the signal even more, so the overall performance of the system only gets better at higher gain settings. As long as you avoid clipping, there's no noise penalty for turning up the gain.

Anyone interested in hearing the original sound files can check them out at SoundCloud:

https://soundcloud.com/beowulf-recor...-noise-vs-gain
Yes this test shows well what you mean by the EIN decreasing as you increase the gain. But do bear in mind that preamps' THD (Total Harmonic Distortion) increases as it approaches the clip point. So a THD+N (Total Harmonic Distortion plus Noise) plot with an AP or similar at different gain values is probably the most meaningful way to understand how noise and distortion changes on a preamp at different gain levels. Unfortunately, very few preamp manufacturers cite their preamps' performance at more than one gain value and frequency, and they most commonly cite the figures where they achieve their best performance. The ideal thing would be to do EIN and THD+N sweeps at different values and for all frequencies. But I think we are the only company that does this for our preamps. Here are our tech specs I speak of if you'd like to have a look. These graphs are from our APx555 audio analyser: http://www.cranborne-audio.com/hubfs..._TechSpecs.pdf

Alternatively, we have our "Camden 500 By the Numbers Video" on Youtube you can check out... There's a little bit of marketing rah-rah in the video but not much. Elliott spends most of the time breaking down the specs themselves with graphs - the same graphics in the Tech Specs PDF I linked above but you get Elliott's mockney accent for narration as well

Happy to answer any questions if anyone finds any of this interesting. Cheers all!

Sean from Cranborne Audio
Old 31st May 2019
  #23
Gear Addict
 

Quote:
Originally Posted by Karpmentalise View Post
Hi Ebeowulf17,

Very interesting thread you started on a topic very near and dear to our hearts here at Cranborne Audio - Equivalent Input Noise of Preamps.




Yes and no. Equivalent input noise is the important spec to understand here. Anyone who knows EIN and what it means knows you are 100% spot-on in that the every preamp will have its best EIN figures at/near Max Gain - provided it is a low-noise gain source which most preamps these days are. At min gain, the EIN of a preamp will be its worst value, which is usually ~-90dB unweighted with a 150 ohm load. Whereas at max gain, most preamps achieve at least -120dB unweighted with 150 ohm with the absolute best low-noise designs achieving -127-128dB EIN unweighted with a 150 ohm load. (Our Camden 500 is 129.5db!)

BUT - the point I would make here is if you give a preamp max gain, you have a larger dynamic range before the noise floor, however, if you are giving a signal 60dB of gain, it will be raising the noise floor of your source by 60dB as well, which could potentially become problematic.




Yes this test shows well what you mean by the EIN decreasing as you increase the gain. But do bear in mind that preamps' THD (Total Harmonic Distortion) increases as it approaches the clip point. So a THD+N (Total Harmonic Distortion plus Noise) plot with an AP or similar at different gain values is probably the most meaningful way to understand how noise and distortion changes on a preamp at different gain levels. Unfortunately, very few preamp manufacturers cite their preamps' performance at more than one gain value and frequency, and they most commonly cite the figures where they achieve their best performance. The ideal thing would be to do EIN and THD+N sweeps at different values and for all frequencies. But I think we are the only company that does this for our preamps. Here are our tech specs I speak of if you'd like to have a look. These graphs are from our APx555 audio analyser: http://www.cranborne-audio.com/hubfs..._TechSpecs.pdf

Alternatively, we have our "Camden 500 By the Numbers Video" on Youtube you can check out... There's a little bit of marketing rah-rah in the video but not much. Elliott spends most of the time breaking down the specs themselves with graphs - the same graphics in the Tech Specs PDF I linked above but you get Elliott's mockney accent for narration as well

Happy to answer any questions if anyone finds any of this interesting. Cheers all!

Sean from Cranborne Audio
First off, thanks for sharing a detailed and insightful response. Lots of great stuff in there! I do have a few questions for clarification.

You mentioned distortion as you approach clipping. Obviously I've always known to avoid clipping, but I didn't really realize how early its more subtle effects kick in as you approach clipping levels, until reading the articles linked above from Tedland Studio. You described distortion variations as a function of gain, but I would've expected distortion to vary as a function of actual levels, not necessarily gain setting. Obviously, for any given input level, gain dictates output level, but with different source levels, is it the gain setting or the actual output levels that affects the distortion (or both?)

In other words, if I have an input signal whose level is such that 50dB of gain delivers the output level I think I want, with an acceptable distortion level, what happens if I feed an otherwise equivalent input signal that's 10dB lower and apply 60dB of gain? The resulting output level is essentially the same and noise levels will be higher, but what about distortion? Is the distortion higher because you've applied more gain, or is the distortion essentially the same (ignoring effect of input noise on distortion measurements) because the output levels are the same, leaving the same headroom below clipping levels?

Similar question for phase response. It honesty never occurred to me that phase response varies as a function of gain until I saw your graphs, so this is new territory for me. Having said that, intuition would've told me that it was actual levels, not gain amounts that would affect phase response. I'm guessing my instincts are wrong here since your published graphs plot based on gain, not levels, but I thought I'd double check.

Finally, can you tell me a little more about frequency response vs. gain settings? I see from your specs and from the Sound On Sound review of your debut preamp that your design is practically perfect in this respect, but I gather that some other designs are not. It may be unfair to ask you to characterize products from your competitors, but if you can give me a better idea what the possible effects of gain setting on frequency response are, I'd love to know more. Is it primarily high frequencies that are affected? Are we talking about smallish changes, or are some of the cheaper preamps rolling off severely at extreme gain settings? I suppose I could try to test this on my current rig and gather some data, but I'm not super well equipped for such things - just doing simple noise comparisons was a stretch for me!

Anyway, thanks again for sharing your knowledge, and good luck with your young business. It looks like you all have developed something pretty remarkable - nice job!
Old 31st May 2019
  #24
Lives for gear
Quote:
Originally Posted by nosebleedaudio View Post
YES...
Like...?
Old 31st May 2019
  #25
Lives for gear
 
12ax7's Avatar
 

Quote:
Originally Posted by Ebeowulf17 View Post
Yes, of course if you turn down your signal level (and since you have the same noise level) then you'll end up with a worse signal to noise ratio. [...]
Exactly.

Reading the OP, I thought all this started as your reaction to what other people were saying, namely:
"I often hear people describe preamps as "getting noisy" when you turn them up too far."
I was just telling WHY they say that:

They say that because when you have a low-output mic, you have to crank the level on the pre (and that makes things noisy).

It is exactly as you say: "If you turn down your signal level then you'll end up with a worse signal to noise ratio."

And so this is what you need to simulate in your test (if it is to bear any relationship to the subject matter that inspired the test in the first place).

...After all, if the only consideration for setting gain on a preamp was the SNR within the box itself, then why even bother putting a knob on the thing anyway?
.

Last edited by 12ax7; 31st May 2019 at 05:38 PM..
Old 31st May 2019
  #26
Gear Addict
 

Quote:
Originally Posted by MediaGary View Post
I commend you for developing a testing protocol for sussing out what you want to know, and working through the collected data. That's the sort of thing that solidifies the basis for your opinions and raises the level of insight that we can all share.

I did a bunch of measurements, and generated charts and graphs when we were experiencing a rash of "bypass the preamp" postings in GS. It turned into a massive 10-article series. Take a tour through it simply because you may benefit from some of the spreadsheet data showing voltage, dbV, dBu, etc, or perhaps some of my commentary may help steer your processes.

Glad you're up to this.
Thanks for the kind words. Also, thanks for sharing your articles. I'm only about halfway through, but there's lots of great stuff in there!
Old 31st May 2019
  #27
Gear Addict
 

Quote:
Originally Posted by 12ax7 View Post
And so this is what you need to simulate in your test (if it is to bear any relationship to the subject matter that inspired the test in the first place).

...After all, if the only consideration for setting gain on a preamp was the SNR within the box itself, then why even bother putting a knob on the thing anyway?
[/INDENT].
I'm already testing with very low signals which warrant very high gain settings. The whole point of this exercise was to demonstrate that if you have a very low signal level, the best performance you can get out of any given preamp is with the highest gain setting (assuming the input level is low enough that max gain doesn't result in an output that approaches clipping levels.)

I don't understand what you think needs to be done differently to make this meaningful. I get that you're saying the most common real world application where you need to worry about preamp noise is with dynamic mics that have low sensitivity. In the case of those mics, the signal is very low, so the signal to noise ratio is automatically much worse than if you had a high quality, high output, quiet condenser.

The point I'm trying to make is that if you find yourself in that situation, TURN IT UP! The way preamp noise "at high gain settings" is often discussed has led some people to believe that they shouldn't turn their preamp up all the way, even when they have a quiet signal that needs every bit of that gain. I'm trying to do my part to dispel that myth.

If you've got a really quiet mic, turn up the gain until you've got the output level you want. If you discover that things are noisy then, you can choose whether or not to buy a nicer preamp. What you should not do in that scenario is turn the gain back down under the delusional impression that you're reducing noise in any useful way.

I have full confidence that you understand all this - I'm not trying to lecture you on gain staging or matching mics with preamps. I'm trying to encourage people to describe noisy preamps in a different way - don't talk about how noisy they are at a certain gain setting, because that will confuse people who don't already have a solid handle on gain staging. Describe them as noisy or not in general, and then it's clear what you're talking about. If you've got hot enough signals, you can get away with a noisier preamp. If you need to work with low signals, you'll get much more benefit from a quiet preamp.
Old 31st May 2019
  #28
Lives for gear
 
12ax7's Avatar
 

Quote:
Originally Posted by Ebeowulf17 View Post
[...] The whole point of this exercise was to demonstrate that if you have a very low signal level, the best performance you can get out of any given preamp is with the highest gain setting (assuming the input level is low enough that max gain doesn't result in an output that approaches clipping levels.) [...]
Well, Hmmm...
...And all along I thought most folks already knew that to start with.
(Just one more piece of evidence showing that I'm as old as dirt.)
.
Old 31st May 2019
  #29
Gear Addict
 

Quote:
Originally Posted by 12ax7 View Post
Well, Hmmm...
...And all along I thought most folks already knew that to start with.
(Just one more piece of evidence showing that I'm as old as dirt.)
.
Everyone should, and I think most do, but I've definitely encountered some confusion around this issue in these forums.

As for being older than dirt, that is helpful. I get the distinct impression that people from just a little before my time, who had to learn how to do everything analog, tended to learn the fundamentals much better than those of us who learned on digital tape. I took an early interest, had a good mentor, and soaked up everything I could, but there were other people in my audio program who graduated without ever learning how to patch two devices together... or even what sequence to power up the studio in. Literally all they knew was plugging in mics and twiddling knobs and faders. That was class of '98. I can only assume things have gotten worse with the proliferation of cheap gear. I'd hope that engineers in real studios know this stuff inside out and backwards, but I'm not so sure about home studios.
Old 31st May 2019
  #30
Here for the gear
 

Quote:
Originally Posted by Ebeowulf17 View Post
Quote:
Originally Posted by Karpmentalise View Post
Hi Ebeowulf17,

Very interesting thread you started on a topic very near and dear to our hearts here at Cranborne Audio - Equivalent Input Noise of Preamps.




Yes and no. Equivalent input noise is the important spec to understand here. Anyone who knows EIN and what it means knows you are 100% spot-on in that the every preamp will have its best EIN figures at/near Max Gain - provided it is a low-noise gain source which most preamps these days are. At min gain, the EIN of a preamp will be its worst value, which is usually ~-90dB unweighted with a 150 ohm load. Whereas at max gain, most preamps achieve at least -120dB unweighted with 150 ohm with the absolute best low-noise designs achieving -127-128dB EIN unweighted with a 150 ohm load. (Our Camden 500 is 129.5db!)

BUT - the point I would make here is if you give a preamp max gain, you have a larger dynamic range before the noise floor, however, if you are giving a signal 60dB of gain, it will be raising the noise floor of your source by 60dB as well, which could potentially become problematic.




Yes this test shows well what you mean by the EIN decreasing as you increase the gain. But do bear in mind that preamps' THD (Total Harmonic Distortion) increases as it approaches the clip point. So a THD+N (Total Harmonic Distortion plus Noise) plot with an AP or similar at different gain values is probably the most meaningful way to understand how noise and distortion changes on a preamp at different gain levels. Unfortunately, very few preamp manufacturers cite their preamps' performance at more than one gain value and frequency, and they most commonly cite the figures where they achieve their best performance. The ideal thing would be to do EIN and THD+N sweeps at different values and for all frequencies. But I think we are the only company that does this for our preamps. Here are our tech specs I speak of if you'd like to have a look. These graphs are from our APx555 audio analyser: http://www.cranborne-audio.com/hubfs..._TechSpecs.pdf

Alternatively, we have our "Camden 500 By the Numbers Video" on Youtube you can check out... There's a little bit of marketing rah-rah in the video but not much. Elliott spends most of the time breaking down the specs themselves with graphs - the same graphics in the Tech Specs PDF I linked above but you get Elliott's mockney accent for narration as well

Happy to answer any questions if anyone finds any of this interesting. Cheers all!

Sean from Cranborne Audio
First off, thanks for sharing a detailed and insightful response. Lots of great stuff in there! I do have a few questions for clarification.

You mentioned distortion as you approach clipping. Obviously I've always known to avoid clipping, but I didn't really realize how early its more subtle effects kick in as you approach clipping levels, until reading the articles linked above from Tedland Studio. You described distortion variations as a function of gain, but I would've expected distortion to vary as a function of actual levels, not necessarily gain setting. Obviously, for any given input level, gain dictates output level, but with different source levels, is it the gain setting or the actual output levels that affects the distortion (or both?)

In other words, if I have an input signal whose level is such that 50dB of gain delivers the output level I think I want, with an acceptable distortion level, what happens if I feed an otherwise equivalent input signal that's 10dB lower and apply 60dB of gain? The resulting output level is essentially the same and noise levels will be higher, but what about distortion? Is the distortion higher because you've applied more gain, or is the distortion essentially the same (ignoring effect of input noise on distortion measurements) because the output levels are the same, leaving the same headroom below clipping levels?

Similar question for phase response. It honesty never occurred to me that phase response varies as a function of gain until I saw your graphs, so this is new territory for me. Having said that, intuition would've told me that it was actual levels, not gain amounts that would affect phase response. I'm guessing my instincts are wrong here since your published graphs plot based on gain, not levels, but I thought I'd double check.

Finally, can you tell me a little more about frequency response vs. gain settings? I see from your specs and from the Sound On Sound review of your debut preamp that your design is practically perfect in this respect, but I gather that some other designs are not. It may be unfair to ask you to characterize products from your competitors, but if you can give me a better idea what the possible effects of gain setting on frequency response are, I'd love to know more. Is it primarily high frequencies that are affected? Are we talking about smallish changes, or are some of the cheaper preamps rolling off severely at extreme gain settings? I suppose I could try to test this on my current rig and gather some data, but I'm not super well equipped for such things - just doing simple noise comparisons was a stretch for me!

Anyway, thanks again for sharing your knowledge, and good luck with your young business. It looks like you all have developed something pretty remarkable - nice job!
Hi Ebeowulf, Sean mentioned your questions to me and I thought I'd tag in for him here because those questions are pretty in depth. You touch on several questions there:

1) "Is it gain or output level that affects distortion?" Well it's actually both. For every gain setting you will have varying distortion mechanisms, several of which get worse as the output level rises. Usually the largest rises are in the last 10dB of the headroom or so and before that the noise floor is the limiting factor on the THD+N figure. I attached a graph I found from a production camden I dug up that shows output level vs THD+N to illustrate this behaviour. As output level rises to the right, the THD improves as the noise floor ceases to mask it then starts to rise. At max (68dB) gain (the top line) the Camden has a best case THD figure of 0.0032% at 22dBu out that then rises to 0.005% at just over 25dBu out. Best case is about 30dB of gain (the bottom line) where the THD figure is 0.00012% at 15dBu rising to 0.0002% at 25dBu. For comparison I've also included a similar graph of a fairly well respected £3000 ish digital console preamp.


This varies with gain though for a number of reasons. It could be larger currents at low gains injected into ground planes upsetting a reference somewhere or running out of open loop gain at higher gain settings reducing the preamps error correction or a few other things but the takeaway is that most preamps have gain settings where they achieve much better THD performance than others and there is usually a sweet spot somewhere between 10dB and 40dB of gain where the performance is at it's very best (And you can usually expect that to be the point picked for the spec).


On top of that it also varies with frequency, again this is usually a result of running out of open loop gain for error correction but there is almost always a significant rise in THD at high frequencies that gets worse at high gains as less loop gain is available for the correction.

Sound on sound took a fairly nice graph of the Schoeps VRS5 preamp here:
https://www.soundonsound.com/reviews...sr5-mic-preamp

if you don't mind converting from dB to % at least, calculator to do it for you here: http://www.sengpielaudio.com/calculator-thd.htm

That shows pretty nicely (at a specific output level) how the gain and frequency affect the distortion performance. Add that on to increasing THD towards clipping and you can see pretty quickly where even a decent preamp that manages 0.002% THD on the spec sheet at 30dB gain @1kHz.... is at 0.01% at 60dB at 1kHz, and gets up as high as 0.1% at 0dB of gain, 0.05% at 60dB gain near 20kHz. Increase the output level to somewhere near clipping and even with most clean preamps you will probably be starting to edge onto 0.5% or even more at the extremes of the gain range and definitely into the point where it can start to colour the sound if the rest of your system is clean. THD in that upper band can start to make things a little hard or edgy sounding, often it's mostly too low to hear until you try to use an EQ in that 10k region to get some air or you have a mic with a resonant peak in that frequency band and then you can't boost too far before it gets a bit zingy/harsh. Lots of the variation in "clean" preamps comes from what is happening at the extremes up here (Shimmer, edge, hardness etc).


2) Frequency response and phase vs gain. This is basically two sides of the same coin. The same processes that roll off the signal also delay the signal and hence add phase. There are separate design criteria that cause the top and low end of the frequency/phase response to get worse at different gain settings. The exact factors in play will vary from preamp topology to topology and with the exact components in use. For both top and bottom of the frequency band though you will get more roll off as gain increases.

Again sound on sound have a few reviews which show these effects in play:
https://www.soundonsound.com/reviews...udio-fox-media
https://www.soundonsound.com/reviews...-pml200e-media
https://www.soundonsound.com/reviews...-hv-32p-hv-35p

These gain vs frequency response graphs don't show the effect that well as they tend to leave the analyzer 10Hz HPF on (As they did in the camden review) but the low end roll off gets steadily worse as gain increases until it surpasses the HPF affect at maximum gain.

The millenia looks to be down about 3dB at 40Hz at max gain in ribbon mode though it keeps it's high end frequency response and I wouldn't expect it's phase at high frequencies to be much worse than a camden (probably not much more than 10 degrees or so) and is in no way a cheap or unworthy preamp against it's peers). It's certainly not an effect limited to cheap preamps as it affects all preamps to some extent. Camden does it as well although we only drop down to -1dB at 2Hz at 68dB of gain. Cheap preamps like those on low level interfaces and the like?...Yeah they can get fairly bad. Easily enough to cause strange phase problems at the high end when you start combining tracks or to make the low end of anything you have gained up become a bit gutless.

Frequency and phase response don't actually tend to vary much with output level unless you are using a transformer which does cause frequency response to alter and will wreak merry havoc on phase as it approaches saturation. Most transformerless audio gear will be largely immune to this kind of variation But for interests sake I attached a couple of traces showing the frequency response and phase response of a transformer coupled input altering as the input signal level changes despite never actually getting very close to saturation (all at the same gain level). The higher the signal level, the more dramatic the effect.

3) The original thread question/topic. All non faulty preamps will achieve their best noise performance at maximum gain. The noise performance will always be worst at minimum gain. However most preamps mostly achieve their best noise floor at a certain gain level before the max and from then on only really increase by tiny fractions while boosting the signal level and therefore cutting off headroom. This would be at about 45dB on mid level console preamps and about 25dB on something like a camden. Theoretically the best sweet spot for noise performance would be to get to that point and then stop, leaving maximum headroom on the table, pushing the preamp as gently as possible and therefore distortion at a minimum and then rely on digital gain in your DAW to boost up the signal level.

However. That would be with the signal coming straight out of the preamp into a perfect noiseless ADC.

Any noise post the preamp, including that of the ADC will sum with the noise of the preamp. If your signal is very quiet and you do not give it enough gain to get the noise floor of mic/preamp above that of following equipment by about 10dB or so then the signal to noise floor ratio will be degraded by having not applied enough gain. Ideally you want to apply as much gain as needed at the front end to just not clip your converters for best signal to noise ratio. For best THD you want to be about 10dB below the clip point of your preamp while keeping the gain as close as possible to the preamps sweet spot gain, for best frequency response and phase response you want the gain as low as possible.

So what do you do? Well you have to prioritise. Is the signal noticeably noisy? well that's going to be the most objectionable problem so apply at least enough gain at the preamp to avoid that (even after you compress it later), start with getting the signal peaks just below clip (plus a little safety margin for artist enthusiasm) on your converters. Is your preamp really clean and the tone good? Great. you are done. Is your recording coming out harsh sounding, or muddy and smeared, or just not giving you what you want somehow? Try backing it off a bit and boosting the level in your DAW and see if it helps.

Hmmm, got a little carried away again, hope that was helpful.

Ed
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump