Topic: The myth of unity gain

[Will_S]

I've often wondered why manufacturers of inexpensive recorders don't set the input-stage gain controls up as described in this thread.  Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?  Why don't manufacturers simply use that point as the starting point of the input gain adjustment so that the input stage will not clip before the ADC?

The only reason I can think of for allowing the input setting range to go lower would be to allow the user to fade-in or fade-out using the input gain control.  Do users actually do that with these recorders, even in the 'larger real world market' outside of TS?

I've wondered the same thing.  The fading in/out might be a part of it, or maybe folks want their peaks below full scale to leave headroom for later mixing or other manipulations (yes I realize they could just lower the volume in post - and folks wouldn't need nearly as much headroom for mixing as some level adjustments allow). I also wonder if in some cases the full numeric range may be usable under some settings but not others (e.g. if on one setting of the mic sensitivity switch but not another), but that doesn't seem to the case with the gear I've used.

"Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?"

Ok - now im confused again - that sounds totally counter-intuitive...lowest gain...before overloading - how does low input gain cause overloading?

runonce:  The low input setting doesn't CAUSE overloading, it MASKS it on a too-hot signal that has already been clipped, but a reduced recording level setting then sends a shrunken down but still clipped waveform to the later parts of the recording chain, including the level meters which will show levels well shy of full scale even though the waveform has been clipped.

So a low recording level setting doesn't cause distortion.  But if your input signal is so hot that you need to go below a certain recording level for it to LOOK below full scale on the level meters, it's a way of knowing that your input signal is simply too hot for that particular recorder to deal with.

[Todd R]

"Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?"

Ok - now im confused again - that sounds totally counter-intuitive...lowest gain...before overloading - how does low input gain cause overloading?

Yes, unless I'm confused, this is where I say it would be more useful to discuss maximum input level rather than lowest gain setting.

Saying "Why should the user have to use test tones to find the maximum input level before overloading the input-stage?" makes total sense to me, and doesn't cause me any confusion.

Personally, I don't want to mess with signal generators and o-scopes to determine when distortion is setting in.  I just wish all manufacturers would provide easy-to-find and accurate specs on things like maximum input level for incoming analog inputs.  I can figure it out from there.

[Will_S]

"Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?"

Ok - now im confused again - that sounds totally counter-intuitive...lowest gain...before overloading - how does low input gain cause overloading?

Yes, unless I'm confused, this is where I say it would be more useful to discuss maximum input level rather than lowest gain setting.

Saying "Why should the user have to use test tones to find the maximum input level before overloading the input-stage?" makes total sense to me, and doesn't cause me any confusion.

Personally, I don't want to mess with signal generators and o-scopes to determine when distortion is setting in.  I just wish all manufacturers would provide easy-to-find and accurate specs on things like maximum input level for incoming analog inputs.  I can figure it out from there.

To some extent I agree, but if you want to give a new taper guidance on how to avoid overload distortion, do you want to get into dBV vs dBu, nominal versus peak levels, and all that jazz?

Plus, for something like a soundbard patch:  even if I know what the spec'd maximum output level of a particular mixer is, I've found that in practice the actual output signal won't be nearly that hot for the shows I record.  So it's handy to know, without bringing along measurement gear to figure out exactly how hot the board is running that night, that if I need to turn my recording levels below XX I better stick my attenuator cable in there but otherwise I have one less connection and piece of gear to fuss with.

[runonce]

I've often wondered why manufacturers of inexpensive recorders don't set the input-stage gain controls up as described in this thread.  Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?  Why don't manufacturers simply use that point as the starting point of the input gain adjustment so that the input stage will not clip before the ADC?

The only reason I can think of for allowing the input setting range to go lower would be to allow the user to fade-in or fade-out using the input gain control.  Do users actually do that with these recorders, even in the 'larger real world market' outside of TS?

I've wondered the same thing.  The fading in/out might be a part of it, or maybe folks want their peaks below full scale to leave headroom for later mixing or other manipulations (yes I realize they could just lower the volume in post - and folks wouldn't need nearly as much headroom for mixing as some level adjustments allow). I also wonder if in some cases the full numeric range may be usable under some settings but not others (e.g. if on one setting of the mic sensitivity switch but not another), but that doesn't seem to the case with the gear I've used.

"Why should the user have to use test tones to find the lowest input gain setting before overloading the input-stage?"

Ok - now im confused again - that sounds totally counter-intuitive...lowest gain...before overloading - how does low input gain cause overloading?

runonce:  The low input setting doesn't CAUSE overloading, it MASKS it on a too-hot signal that has already been clipped, but a reduced recording level setting then sends a shrunken down but still clipped waveform to the later parts of the recording chain, including the level meters which will show levels well shy of full scale even though the waveform has been clipped.

So a low recording level setting doesn't cause distortion.  But if your input signal is so hot that you need to go below a certain recording level for it to LOOK below full scale on the level meters, it's a way of knowing that your input signal is simply too hot for that particular recorder to deal with.

Understood - I think thats what we call the classic brickwall. Im sure we've all experienced a show where the meters just didnt look quite right. Predictably - the waveform will look flat, but still show plenty of headroom.

We're usually talking about this relative to the mic>preamp connection - but now "unity myth" takes it to the preamp>recorder context.

At that point it gets harder to talk about because I think everyone is thinking of their own gear...

[Todd R]

Will_S:  agreed, it's just if you dumb things down too much, people don't learn the basics, and we have all these myths and misconceptions.

I think it is important for tapers to learn the basics, and in this case that is that you can send too hot of an analog signal into your preamp or recorder, and cause distortion.  Then after that, we can discuss the shortcut of determining whether this input-overload situation is occurring is when you have to turn your recorder down below level setting 2 or whatever it is.  The problem seems to be that if we skip to the shortcut on how to determine if there is a problem, there ends up being a lot of people who don't understand what the basic problem is at hand.

Agreed though: it is very useful to know for the various recorders where that minimum level setting that indicates input overload is -- very helpful bit of field advice to have at your disposal.

[stevetoney]

Dang it.  I'm really dense today.  LOL. 

Will, Todd, Jon, et al.

I thought I always understood how input signal could be peaking without that showing on your recorder levels. 

However, it was always my understanding that the electronics of the recorder itself could cause us to mess up our recording because it can add too much gain and that was a concern AS WELL because it could also mess up a clean input signal.  This context was discussed without any regard to whether the input is clean or distorted.  So, just to help those of us that are confused/stuck, please answer the following question...

If one knows beyond any doubt that their input signal is clean, can they mess up their recording by setting the levels on their recorder too high, even if the meter on the recorder isn't peaking? 

(Please, if possible to help avoid further confusion, I'm just looking for a basic answer of yes or no with some really noob-ish language about why the answer is yes or why it's no.  All the technical discussion just confuses this basic question for me...and I think others.)

Thanks.

[Gutbucket]

Yes, unless I'm confused, this is where I say it would be more useful to discuss maximum input level rather than lowest gain setting.

Saying "Why should the user have to use test tones to find the maximum input level before overloading the input-stage?" makes total sense to me, and doesn't cause me any confusion.

^^^
This makes sense to me and does seem less confusing.  I'll try and adopt the phrase maximum input level before overloading the input when I describe this.  The next step will then be to explain the difference between that and the input gain setting.

Personally, I don't want to mess with signal generators and o-scopes to determine when distortion is setting in.  I just wish all manufacturers would provide easy-to-find and accurate specs on things like maximum input level for incoming analog inputs.  I can figure it out from there.

Yes!  That shouldn't be too much to ask, simply providing information on the recorders as they now exist.

However, I really don't understand why manufacturers don't set up the input gain setting range as I've described.  It would make problem free recording much simplier all users of the device, especially less techical users that would be complely lost in following this discussion, much less care about it.. except for being confused and disappointed when they get brick-walled recordings even though they've adjusted the recording level of the recorder so that everything looks right on the meters.

[Gutbucket]

If one knows beyond any doubt that their input signal is clean, can they mess up their recording by setting the levels on their recorder too high, even if the meter on the recorder isn't peaking? 

Not really.  I'm temped to say no outright, but yes if you consider too much hiss as messing up the recording. If the preamp stage built into the recorder is not very quiet, setting the input gain level too high will introduce that excess noise (preamp hiss).  But that's less of a catastrophic problem than clipping the input stage (the opposite scenario we've been considering here), plus the level meters should accurately reflect the recording level in that case and won't lie, so there is far less possibility for confusion. 

It would not be an issue at all if only using the internal preamp.  Using an external preamp that has adjustable gain, you can trade off how much gain you add with that verses how much gain you add with the internal circuit.  If you add only a minimal amount with the external preamp and a lot of gain with the noisier internal circuit, you will introduce more noise as hiss, messing up your recording that way.  But that is pretty obvious to most users around here and counter to how they typically operate things, plus as mentioned, even if someone was to set their gains that way, the recorder's meters should still accurately indicate any overload peaks.

[Todd R]

If one knows beyond any doubt that their input signal is clean, can they mess up their recording by setting the levels on their recorder too high, even if the meter on the recorder isn't peaking? 

I'd say, practically speaking, no.

The clean input signal (if too hot) can be messed up by the recorder (since it cannot accept that high of an input signal), but setting levels would have nothing to do with this:  it will be screwed up, but not because of what happens with setting levels.

Setting levels could screw up a clean signal even without the recorder peaking (ie, trying to go above 0dbFS) if a recorder was designed very poorly (this is the second type of distortion I discussed in my post above). 

But that is a foolish way of designing a recorder -- if there is an upper limit on how hot of an output your active gain stage can accomplish, since that active gain stage comes right before the A/D converter and is all inside the box of the recorder, it would be stupid to design your recorder such that that max output level can only achieve a digital level of say -4dbFS -- then you're just asking someone to overload the output of the gain stage by trying to get to -1dbFS.  Instead, you would design the recorder such that the A/D stage would output 0dbFS before you got to that max output level.  Thus, in practical terms, you would not have anyway of distorting your recording by setting the input level above a certain point, with the exception of going above (trying to go above) 0dbFS -- that is besides peaking your recording as you call it.

This is why I say the issue isn't learning where the max input setting is -- there really is no such thing, or perhaps rather, avoiding what that level setting is simply means you should not go above 0dbFS or have clipping -- and usually there already is a clip meter/LED on recorders to show you this, so it would be obvious.

[stevetoney]

If one knows beyond any doubt that their input signal is clean, can they mess up their recording by setting the levels on their recorder too high, even if the meter on the recorder isn't peaking? 

Not really.  I'm temped to say no outright, but yes if you consider too much hiss as messing up the recording. If the preamp stage built into the recorder is not very quiet, setting the input gain level too high will introduce that excess noise (preamp hiss).  But that's less of a catastrophic problem than clipping the input stage (the opposite scenario we've been considering here), plus the level meters should accurately reflect the recording level in that case and won't lie, so there is far less possibility for confusion. 

It would not be an issue at all if only using the internal preamp.  Using an external preamp, you can trade off gain with that verses gain with the internal circuit, so if you added a minimal amount of gain wth the external preamp and alot of gain with the noisier internal circuit, you would get more noise as hiss, messing up your recording that way.  But that is more obvious to most users, and as mentioned, the recorders meters should accurately indicate any overload peaks.

Thank you.  That TOTALLY explains my confusion then because it was my apparently incorrect understanding that the recorder itself could add distorted gain...thus when discussing unity gain, my incorrect conclusion was that you'd never want to go over that in order to be safe against the potential for adding distortion from the recorder. 

In the words of Stan from Southpark, I learned something today.

[stevetoney]

If one knows beyond any doubt that their input signal is clean, can they mess up their recording by setting the levels on their recorder too high, even if the meter on the recorder isn't peaking? 

I'd say, practically speaking, no.

The clean input signal (if too hot) can be messed up by the recorder (since it cannot accept that high of an input signal), but setting levels would have nothing to do with this:  it will be screwed up, but not because of what happens with setting levels.

Setting levels could screw up a clean signal even without the recorder peaking (ie, trying to go above 0dbFS) if a recorder was designed very poorly (this is the second type of distortion I discussed in my post above). 

But that is a foolish way of designing a recorder -- if there is an upper limit on how hot of an output your active gain stage can accomplish, since that active gain stage comes right before the A/D converter and is all inside the box of the recorder, it would be stupid to design your recorder such that that max output level can only achieve a digital level of say -4dbFS -- then you're just asking someone to overload the output of the gain stage by trying to get to -1dbFS.  Instead, you would design the recorder such that the A/D stage would output 0dbFS before you got to that max output level.  Thus, in practical terms, you would not have anyway of distorting your recording by setting the input level above a certain point, with the exception of going above (trying to go above) 0dbFS -- that is besides peaking your recording as you call it.

This is why I say the issue isn't learning where the max input setting is -- there really is no such thing, or perhaps rather, avoiding what that level setting is simply means you should not go above 0dbFS or have clipping -- and usually there already is a clip meter/LED on recorders to show you this, so it would be obvious.

Wow, now that I've got the concept through my crazy thick brain, I understand what you've saying Todd.  Scary. 

[Todd R]

   Glad we've penetrated your cerebral cortex Steve.  With many of us having yesterday off as a holiday, it is effectively Monday today, so I wouldn't worry about your thick-headedness.

[Church-Audio]

People that say unity gain will avoid distortion dont understand like Dsatz said the overload is happening in the front end of the preamp before any attenuation. That's why on mic preamps on consoles at least good ones there is always a pad control to pad the input down before any buffering of the signal has occurred. The problem is now most gain controls in small recorders are digital and thus will overload even when the control is below unity gain. Unity gain assumes that what comes in goes out. But when there is a buffer amp in the front end of a preamp its already set to unity gain or in some cases it might have a 10 or 20 db gain in the that section to help increase signal to noise ratio its in this section that overload will occur and with digital attenuation there is no real potentiometer clamping down on the signal just a ic encoder chip and vca chip. Unity gain is really a term that applies more to recording or live console gain structure than it does to portable recording devices. In the end you should just always aim for -10 or so on your meters and know what the overload limit is of your gear. It can be pretty hard to figure that out with some of the test gear people like me have at my disposal. Generally speaking you want to look for the max gain in at a percentage of THD% It should be ideally less than 1% Its a complicated topic that Dsatz very easily spelled out in plain English Like he always does. I wish I could convey my thoughts as easily as he does its truly a rare talent for technically minded people like us, I have never mastered it.

[Gutbucket]

Steve, since it's effectively Monday and I'm only partially working , I'll try and take this one step further to shine some light on another commonly used technical term, which has been used in this thread, and better describes the actual thing users are trying to optimize when misapplying the term 'unity gain'.  By that I mean determining the limits of a comfortable range of gain settings which provide a good signal-to-noise ratio.  A good signal-to-noise ratio simply means getting more of what we want (signal) and less of what we don't want (noise).  'Noise' in this sense is anything we don't want, and it refers in this case specifically to two forms of noise which show up at opposite extremes of that comfortable range of gain settings- Noise in the form of overload distortion or clipping from a signal which is too hot at one extreme and excess hiss from needing to use too much gain on a signal that is too weak at the other.

There is a range of possible gain settings which will provide a good signal-to-noise ratio, not just a single setting.  The important thing is to find the comfortable 'safe zone' of the gain settings which avoid the problematic extremes.  It's not a single setting but a range, with brick-walling or clipping at one end and excess noise at the other.


The way in which the term 'unity gain' has been misused around here implies a single specific input-gain setting, and so in addition to being the incorrect term for what people are actually trying to optimize, it also is misleading in implying that there is only a single 'optimal' setting.

At the risk of confusing things again, I'll point out that even that mis-implication is incorrect.  As established in this thread, 'unity gain' simply means that the input voltage equals the output voltage, which DSatz pointed out somewhere earlier. [edit- and as Chirs just pointed out as well with regards to gain through mixing consoles] Most small recorders have separate controls for both input gain and for output gain of the line-out or headphone jack.  You can set the input gain way low but compensate by setting the output gain high enough so that the device would still be at ‘unity gain’ as long as the input voltage = the output voltage.  You could also do the opposite and set the input gain too high and the output gain low enough to achieve 'unity gain'.  In either case your signal-to-noise ratio would not be a good as a more reasonable setting for both, and that’s what we’re really concerned about.

Notice that the above description on different settings that all achieve ‘unity gain’ says nothing about actually recording anything on the machine, only about a signal passing through it having the same voltage going out as it did on the way in.  That’s the only way ‘unity gain’ applies for end-users.  When we get more technical and talk about the different stages of circuitry inside the device and how they interact, we might apply the ‘unity gain’ terminology in describing the gain through those internal circuits, but that doesn’t usually apply to anything the end user can adjust on these small recorders. [edit- Chris's example of a mixing console is a good one, since in that case the user can adjust signal gain at various points along the signal path on the console.]

[F.O.Bean]

Isn't the M10 a lil different tho, since if its MIC-IN or LINE-IN can go as low as possible to avoid clipping and/or levels below 0db I cant put into words what I'm really trying to say