Become a Site Supporter and Never see Ads again!

Author Topic: The myth of unity gain  (Read 30694 times)

0 Members and 1 Guest are viewing this topic.

Offline DSatz

  • Site Supporter
  • Trade Count: (35)
  • Needs to get out more...
  • *
  • Posts: 3347
  • Gender: Male
The myth of unity gain
« on: December 25, 2011, 02:02:01 PM »
Hi. May I post a Christmas rant that I hope will help someone in the long run?

I just read a discussion thread in which someone advised a new user to keep the record level control on a certain recorder set higher than about 4 or 5, since that was that recorder's "unity gain" point, and lower settings might lead to overload distortion even if the meters were below 0 dB and the red lights weren't flashing.

Many people here give advice like that. The advice could well be OK in many cases. But in other cases it could be off by enough to make someone end up with an overloaded recording despite their best efforts to avoid it. To understand why I say that, please follow me through the following set of reasons which I've compulsively organized into three blocks of three statements each:
  • Nearly all recorders have active first-stage amplifiers at their inputs rather than passive, variable resistors.
  • Every active circuit has an input overload limit--a signal voltage high enough to force the circuit out of its linear range.
  • Thus there is nearly always some input voltage high enough that if it occurred, it would overload the recorder's input stage and cause distortion.
  • Recording level controls generally affect incoming signals only after those signals have passed through the input stage.
  • (Record level controls are misnamed; they really only control the voltage gain at one point in the recorder's signal path. Recording level is a function of the input voltage, the voltage gain set by the control, and the fixed gain in the rest of the recording circuitry.)
  • If distortion occurs in the input stage, the action of the recording level control will be too late to stop it. In that case your "record level" setting will only determine the level at which you record the distortion. You can set it so that the meters only go to -20 and the overload LEDs never come on; you'll still have distortion due to input overload.
  • If there is a certain voltage above which input stage overload will occur, then there must also be a setting of the record level control at which a signal having that voltage would reach exactly 0 dB (full scale) on the recorder's meters.
  • If you find yourself wanting to set the record level control any lower than that setting in order to avoid reaching 0 dB on the meters, you probably have an input overload problem that needs a separate solution.
  • Thus it's well worth knowing what that control setting is for any recorder that you're using.
However (and this is my point), that setting of the record level control has nothing directly to do with the so-called "unity gain" setting of the device. The two things are independent and can very well differ from one another. I don't have any specific examples but in a given instance, setting a recorder's level control to its "unity gain" setting could still allow input overload to occur before 0 dB (full scale) is reached and before the "overload" indicators come on. Thus if you tell someone that the "unity gain" setting is the lowest safe setting, you may be giving them a false sense of safety.

The only reliable way to find the lowest safe setting is first to find the actual input overload point with a signal generator--then feed that level of signal (or just slightly below it) in to the recorder and observe the level setting at which 0 dB is reached on the meter. That's the safe setting limit for the control on that recorder. The voltage that happens to appear at the line level outputs of the recorder at that point is irrelevant; it may be similar to, or it may be higher than or lower than the voltage being fed in.

I urge everyone who refers to "unity gain" in this context to stop doing so, and replace the concept in your thinking with what really goes on.

--best regards
« Last Edit: December 25, 2011, 03:19:03 PM by DSatz »
music > microphones > a recorder of some sort

Offline Cheesecadet

  • Trade Count: (21)
  • Needs to get out more...
  • *****
  • Posts: 3016
  • Gender: Male
  • http://db.etree.org/cheesecadet
Re: The myth of unity gain
« Reply #1 on: December 25, 2011, 03:11:13 PM »
 :clapping:
AKG 481's, DPA 4061's (Matched), AT ES933's w/ AT853-ELE's (SC/C/O), ECM-19B's, MixPre-3, A10, M10 x 2, Hi Ho Silvers, Various Darktrain & GAKables

Vinyl:
Fluance RT83 Reference > MCS 3230 Receiver > Realistic MC-500's (NOS)

Upcoming:
02/10 J. Dilla
02/14 Wiliie Waldman & The Chocolate Factory
02/15 Gabe & Giz
02/16 GM5
02/28 Grace Potter

Offline SmokinJoe

  • Trade Count: (63)
  • Needs to get out more...
  • *****
  • Posts: 4210
  • Gender: Male
  • "75 and sunny"... life is so much simpler.
    • uploads to archive.org
Re: The myth of unity gain
« Reply #2 on: December 25, 2011, 03:51:24 PM »
Good explanation. Your rants are always a great source of information.
Mics: Schoeps MK4 & CMC5's / Gefell M200's & M210's / ADK-TL / DPA4061's
Pres: V3 / ST9100
Decks: Oade Concert Mod R4Pro / R09 / R05
Photo: Nikon D700's, 2.8 Zooms, and Zeiss primes
Playback: Raspberry Pi > Modi2 Uber > Magni2 > HD650

Offline StuStu

  • Trade Count: (40)
  • Needs to get out more...
  • *****
  • Posts: 2860
  • Gender: Male
Re: The myth of unity gain
« Reply #3 on: December 25, 2011, 06:05:02 PM »
Good explanation. Your rants are always a great source of information.
MK5, MK8, MK41, KM184D, CK77, B3 ---CMD 2U XT, KC5, KCY, AKI---KCY Tinybox, Ugly BB---AETA 4MinX, PMD661 MKII, R-26, M-10, MR-1

Offline F.O.Bean

  • Team Schoeps Tapir that
  • Trade Count: (126)
  • Needs to get out more...
  • *****
  • Posts: 40690
  • Gender: Male
  • Taperus Maximus
    • MediaFire Recordings
Re: The myth of unity gain
« Reply #4 on: December 25, 2011, 07:03:00 PM »
Good explanation. Your rants are always a great source of information.

I had trouble understanding most of it tho :P ;D
Schoeps MK 4V & MK 41V ->
Schoeps 250|0 KCY's (x2) ->
Naiant +60v|Low Noise PFA's (x2) ->
DarkTrain Right Angle Stubby XLR's (x3) ->
Sound Devices MixPre-6 & MixPre-3

http://www.archive.org/bookmarks/diskobean
http://www.archive.org/bookmarks/Bean420
http://bt.etree.org/mytorrents.php
http://www.mediafire.com/folder/j9eu80jpuaubz/Recordings

Offline flipp

  • resident curmudgeon
  • Trade Count: (17)
  • Needs to get out more...
  • *****
  • Posts: 4285
Re: The myth of unity gain
« Reply #5 on: December 25, 2011, 08:53:31 PM »
Happy to read your Christmas rant. I think it ties in nicely with this post you made earlier this year, particularly the final paragraph; and also from earlier this year - this post.

Like Bean, I don't always understand everything you mention in your posts, and sometimes almost nothing. In those cases you provide the seed for me to do further exploration/research.
« Last Edit: December 25, 2011, 09:53:54 PM by flipp »

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #6 on: December 25, 2011, 10:08:43 PM »
Many people here give advice like that. The advice could well be OK in many cases. But in other cases it could be off by enough to make someone end up with an overloaded recording despite their best efforts to avoid it.

I also don't understand your post.  However, my response is not to debate your post on technical merit.  I think you're missing the point of how people, for better or worse, are applying the unity gain term.

Unity gain, as presented in the discsussions I've read elsewhere, is a concept that's only a rule of thumb or guideline to help people understand how to better set their record levels.  I find that the concept is really useful to help to understand on a macro level the concepts, but it particularly helps out noobs in terms that are understandable and once noobs DO understand, they have a much higher success rate for getting good recordings. 

Thus if you tell someone that the "unity gain" setting is the lowest safe setting, you may be giving them a false sense of safety.

I think that your explanation, which I don't get, overcomplicates a simple concept which IMHO, for 99% of users, doesn't need any technical elaboration.  If I use a Sony M10 and it overloads at a loud rock show with a setting of 4, then the next time I set it at 3.  If my explanation for the distortion was that my setting was above 'unity gain', whether technically accurate or not, why does it matter?
 
The only reliable way to find the lowest safe setting is first to find the actual input overload point with a signal generator--

Really?  Who has access to a signal generator?  Not to be cynical, but why would I bother and why do I care? 

If my recordings never break up at 3 but they do at 4, then I know to set it at 3 because 3 is probably below the 'unity gain' point. 

Perhaps people are using some poetic license then if they were to call '4' the unity gain point because they haven't done the bench testing you suggest, but really why is that such a big deal in the context it's been used?


I hope I don't come off as a jerk with my responses above, but I'm just not getting your points about why the concept of unity gain, as it's been used here in the past, is a myth and why it's not valid.

Offline Will_S

  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 2217
Re: The myth of unity gain
« Reply #7 on: December 25, 2011, 10:28:37 PM »
Many people here give advice like that. The advice could well be OK in many cases. But in other cases it could be off by enough to make someone end up with an overloaded recording despite their best efforts to avoid it.

I also don't understand your post.  However, my response is not to debate your post on technical merit.  I think you're missing the point of how people, for better or worse, are applying the unity gain term.

I believe his point is that what many folks call "unity gain" is not, in fact, unity gain. Nor is unity gain (as correctly defined) a particularly important thing for our purposes.  I thought most engineers were sticklers for correct terminology.   :D

I think that your explanation, which I don't get, overcomplicates a simple concept which IMHO, for 99% of users, doesn't need any technical elaboration.  If I use a Sony M10 and it overloads at a loud rock show with a setting of 4, then the next time I set it at 3.  If my explanation for the distortion was that my setting was above 'unity gain', whether technically accurate or not, why does it matter?

Why not just call it what it is - the minimum safe setting for the input level?  Rather than unity gain, that has an actual meaning, but that actual meaning is different.

Also, how do you know setting it at 3 next time will help?  Perhaps what you should be learning about your gear is that if the levels are too hot at 4, you need an external attenuator, turning it down to 3 won't help.  (This is arbitrary, I don't know what the safe level is on an M10.  But I do know that with a Tascam DR2D's mic in, if your levels are too hot at 67, don't bother turning it down to 60 - you need an external attenuator.)
 
The only reliable way to find the lowest safe setting is first to find the actual input overload point with a signal generator--

Really?  Who has access to a signal generator?  Not to be cynical, but why would I bother and why do I care?

Just about anybody reading this site has access to a signal generator.  Any decent audio editing software can generate all the test tones you could every need, and most of us have our computers set up for reasonable playback.  In terms of actual voltages, who cares indeed.  But it can be very useful to fire up a test tone (so you can very easily recognize distortion in the recorded wav file), send it into your recorder at various levels, and see at what input levels you can get an unclipped waveform back and what levels you can't, even if the recording meters aren't hitting "over" as you record.  (Do be aware though that some recorders can take hotter inputs than a lot of consumer gear can put out without clipping, so you do have to be sure you're overloading your recorder and not simply overdriving your source).

Quote
Perhaps people are using some poetic license then if they were to call '4' the unity gain point because they haven't done the bench testing you suggest, but really why is that such a big deal in the context it's been used?

I hope I don't come off as a jerk with my responses above, but I'm just not getting your points about why the concept of unity gain, as it's been used here in the past, is a myth and why it's not valid.

I think the problem is that people are conflating minimum safe input levels with "unity gain", where the output voltage is equal to the input voltage.  But what really matters for recording is making sure that you don't overload any intermediate stages hidden deep inside the recorder.  I agree that in practical terms it is not the actual numeric value of the voltage that matters - it's the record setting that is safe.  But that is not unity gain!

There also seems to be a myth that at the magical "unity gain" setting, you are bypassing all of the internal gain stages of the recorder and getting something purer.  With most consumer gear, that is not the case - there are usually gain stages that you just can't bypass, so there's nothing magical about unity gain on a typical digital pocket recorder.  The concern should be minimum safe input levels, and knowing that you will need an external attenuator if the incoming signal is too hot at those minimum levels. 

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #8 on: December 25, 2011, 11:00:51 PM »
^ Good response Will.  I understand your points and they for sure helps to explain David's initial points, I think.  If terminology could be better applied, then obviously I'm on board with that.  On the other hand, it seems as if the vernacular use of the term was serving a reasonable purpose, even if it wasn't being used with complete accuracy.  The bottom line of my earlier comment is that I know what most people here meant when they used the term.

« Last Edit: December 25, 2011, 11:15:17 PM by tonedeaf »

Offline Will_S

  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 2217
Re: The myth of unity gain
« Reply #9 on: December 25, 2011, 11:28:57 PM »
^ Good response Will and it helps to explain David's initial point, I think.  If terminology could be better applied, then that's cool.  On the other hand, it seems as if the vernacular use of the term was serving a reasonable purpose, even if it wasn't being used with complete accuracy.  It's not my call one way or the other, but I know what most people mean when they use the term (up until now anyway ;)).

Right, and I agree that the bottom line is getting the right result.  But...

 I think in some cases people really were looking for true unity gain (I'll give an example in a second), and there were definitely cases of people writing the manufacturers and asking them what unity gain was for device XX.  Now if the manufacturer gave a response that was correct (for the correct definition of the term), that would not necessarily mean that folks could turn around and use that level as a safe minimum without running into trouble -- so it seems worth making sure that people are using the right terms to avoid disappointment down the line.

It's also problematic because folks were also looking for unity gain in another context - but as DSatz pointed out even then they may not really be getting what they were hoping for.

Remember when the Edirol R09 first came out?  All of a sudden there was a cool recorder with a respectable A/D converter and all kinds of other nice features, but only analog inputs and mediocre microphone preamps.  What was a V3>D8 taper to do without their trusty digital input?  The R09 offered so many obvious advantages over DAT, but how did you make the best use of the V3's better preamps when recording with the R09?  If you set the R09's input level "too high", you could still get decent levels by applying only a little gain with the V3, but then you were "wasting" the good preamps in the V3 and adding more noise than necessary with the R09's lesser gain stage.

At the same time, if you set the R09's levels "too low", at worst you might be brickwalling the R09 (although it could take a pretty hot signal - we're talking hypotheticals here) and even if you weren't brickwalling it, hypothetically you might be activating an un-necessary additional attenuation stage that might itself introduce noise, and meant you were adding more gain (and thus potentially more noise) than you really needed to be adding with the V3.

So "in theory", it made sense to look for some magical "unity gain (sic - meaning this isn't really unity gain either)" level where the R09 would be passing on the V3's signal to its A/D converter with as little additional mucking about as possible.  But finding the level where the R09's line output voltage matched the input voltage wasn't necessarily that magical level - for one thing what really matters is the internal voltage reaching the recorder's internal A/D relative to its nominal level -- the output levels for the recorder are irrelevant.  But also as mshilarious pointed out you often have sequences of attenuators and gain stages, only some of which can be bypassed - so given an un-bypassable stage, "correcting for it" to get back to unity is actually resulting in more mucking about with the signal than the minimum amount of mucking possible with that recorder.
« Last Edit: December 25, 2011, 11:53:08 PM by Will_S »

Offline DSatz

  • Site Supporter
  • Trade Count: (35)
  • Needs to get out more...
  • *
  • Posts: 3347
  • Gender: Male
Re: The myth of unity gain
« Reply #10 on: December 26, 2011, 01:01:55 AM »
If a given setting of the record gain control results in the recorder having voltage in = voltage out, that's unity gain, and I'm not arguing with that use of the term.

I'm only saying that the "lowest safe setting" of the record level control on a given recorder isn't necessarily the same as its unity gain setting. There's simply no inherent reason why the two should coincide, other than that "it would be nice." If you know the unity gain setting for your recorder, but you treat that bit of information as if it told you the lowest safe setting, then with some recorders, you could still get distorted recordings due to input overload even when 0 dB isn't being reached. Or conversely, you could be missing the chance to set levels that would give your recordings a better signal-to-noise ratio.

So we need to find out the lowest safe setting for each recorder (which is easy to do) and pass that information around, instead of finding the unity gain setting and pretending that it tells us anything about the recorder's safe range of settings.

Is that a little clearer, I hope?

--best regards
« Last Edit: December 26, 2011, 01:05:46 AM by DSatz »
music > microphones > a recorder of some sort

Offline F.O.Bean

  • Team Schoeps Tapir that
  • Trade Count: (126)
  • Needs to get out more...
  • *****
  • Posts: 40690
  • Gender: Male
  • Taperus Maximus
    • MediaFire Recordings
Re: The myth of unity gain
« Reply #11 on: December 26, 2011, 01:08:47 AM »
Yeah, thats a lil better!
Schoeps MK 4V & MK 41V ->
Schoeps 250|0 KCY's (x2) ->
Naiant +60v|Low Noise PFA's (x2) ->
DarkTrain Right Angle Stubby XLR's (x3) ->
Sound Devices MixPre-6 & MixPre-3

http://www.archive.org/bookmarks/diskobean
http://www.archive.org/bookmarks/Bean420
http://bt.etree.org/mytorrents.php
http://www.mediafire.com/folder/j9eu80jpuaubz/Recordings

Offline John Willett

  • Trade Count: (0)
  • Taperssection All-Star
  • ****
  • Posts: 1550
  • Gender: Male
  • Bio:
    • Sound-Link ProAudio
Re: The myth of unity gain
« Reply #12 on: December 26, 2011, 08:07:34 AM »
Hi. May I post a Christmas rant that I hope will help someone in the long run?

I just read a discussion thread in which someone advised a new user to keep the record level control on a certain recorder set higher than about 4 or 5, since that was that recorder's "unity gain" point, and lower settings might lead to overload distortion even if the meters were below 0 dB and the red lights weren't flashing.

Many people here give advice like that. The advice could well be OK in many cases. But in other cases it could be off by enough to make someone end up with an overloaded recording despite their best efforts to avoid it. To understand why I say that, please follow me through the following set of reasons which I've compulsively organized into three blocks of three statements each:
  • Nearly all recorders have active first-stage amplifiers at their inputs rather than passive, variable resistors.
  • Every active circuit has an input overload limit--a signal voltage high enough to force the circuit out of its linear range.
  • Thus there is nearly always some input voltage high enough that if it occurred, it would overload the recorder's input stage and cause distortion.
  • Recording level controls generally affect incoming signals only after those signals have passed through the input stage.
  • (Record level controls are misnamed; they really only control the voltage gain at one point in the recorder's signal path. Recording level is a function of the input voltage, the voltage gain set by the control, and the fixed gain in the rest of the recording circuitry.)
  • If distortion occurs in the input stage, the action of the recording level control will be too late to stop it. In that case your "record level" setting will only determine the level at which you record the distortion. You can set it so that the meters only go to -20 and the overload LEDs never come on; you'll still have distortion due to input overload.
  • If there is a certain voltage above which input stage overload will occur, then there must also be a setting of the record level control at which a signal having that voltage would reach exactly 0 dB (full scale) on the recorder's meters.
  • If you find yourself wanting to set the record level control any lower than that setting in order to avoid reaching 0 dB on the meters, you probably have an input overload problem that needs a separate solution.
  • Thus it's well worth knowing what that control setting is for any recorder that you're using.
However (and this is my point), that setting of the record level control has nothing directly to do with the so-called "unity gain" setting of the device. The two things are independent and can very well differ from one another. I don't have any specific examples but in a given instance, setting a recorder's level control to its "unity gain" setting could still allow input overload to occur before 0 dB (full scale) is reached and before the "overload" indicators come on. Thus if you tell someone that the "unity gain" setting is the lowest safe setting, you may be giving them a false sense of safety.

The only reliable way to find the lowest safe setting is first to find the actual input overload point with a signal generator--then feed that level of signal (or just slightly below it) in to the recorder and observe the level setting at which 0 dB is reached on the meter. That's the safe setting limit for the control on that recorder. The voltage that happens to appear at the line level outputs of the recorder at that point is irrelevant; it may be similar to, or it may be higher than or lower than the voltage being fed in.

I urge everyone who refers to "unity gain" in this context to stop doing so, and replace the concept in your thinking with what really goes on.

--best regards

Well said DSatz - a very good and well argued post - I totally agree.

Many video cameras suffer from this problem - the level control is after the first amplification stage and it was very easy to distort this first stage even though the meters were showing well safe.  This is why Sennheiser did a special padded-down version of the K6 for video cameras to prevent overload of the first stage.
 

runonce

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #13 on: December 26, 2011, 08:40:40 AM »
If a given setting of the record gain control results in the recorder having voltage in = voltage out, that's unity gain, and I'm not arguing with that use of the term.

I'm only saying that the "lowest safe setting" of the record level control on a given recorder isn't necessarily the same as its unity gain setting. There's simply no inherent reason why the two should coincide, other than that "it would be nice." If you know the unity gain setting for your recorder, but you treat that bit of information as if it told you the lowest safe setting, then with some recorders, you could still get distorted recordings due to input overload even when 0 dB isn't being reached. Or conversely, you could be missing the chance to set levels that would give your recordings a better signal-to-noise ratio.

So we need to find out the lowest safe setting for each recorder (which is easy to do) and pass that information around, instead of finding the unity gain setting and pretending that it tells us anything about the recorder's safe range of settings.

Is that a little clearer, I hope?

--best regards

When ever I use the term, I mean exactly what you state in your first sentence. But - I think I(we) use the term loosely - and without regard to any true measurements.
Also - note - when we use the term "unity", we are almost always referring to a preamp feeding a line level to a recorder.
I think what we're saying is that a "theoretical" unity is good starting point - and NOT a lowest safe setting.  If anything, I think of unity as the "highest safe setting"

I used to tell people just to record a line level from CD player - and get a ball park idea where "unity" was and use that for a starting point.
Your first post sounds more like a desciption of "brickwalling." (a term I think you've claimed we made up or doesnt really exist ;))

I used to have device by Midiman called the Flying Calf A/D - it had no level controls - just meters and inputs. You drove it from your mixer...I think thats the model we are going for here.

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #14 on: December 26, 2011, 10:50:59 AM »
Thanks guys for the follow-up posts that help clarify the issue.  Much appreciated.  After getting a better understanding of the differences between the terms 'unity gain' and 'recommended input setting', for accuracy I'll carry the recommendations forward in my net conversations.

I'm still a little bit confused though.  In the follow-up discussion, the focus was on ensuring the proper use of the term 'lowest recommended input setting'.  In the context of the discussions in which the unity gain term was improperly applied, the goal is to make sure that the setting on your recorder is BELOW a specific point to make sure that the recorder isn't either overloading or mucking up the more desirable external pre-amp input gain, wouldn't the proper use of the term be to specify the 'highest recommended input setting'?
« Last Edit: December 26, 2011, 11:09:30 AM by tonedeaf »

 

RSS | Mobile
Page created in 2.766 seconds with 43 queries.
© 2002-2024 Taperssection.com
Powered by SMF