Become a Site Supporter and Never see Ads again!

Author Topic: The myth of unity gain  (Read 30679 times)

0 Members and 1 Guest are viewing this topic.

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 #45 on: December 28, 2011, 03:13:20 AM »
I actually understood SOME of that. I have been  recording digitally for around 15 years now, and I don't know SHIT about the technical jargon :P :) I just know how to use the gear to get the best recording possible :)
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

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #46 on: December 28, 2011, 07:44:44 AM »
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 :P

Don't even go there Bean unless you want to drive yourself crazy like I have the last couple days.  That's what I've been trying to sort out in this thread for myself, but also for others that are confused by this...because I know there's a ton of people that have had a ton of discussion about this for at least a couple of years now when the discussion was 'unity gain' on the R-09, R-09HR, M10, PMD-620 and a couple others that I'm not coming up with.  According to the answers I've gotten in this thread, the answer is 'no'.

Offline aaronji

  • Site Supporter
  • Trade Count: (9)
  • Needs to get out more...
  • *
  • Posts: 3861
Re: The myth of unity gain
« Reply #47 on: December 28, 2011, 10:48:55 AM »
^^^ On the M10, if you need to go below one on the rec level dial to keep from going over on the meters, you might be distorting at the input.  At least, that's where guysonic put it (http://taperssection.com/index.php?topic=130924.msg1722942#msg1722942)...

In that post, he also responds to a request about the M10's unity gain setting.  In his response, he says that, at 6, the "output jack signal equaled the input giving unity 'system gain.'"  I thought the word choice was interesting in light of this thread!  He doesn't elaborate, but he is definitely suggesting that there's more to the "voltage in equals voltage out" point than meets the eye...On some recorders, you can even get a pretty good feel for what's being added/attenuated internally from the specs (the DR-07, for example).

Offline vanark

  • TDS
  • Site Supporter
  • Trade Count: (29)
  • Needs to get out more...
  • *
  • Posts: 8510
  • If you ain't right, you better get right!
    • The Mudboy Grotto - North Mississippi Allstar fan site
Re: The myth of unity gain
« Reply #48 on: December 28, 2011, 11:03:35 AM »

In the context of most conversations where the 'unity gain' term was used, it was in reference to a setting on a handheld...R-09, R-09HR, M10, PMD-620, and a maybe a couple others in which people (mostly noobs) were trying to gauge where they should set their recorders.  As you know, handheld recorders have one setting only...a control with a correponding numeric range that goes from 0 to X. 

Is it OK or is it not OK to state that the 'highest recommended setting' (or whatever term will be used in place of the formerly incorrecly applied 'unity gain' term) on the numeric scale of the recorder should be Y to avoid distortion, with Y being a value that has been determined to be roughly where a noob should set his recorder so that there is a nearly 100% chance that the recorder won't contribute any distortion? 

Isn't the false sense of security you're talking about removed, for example, by setting an m10 at 3 instead of 4 (because maybe 4 was felt to be the optimal setting because that's where someone incorrectly thought that voltage in = voltage out).

Without regard to the technical discussion happening in this thread or whether it is the minimum safe setting or the highest recommended setting, the previous discussions I've read on the R-09, R-09HR and M10 have been how to minimize the impact (either amplification or attenuation) of the recorder's preamp.  In doing so, it was a hope that the noisier (?) internal preamps of the recorder could be avoided and the external preamp utilized instead.  While this may not be unity gain, I don't think we should lose sight of the desire to find the most neutral input setting on the recorder.  That is what I wanted to know - the most neutral recommended input setting.

If you have a problem relating to the Live Music Archive (http://www.archive.org/details/etree) please send an e-mail to us admins at LMA(AT)archive(DOT)org or post in the LMA thread here and we'll get on it.

Link to LMA Recordings

Link to Team Dirty South Recordings on the LMA

Mics: Microtech Gefell M21 (with Nbob actives) | Church Audio CA-11 (cards) (with CA UBB)
Pres: babynbox
Recorders: Tascam DR-60D | Tascam DR-40 | Sony PCM-A10 | Edirol R-4

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #49 on: December 28, 2011, 11:32:57 AM »

In the context of most conversations where the 'unity gain' term was used, it was in reference to a setting on a handheld...R-09, R-09HR, M10, PMD-620, and a maybe a couple others in which people (mostly noobs) were trying to gauge where they should set their recorders.  As you know, handheld recorders have one setting only...a control with a correponding numeric range that goes from 0 to X. 

Is it OK or is it not OK to state that the 'highest recommended setting' (or whatever term will be used in place of the formerly incorrecly applied 'unity gain' term) on the numeric scale of the recorder should be Y to avoid distortion, with Y being a value that has been determined to be roughly where a noob should set his recorder so that there is a nearly 100% chance that the recorder won't contribute any distortion? 

Isn't the false sense of security you're talking about removed, for example, by setting an m10 at 3 instead of 4 (because maybe 4 was felt to be the optimal setting because that's where someone incorrectly thought that voltage in = voltage out).

Without regard to the technical discussion happening in this thread or whether it is the minimum safe setting or the highest recommended setting, the previous discussions I've read on the R-09, R-09HR and M10 have been how to minimize the impact (either amplification or attenuation) of the recorder's preamp.  In doing so, it was a hope that the noisier (?) internal preamps of the recorder could be avoided and the external preamp utilized instead.  While this may not be unity gain, I don't think we should lose sight of the desire to find the most neutral input setting on the recorder.  That is what I wanted to know - the most neutral recommended input setting.

Yeah, that's a good point and as has been pointed out, you start running into problems when you roll your recorder levels back too far in hopes of minimizing the noisier recorder gain and maximizing external preamp gain because you could be overloading the external preamp.  You wouldn't realize it because the levels on your recorder may only be peaking at -12db or something like that...e.g. the infamous brickwall situation. 

My entire misunderstanding earlier in this thread (and what I was having trouble getting a grip on) was that I thought 'unity gain' was also important from the recorder perspective in the other direction...that it's also possible to distort your recordings (without levels peaking on the meter) simply because you take your recorder too far over that 'unity gain' point, but Todd and Gut are saying that's not going to happen without the recorder showing levels peaking...which now that I understand the whole discussion better makes complete sense to me.

Offline Church-Audio

  • Trade Count: (44)
  • Needs to get out more...
  • *****
  • Posts: 7571
  • Gender: Male
Re: The myth of unity gain
« Reply #50 on: December 28, 2011, 11:39:45 AM »

In the context of most conversations where the 'unity gain' term was used, it was in reference to a setting on a handheld...R-09, R-09HR, M10, PMD-620, and a maybe a couple others in which people (mostly noobs) were trying to gauge where they should set their recorders.  As you know, handheld recorders have one setting only...a control with a correponding numeric range that goes from 0 to X. 

Is it OK or is it not OK to state that the 'highest recommended setting' (or whatever term will be used in place of the formerly incorrecly applied 'unity gain' term) on the numeric scale of the recorder should be Y to avoid distortion, with Y being a value that has been determined to be roughly where a noob should set his recorder so that there is a nearly 100% chance that the recorder won't contribute any distortion? 

Isn't the false sense of security you're talking about removed, for example, by setting an m10 at 3 instead of 4 (because maybe 4 was felt to be the optimal setting because that's where someone incorrectly thought that voltage in = voltage out).

Without regard to the technical discussion happening in this thread or whether it is the minimum safe setting or the highest recommended setting, the previous discussions I've read on the R-09, R-09HR and M10 have been how to minimize the impact (either amplification or attenuation) of the recorder's preamp.  In doing so, it was a hope that the noisier (?) internal preamps of the recorder could be avoided and the external preamp utilized instead.  While this may not be unity gain, I don't think we should lose sight of the desire to find the most neutral input setting on the recorder.  That is what I wanted to know - the most neutral recommended input setting.

Yeah, that's a good point and as has been pointed out, you start running into problems when you roll your recorder levels back too far in hopes of minimizing the noisier recorder gain and maximizing external preamp gain because you could be overloading the external preamp.  You wouldn't realize it because the levels on your recorder may only be peaking at -12db or something like that...e.g. the infamous brickwall situation. 

My entire misunderstanding earlier in this thread (and what I was having trouble getting a grip on) was that I thought 'unity gain' was also important from the recorder perspective in the other direction...that it's also possible to distort your recordings (without levels peaking on the meter) simply because you take your recorder too far over that 'unity gain' point, but Todd and Gut are saying that's not going to happen without the recorder showing levels peaking...which now that I understand the whole discussion better makes complete sense to me.

If your recorders gain controls are set to unity gain.. And you overload the input they should show the overload in the meter section.. Why because the input is not being attenuated by the gain control so if we are sure the input buffer is at unity and in most cases it is. And we know our gain control is at unity then we should be able to see a "real" level being reflected by the meter providing its accurate :)

So yes you are correct. Unity gain will show us that the input is being overloaded. The problem for overload comes in when you are using the gain to "attenuate" the input below unity gain then you can "trick" The meters into showing a signal as being lower than it really is.
« Last Edit: December 28, 2011, 11:41:30 AM by Church-Audio »
for warranty returns email me at
EMAIL Sales@church-audio.com

stevetoney

  • Guest
  • Trade Count: (0)
Re: The myth of unity gain
« Reply #51 on: December 28, 2011, 11:45:34 AM »
LOL.  If we get this basic concept repeated about 30 different ways in 20 different languages, at some point I'll be able to pass an essay test instead of failing a multiple choice. 

 :coolguy:
« Last Edit: December 28, 2011, 11:47:15 AM by tonedeaf »

Offline it-goes-to-eleven

  • Trade Count: (58)
  • Needs to get out more...
  • *****
  • Posts: 6696
Re: The myth of unity gain
« Reply #52 on: December 28, 2011, 11:57:22 AM »
It all comes back to what I have said for years - use the level that sounds best.  Though it can help to rule out obvious non-starters with simple testing - like the r09 and r09hr's problems distorting signal at lower trim settings.

It may be important to note that some recorders distort input at low trim settings even if they are not brickwalled.  That is my recollection, and it isn't worth the time to re-check.  In the case of the r09/r09hr, you are better to let it clip than to reduce the gain below the "line of death".

And to throw more gas onto the fire..... Some instruments are notorious for producing asymmetric waveforms.  Horns, for example, often produce waveforms with large peaks but shallow bottoms.  How does the recorder handle those at various settings?  Does the digital attenuation introduce artifacts?  Each recorder is different.

Less talk, more testing and comp'ing :P

Offline Todd R

  • Over/Under on next gear purchase: 2 months
  • Trade Count: (29)
  • Needs to get out more...
  • *****
  • Posts: 4901
  • Gender: Male
Re: The myth of unity gain
« Reply #53 on: December 28, 2011, 12:41:13 PM »
^^

Yes, a lot of wisdom there.  I was thinking of responding with a ton of uber-geek stuff (and might still), but this is the bottom line.

I think this thread started off with the idea of being sure the recorder was acting "normally" -- for example, not getting into the realm of distortion or "brickwalling" or whatever -- but beyond that, you just want to have your recorder, and your whole recording system, sound good.  That means testing, trying things out, listening, and deciding what you think sounds best.
Mics: Microtech Gefell m20/m21 (nbob/pfa actives), Line Audio CM3, Church CA-11 cards
Preamp:  none <sniff>
Recorders:  Sound Devices MixPre-6, Sony PCM-M10, Zoom H4nPro

Offline Todd R

  • Over/Under on next gear purchase: 2 months
  • Trade Count: (29)
  • Needs to get out more...
  • *****
  • Posts: 4901
  • Gender: Male
Re: The myth of unity gain
« Reply #54 on: December 28, 2011, 04:05:55 PM »
On Vanark's point about finding the most neutral input setting, I think that is a good idea to do, I just don't know if it is necessarily the unity gain point.  Like itgoes11 says, the way to get there might better be to listen to what sounds best.

The signal path in a recorder or preamp could be going through a number of capacitors, passive attenuators/resistors, active gain stages, etc.  Who knows how the signal is affected by all those components, and what sounds best?  I wouldn't say that it isn't necessarily a bad starting point to begin setting levels at unity gain, and listening to what it sounds like there and then at other gain settings, but I also don't know or assume that the unity gain point is where it will sound most neutral.

Ok, here goes the uber-geeky stuff, but I'll try to walk through it slowly (and I'll use pictures!).  Like I said, there could be lots of elements in the signal path, but consider the op amp gain stage (probably most/all of the recorders we use will have at least one active gain stage using an op amp or something akin to it). 

There are lots of factors affecting the sound of an op amp, and most of these are characterized by the manufacturers.  Among other things, this might include noise performance, slew rate, signal response, etc, and these characteristics might/will change depending on what gain setting you have on the op amp -- that is, unity gain (gain factor of 1, or 0db gain), gain of 2x (6db gain), gain of 10x (20db gain).

Below I've got some graphs and oscilloscope outputs provided in the datasheets for the Burr-Brown OPA2132 op amp and Analog Devices AD823 op amp (good op amps, though probably a bit dated).  These graphics represent how the op amp manufacturers characterize the response of their op amps.

First:  noise performance.  In the figure below you can see that at a gain of 1 (unity gain, 0db gain), the noise performance is about 0.0001%, whereas at a gain of 10 (20db of gain), the noise performance is about 0.001%.  That is, at 20db of gain, the noise performance is 10 times worse than it is at unity gain -- though the noise performance in either case is very, very good (and you'd be unlikely to discern a difference).




Next up: slew rate.  Slew rate represents how fast the output voltage changes in response to changes in the voltage of the input signal.  Higher values are better.

In the figure below, you can see that the slew rate at a gain of +10 (20db gain) is higher (better) than it is at a gain of +2 (6db of gain), which are both higher than the slew rate at a gain of 1 (unity gain).  In all cases, the slew rate is sufficiently high that the differences wouldn't have any discernible difference in sound (I wouldn't think anyway), but the slew rate response is better at higher gains than it is at unity gain.



Next figures:  small and large signal response, and overshoot.  If for instance you send an input of a square wave, you would ideally want to get an output of a square wave.

The figure below is an oscilloscope output of a small signal response of an op amp set to unity gain.  As you can see, what should be a square wave isn't perfectly square:  The signal rise and fall points should be straight up and down, but instead they are actually a steep ramp up and a steep ramp down.  On top of that, you can see the results of "overshoot" where the output signal rises too far and needs to correct itself, and falls too far and needs to correct itself.



The figure below shows overshoot as a percentage for small signal responses, for gains of 1 (unity) and 10 (20db) -- higher means more overshoot, or worse performance.  As you can see, performance is worse at unity gain than it is for a gain of 10, or a 20db gain.




Finally, large signal response.  Again, these use square wave inputs (on top) and show the resulting output (on bottom), which ideally should also be a square wave.  The first figure is the large signal response at a gain of 1 (unity gain), the second figure shows the large signal response at a gain of 10 (20db gain).  As you can see, the large signal response is better (more like a square wave) when the op amp is set for a +20db gain than when it is set for unity gain.





Bottom line, the info from the manufacturers spec sheets indicate that the op amps, not surprisingly, aren't perfectly ideal -- as in, perfectly neutral.  But the op amps are more neutral if you will, when considering signal noise when set for unity gain than when set for a gain of 10 (20db of gain).  On the other hand, the op amps are better/more neutral when set at higher gains rather than being set at unity when you are looking at slew rate, overshoot, and small and large signal response.  I'd guess though that in all cases, it is all just a matter of degree and that for all these performance metrics the op amps perform perfectly good regardless of whether they are run at unity gain or at +20db gain.

So which of the different op amp characteristics most impact the resulting sound, and lead to it sounding more neutral?  I couldn't tell you, but it seems like op amp response isn't universally better at unity gain than it is when set for a gain of +20db.  So I wouldn't assume it will sound more neutral at unity gain -- need to just listen for yourself at unity and at different gain settings and make your own judgements.

Hopefully I've got all that right with interpreting the spec sheets and mshilarious won't come in here saying I've screwed it all up.   :P
Mics: Microtech Gefell m20/m21 (nbob/pfa actives), Line Audio CM3, Church CA-11 cards
Preamp:  none <sniff>
Recorders:  Sound Devices MixPre-6, Sony PCM-M10, Zoom H4nPro

Offline hi and lo

  • Trade Count: (38)
  • Needs to get out more...
  • *****
  • Posts: 2294
Re: The myth of unity gain
« Reply #55 on: December 28, 2011, 04:23:46 PM »
Great post, Todd! The first convincing argument to completely throw the concept of unity gain out the window. I think running at unity gain is still good as a best practice concept, but one cannot assume it is the absolute sweet spot in terms of performance or coloration.

Offline it-goes-to-eleven

  • Trade Count: (58)
  • Needs to get out more...
  • *****
  • Posts: 6696
Re: The myth of unity gain
« Reply #56 on: December 28, 2011, 05:06:27 PM »
That's an excellent summary of the vagaries of op-amp performance.  Unless I am mistaken, one further gotcha is that most of the handhelds don't use op-amps in the traditional sense - they use highly integrated multi-function chipsets that include the a/d.  That's how they get the tiny form-factor and low power consumption.  There is also some question whether any of the gain is done in the digital domain.   Sound Devices has historically declined to reveal how much of their gain is analog vs. digital.

Another sore point with basing unity off the headphone output is that it assumes a lot about the headphone output gain structure.  And even whether the headphone output is before the a/d, or after.  In the 7xx recorders the headphone output signal path is a/d > d/a > amp > headphone out.



Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: The myth of unity gain
« Reply #57 on: December 28, 2011, 05:58:31 PM »
I think running at unity gain is still good as a best practice concept, but one cannot assume it is the absolute sweet spot in terms of performance or coloration.

Remember that you can enter settings on your device which achive unity gain though it, but which may be very poor performing and terrible choices from a signal-to-noise perspective.  As mentioned previously, on a device with both input and output gain controls, there is not just one unity gain point, but a whole range of settings that offset each other to achieve unity gain.  Some of them are very poor choices.  (For instance, on the original R-09 you could turn the input gain all the way up to 30 and the line-out/headphone gain down low enough to compensate for that, so the same signal level was present at the line-input jack and line-out/headphone jack.  But that would be very noisy)

Another sore point with basing unity off the headphone output is that it assumes a lot about the headphone output gain structure.  And even whether the headphone output is before the a/d, or after.  In the 7xx recorders the headphone output signal path is a/d > d/a > amp > headphone out.

All true, but 'unity-gain' just means no gain change of the system, so input level is equal to output level.  The 'system' might be be a single op-amp, or an entire collection of circuits in a box.  Regardless, the system is at unity-gain when it's input level = output level. That the term 'unity-gain' doesn't really apply to a device like a recorder where the output is a digital file and not an electrical signal at an output bus or jack is the entire point of this thread.

Getting closer to what we are looking for, there is a large range of settings that achieve a good signal-to-noise ratio and will all work equally well (or nearly so in pratical terms), not just a single setting.  (Think input gain settings somewhere between 9 and 18 on the original R-09, the safe zone between between the extreme settings that cause distortion or excess noise)

If what you want to find is a single reference input-gain setting for an all-in-one recorder, and the manufacturer doesn't tell you what that setting is >:(, the way to go about it is to 1st select the appropriate reference level and choose a desired amount of headroom, make sure that reference level is within the good signal-to-noise range of the particular input and mode chosen on the recorder, input a test signal of that reference level and adjust the input gain until the meter reads your desired amount of headroom below 0dBFS. (Think input gain setting of 13 on the original R-09.  Though I'm not sure exactly what reverence level that setting is, it was determined to be the defacto 'reference input gain setting' for that machine, comfortably within the good signal-to-noise range, and incorrectly identified around here as the 'unity-gain' setting)
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

Offline kleiner Rainer

  • Trade Count: (0)
  • Taperssection Regular
  • **
  • Posts: 137
  • Gender: Male
Re: The myth of unity gain
« Reply #58 on: December 31, 2011, 09:38:32 AM »
Hi all,

remember my post in the "hiss with preamp" thread lately?

"There is another point that irritates me: the obsession with "unity gain". This is only meaningful if you have access to line input AND line output and if you want to use your device in a chain of other devices as a kind of "signal repeater".

For all other purposes, the only relevant level is 0dBFS, the point where hard clipping starts. Whenever I record a sound source, I set the gain control until I have optimum level with enough headroom to 0dBFS to avoid clipping. The *exact* value of gain or attenuation is not relevant for me, because the S/N ratio is specified *at the A/D input*. I use *optimum gain* instead of unity gain, and that concept is also used by all professionals I know.

On my PCM-M10 I set its line input level so that the 0dB test tone from my MixPre D gives a VU meter display of -24dBFS on the M10, this leaves enough headroom for unexpected level excursions even when using the limiter of my MixPre D. I then set the gain on the MixPre D to whatever the situation requires.

Greetings,

Rainer"

I feel vindicated by DSatz  >:D

BTW the mic preamp in the Sony PCM-D50 uses variable gain: the level control is a four-gang pot, two of the pots are used to set the mic preamp gain (non-inverting amp with opamp, variable resistor between output and inverting input), the mic attenuation switch changes between two resistor values from the inverting opamp input to ground.
The two other pots attenuate either the line input or the mic preamp output, and feed the resulting signal to the A/D converters via buffer amps. The mic pre is an Analog Devices ADA4841, not a bad chip, especially when powered with enough voltage to achieve excellent headroom.

The schematic for the input stage of the D50 can be found on page 25 of the service manual.

I suspect the M10 uses a similar input stage, because the same four-gang pot is used. I felt no urge to open my M10 and check it, but somebody else placed photos of a dismantled M10 on the net where this gain pot is clearly visible.

Both devices are known for rather robust mic inputs, that tolerate very high input levels. I measured the following input levels on the D50, the onset of clipping was verified by ear (and the ear was calibrated with RMAA  ;D)

D50, mic attenuator "0dB", clipping level 350mVRMS
VU level 0dB: Gain setting 10 0.5mVRMS   5 17.4mVRMS  3 191.6mVRMS

D50, mic attenuator "-20dB", clipping level 1.89VRMS (!)
VU level 0dB: Gain setting 10  5.5mVRMS  5 141.7mVRMS  3 1243mVRMS

D50, line in, clipping level not found since the signal generator only reached 2.3VRMS (-14dB on the VU meter)
VU level 0dB: Gain setting 10 369mVRMS  5 2160mVRMS  3 not enough signal generator output!

I will repeat those measurements with my M10, if there is interest in this group.

BTW I would not go below "1" on the gain setting, because the stereo tracking of ganged audio pots is not very good at that setting and below.


Greetings,

Rainer
recording steam trains since 1985

Offline Brian E.

  • Trade Count: (2)
  • Needs to get out more...
  • *****
  • Posts: 4095
  • Gender: Male
  • is chicago.... is not chicago....
Re: The myth of unity gain
« Reply #59 on: July 16, 2015, 12:40:26 AM »
bookmarking to read later
my tapes:  The Archive | Dime | Etree

Recorder - Sony PCM A-10 | Cans - Shure SE535 | Mics - CA-14 Cards | Canon EOS 5D Mark II 17-40L f4 50 f1.4 70-200L f2.8 IS II 430EX II

 

RSS | Mobile
Page created in 0.072 seconds with 40 queries.
© 2002-2024 Taperssection.com
Powered by SMF