Desirable Signal to Noise Ratio

Discussion in 'Location Recording' started by John Stafford, May 3, 2005.

  1. John Stafford

    John Stafford Well-Known Member

    Just wondering if anyone has any words of wisdom on this topic. I know it's not a simple question to answer, but I don't have any mics that have a worse signal to noise ratio below the mid 70s A-weighted. In practice, is a mic with an S/Nr in the low 70s going appear very noisey? I really notice the difference between a U87 with 80 dB when compared with my AT4060 that has a ratio of 75 when I A/B them, but I find the AT's noise floor perfectly acceptable, although one reason for this is the nature of the noise -it doesn't seem to get in the way of the recording.

    If I have a mic with an S/N figure of say 72 (such as the AKG C34) distance mic'ing a choir, am I likely to notice a lot of noise during the quiet passages?

    I'll be doing some work during the year where I require an ultra low noise floor, and it's great that there are so many new mics out there with very low noise and a ratio well into the 80s. I'm glad to have a cheaper alternative to the TLM-103 for this sort of work.

    Many thanks

    PS it's a pity the U87 isn't a little quieter. It's absolutely amazing for birdsong recordings!
  2. DavidSpearritt

    DavidSpearritt Well-Known Member

    John, I personally think the important indicator for S/N is mic self noise in dBA. This is an equivalent A-weighted SPL of the noise floor so should be independent of the sensitivity of the mic if it is measured correctly. The lower, the better.

    I find this one of the most important specs when choosing a mic, but some disagree, not sure why. There was some rumours around that certain mic manufacturers were spiking the results by ensuring that the mic self noise shape was the same as the inverse of the A-weighting curve, but to me, its still an A-weighted result and hence, if its a low dBA its quiet.

    Some of the new generation mics from Neumann and AKG are less than 10dBA, the MKH's have always been quiet at around 10dBA.

    The SPL at clipping minus the self noise gives the effective dynamic range of the mic. I think the figures you are quoting is 1Pa of pressure on the diaphram at 1kHz (single freq), ie 94dB minus the self noise at that frequency, hence around the 70dB odd.

    Very interested in other's views on this, its an interesting topic.
  3. John Stafford

    John Stafford Well-Known Member

    Thank you very much for your reply. I know that low noise is very important, but my main concern would be the relationship between sensitivity and self noise on the assumption that with high sensitivity the noise will be less apparent as it will allow me to use lower gain.

    On the other hand, I can understand your point of view on the grounds that self-noise constant (I know this is an over-simplification) but sensitivity varies according to frequency, angle and so many factors, so that mics with similar self noise and sensitivity specs can sound different from each other in terms of apparent sensitivity. I would imagine that in the real world, maximum SPL might appear different as well for the same reasons. I'm thinking of the difference between an ideal hypercardioid and an ideal omni shoved in front of a wide sound source. I know that mics with similar pattern types will also have pattern variations; U87 in cardioid and single pattern AT4060 and AT4047 do not share the same polar pattern in my opinion. Anyway I'm beginning to stray from the point, but I can see your point of view -unless I took it up wrongly :wink:

  4. recordista

    recordista Active Member

    A-weighted noise specs have become very deceiving, driving condenser mic manufacturers to create products designed to test well.

    Kind of like the mandatory testing game we're playing in schools these days. Just because a student tests well does not mean s/he actually has critical thinking ability.
  5. John Stafford

    John Stafford Well-Known Member

    I think the analogy is a good one. I remember when I was in school some people attended extra classes in the evening for 'exam technique', where they discussed exam papers from previous years. I don't think they learned any more about their subject.

  6. DavidSpearritt

    DavidSpearritt Well-Known Member

    I have convinced myself that self noise is independent of mic sensitivity, as long as you "assume" the preamp gain noise is not as significant, which it usually isn't with quality preamps. Because self noise is an equivalent SPL, then it will make no difference if your preamp is set on 40dB or 60dB, you still have a lower limit of an SPL that you can record and not an electrical amplitude of some sort. Mics measure SPL not millivolts. This is why I look at self noise with great interest.

    Regarding mic manufacturers making things test well, I am not as cynical as some. There is a limit to what they can do without stuffing up the sound quality, so even if they make the self noise test results good, if the mic sounds bad to do this then it will not sell.
  7. FifthCircle

    FifthCircle Well-Known Member

    Personally, I'm not too concerned with self-noise ratings until I put a mic up and it becomes a distraction.

    For example, ribbon mics don't have much self noise as they are a passive mic. However, your preamp better be quiet when you have the gain cranked as that is where you'll notice the self noise.

    Some condensers are very quiet, most notable that I've recently used are probably the Sennheiser MKH series mics and the MG M930s. All of those similar quality mics, though will be very quiet (ie Neumann, Sennheiser, MG, Schoeps, DPA, etc...).

    A lot of the older tube mics out there sound fantastic, but the noise floor is a bit higher. You listen to the old analog recordings from 40-50 years ago and you aren't thinking noise, rather you hear an orchestra that is recoreded very well.

    Much of the issue is also based on what you are recording. If you are recording solo lute or clavichord, noise floor can be a big issue. An orchestra playing Strauss, though, it becomes less of an issue...

    Oh well... my rambling $0.02.

  8. zemlin

    zemlin Well-Known Member

    95% of the time or so (for me), environmental noise is a bigger factor than mic noise. I don't have a nice quiet studio to work in, and my bigger projects are location recordings.
  9. John Stafford

    John Stafford Well-Known Member

    I was listening to one of those wonderful old Decca recordings, with Janet Baker singing Dido. While this is a recording I have known and loved for a long time, I had never really noticed until recently the level of noise. It such a great record, and sounds amazing. All I ever noticed was a beautiful recording.

    As Karl points out, environmental noise is a bigger problem in the real world.

    The reason I mentioned noise in the first place is that there's a particular venue where everything has to be mic'ed from a distance. The choir is at the front of the church and the organ is at the back. They don't want mic stands all over the place, so placing a single stand in the middle is the best compromise. I was thinking of using a single stereo mic. I'm afraid that the noise of something like a C34 would be very apparent in this situation.

    This is one of those situations where there is no easy answer given that quoting a figure like 72 dB is an over-simplification. Anyhow, the results will probably be incorrect unless the mic is new or in excellent condition.

  10. Cucco

    Cucco Distinguished Member

    Yep, I agree with Karl.

    Very few halls anywhere are truly quiet enough to reveal a mic's self noise and certainly very few playback systems are really all that accurate to portray a mic's self-noise either.

    Of course, if taking into account the self-noise, one must also consider the sensitivity. A highly sensitive mic with a high noise floor will be far less noisy potentially than a "quiet" mic with very low sensitivity. But again, I don't find this to be much of an issue. Even on relatively unsensitive mics with somewhat high noise specs (look at the M50 as an example), the noise rarely poses a large problem. You are much more likely to hear the noise from air-handlers in the concert hall or the stupid trumpet player emptying his spit during the horn solo in Tchaik 5 (yes, I'm bitter.)

    Of course, the A-weighting refers to a specific frequency response (which I don't currently recall the exact numbers and I'm feeling too lazy right now to look it up) but compared to C rating, it lops of low frequencies and high frequencies to concentrate on the critical mid-range. Some of the most frustrating noise I've actually ever heard from a microphone is in the lowest of frequencies (50 Hz and lower) which isn't measured by the A-weight scale.

    So, in all my random babbling above, I guess my point is - I aim for the lowest possible noise floor in any situation and I've never really had a problem with the mic getting in the way of that. In almost every situation, it's environmental variables which cause the problem.

    My .02

    J. :cool:
  11. DavidSpearritt

    DavidSpearritt Well-Known Member

    I am not sure this is correct. Take 3 mics, with the following specs

    Mic Sens (mV/Pa) SelfNoise (dBA)
    A 25 15
    B 12 15
    C 25 20

    Now imagine you setup to record a lute with an average SPL of 20dBA. Recordings A and B will record the lute level 5dBA above the noise floor. The fact that preamp A has to be set on 40dB gain and preamp B has to be set on 46dB gain is not important.

    If you setup mic C, you will still end up with a recording of noisy lute, even though your preamp is set on a low gain.

    This is precisely why self noise is in dBA and not millivolts. All this assumes that the preamp noise is much less significant than the mic noise.

    I agree with all the conclusions that self noise in most live classical recordings is a non-issue, due to the miriad of noises in the hall.

    But for nature recordings or similar it is vital to get a low noise mic and if its sensitive as well then all the better, but this does not influence the noise floor unless the preamp noise dominates.
  12. John Stafford

    John Stafford Well-Known Member


    Forgive me if I'm missing the point, but is it not true that using a mic with the same noise spec. as mic C but with higher sensitivity, the result will be the same as mic C recording a louder lute?

    Another thing; I'd like to know is whether the decibel scale and voltage scale are of a similar type. I have always assumed they are not, in which case I think I understand.

    I've just bought an AKG C34 on ebay. It has both low sensitivity and seems to have a very high noise floor. I was about to buy a Brauner VM-1, but unfortunately I had to restrain myself as I don't actually need the Brauner, but the stereo mic will solve more pressing problems. Some of my favourite pop vocals were done using the AKG C34 so it has its uses even if it proves to be less than ideal for other purposes!

  13. DavidSpearritt

    DavidSpearritt Well-Known Member

    Let's try another approach. The goal, when recording, is to fit the dynamic range of the SPL of the performers into the dynamic range on the digital recorder by adjusting the "gain knobs" on either the recorder or preamp or both.

    Now, you could argue that the digital recorder as being a voltage recorder, which it is of course, but its useful to think about it being an SPL range recorder. My recorder gain knobs are actually marked as SPL and not "gain" from 1 to 10.

    So lets say you have a small chamber orchestra that plays from 20dBA to 80dBA. You will try to set your recorder to clip at 90dBA, and so the lower recorder limit is perhaps 0dBA.

    When the music is playing at 20dBA, mic self noise is easily audible at say 15dBA if comparing any two mics with this spec. All the mic sensitivity issue does is change the gain ratio (pot rotation) on the recorder/preamp to fit the required 60dBA into the same optimium area on the recorder.

    For a sensitive mic this rotation of the gain knob will be less than for a low output mic, but the final "SPL on tape" is the same and so the self noise can be compared consistently between mics, independent of sensitivity.
  14. Zilla

    Zilla Active Member

    attempting the math on the above proposed scenario:

    20dBA(spl) = .0002Pa
    Mic Signal A = .0002Pa*25mV = 5uV
    Mic Signal B = .0002Pa*12mV = 2.12uV

    Preamp A Signal output signal = 20log(5uV * 100/.775V) = -64dBu
    Preamp B Signal output signal = 20log(2.12uV * 200/.775V) = -65dBu

    Both pre-amps have approximately the same signal output level.
    However, pre-amp B has twice (6dB) the noise gain as pre-amp B due to it's higher setting. Therefore A's output will have a better realized signal to noise ratio than B for the same signal output.

    Conclusion: Mic-pre gain is relevant to the final recorded signal to noise ratio. To get the full picture, both Sensitivity AND Self Noise should be evaluated. One is not more important than the other, rather they are inter-related.
  15. DavidSpearritt

    DavidSpearritt Well-Known Member

    All true, but I did state all along in my discussion that preamp gain noise is assumed to be less significant than mic noise, and in practice I have found this to be so.

    When I hear a noisy recording, I am hearing mic noise in the final recording, not preamp noise. The most often effective solution, change the mic, not the preamp.

    I wonder what the effective SPL of preamp self noise is, in a typical high quality preamp.
  16. Zilla

    Zilla Active Member

    My analysis of your fictitious scenario made the assumption of an ideal, noiseless preamp. But to be technically accurate, the final recording will have BOTH mic and preamp noise. I agree with you that mic self noise is the dominating factor with quality equipment. This makes sense because the total mic noise is amplified by the gain of the pre-amp, while only the EIN of the pre-amp's electronics is amplified by the gain. But if you have a low quality pre-amp, it's noise could easily swamp a good mic.
  17. DavidSpearritt

    DavidSpearritt Well-Known Member

    Yes again, your analysis is absolutely correct, the essence of my argument is one of significance, and the point that self noise specs of mics can be consistently compared pretty safely without really having to consider the preamp.

    Also in your example, while the preamp A gain noise is 6dB better for the sensitive mic, doesn't the higher mic A output just cancel this out, ie Mic A's self noise is more twice the uV of mic B. Its the location of the noise gain thats moved, either the noise gain is in the mic or the preamp in this case, so the effective noise is still the same ultimately on the recorder.

    I must read the DIN standard to see how they measure self noise and try to understand why no sensitivity settings are quoted with the self noise specs.
  18. Zilla

    Zilla Active Member

    Sorry, but I must be misunderstanding you because I don't see how you can afirm my analysis and still make that statement.
    No. Disregard the preamp from the discussion for now. Mic A has greater sensitivity, so it ouputs more signal for the same given SPL than Mic B. Both mics have the same self noise. Therefore: Mic A has a greater S/N ratio than mic B. Straight up. Any ratio involves TWO values. The self noise figures alone are not enough to make this evaluation (which is my understanding of what this thread is all about.)
    Noise gain does not move. It is present in the amplifying stage (ie. the preamp). What ever gain is set, that will be the noise gain at the preamp's input (the mic+the pre's own EIN). Thus, the effective noise changes with preamp settings and is a result of the noise of BOTH devices.
  19. DavidSpearritt

    DavidSpearritt Well-Known Member

    But Mic A also outputs more volts for its noise as well. I am interested in your answer to this ... Assuming both mics have a self noise of 15dBA and we put both of these mics into a room with a 20dBA lute. Will you not still have a S/N ratio for both recordings of 5dBA?
  20. Zilla

    Zilla Active Member

    I don't think so. If both mics have the same self noise then their noise voltage should be equivalant.
    Ah, I think I see how I have been misunderstanding you. You are trying to express the final S/N as a "dBA(spl)" value. Well, I cannot refute the accuracy of your statements in those terms because I am not informed enough about that standard. However, it is my intuition that doing so is possibly an erroneous use of terms.

    All I can do is try to answer your question in terms of electronic signals. Again, my attempt at a mathematical analysis:

    A Lute produces 20dBA(spl) = .0002Pa

    Mic Signal A = .0002Pa*25mV/Pa = 5uV
    Mic Signal B = .0002Pa*12mV/Pa = 2.12uV
    Mic Noise A or B = 15dBA(spl) = 0.7uV (IEC 179A)

    Mic A S/N = 20log(5uV/.7uV) = 17dB
    Mic B S/N = 20log(2.12uV/.7uV) = 9.6dB

    So unless someone can show the error of my ways, I believe the numbers above demonstrate that our two fictitious mics indeed have different S/N ratios.

Share This Page