Loading ...
Sorry, an error occurred while loading the content.

RE: [micbuilders] 20dB gain 1nV/rtHz Mic preamp

Expand Messages
  • umashankar mantravadi
    i do use transformers when the object is very low noise - that 20 db of noise free gain is not to be scoffed at.i use parabolic reflectors too, for noise free
    Message 1 of 13 , Oct 1, 2006
    • 0 Attachment
      i do use transformers when the object is very low noise - that 20 db of
      noise free gain is not to be scoffed at.i use parabolic reflectors too, for
      noise free gain.

      one rarely needs to use the same pre for all purposes. i like the cleanness
      of the sound i get from fairly simple transformerless designs. when they are
      not aiming at ultranoise it is not very difficult to keep them simple.

      did any of you get to look at the drawing i posted?

      umashankar


      >From: "Thorsten Loesch" <Thorsten.Loesch@...>
      >Reply-To: micbuilders@yahoogroups.com
      >To: <micbuilders@yahoogroups.com>
      >Subject: RE: [micbuilders] 20dB gain 1nV/rtHz Mic preamp
      >Date: Sat, 30 Sep 2006 11:37:59 +0100
      >
      >Guy's,
      >
      >Why settle for something that EASY to do....
      >
      >Let's call it the "sub 1nV|/Hz @ Unity Gain" challenge, for the time being
      >strictly theoretical.
      >
      >But then again, do I really care about super low noise once my noise factor
      >is down to 3db? I trend to always apply the 3db rule, that is if I spend
      >money/effort and I do not get a 1.41 potential improvement (be it in
      >subjective quality points on my personal scale, or measured) it is usually
      >not worth doing. But that is just me.
      >
      >BTB, IIRC the Coles Ribbon Mike has a build in transformer that steps up
      >the
      >impedance and voltage, if it offers a 1:10 stepup then the 300R are
      >likely
      >mainly the resistive losses in the transformer primary and it would explain
      >the 20db gain difference too.
      >
      >In that case using a mike amp with "only" 2.2nV|/Hz Ein will do fine. If I
      >have another transformer on the Input (as it was common in the olde days, a
      >1:5 for example) of the Preamp I can manage a Noise Factor of 3db by having
      >an active circuit with 11nV|/Hz, which is trivial and I can get a sub 1db
      >Noise Factor using fairly common integrated circuits.
      >
      >Just goes to show how much the irrational fear of using a transformer
      >(can't
      >be cheapness, as the cost of a Coles 4038 is � 700 each anyway, decent Mike
      >Stepups are available < � 100) compromises design.... ;-)
      >
      >Ciao T
      >
      >
      >
      >
      >[Non-text portions of this message have been removed]
      >
    • userno232000
      ... Ohm ... transformer. Eric has an impedance measurement in our Files for a Royer SF12 ribbon which shows about 500R above 1kHz and rising to more than 1k5
      Message 2 of 13 , Oct 1, 2006
      • 0 Attachment
        --- In micbuilders@yahoogroups.com, Richard Lee <ricardo@...> wrote:
        >
        > >I would assume the 300R spec is the output impedance of the
        > transformer since the ribbon itself is usually a fraction of an
        Ohm
        > resistance.
        >
        > 300R is the impedance seen if you look into the output
        transformer. Eric has an impedance measurement in our Files for a
        Royer SF12 ribbon which shows about 500R above 1kHz and rising to
        more than 1k5 at resonance.
        >
        > >The spec you quoted is about .5V peak signal @130dB so a 3K
        feedback would give 20dB of gain and 5V peak out.
        >
        > 20dB gain would give good results with my example and your 10V FS
        24b A/D and the line inputs of professional mixers.
        >
        > But your 3K feedback implies 333R between the input emitters. If
        the rest of the amp had no noise, you would have degraded the noise
        floor of your 300R microphone by more than 3dB. (assuming the 300R

        No, I was thinking of that diff-amp structure like the one from that
        company in Texas. It has internal common mode feedback so the trans
        former goes across the inputs and the feedback goes directly the
        ends of the transformer secondary. You would need to add some "grid
        leaks" if the transformer is not center-tapped. I would do a roll
        your own with JFET inputs. Of course there would be some
        equalization needed because the gain is now RF/Ztran and DC offset
        could be problematic if the DCR of the tran is really low. There
        also could also be problems if the Ztran gets funky out of band. A
        nice design project but specific to this mic only.



        > >The guys from SSM said that it was a sticking point that when
        mixing consoles were not used or micing pros insisted that they
        support 0dB min gain.
        >
        > Scott, I'm not sure I understand this. Could you say it again
        slowly?
        >
        > Thanks

        Little too quick there. I meant pros want the option of using the
        inputs at line level and insist on unity gain as an option. There
        are some cute gain of 5 and up diff-amp circuits that can be made
        very cheaply but can't be used in this application.

        > --
        > No virus found in this outgoing message.
        > Checked by AVG Free Edition.
        > Version: 7.1.407 / Virus Database: 268.12.10/459 - Release Date:
        29/09/06
        >
      • Richard Lee
        ... Should be ... 420R above 1kHz. ... Sorry! Should be 20dB gain. -- No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.1.407 /
        Message 3 of 13 , Oct 1, 2006
        • 0 Attachment
          >Eric has an impedance measurement in our Files for a Royer SF12 ribbon which shows about 500R above 1kHz and rising to more than 1k5 at resonance.

          Should be ... 420R above 1kHz.

          >Then for 30dB gain, your feedback resistors would be 2 x 65R5

          Sorry! Should be 20dB gain.




          --
          No virus found in this outgoing message.
          Checked by AVG Free Edition.
          Version: 7.1.407 / Virus Database: 268.12.10/459 - Release Date: 29/09/06
        • Richard Lee
          ... This is easy to do. Just make the feedback resistors 0R. The resulting amp will have unity gain and Ein determined only by the input devices. If you
          Message 4 of 13 , Oct 1, 2006
          • 0 Attachment
            >the "sub 1nV|/Hz @ Unity Gain" challenge

            This is easy to do. Just make the feedback resistors 0R. The resulting amp will have unity gain and Ein determined only by the input devices. If you speak nicely to Audio Precision, they might tell you how to do this with the evil AD797 w/o the free oscillator.

            It's 6 - 20dB gain with 1nV/rtHz which is Unobtainium w/o transformers.

            >do I really care about super low noise once my noise factor is down to 3db?

            My view is that of a recording engineer. There are (a few) times when I want every -dB of Ein. My moderately noisy cathedral is one. The 4038 is another.

            An SB TX103 transformer is a very good way of dealing with these. But I can't afford the money or space.

            >if it offers a 1:10 stepup then the 300R are likely mainly the resistive losses in the transformer primary

            The step up is MUCH more than 1:10. If the 300R is mainly resistive losses in the transformer, it would be a crap transformer and the 4038 a crap mike rather than one of the best in the known universe.

            But then it was designed by BBC recording engineers who were also mike & speaker designers. The Golden Pinnae claim these people never go to live concerts, so anything they do must be crap.

            However, if you are not a Golden Pinnae, here are pearls of wisdom

            www.bbc.co.uk/rd/pubs/archive/pdffiles/monographs/



            --
            No virus found in this outgoing message.
            Checked by AVG Free Edition.
            Version: 7.1.407 / Virus Database: 268.12.10/459 - Release Date: 29/09/06
          • Thorsten Loesch
            Hi, ... ... Actually, I should formulated my challenge more precisely as 1nV|/Hz at untiy gain in a microphone preamp that has adjustable
            Message 5 of 13 , Oct 2, 2006
            • 0 Attachment
              Hi,

              > >the "sub 1nV|/Hz @ Unity Gain" challenge
              >
              > This is easy to do.

              <snipped for clarity>

              > It's 6 - 20dB gain with 1nV/rtHz which is Unobtainium
              > w/o transformers.

              Actually, I should formulated my challenge more precisely as 1nV|/Hz
              at untiy gain in a microphone preamp that has adjustable gain in the
              0db to 60db range.... And I agree, Transformers offer a neat solution.

              > The step up is MUCH more than 1:10. If the 300R is mainly
              > resistive losses in the transformer, it would be a crap
              > transformer

              Actually, you would be surprised. A decent transformer has usually a
              few Ohm primary DCR unavoidably, in a 1:10 stepup transformer this
              easily translates into a a few 100 ohm output impedance, if you
              reduce the turns ratio the primary DCR becomes more critical and
              there is a limit in terms of physical transformer size, wire
              thickness and bandwidth that make a "lossless" transformer unlikely...

              Royer list the ribbon elements as ~ 30db less sensitive than
              Condensor mikes, so if 10mV/94db is 'average' condensor mike, then
              the ribbon needs a 1:30 stepup and thus an impedance stepup of 1:900,
              then for a 300R output impedance of the transformer the DCR in the
              primary (assuming a lossless secondary) must not be larger than 0.33
              Ohm.....

              > However, if you are not a Golden Pinnae, here are pearls of wisdom
              >
              > www.bbc.co.uk/rd/pubs/archive/pdffiles/monographs/

              I am very familiar with the content of these (I actually downloaded
              the lot where pertinent to audio a while back, when the BBC started
              to post them and have been following), however, just because it is
              written in a BBC monograph does not automatically make it any more
              true than being written in the JAES or in the IEEE Transactions.

              And your banging on about "golden pinnae" becomes extremely tiring.
              Just as I do not assume people to be deaf just because they have an
              engineering degree it may be worth for you to consider that others
              who do not hold one may still have decent hearing.

              BTW, I hope you can easily agree with me that the complex
              interactions and noise loops in unbalanced audio systems containing
              more than one integrated unit actually result in conditions under
              which mains and interconnection cables of sufficiently differing
              construction (electrincal properties) are almost invarbialy forced to
              make an easily measured difference to noisemodulation and noisefloor,
              which in turn may easily approach or exceed audibility?

              Ciao T
            • Richard Lee
              ... Wow! A floating virtual earth ! Have you a link to that company in Texas ? Virtual Earth is the best configuration for noise. No evil noise resistors.
              Message 6 of 13 , Oct 2, 2006
              • 0 Attachment
                >No, I was thinking of that diff-amp structure like the one from that company in Texas. It has internal common mode feedback so the transformer goes across the inputs and the feedback goes directly the ends of the transformer secondary.

                Wow! A "floating virtual earth"! Have you a link to "that company in Texas"?

                Virtual Earth is the best configuration for noise. No evil noise resistors. Definitely the way to go if your source dun mind 0R loading. Unfortunately, among common sources, only Moving Coil phono cartridges would fall into this category. Have a look at Eric's impedance measurement of Royer SF12 for what happens with dynamic mikes (including ribbons). The response would be the inverse of the impedance curve cos " the gain is now RF/Ztran"

                Audio magazine used to publish impedance plots of mikes IIRC. Anyone know if these are now on the net?

                If I wasn't just a pseudo guru, I would now pontificate on reciprocity & show a design of a leaky Unobtainium transformer or gyrator optimised for this application ala P Baxandall. "Loudspeakers as High Quality Microphones" from our Files\ricardo directory. Would the true gurus like Dick Campbell care to oblige ?

                The SSM2019 type ICs aren't optimised for this. Doing so would operate the input devices in common base which has a best NF of 3dB. The INA103 might be suitable as the feedback resistors are accessible so you could cross couple to the bases of the inputs. Depends on how much internal cross coupling internally.

                Mr Wurcer, have you any insider friends at Texas who could enlighten us on this and also on ...

                >Anyone know if the Unity Gain x1 buffers shown on the SSM2019 datasheet are also in the INA103/163?

                >Mr Wurcer, are these low-impedance points?

                >If they aren't, we might be able to parallel further LM394s / MAT02/3s on the input and still have the INA103 give us low distortion in a small package.
                ___________

                Incidentally, this means for the SSM2019 / INA103 used as intended, the important matching is to the emitter emitter resistor rather than the source cos this is effectively on a common base input which has a max NF of 3dB while the source is common emitter which can have much greater NF. Prof Cohen runs his input current to match this in

                "Double Balanced Microphone Amplifier" - Graeme Cohen, AES preprint 2106
                ___________

                >I meant pros want the option of using the inputs at line level and insist on unity gain as an option.

                At Calrec in the 1980s, the mike inputs were transformer coupled and we would switch to a different transformer coupled or electronic balanced circuit as the gain dropped. Then, even the Broadcast organisations, who were BY FAR the most demanding customers, were happy with +30dB more dynamic range on top of 16b processing, which we did quite easily (?!) under "all operating conditions". This little phrase was the big difference between broadcast desk and a "music" desk cos one of the tests for this on a 72 channel desk was feeding a signal of one strip into the next until it passed through all 72 channels. There were the inevitable arguments about which knobs you could twiddle under "all operating conditions".

                I'm not sure how the best desks do things today as we should try to accomodate 24b recorders. Fortunately (?), most (practical) 24b A/D implementations have only slightly better dynamic range than a good 16b converter like the old Sony PCM-F1.

                A quick look at the RMAA site shows that >90dB A wt is excellent for 16b and >110 is excellent for 24b loopback on soundcards. This hold for several highly regarded 24b "professional" digital recorders too. Scott mentioned his experience with a true Panasonic 24b A/D prototype which was declared unmanufacturable.
                ____________

                Umashankar, your circuit is a form of "virtual earth" but I'm not sure what the advantage is apart from non 0R. It will have good noise at low gains. But you still have the evil 1K noise resistor to earth. It looks typical Baxandall though. Do you have a reference to the article where it came from?


                --
                No virus found in this outgoing message.
                Checked by AVG Free Edition.
                Version: 7.1.407 / Virus Database: 268.12.10/459 - Release Date: 29/09/06
              • Roy Singleton
                The Golden Pinnae claim these people never go to live concerts . Went to an orchestral concert recently and was a little afraid, as my seat was four rows from
                Message 7 of 13 , Oct 3, 2006
                • 0 Attachment
                  "The Golden Pinnae claim these people never go to live concerts".
                  Went to an orchestral concert recently and was a little afraid, as my seat was four rows from the front, whereas I'm used to sitting in the "ideal" (i.e. expensive) seats.
                  The music started (Egmont overture) and the thought struck me, "My God! It sounds just like electrostatics!"

                  [Non-text portions of this message have been removed]
                • umashankar mantravadi
                  i have found the article (distortion in low noise amplifiers) and will post it shortly. what it is supposed to do is to give the low noise performance of a
                  Message 8 of 13 , Oct 3, 2006
                  • 0 Attachment
                    i have found the article (distortion in low noise amplifiers) and will post
                    it shortly. what it is supposed to do is to give the low noise performance
                    of a noninverting amp and the overload performance of an inverting amplifier
                    in one go. the evil feedback resistor doesnot have to be 1 k (it usually is
                    220 ohms in my circuits).

                    umashankar

                    >
                  • Thorsten Loesch
                    Hi, ... I often get the same feel for Tannoy s, my open baffles or my new monitor designs. The reason is simple, in the front rows you get a sonic picture
                    Message 9 of 13 , Oct 3, 2006
                    • 0 Attachment
                      Hi,

                      > The music started (Egmont overture) and the thought struck me,
                      > "My God! It sounds just like electrostatics!"

                      I often get the same feel for Tannoy's, my open baffles or my new
                      monitor designs.

                      The reason is simple, in the front rows you get a sonic picture
                      somewhat closer to what the mikes pick up when recording and when the
                      replay system and listening room do not fubar the recording too much
                      than front row seats indeed sound a lot like really good HiFi, if you
                      come from the seats that get served expensive purely reverb
                      sauce.... ;-)

                      I actually appreciate front row seats...

                      Ciao T
                    • Richard Lee
                      ... A good & easy way to do this Virtual Earth trick and get the best noise performance from your OPA is to use the 4 resistor 1 OPA differential amp but omit
                      Message 10 of 13 , Oct 4, 2006
                      • 0 Attachment
                        >Wow! A "floating virtual earth"! Have you a link to "that company in Texas"?

                        >Virtual Earth is the best configuration for noise. No evil noise resistors. Definitely the way to go if your source dun mind 0R loading.

                        A good & easy way to do this Virtual Earth trick and get the best noise performance from your OPA is to use the 4 resistor 1 OPA differential amp but omit the 2 input resistors.

                        This is balanced but not floating. Excellent noise performance.

                        But it doesn't have the very high CM input Z which Bill Whitlock, Jensen points out is necessary for the best practical CMRR. Not so important if your source is fully floating.

                        "Subtleties count in wide-dynamic range analog interfaces" EDN 4jun98

                        from a link of Thorsten's.

                        Mr Wurcer, what's that company in Texas. Not the dreaded TI ? ... 8>D


                        --
                        No virus found in this outgoing message.
                        Checked by AVG Free Edition.
                        Version: 7.1.407 / Virus Database: 268.12.12/462 - Release Date: 3/10/06
                      Your message has been successfully submitted and would be delivered to recipients shortly.