And I'm saying these concepts don't apply here. We are not dealing with EM waves and match terminated systems. For simple transducers the real part of the impedance and it's voltage noise is what matters. If you choose to use the concept of noise figure the optimum is 0dB in this case, in actuality asymptotic.
Doesn't make much sense. I'd rather believe that beyond 3mA your capability of measuring noise was limited. The noise improvement by increasing the collector current is obvious in that range, since 1/2gm is still far larger than Rbb. If any, high injection level effect should appear at much higher currents.
I quote this in my MicBuilders document and it applies to specific Ortofon cartridges.
The context is that I believe Duraglit & Gerhad represent the SOTA for real-life devices though Gerhad has never tried his in a MC vinyl playing system. In the MicBuilders document, I also point out that Gerhad's amp will have a NF of 2dB for those Ortofon cartridges so that's the maximum noise 'improvement' possible.
While Duraglit does well in a Ein willy-waving comp (as shown by Bonsai) it is really the performance with real-life MC cartridges that is my main interest .. and likely the interest of those here who want to build something useful. (Cue 2 pages of Guru Wurcer on record noise drowning ...)
Please post a link to this much quoted MicBuilders document. I can't find it.
Please define "SOTA for real-life devices".
Please explain why "a NF of 2dB for those Ortofon cartridges" is the "maximum noise 'improvement' possible". It doesn't make sense to me, together with using NF in this context. NF is a useless metric when it comes to audio reproduction since, as others already mentioned, the concept of maximum power transfer by matching impedances doesn't apply.
Please explain why "Duraglit" would perform better with real-life MC cartridges, compared to other approaches. I don't see any technical reasons why.
Are you sure? Please provide a link or any other relevant reference to this Audio Magazine, showing a greater than 10dB improvement. Improvement in what parameters?
Your question was already answered:
Since the current mirror version has the noise very close to the theoretical value and distortions are as good as one can get from a bipolar open loop design, I find this version much better from a practical perspective.
LTSpice won't help with any of the solar cell aspects.
As others already mentioned, the concept of matching impedances for maximum power transfer is nonsense for audio. I don't see any serious technical reason for applying this concept to the "Duraglit", whatever formulas you are quoting and whatever they mean to you. Quoting Baxandall is here out of context as well.
Please explain why "lesser common base designs have other noise sources so this 'optimum' isn't so clear cut". According to the common body of knowledge, intrinsic device noise sources are the same, disregarding the circuit topology.
Please explain why "reducing Duraglit's input impedance by running more current improves THD with the cartridge" and anyway, why is this true only for the "Duraglit". BTW, you just mentioned a few messages up that the concept of optimal MC cartridge loading is not important.
As you can see, your answers raised more question than they addressed, so if you feel like answering, please be concise and at the point. Unfortunately I'm not a beach bum yet, so my time of going through and filtering the chaff from your postings is limited.
Here is a table of amplifier noise and source resistance noise. It shows the combines amplifier vs source resistance noise.
Select your source resistance on the x axis and your amplifier equivalent noise resistance on the y axis. The intercept is the total input referred noise.
Select your source resistance on the x axis and your amplifier equivalent noise resistance on the y axis. The intercept is the total input referred noise.
Attachments
Last edited:
- Distortions for the circuit shown in https://www.diyaudio.com/forums/ana...tra-low-noise-mc-head-amp-38.html#post5841706 From bottom up, Vin=0.5mVeff to 2.5mVeff in 0.5mVeff steps. 0.007% for a typical MC cartridge input. Can't get better than this with bipolars, open loop, unless some heavy degeneration is used (at the price of noise). Mr. Duraglit will of course note that other than the topology, the circuit is different from the Leach mirror circuit; degenerations are 1ohm and the circuit runs at 5mA rather than 125uA. Hence, the Leach comments about a higher noise in the current mirror circuit does not apply.
- Distortion spectrum @1KHz and 2.5mVeff, fundamental notched. A little of 2nd and 3rd harmonic, everything else is in the noise floor.
Attachments
You can of course rapidly end up at the point where the cartridge itself is picking up more noise than the preamp. Shielding that becomes more challenging. As I discovered helping a friend out recently, ethernet over the mains boxes are evil and put out all sorts of grot that the cartridge will happily pick up.
100 times less that the best case distortion from the vinyl. Personally would mark that as transparent.
The MoD and his family covered it up cos Dick Duraglit was a homosexual.
Now that the 50 yr. Official Secrets Act time frame is over for that period, we might see the truth emerging.
Last edited:
All true Guru Wurcer. And I'm suitably reprimanded for using Zsource when I should have been using Re(Zsource) bla bla though for MC cartridges & Duraglit, both Zsource and Zin are essentially resistive
For noise, the formal matching condition is of course
Rsource = Roptimum = rt(Rnv * Rni) please excuse my not using Re(bla bla)
It just so happens, for common base, Roptimum = 1/gm
Yes. Gerhad's amp is just over 2dB from this optimum with an Ortofon cartridge and mine nearly a dB worse.
Last edited:
I still don't know what you are talking about, take a Denon DL103R 14 Ohms IIRC, the net RTI noise is still going down at 1/gm = 5 Ohms. Optimum means a cusp, worse below and above a certain bias point.
Last edited:
Richard, you still don't get it. There is no noise source/load optimum for audio. That's because the notion of maximum power transfer doesn't make any sense.
Your circuit, yes. Though "noise figure" in this context means the output voltage as if it was driving a matched load since noise figure is a power based figure of merit and not applicable in this case.
For this situation I got about .38dB at 12mA of collector current for 2 Ohm RB, beta of 250, and 14 Ohm source as the optimum.
The bandwidth, noise, and gain of that circuit are all functions of beta and RB to some extent. There is also peaking at low frequencies with the values everyone is using, the self biased transistor is still a feedback system.
Please post the *.ASC file you used.
Guru Wurcer, what would you use for common base Rni ?
I'm using NF in the same sense as designing a mike preamp. ie what is the increase in noise compared to a noiseless preamp for a specified Rsource. eg Noise figure - Wikipedia which makes no mention of matched load.
As a newbie, wannabe LTspice guru, I haven't got to grips with its noise sims. I need to resurrect another computer to run LTspice which I haven't really touched in the last 2 yrs. Need to get some beach bum issues out of the way.
Thanks to those of you who have identified the subsonic problems. I must update the MicBuilders document with that and some of the other stuff that has come up including syn08's Solar Cell experiments. In the meantime, I would choose the critical base electrolytics first (by ESR and listening), then twiddle the other values to tame any subsonic peak.
My somewhat lame excuse for not seeing this in 1980 is that I probably had the B&K voltmeter's 20Hz brickwall filter engaged. Mea maxima culpa
Last edited:
My apologies Guru Wurcer. Serves me right for pre10ing to reed en rite.
May I ask that we relax the matched load requirement and simply use NF as the increase in noise over rt(4kTBR) multiplied by the effective gain of the amplifier? This is simple and easy to understand for audio .. hence used in specifying eg mixers & mike preamps. If there is a formal name for this metric, please let us know.
________________________
Guru Wurcer, what would you use for common base Rni ?
Last edited:
Why inventing a square wheel instead of using the standard input referred noise?
syn08, could you post more details about your Solar Cell PSU? What did you use and where did you buy it? Same for the LEDs.
Pics of how you have cobbled them together to minimise noise from ambient light and construction details would be great.
A new & useful addition to the fount of human knowledge and of particular interest in those who wish to construct a MC head amp with SOTA noise .. but lack the skills & facilities possessed by Gerard & your good self
This would help improve S/N ratio .. which as you rightly point out .. has fallen somewhat of late.
Pics of how you have cobbled them together to minimise noise from ambient light and construction details would be great.
A new & useful addition to the fount of human knowledge and of particular interest in those who wish to construct a MC head amp with SOTA noise
.. but lack the skills & facilities possessed by Gerard & your good self This would help improve S/N ratio .. which as you rightly point out .. has fallen somewhat of late.
Last edited: