Richard Lee's Ultra low Noise MC Head Amp

I think you are missing the fact that the transformer AND the amplifier input Z matter?
Actually for noise, its Ropt = rt(Rni x Rnv) for the 'OPA' that matters. You can set amplifier input Z with resistors which aren't as EVIL as usual cos you are using transformers. :D

I think you are mixing up relative vs absolute noise here?
Lepaisant et al achieved 3dB NF. I managed 1.7dB so there must have been Neve designers who managed that too along with several other Calrec designers.

These would be people who could design & specify LN transformers for best performance with their 'OPAs'. They would not find 0R5 source too far-fetched though Old Man Sowter would probably say rude things when presented with our demands.

3dB - 1.7dB = 1.3dB .. so slightly more than 1dB improvement over Lepaisant et al. I'm assuming da Neve wallahs can equal but not beat me :)

Been done. Winner is Grado if low noise is your goal.
I'm sure it is as it has the least inductance of any MM/MI cartridge.

But how do you get similar noise performance from the high inductance ones? You'd have to break even more Golden Pinnae myths ... probably incorporating 'active loading' ... And to correct the ensuing wonky responses, matched digital EQ/RIAA from the thread I haven't visited in 2 yrs :eek:

The way you have described I still don't know if its power or voltage.
Does it matter?

If you mean .. 'Is it 10xlog or 20xlog?' .. as you are measuring voltages, it's 20xlog
 
Last edited:
www.hifisonix.com
Joined 2003
Paid Member
I would like to see some actual noise measurements in modern units. The BBC 200 Ohm example doesn't wash against a measurement with a 0.5Ohm source.

But why not?

If the source resistance is specified, to me it seems quite clear.

I much prefer a total noise (that is to say total current noise and voltage noise) spec in nV/rt Hz, but can see the merit in the NF approach.
 
Member
Joined 2014
Paid Member
Because to actually make a valid comparison you have to calculate the noise of both sources (0.5 Ohm and 200Ohm) and then work out the noise of the preamp from the NF to be able to compare. 200Ohms all else being equal has 20 times the RMS noise of 0.5Ohms so a 1.7dB NF suddently doesn't seem that impressive.
 
And if source resistance is defined (as is the case with the BBC spec: 200R) you can easily calculate e.g. a 3dB NF spec into a 1.8nV/rtHz input noise spec of the amp.
And of cause, a 3dB noise spec for a .5R source is quite a different animal as this translates into 90pV/rtHz input ref. amp noise
 
Because to actually make a valid comparison you have to calculate the noise of both sources (0.5 Ohm and 200Ohm) and then work out the noise of the preamp from the NF to be able to compare. 200Ohms all else being equal has 20 times the RMS noise of 0.5Ohms so a 1.7dB NF suddently doesn't seem that impressive.


As said, dB and noise figures are no absolute values. 1.7dB noise figure at 200 Ohms is not impressive at all (but it was 40 years ago) while for .5R it is
 
These would be people who could design & specify LN transformers for best performance with their 'OPAs'. They would not find 0R5 source too far-fetched though Old Man Sowter would probably say rude things when presented with our demands.

Richard you keep missing the point, there is no need for match termination in audio. A transformer is perfectly good at stepping up voltage alone i.e. my microphone has a 2000:200 Ohm output transformer but I don't terminate it in 200 Ohms at the other end or I get 1/2 the output voltage and the pre-amp "sees" 100 Ohms.

A MC step up transformer works fine with a wide range of cartridge resistances, in fact show me a catalog with many transformers paired with particular cartridges. Look at the graph in the article again (low frequency anomaly aside) they get 0.06nV for a wide range of source resistances.
 
Sigh ... its a Transformer example. I assume (like GG Baxandall) anyone designing a transformer i/p LN amp is versed in the art of designing/specifying/making an appropriate transformer to match any source .. whether 0R1, 10R or 200R

On the MC front, if I claim a certain Duraglit variant has 3dB NF with Ortofon MC20, I'm also claiming you can't better this performance by more than 3dB ... regardless of complexity, number of bits or zillion mA

On the transformer i/p LN design side, a closer look at Lepaisant et al shows they have 8 & 10T windings to match 0R5. This means their core isn't as good as the mu-metal ones I'm used to and favoured by Lundahl, Sowter & Jensen. This is certainly one reason why they only achieve 3dB NF

Don't get me wrong. I think its a good, if over complex, article .. but their implementation can be bettered by any number of old fogey Broadcast Mixing Desk Preamp designers .. (but only by slightly more than 1dB)

Marik uses 'ferrite' cores in his ribbon mikes so the ones he uses must outperform the usual mu-metal ones. I just wish I knew what he used. :( But I'm certain it's not Siemens N48

Ok, now I’m even more confused. I have no idea what “matching” you are talking about.
 
From a manufacturer's mouth...

THE IMPEDANCE MYTH
A common misconception is that the impedance of the preamp input should match the same impedance as the microphone. This idea of “matching” was born out of early telecom technology when matching the mic’s impedance to the preamp impedance was necessary to get the highest power transfer. This is no longer true.

The rule today is in order to deliver its full frequency response, a passive ribbon mic needs to see an input impedance of at least five times its output impedance. That means if the mic has an output impedance of 300 ohms, the preamp should have at least an input impedance of 1200 to 1500 ohms. We recommend a preamp with an input impedance of at least ten times the output impedance of the mic to ensure optimum performance.

"Matching" can only relate to the fact that the primary side has to have a series resistive loss less than the resistance of the transducers.
 
Last edited:
I got some private requests about the Gerbers for #375. Attached is the latest version for Zetex devices pinout. It's a mono board so one needs two boards for stereo. White LEDs are the much cheaper DigiKey 1510-1552-1-ND devices.

No further development is expected with #375. Use the attached for your own MC pre development.
 

Attachments

  • MCPRE-14.zip
    30.8 KB · Views: 63
  • schematic.jpg
    schematic.jpg
    89.3 KB · Views: 260
The latest thing from the East is 6000mAH USB battery packs with one side covered with solar cells (I assume they aren't fake like the "solar" calculators). A quick look found some of the cells going for as little $0.66.

I've experimented a little with all kind of solar cells; large area cells have a significant issue for this particular project: they need a more or less uniform illumination to deliver the required current. This means either lots of LEDs to cover the surface, or placing the cell far enough from the LEDs, otherwise the short circuit current is a dismal (and strongly relative position dependent, needing adjustment). Also, most of the available cells don't have an antireflective layer that would "trap" the light inside.

I think these IXYS cells covers it pretty well: small area (also saves PCB area) so they need only four LEDS, one for each quadrand, antireflective layer, 2.8V open voltage, >50mA short current. They are also not expensive. Unfortunately they don't seem to manufacture any smaller model with the same voltage but lower short circuit current (20mA would be enough).
 
Last edited:
www.hifisonix.com
Joined 2003
Paid Member
I would like to see some actual noise measurements in modern units. The BBC 200 Ohm example doesn't wash against a measurement with a 0.5Ohm source.

The point is the source impedance is specified - and in most cases where noise is spec'd as a NF, the source conditions are also specified.


But, of course you can always back calculate the NF to actual total equivalent input noise given the generator resistance.


Anyway, I prefer noise to be spec'd in V/rt Hz so this interaction is really a moot point.
 
I got some private requests about the Gerbers for #375. Attached is the latest version for Zetex devices pinout. It's a mono board so one needs two boards for stereo. White LEDs are the much cheaper DigiKey 1510-1552-1-ND devices.

No further development is expected with #375. Use the attached for your own MC pre development.

Sorry, the schematic shows 9 LEDs; the attached board requires only 4 LEDs of the indicated type. The schematic is for a version fed by a 5V solar cell, that project was abandoned before creating a PCB.
 
I've experimented a little with all kind of solar cells; large area cells have a significant issue for this particular project: they need a more or less uniform illumination to deliver the required current. This means either lots of LEDs to cover the surface, or placing the cell far enough from the LEDs, otherwise the short circuit current is a dismal (and strongly relative position dependent, needing adjustment). Also, most of the available cells don't have an antireflective layer that would "trap" the light inside.

I think these IXYS cells covers it pretty well: small area (also saves PCB area) so they need only four LEDS, one for each quadrand, antireflective layer, 2.8V open voltage, >50mA short current. They are also not expensive. Unfortunately they don't seem to manufacture any smaller model with the same voltage but lower short circuit current (20mA would be enough).


I think the IXYS cells are already belonging to the most optimal group of cells. Beside the pros you already mentioned, the fact that they are also monocrystaline may also have a positive effect for noise. The majority of cells today are of the poly type (cheaper to produce for large areas) and I can imagine that noisewise these are not as good.
 
I think the IXYS cells are already belonging to the most optimal group of cells. Beside the pros you already mentioned, the fact that they are also monocrystaline may also have a positive effect for noise. The majority of cells today are of the poly type (cheaper to produce for large areas) and I can imagine that noisewise these are not as good.

I don't know much about solar cells, other than the basic (physics) plot, therefore the first attempt for illumination was with four power white LEDs @600mA. The IXYS solar cell wideband output noise was horrific, I suppose of the GR type. @50-100mA through the same LEDs, with the same cell, they are completely silent. Hence I switched from high power LEDs to medium power LEDs @75mA (50mA also works fine) and didn't look back since.