Yes. I've examined that in great detail. I was hoping for more info .. so that one of the unwashed masses could replicate its performance.
Yes - indeed. To use the table as is you would have to first manually compute the total input referred amplifier noise (Ven and IenRg).
I’ll beef the spread sheet up to show this and put a link in.
Last edited:
No you don’t, that’s it. The only detail not obvious is spectral matching, choose white LEDs that have the emission spectra as close as possible to the solar cell sensitivity. These are usually warm white LEDs.
Use one LED per solar cell diode (there are 4 diodes in a 2.7V solar cell panel) to provide an uniform illumination. Expect a power efficiency of about 10-15% (LED power/solar cell output power).
A matched load for Riaa Amps is rather pointless.
More important is to keep in mind that A-weighted noise after Riaa above 75dB-A doesn't make any sense.
For a Denon DL103R with 0.27mV@5cmsec/1KHz, this means an equivalent RTI of 48nV after Riaa and A weighting, corresponding to 119nV flat.
119nV@20Khz flat is 0.84nV/rtHz or 42R7.
Subtracting the 14R from the Denon leaves 28R7 or 0.69nV/rtHz RTI for the MC Amp.
Everything below 0.69nV/rtHz without the DL103R Cart is rather over the top.
Hans
More important is to keep in mind that A-weighted noise after Riaa above 75dB-A doesn't make any sense.
For a Denon DL103R with 0.27mV@5cmsec/1KHz, this means an equivalent RTI of 48nV after Riaa and A weighting, corresponding to 119nV flat.
119nV@20Khz flat is 0.84nV/rtHz or 42R7.
Subtracting the 14R from the Denon leaves 28R7 or 0.69nV/rtHz RTI for the MC Amp.
Everything below 0.69nV/rtHz without the DL103R Cart is rather over the top.
Hans
Attachments
Can you tell us which Solar Cells & LEDs you used ... and where you got them from please.
Humour an old beach bum who is now one of the unwashed masses. The crocs don't like you washing in Cooktown. I know you guys think I'm joking but you can confirm this with Prof Angelo Farina who visited Cooktown Recording and Ambisonic Productions in 2006 or 2008
Last edited:
Clearly indicated on the schematic in #375. DigiKey is your friend. For the rest, I have no idea what you are talking about.
Hi Richard,
Just back from Hamilton Island.
About 20 years ago my cousin managed a station call Holland’s Bay out of Cooktown. It was over a days ride along the fence boundary. Is the Postoffice still the Bar? Perhaps some member don’t appreciate this website was founded by an Australian.
On the phono l think it makes sense to start at the start.
l had read through “The Sound of Silence” on the way back from Cooktown.
My impression is that on a sunny day scenario the S/N of a vinyl pressing is about 70-72 dB and rainy day scenario 60-62 dB depending on the pressing and manufacturing process. Does that sound right?
Can you provide any insights on the above?
Just back from Hamilton Island.
About 20 years ago my cousin managed a station call Holland’s Bay out of Cooktown. It was over a days ride along the fence boundary. Is the Postoffice still the Bar? Perhaps some member don’t appreciate this website was founded by an Australian.
On the phono l think it makes sense to start at the start.
l had read through “The Sound of Silence” on the way back from Cooktown.
My impression is that on a sunny day scenario the S/N of a vinyl pressing is about 70-72 dB and rainy day scenario 60-62 dB depending on the pressing and manufacturing process. Does that sound right?
Can you provide any insights on the above?
Last edited:
My analysis of the simple complementary common-base head amp follows. There are more subtle complications than one might think including some that might not be intuitive. For this exercise the standard low frequency instrumentation concepts of referred to input and referred to output error are used.
I started with the simplest form and used idealized transistor models to avoid confounders, only the Vbe equation (Is), RB, RE, RC, and beta were used.
Tabulated bias conditions from the LTSpice log (gain from simulation);
Gain @ 1kHz – 27.7dB
gm - 0.463 S
RB – 2 Ohms
Ic – 12mA
Referred to input noise contributions;
Total output noise (RTI) - 0.525nV rt-Hz
Contribution from cartridge (14 Ohms) – 0.48nV rt-Hz
Q1 and Q2 – 0.13nV !!!
The load resistor – 0.09nV (the rest of the resistors are trivial)
The first takeaway is that lowering the Ic on the transistors to match 1/gm to the source resistance only makes the noise worse and the noise at the output is almost equal to the thermal noise of the cartridge series resistance (14 Ohms) at 1/2gm of ~1 Ohm without much room for improvement.
The question is why do the transistors contribute so little noise. At this level of Ic and input resistance the cartridge acts as a degeneration resistance so currents and voltages divide at the input into common mode and differential components. Using probes to compute the noise gain from voltage at the base of each transistor and the noise gain of the collector shot noise referred to the input current gave interesting results. The noise gain for voltage injected at the base was ½ to the output, which means the effective noise is from RB/4, while the collector shot noise is a factor of 10 lower or the shot noise of Ic/100. These numbers seemed somewhat non-intuitive but plugging in all the numbers gave almost exactly the answer from LTSpice. I think this is a hidden benefit of the floating Leach circuit that I have never seen pointed out.
@Hans I don't see the point with the RIAA or A weighting they apply to all setups assuming all noise is flat levels the playing field.
I started with the simplest form and used idealized transistor models to avoid confounders, only the Vbe equation (Is), RB, RE, RC, and beta were used.
Tabulated bias conditions from the LTSpice log (gain from simulation);
Gain @ 1kHz – 27.7dB
gm - 0.463 S
RB – 2 Ohms
Ic – 12mA
Referred to input noise contributions;
Total output noise (RTI) - 0.525nV rt-Hz
Contribution from cartridge (14 Ohms) – 0.48nV rt-Hz
Q1 and Q2 – 0.13nV !!!
The load resistor – 0.09nV (the rest of the resistors are trivial)
The first takeaway is that lowering the Ic on the transistors to match 1/gm to the source resistance only makes the noise worse and the noise at the output is almost equal to the thermal noise of the cartridge series resistance (14 Ohms) at 1/2gm of ~1 Ohm without much room for improvement.
The question is why do the transistors contribute so little noise. At this level of Ic and input resistance the cartridge acts as a degeneration resistance so currents and voltages divide at the input into common mode and differential components. Using probes to compute the noise gain from voltage at the base of each transistor and the noise gain of the collector shot noise referred to the input current gave interesting results. The noise gain for voltage injected at the base was ½ to the output, which means the effective noise is from RB/4, while the collector shot noise is a factor of 10 lower or the shot noise of Ic/100. These numbers seemed somewhat non-intuitive but plugging in all the numbers gave almost exactly the answer from LTSpice. I think this is a hidden benefit of the floating Leach circuit that I have never seen pointed out.
@Hans I don't see the point with the RIAA or A weighting they apply to all setups assuming all noise is flat levels the playing field.
Attachments
In the current mirrored circuit in #375, the funniest property I observed in this common base circuit is that the gain does depend only on the power supply voltage (For Rs=0)
Gain=(Vcc-Vbe)/Vt where Vt=kT/q so is independent on transistor beta, collector current, load resistor, etc...
that's because Gain=gm*RL but in turn gm=(Vcc-Vbe)/(RL*Vt)
Kind of the ultimate zero PSRR
.I am not surprised. Warning, napkin calculation ahead:
Transistors appear to work at some 12mA a pop, so gm=480mA/V and 1/2gm~1ohm
These 1ohms plus 2ohms RB makes 3 ohm, divided by two for the pair is 1.5ohm noise equivalent. 1.5ohm is 0.15nV/rtHz transistors contribution, which is not far from what the simulator is telling you.
What I find interesting in the common base circuit is that increasing the collector current decreases 1/2gm (so the noise), also increases the gain, proportional to gm, so the output noise should stay constant. However, increasing the collector current also decreases the input impedance, so the signal gain decreases wit 1/(1+Rs*gm) Rs is the source (cartridge) impedance. Therefore the signal gain is gm*RL/(1+gm*Rs) or ~RL/Rs also about constant. So the signal/noise ratio powers at the output is independent on the transistor bias!
Last edited:
That was 0.13nV each from the simulator still small. It is six of one half dozen of the other 2||2 is the same as 2/4 + 2/4.
Last edited:
Rechargeable batteries (if this is what you mean) are an option, but a circuit to recharge is required. NiCd cells can be conveniently recharged using a trickle current (so a simple resistor), however lithium based rechargeables do not, at least not without significantly decreasing their lifetime.
Batteries, solar cells, rechargeables, all good (make sure batteries have low noise, since the PSRR of these head amps is zero), use one depending on your tolerance to possible usage interruptions. I keep my RIAA preamp powered up 24/7, ready for action, so a solar cell approach to build a floating power supply is logical. BTW, Hawksford used in the LFD preamp a 2V acid cell
.
Last edited:
75dB-A as a sensible upper limit applies to all Riaa amps, MM or MC.
In the example I used a Denon DL103R to show that the Carts output voltage is the important parameter.
To achieve f.i. a 75dB-A S/N for a 0.27mVolt cart, Rcart + Req amp should be ca. 43 R.
This will determine the Amps RTI, matching loads does not apply.
As an intellectual challenge it's nice to see how deep you can go and what factors are playing a role, but the practical use for a low output MC Cart is limited IMO.
Hans
Richard said if he was doing it again he'd do it with solar panels instead of batteries. So for the sake of discovery it was done. And proved to be possibly less of a good idea than it first appeared!
But some people don't like batteries and I can respect that.
Edit: ignore me, I see Syn08 answered far more lucidly whilst I was typing.
Last edited:
Each, that's small. I thought you got 0.13nV/rtHz for input referred for both transistors contribution. I don't know...
Last edited:
Thanks Hans. I have an inverse RIAA that I’ll try to turn into a block as well and put up.
I had a Hana ML (400 uV output) MC for 2 weeks as the source for a show we’ve just done here. I’ve gone back to my Ortofon MM Red and I am struggling . . . I don’t know if all MC are more open than MM’s (probably one Bill is best positioned to answer) but the ML sounded wonderful with loads of detail.
1000 bucks mind you and IIRC you get about 1000 hours out of the stylus. Not cheap.