This is basically what I pointed out above as one of the potential limitations of the technique. Using this technique may result in a peak in the response where the cartridge being used has a faily low cantilever resonance frequency that is not well damped. A good MM cartridge with a mechanical resonance near 30 kHz, like the V15 V, will not have much of this limitation.
It is important to bear in mind that in the heavy-load approach, the 75us RIAA time constant is achieved right at the cartridge input interface, and the normal electrical 75us time constant in the following electronics is disabled. There are important noise tradeoffs and optimizations in this approach.
MM cartridges that rely heavily on two resonances working together to achieve a flat response are largely unreliable in achieving a consistently flat HF response without impairments to the transient response and introduction of ringing. Such cartridges are particularly sensitive to the R-C cartridge loading choice, making accurate RIAA EQ in the 15-20kHz range a rarity in the real world. The R-C loading choice then done by ear makes it a virtual very high frequency tone control.
It is notable that many MC cartridges are also limited in their HF performance by their cantilever resonance, although some are able to push it to a quite high frequency.
In comparing MM and MC cantilever resonance frequencies, for a given cantilever material, construction and length, the relative cantilever resonance frequency is largely determined by the relative mass of the MM magnet as compared to the MC coil.
Cheers,
Bob
Hi,
MVG presented IIRC around 2dB practical noise improvement using this method. I cannot remember what the theoretical limit was if everything was "just so" (perfect zero tolerance resistors, perfect Op-Amp's etc) but it was not an improvement I judged worthy, in relation to the additional effort.
On sonic grounds, as remarked, I found that using a real resistor instead of one imperfectly synthesised with assistance from an imperfect Op-Amp was much preferable.
Ciao T
IIRC, there was a previous discussion about using a feedback capacitor with an amp whose gain rolls off 6dB/oct to synthesize e.g. a 47K input impedance. The conclusion then was that there wouldn't be a noise advantage. (I think Scott had the explanation).
MVG presented IIRC around 2dB practical noise improvement using this method. I cannot remember what the theoretical limit was if everything was "just so" (perfect zero tolerance resistors, perfect Op-Amp's etc) but it was not an improvement I judged worthy, in relation to the additional effort.
On sonic grounds, as remarked, I found that using a real resistor instead of one imperfectly synthesised with assistance from an imperfect Op-Amp was much preferable.
Ciao T
He did a lot of innovative designs. I am only the messenger. It is often forgotten or ignored that a great many of the breakthroughs in analog circuit design happened in the 1960's and 1970's, especially in the USA, in fact more in the USA than other places, in that time period.
What we are discussing here was probably independently developed in E. Europe at some time, but perhaps not first.
What we are discussing here was probably independently developed in E. Europe at some time, but perhaps not first.
You are probably correct about an improvement of only about 2 dB when feedback-synthesizing the 47k cartridge load resistor in a conventional RIAA equalized phono preamp.
However, in the heavy-load scheme I described above, if you just heavy-load the cartridge with a real shunt 5k (or so), you will suffer a noise penalty as compared with conventional RIAA equalization. You can get most of this penalty back by synthesizing the 5k resistor with a larger feedback resistance of 47k (or even more). These are thus two different situations, and the use of the synthesized load resistor in the heavy load architecture is more significant noise-wise than in a conventional architecture.
Bear in mind that in the heavy-load architecture the 75us HF RIAA roll-off (attenuation) occurs BEFORE the input amplifier and there is no 75us rolloff AFTER the input amplifier.
Cheers,
Bob
I'm not sure I understan how this could reduce input referred noise. Perhaps the shunting effect will create a lower source resistance at higher frequencies.
This will also demand more HF performance since higher gain is required at higher frequencies.
Keith Johnson uses a similar scheme on the tape head preamp, but that is not quite the same problem.
The Barney Oliver preamp used the inductance of the cartridge for part of the EQ as well.
Sent from my T-Mobile G2 using Tapatalk
This will also demand more HF performance since higher gain is required at higher frequencies.
Keith Johnson uses a similar scheme on the tape head preamp, but that is not quite the same problem.
The Barney Oliver preamp used the inductance of the cartridge for part of the EQ as well.
Sent from my T-Mobile G2 using Tapatalk
Yes, and I somehow replaced Dick Burwen's input to Mark Levinson, piece for piece, for several years. Only much later, did Mark again retain Dick Burwin, (a wise choice) years after he and I separated from working with each other, forever.
Core Ideas are different from exceptional circuit designs. This is because Blumlein did not have complementary devices to work with, and for example, Peter Walker used my PATENTED JC-1 topology without paying me or giving me credit.
Core Ideas are different from exceptional circuit designs. This is because Blumlein did not have complementary devices to work with, and for example, Peter Walker used my PATENTED JC-1 topology without paying me or giving me credit.
Last edited:
For a moving magnet cartridge, isn't it the usual case that thermal noise is dominated by the 47K load resistor, only partially shunted by the cartridge because of the cartridge's LC resonance inside the passband? Although this is fairly low Q and somewhat attenuated by RIAA rolloff, it's still a big resistor to start with.
Thanks,
Chris
Thanks,
Chris
I'm probably not going to a good job describing this, but I'll give it a try. Most of this can be verified by SPICE noise simulation.
The use of a feedback-synthesized load resistance of, say, 5k does not reduce the input-referred noise of the input amplifier stage, per se. Remember, we are talking about signal-to-noise ratio at the end of the day. The noise disadvantage that we are trying to overcome results from the fact that the conventional 5k shunt reduces the size of the signal coming from the cartridge at high frequencies. Thus, for a given input-referred noise for the input amplifier (say, 3 nV/rt Hz), the S/N gets worse.
When we synthesize the load resistor with, say, a 47k feedback resistor, the input voltage AVAILABLE at the input of the input stage is effectively higher, just as with a conventional 47k shunt load. When we suppress this available signal input voltage using feedback, we do not get the same noise penalty.
Some of these assertions may depend on the model you use for the phono preamp. The model I am using is a simple flat-gain X10 input stage. An inverted version of this is fed back through the feedback resistor to load the cartridge in the heavy-load mode. In my model, in conventional mode (simple 47k shunt load at the input with no feedback for impedance synthesis), I assume a passive 75us LPF at the output of this stage. In heavy-load mode, the feedback synthesis is used to get the low effective input loading resistance and the 75us LPF at the output of the first stage is removed.
This approach can actually improve HF headroom and reduce HF distortion in the heavy-load mode because the actual voltage output of the first stage is smaller because the 75us LPF is already in there at the input node. There is no higher HF gain required in the signal path.
I know that some of this is hard to get one's head around, and this is where SPICE simulation is useful to gain confidence in the understanding of the actual workings.
Also, YMMV with different assumed preamp/EQ architectures.
Cheers,
Bob
Chris, if you can buy/beg/steal/borrow a copy of National Semiconductor's Audio Radio Handbook (1980), they do a terrific treatment of that noise calculation, taking into account cartridge DCR, L, and preamp input R and C. In 1980, it was a by-hand method; we can do it more easily now. When I get a chance, I'll put it into spreadsheet form and post it.
Thanks! Got one coming from Powells. My take-away on the MM phono thermal noise question is that it's easily possible to make a 200 ohm equivalent amplifier + feedback resistor circuit, and that we should maybe define phono amplifier noise in "noise figure" above an agreed/guesstimated average cartridge-plus-load's noise source, maybe about 0.5H and 500Ohms, with 200pF and 47KOhms (or something like that).
When we see that our amplifiers' noise figures are less than 1dB we can move on.
Thanks,
Chris
When we see that our amplifiers' noise figures are less than 1dB we can move on.
Thanks,
Chris
Last edited:
Thank you, John. I believe it is not bad, and I like the sound. I have a direct comparison with OPA627 + buffer based preamp (my older design), and I believe you can imagine which one I prefer. Anyway, I have learned a lot in discussions with you here, about circuit topologies, and without that I probably would not design a discrete phono preamp.
- Status
- Not open for further replies.