Moving Coil preamp design

[paul joyce]

Hi DIY fans,

I am a newbie on this forum (but not to DIY audio) and have been looking through various threads and links on moving coil preamps, and I have not really found one yet that I like.
I have a low output Ortofon MC20 MkII - around 90uV @ 5cm/s.
Needs around 32dB of gain (like the Ortofon transformers or ancient MCA76 preamp)
From an engineering perspective, I would think the following criteria are important:

1) Minimal DC current in the cartridge coils. If we think of the normal output of 90uV into 10 ohm (the minimum load specified by Ortofon) this equates to approx 13uA peak. How much DC can the cartridge tolerate, maybe 1,000th of this , say 13nA? This rules out any single supply circuit with a large electrolytic cap on its input (leakage easily larger than the signal!)

2) No shunt resistors across the input. With the tiny signal current, why would we want to waste it in a resistor? This is the premise of the phono clone, itself an old idea used, I believe, 30 years ago in the Lentek moving coil preamp. Requirement in 1) above would need care to meet though. The virtual ground is also very sensitive to RFI and noise (as found by Phonoclone builders)

3) Transformers seem attractive, in that they avoid 1) and 2) above, but you have little control of the input impedance, since it is a reflection of the following stage.

4) Equivalent Input noise for a reasonable 60dB down on normal listening level needs to be 90nV or less (I hate hiss). This equates to a device noise of approx 0.7nV/rtHz. OP27 is not going to be good enough (Phonoclone). Seems to favor the multiple parallel fets used in the Pass phono amp and others.

5) No input capacitor. With signals this tiny, I would expect capacitor imperfections could dwarf low level detail.

I appreciate any feedback (no pun intended) on these ramblings.

Paul J

[AKSA]

Hi Paul,

You raise pivotal issues.

1) Minimal DC current in the cartridge coils. If we think of the normal output of 90uV into 10 ohm (the minimum load specified by Ortofon) this equates to approx 13uA peak. How much DC can the cartridge tolerate, maybe 1,000th of this , say 13nA? This rules out any single supply circuit with a large electrolytic cap on its input (leakage easily larger than the signal!)

This output is the lowest I've seen; not even 0.1mV. Further, consider the optimum active device for this job; a JFET. Bias currents are around 3000 times less than a bipolar, so we might assume around 300K times less than the stage current. If this is, say, 10mA, we are talking of 0.03uA, which is around 400 times less than the cartridge output. It is unlikely a bias current of 30nA would appreciably disturb the cartridge output integrity, and in fact it would be dwarfed by the second harmonic created by asymmetric mechanical action.

2) No shunt resistors across the input. With the tiny signal current, why would we want to waste it in a resistor? This is the premise of the phono clone, itself an old idea used, I believe, 30 years ago in the Lentek moving coil preamp. Requirement in 1) above would need care to meet though. The virtual ground is also very sensitive to RFI and noise (as found by Phonoclone builders)

Note that Ortofon specify a 10R load; they do this presumably for a reason, and while it may sound different, even better, to use a higher value than 10R, I don't believe it's best to have no loading at all. I'd be tempted to offer a choice; 10R, 100R, 500R and 1K would seem a reasonable range.

3) Transformers seem attractive, in that they avoid 1) and 2) above, but you have little control of the input impedance, since it is a reflection of the following stage.

Transformers are very expensive, and a potential source of hum intrusion. I've heard it said by many they could never achieve high resolution with trafos, but I cannot say I necessarily agree. The chief criticism would appear to be cost.

4) Equivalent Input noise for a reasonable 60dB down on normal listening level needs to be 90nV or less (I hate hiss). This equates to a device noise of approx 0.7nV/rtHz. OP27 is not going to be good enough (Phonoclone). Seems to favor the multiple parallel fets used in the Pass phono amp and others.

Yes, I agree conclusively. However, much of the noise comes from the Vcc of the active device; normally, the load for this device is referred to the rail, and that could be a major point of noise intrusion. If the load is referred directly to earth, and the power supply completely decoupled from the load, this might be reduced.

5) No input capacitor. With signals this tiny, I would expect capacitor imperfections could dwarf low level detail.

Yes, you need several hundred uF, and this means either very large film caps - with resulting noise intrusion through large antenna surfaces - or electrolytics, with all the implications of poor resolution you mention. If you use a JFET, there are some nA bias issues, but by and large direct coupling to the gate can be effectively used.

Cheers,

Hugh

[phn]

Transformers are very expensive, and a potential source of hum intrusion. I've heard it said by many they could never achieve high resolution with trafos, but I cannot say I necessarily agree. The chief criticism would appear to be cost.

Hum can be a problem, but can always be solved.

Transformers are used in high-bandwidth apps like routers. Audio is an ultra low-bandwidth world. The DVD-A is only "hi-rez" compared to the CD. The DVD-A is absolutely nothing in the real world.

Virtually every European record cut between 1966 and 1975 was cut on the Tab U 73 limiter. None of those albums can have higher resolution than those Telefunken input balancing trafos, albeit the best step-up transformers ever made.

[Onvinyl]

Hi Paul, Hugh and phn,
I’m in the process of improving the phonoclone, so I’m asking the same questions for some time.
First, 2)
I’m taking the following viewpoint, knowing only few sharing it:
The cart produces current, that is changed into voltage over the ‘shunt’ resistor (yes, it’s a simplification, you have to take into account the source resistance and all those parasitic L,C,R of the whole assembly. Has anybody a model of the DL103 with all those parameters for me? ;-)
So you indeed loose some of the real stuff over that resistor, no wonder it should have the highest quality available/affordable.
Using an inverting topology as an I/V circumvents that (at last in my theory…)
I exchanged the op27 with an AD811 (cf-opamp, consequently) which gives you much more resolution but leads us to
1)
Apart from that resolution, it sounded quite dull and damped. You guess it, my cheap DVM read 0.2mV input offset. A quick check with a huge foilcap, and everything clears up. If you can cope with the offset in another way, you make a big step forward. So far, I’m fiddling with a pot, but: how much I and V are tolerable both soundwise and cartridge-lifetimewise? I don’t know…
Noise and RFI: yes, I’m playing with layout since days because of that…
Maybe one should go balanced in a later stage, having two inputs here…
3) Transformers? I’m biased, because the best (to me) phonostages I’ve heard until today didn’t have those…
4) So far, noise is not the biggest issue: it’s for more quiet if you disconnect the cartridge (I have unshielded wire, which worked well with a balanced circuit) I would tolerate this if anything else were perfect. But, ugh, it gets in the way with those tiny currents the cart delivers…
5) The big Jensen input trannies might be cheaper than the huge high quality caps one would need, I guess.
Rüdiger

[tschrama]

For a THD-analyzer project, I build a 0.350nV/srqtHz preamplifier from 4 parrallel JFET, powered by 9Volt battery. I think gain was about 30dB.. I would be good solution for your problem I guess.

By lowering the drain-to-source voltage on the JFETs, you can lower the tiny input current to a minimum, and if that's not enough, use a cap in series for the input to lower it even further.

Anyway, I would think that JFETs are the way to go.


regards,
Thijs

[jcx]

moving coils are not current sources

with any reasonable permanent magnet& practical pole structure the mag field in the air gap is virtually unaffected by coil current short of meltdown - so the better model is a Voltage generator (Faraday's Law anyone?) in series with a really miniscule R and L

in general phono carts are electromechanical transformers and electrical load is reflected as mechanical damping - a reason to use specified load R

but they are probably inefficient electromechanical transformers so the load R may not be of mechanical significance, but the electrical damping of LC resonance is important too- so still use the recommended Rload

since mc coils are full of nice magnetically linear air, not iron, they are much less sensitive to bias current than mm/var reluctance types - I think you are being pointlessly obsessive by focusing on sub-uA bias current for mc amplifiers

Low noise Bipolar op amps like AD797, LT1028 are the easiest approach but discrete single ended front ends achieve the lowest noise in active amplifiers - bjt LM394, MAT-02 or very large area jfets ( http://www.linearsystems.com/index.html )

parametric amps are reputed to have the lowest possible noise but I haven't seen a phono pre - or any modern implementation for that matter

[Onvinyl]

Hi jcx,
what I don't get in your argument is:
If a voltage is present, then a current has to be present, otherwise there couldn't be a form of electrical energy (I'm no engineer, did you guess that? )
So, wether one deals with current or voltage depends on the device the measures that 'energy'. My basic understanding is, that a current-to-voltage-amplifier does sense the 'current portion' of that energy.
But maybe what you say is, that this approach is not optimal, due to the lack of a load resistor?
My final goal then is not to claim the chicken/egg-thing solved, but to 1) understand why the phonoclon topo works (so good), and, in order to improve further, 2) to understand why. One clue might be the absence of that initial 'load' resistor, which was a clue of Paul's initial post as well and made me post my thoughts.
At the end of the day, one feels better should practical observations and theorie fall together, so I'm open to learn and eager to try further experiments...
Rüdiger

[jcx]

you can use the Norton equivalent to calculate with, in a virtual gnd input amp this can be handy

but the physics of the mc transducer are preferably described in terms of a voltage source from the E field caused by Faraday's dB/dt * A, alternatively the Lorentz? B x qv law [edit: i'll get the name right eventually] gives the force on the electrons in the moving coil, the integral over the length of the coil gives the work done moving a electron from one end to the other which is again a V - Stokes theorem show the equivalence of the 2 views but both give a V source in series with the coil - not a current

this can be seen with a "thought experiment": same coil geometry, mag feild, velocity but different resistance wire in the coil - the open circuit Vout is unchanged

If you adopt a Norton equivalent explanation you have to hypothesize that the current generator changes with wire resistance in order to explain the constant open circuit Vout at a given constant dB/dt ~= coil velocity

we can guess that the coil is a very linear low value resistor at audio frequencies so the short circuit current is a good measurement of the induced V - as long as the short circuit doesn't upset the electromechanical damping

a way 2nd order effect is that the extra current in a shorted coil does increase the mag field generated by the coil and could be expected to cause more interaction with any magnetic circuit nonlinearity – which should be vanishingly small with the large air gap of a mc cart (eddy current "drag" from coil field induced currents in pole pieces?)

[Flodstroem]

Hi paul joyce
Could one of these circuits be a winner???
Marshall Leach three circuits for head-amps

http://users.ece.gatech.edu/~mleach/headamp/

Personally I have tried and uses this circuit with great success: The Common-Base Circuit
It is extremely silent (with 2N4401+4403) and there is no interference's due to switching the battery on and of. I uses Tantalum caps for this circuits (100µF/10V).

Regards

[MichaelK]

I have also used the Common-base amplifier for many years with a Dynavector MC pickup. Mine was powered by a single AA cell for each channel. No hum and very pleasing sound