Low-output vs. high-output moving coil cartridges - is the former really better?

[Michael Fidler]

EDIT - Not sure if this belongs in 'general audio discussion' or '...analog audio'. Admins please move if it's in the wrong place!

Hello all!

I've been thinking for quite some time about whether or not there is any real advantage to low-output moving coils with nominal outputs in the region of 0.5mV compared to high-output types of 2mV or more, intended for moving-magnet inputs. The conventional audiophile wisdom regarding the latter type seems to be that it is an inferior system, making a compromise of more moving mass attached to the cantilever due to the higher number of turns required to produce a more substantial output voltage. I have some considerable scepticism towards this point of view, as for a start cartridge manufacturers who make models that feature low and high output versions specify identical frequency response and channel matching figures in the specification, which would not be the case if the moving mass (and hence tracking ability) was degraded by the addition of the extra windings. They're so very close to the pivot point/cantilever suspension that their momental inertia seems likely to be insignificant.

There do not appear to be any controlled measurements (using the same turntable and a preamplifier with at least similar frequency response/noise/distortion characteristics), or low vs. high output MC cartridges in the same model range. It would be most interesting to see if there is any difference beyond what could be expected in terms of device-to-device manufacturing tolerances, although perhaps a large sample size might be necessary to make a conclusively finding. This would probably be quite an expensive undertaking if the cartridges were not already in the possession of the tester .

From a purely electronic perspective, the thermal noise generated by the increased coil resistance of high-output types actually gives a considerable advantage over the low-outputs when juxtaposed against their nominal output levels at 5mV/s. For instance, let's compare a few examples in the wild. As the coil resistance dominates the impedance of the cartridge in the audio band by at least an order of magnitude, a simple back-of-a-napkin resistive analysis will do:

Cartridge	Nominal output (mV)	Coil resistance (Ohms)	Thermal noise (nV) 20Hz-22kHz	Flat SNR (dB)
Hana ML - low	0.4	8	51	77.9
Hana MH - high	2	130	205	79.8
Dynavector DV20-X2L - low	0.3	5	40	77.5
Dynavector DV20-X2H - high	2.8	150	220	82.1
Dynavector DV10X5 Mk2 LOW	0.5	32	102	73.8
Dynavector DV10X5 Mk2 - high	2.8	150	220	82.1

If we apply RIAA equalisation to these figures, with a measurement bandwidth of 220Hz-22kHz (useful in real-world applications as it precludes flicker noise which is far less audible around 50Hz), we will see a further reduction from the flat figure of about 7.6dB thanks to the HF rolloff - meaning that the Dynavector cartridges are pushing a potential SNR of almost 90dB straight out of the gate. This is further improved by the fact that it's far easier to produce a front end that comes within 6dB of the theoretical maximum for a source resistance of 150 ohms than one of 5 ohms; the job can be done with a single IC such as the OPA1612, without the need for parallel discrete transistors and hybrid amplifier topologies.

Making a decent front end using discrete transistors for a very low source impedance can be quite a tricky business, as it's far more prone to detecting RFI than a simple op-amp input, not only due to the much higher gain, but also because the overall level of frequency compensation is increased to include all the extra devices in the loop which reduces the bootstrapping effect on the PN junction on the input where RF starts to rear up. This can lead to problems in setups where there are high levels of RFI - dodgy SMPS modules driving LED lighting, either operating without any EMC components or simply generating spurious transients as they approach failure, or ethernet-over-power can cause unwanted interjections. The conventional approach to dealing with this is to just use an IC input and maybe a stopper resistor, resulting in an SNR below 70dB after RIAA. The market has low expectations 'because the gain is so high', so manufacturers have been getting away with this for ages, but it's very poor engineering.

The setup itself is also far more prone to tiny currents here and there causing big problems with low-output cartridges, especially in relation to the floating ground drain/return of the cartridge. Setup hum and buzz might be below the noise floor for an output some 18dB higher. Making a low-noise input exacerbates the problem further: if the user couldn't hear these issues before, or simply compared them to the background hiss, the introduction of a phono-stage with a 10dB noise improvement might make these otherwise hidden problems noticeable which is not fun from a customer service perspective. For these reasons, I'm taking a little hiatus from making products for low-output MC (you can even find the schematic for the MC PRO attached to this post as a result).

If an OPA1612 with a 30 ohm series-feedback resistor (total noise 1.3nV/sqrtHz) was deployed as a front end for the Dynavector high-output cartridges, we would see a super post-RIAA SNR of 87.44dB, far better than the theoretical maximum of most low-output cartridges based on coil resistance alone. Even if we feed this into a moving magnet-optimised input based on an NE5534, we still get over 80dB. I'm fighting with everyone who contacts me now to go down this route if they've not already acquired a low-output MC!

Taking a reckless plunge into the murky depths of speculation, we could consider why the low-output types are lauded to be superior in the realm of audiophilia. Here are a few of my thoughts in this regard:

• High-output cartridges with a greater number of hand-windings are going to be more difficult to produce, incentivising manufacturers to promote the low-output type
• High-output cartridges are seen as a compromise, with some sort of trade-off for the increased level/ use with MM inputs; a good story about more coils and more moving mass
• Low-output cartridges are more specialist, requiring a dedicated pre-amplifier or step-up transformer, hence a more 'serious business' approach
• Low-output cartridges may be easier to produce with greater production tolerances, a big plus point for the exotic/high-end manufacturer in a shed somewhere
• The requirement for a step-up transformer allows the manufacturer to make an easy killing for a special sale, especially for users of valve equipment that can't impedance match the noise characteristics with a direct connection, hence more scope for promotion

So, all we really need now are some measurements in a controlled setup. Does anyone here have low and high output versions of the same model? If not, I'd better start saving up... It would be great to be able to have some proper hard evidence to present on this one.

Let me know what you think anyway - is there something that I might be missing here?

 

[Soniclife]

I've wondered if very low output MC are preferred because they amplify the reverb you get with vinyl playback if the TT and speakers are in the same room. Vinyl is riddled with defects, anyone still using it hopefully enjoys the effect, so logically more defects are better.

Happy very low output MC user.

 

[Soniclife]

PS Whilst trying to educate the market is great, trying to control the market isn't going to work, I say keep making the full range of products. Vinyl usage isn't driven by logic and facts, so it won't respond to them.

 

[Michael Fidler]

PS Whilst trying to educate the market is great, trying to control the market isn't going to work, I say keep making the full range of products. Vinyl usage isn't driven by logic and facts, so it won't respond to them.

I'm not really interested in controlling the market, but seeing as things are going well enough with MM devices, I'd really rather take a break from LOMC, or at least have a mind to include LOMC functionality as a side-feature of an MM product. These are troublesome devices, so it's always better to avoid them from the supplier side if you can afford to.

 

[genesisaudiorack]

i personally like mc low output. the only time i liked mm higher output was for lower mastered records at first. then i tried low output mc, and i liked it more.

 

[SIY]

The lowest effective mass cartridge I'm aware of was a MM. For what that's worth.

The biggest advantage of LOMC that I can see is the insensitivity to loading. This is the Achilles heel of a lot of MMs and many HOMCs- I had to design a special preamp to accommodate the finicky A-T 150MLX because of its unhappiness with even moderate levels of load capacitance.

 

[Michael Fidler]

The lowest effective mass cartridge I'm aware of was a MM. For what that's worth.

The biggest advantage of LOMC that I can see is the insensitivity to loading. This is the Achilles heel of a lot of MMs and many HOMCs- I had to design a special preamp to accommodate the finicky A-T 150MLX because of its unhappiness with even moderate levels of load capacitance.

Is HOMC really sensitive to loading (in the audio band), though? Especially when connected to a high-impedance MM load and a reasonably fast/linear input stage?

LOMC has RLC resonance in the MHz region, though, a bit prone to RFI to say the least!

 

[SIY]

Is HOMC really sensitive to loading (in the audio band), though?

More so than a LOMC with significantly lower inductance.

 

[Michael Fidler]

More so than a LOMC with significantly lower inductance.

OK, but the load resistance is typically 400x higher - is it a problem at audio frequency?

A peak in the response at 200kHz isn't amazing, but any sensible high-impedance IC input (such as a 5534) will handle it without detection effects.

Far better a peak at 200kHz with a fast IC (GWB of 1MHz), than at 1MHz with a discrete/compensated transistor input. The wiring to the cartridge is a more effective antenna as the resonant frequency increases + the extra gain makes the tiny amount of detection required far more noticeable (especially if you have a low voltage noise that doesn't mask it!).

 

[AaronJ]

I'm not really interested in controlling the market, but seeing as things are going well enough with MM devices, I'd really rather take a break from LOMC, or at least have a mind to include LOMC functionality as a side-feature of an MM product. These are troublesome devices, so it's always better to avoid them from the supplier side if you can afford to.

I would personally like to see more phono stages utilizing built-in SUT's like Luxman E-250 and Bryston BP1.5.

 

[Michael Fidler]

I would personally like to see more phono stages utilizing built-in SUT's like Luxman E-250 and Bryston BP1.5.

Not a big fan of these, for the following reasons:

• Winding resistance generates thermal voltage noise - not noise-free as so often advocated! EDIT - shown to be negligible
• Frequency response at either end of the audio spectrum frays a bit/dependent on loading, due to resonance at the top and LF roll-off at the bottom
• Being magnetic devices, they have to be screened with exceptional care/mu metal and kept away from magnetic fields, as the gain of the following stage is going to be about 60dB at the mains frequency
• Good ones are very expensive compared to solid-state active front ends + quite bulky, precluding a compact design

Giving the devil his due, though:

• Frequency response limitations/internal shielding prevent RFI at the LOMC cartridge inductive resonance point from making it through
• Multiple taps allow for easy gain adjustment/transforming the current/voltage noise characteristics of an MM front end to any MC impedance
• Galvanic isolation prevents turntable mis-wiring/shorts between cartridge and turntable ground from causing issues with hum through the cartridge return path
• Can work as either a balanced or single-ended front end for MC without the 3dB noise penalty of solid state - inherently flexible
• Power efficient, highly robust to overvoltage/static zaps
• Highly marketable - with great expensive comes great margins!

Hmm... Maybe not so bad, but not really my style.

 

[SIY]

OK, but the load resistance is typically 400x higher - is it a problem at audio frequency?

A peak in the response at 200kHz isn't amazing, but any sensible high-impedance IC input (such as a 5534) will handle it without detection effects.

Far better a peak at 200kHz with a fast IC (GWB of 1MHz), than at 1MHz with a discrete/compensated transistor input. The wiring to the cartridge is a more effective antenna as the resonant frequency increases + the extra gain makes the tiny amount of detection required far more noticeable (especially if you have a low voltage noise that doesn't mask it!).

I was trying to find the inductance of a Blue Point (since it's extremely popular) and failed. That's the key issue wrt loading sensitivity. Would you happen to know what it is?

Winding resistance generates thermal voltage noise - not noise-free as so often advocated! Can be beaten with discrete transistors

The Sowter I use in my MC preamp (balanced input) has a primary DCR of 0R3. Secondary is 60R, which is equivalent to 0R6 referred to the primary. I think it's safe to say that this is pretty negligible as far as thermal noise. And this is not a particularly expensive transformer ($125 per channel). I think the equivalent from Cinemag is even cheaper.

Frequency response at either end of the audio spectrum frays a bit/dependent on loading, due to resonance at the top and LF roll-off at the bottom

This is indeed often neglected, especially by people using outboard transformers. Not just the secondary loading, but also the variability with cartridge source resistance. The fix is easy- a series RC across the secondary. But of course, a lot of external step-up peddlers don't do that.

 

[Michael Fidler]

I was trying to find the inductance of a Blue Point (since it's extremely popular) and failed. That's the key issue wrt loading sensitivity. Would you happen to know what it is?

The Sowter I use in my MC preamp (balanced input) has a primary DCR of 0R3. Secondary is 60R, which is equivalent to 0R6 referred to the primary. I think it's safe to say that this is pretty negligible as far as thermal noise. And this is not a particularly expensive transformer ($125 per channel). I think the equivalent from Cinemag is even cheaper.

This is indeed often neglected, especially by people using outboard transformers. Not just the secondary loading, but also the variability with cartridge source resistance. The fix is easy- a series RC across the secondary. But of course, a lot of external step-up peddlers don't do that.

We can guess the inductance if we assume that the coil is wound on the same diameter former. We just need to square the voltage ratio. Voltage output is directly proportional to the number of turns, while inductance is proportional to the square of turns.

If the coil resistance increases in a manner that is directly proportional to the increase in output voltage from low to high output types, then we can very safely assume that the same wire gauge is being used for the coil winding. The voltage-ratio-squared approach is going to hold up very safely here. In this case, the voltage output increases directly with the number of turns, but the resistive thermal noise is only increasing by the square root of the turns number - another plus point for HOMC!

So let's say our HOMC cartridge has 8 times the output voltage of the LOMC version, we would see an increase in inductance of 64 times. Is this a problem? Most likely not, as we are going to connect it to a load resistance some 400 times higher than a typical LOMC load (120 ohms vs. 47k ohms).

OK - looks like I'm wrong on the winding resistance front, I must have been looking at the wrong figures. It does occur, but will be evidently be quite negligible! $250 for stereo is rather pricey if you live in a 2nd-world country (like the UK).

 

[SIY]

Most likely not, as we are going to connect it to a load resistance some 400 times higher than a typical LOMC load (120 ohms vs. 47k ohms).

The issue is the resonance with the input capacitance (including cabling)- the high load DCR only means that the Q of the resonance is even higher than with a low load. So it would be nice to have some actual inductance numbers...

As an example, here's a sim of what happens with the aforementioned 150MLX with the combined capacitance of cabling and a typical MM stage using conventional circuitry having a moderate Miller effect (from my article in the late, great Linear Audio). I'd want to run a similar calculation on some HOMC if only I knew what their impedances were...

 

[Michael Fidler]

The issue is the resonance with the input capacitance (including cabling)- the high load DCR only means that the Q of the resonance is even higher than with a low load. So it would be nice to have some actual inductance numbers...

As an example, here's a sim of what happens with the aforementioned 150MLX with the combined capacitance of cabling and a typical MM stage using conventional circuitry having a moderate Miller effect (from my article in the late, great Linear Audio). I'd want to run a similar calculation on some HOMC if only I knew what their impedances were...

View attachment 413611

Yes, but that's an MM cartridge? Pretty much par for the course with loading as high as 500pF!

If I recall correctly, the AT cartridges tend to have a cartridge inductance around the 400mH mark?

So let's do some guesstimating: It seems to be the case that most LOMCs with outputs around 0.4mV have a coil inductance of 25uH.

If we consider that most HOMCs hover around the 2.5mV mark, we can assume that the inductance is going to be:

25uH*(2.5/0.4)^2 = 977uH

So some 400 times less than the typical coil inductance of an MM cart! (Luv me 400s in this thread)

Where our 400mH MM cartridge would have a resonance point of 18kHz when coupled to a reasonable 200pF load capacitance, our HOMC at 1mH would give us 355kHz.

This would seem to be a sweet spot (as long as our RIAA response continues to fall at ultrasonic frequencies), not so high a resonance point as to cause RF detection with a sensible IC input circuit or allow the antenna effect in the connecting leads to effectively absorb RFI, but high enough to be totally irrelevant at 20kHz.

Should I fire up the simulator, boys?

 

[Michael Fidler]

That is a rather spicy peak - +26dB! It doesn't seem to cause much bother in practice. The cartridge isn't capable of substantial enough output above 50kHz. Frequency is low enough not to provoke detection in a competent IC front end IMHO.

Less than 30mdB deviation at 20kHz, no issues in the audio band.

 

[cgallery]

The biggest advantage of LOMC that I can see is the insensitivity to loading. This is the Achilles heel of a lot of MMs and many HOMCs- I had to design a special preamp to accommodate the finicky A-T 150MLX because of its unhappiness with even moderate levels of load capacitance.

OTOH, one of the greatest advantages of MM cartridges is that one can shape their response via loading.

 

[LTig]

About the improved SNR of HOMC: I used 2 LOMCs so far (Van den Hul MC One Special, 0.65 mV and AT33 PTG/II, 0.4 mV), with a DIY phono preamp (a rather simple and straight forward design with LT1028 or AD797 in stage 1 and OPA134 or LME49710 in stage 2) and now a Cambridge Audio Duo.

In all combinations amplifier noise was far lower than the noise of the quietest vinyl I own. If I play vinyl at maximum SPL possible I hear no noise when I lift the arm up. Hence I think reducing SNR gives no audible improvement due to the limits of vinyl pressings.

 

[SIY]

Yes, but that's an MM cartridge? Pretty much par for the course with loading as high as 500pF!

But not atypical with FET input (non-cascode) or triodes, especially with cables added in. The 150MLX is absolutely an MM, I was just using it for illustrative purposes re: why inductance is important.

25uH*(2.5/0.4)^2 = 977uH

Possible, which removes loading as a critical factor, but it would be nice to verify before jumping into sims.

 

[Michael Fidler]

But not atypical with FET input (non-cascode) or triodes, especially with cables added in. The 150MLX is absolutely an MM, I was just using it for illustrative purposes re: why inductance is important.

Possible, which removes loading as a critical factor, but it would be nice to verify before jumping into sims.

I can't see how it wouldn't be the case, although I might be getting a couple of Dynavectors soon, so can report back with some genuine measurements.

IMO: Triode inputs should be cascode (either traditionally, or feeding into a very low-impedance amplifier stage from the anode to prevent significant signal voltage such as EAR 834P), or simply pentodes (quite noisy though). But should we really consider wacky designs that have such flaws as hundreds of pF of Miller capacitance? Even the ubiquitous two-transistor 'Dinsdale' circuit of the 1960/70s didn't have such a flaw.

Q101 collector stays pretty much fixed to 1.6V, working as a current source into the base of Q102 - no appreciable Miller effect.

About the improved SNR of HOMC: I used 2 LOMCs so far (Van den Hul MC One Special, 0.65 mV and AT33 PTG/II, 0.4 mV), with a DIY phono preamp (a rather simple and straight forward design with LT1028 or AD797 in stage 1 and OPA134 or LME49710 in stage 2) and now a Cambridge Audio Duo.

In all combinations amplifier noise was far lower than the noise of the quietest vinyl I own. If I play vinyl at maximum SPL possible I hear no noise when I lift the arm up. Hence I think reducing SNR gives no audible improvement due to the limits of vinyl pressings.

Yes, but you can always have more!