As We See It

Dispelling a myth (?) about phono-cartridge loading

Jim Austin  |  Dec 12, 2023

Pink noise spectrum of Ortofon Windfeld Ti phono cartridge loaded with 47k ohms (blue) and 30 ohms (red).

In the midst of his December 2023 Gramophone Dreams column, Herb Reichert presented the results of an experiment. He was listening to the most recent version of Zu Audio's Denon DL-103, installed on his new-old Lenco. He hooked it up to the moving coil input of his SunValley SV-EQ1616D phono preamp, which apparently is intended for use with low-output MC cartridges since it loads them down with a 50 ohm shunt resistor—a heavy load for all but the lowest-impedance MCs. The rough rule of thumb for loading an MC cartridge, as many readers are aware, is that the load resistance should exceed the cartridge's internal impedance by about a factor of 10.

As MC cartridges go, the DL-103 has a high source impedance, around 40 ohms—almost equal to the load impedance imposed by the SunValley phono pre. The result—loading at nearly 1:1—is very unconventional!

So, Editor Jim, doing his editorial due diligence, started asking questions. He found himself involved in a three-way conversation, with Herb plus Dave Slagle, proprietor of Intact Audio and partner in EMIA. Herb brought to my attention a 1980 article by J. Peter Moncrieff, published in International Audio Review #5. Moncrieff started by noting the "common speculation in the audio community that a moderately low-value resistor ... damps the high-frequency peaks which most moving coil cartridges exhibit in their frequency response." Yep, that's what I thought. While admitting that it's plausible, Moncrieff concludes, "in this case it turns out that common speculation is substantially untrue." He presents measurements of cartridge response as a function of loading from 100 ohms down to 5 ohms, with no change in frequency response.

Slagle became involved in our conversation because, as Herb already knew, he has independently corroborated Moncrieff's result. Both Moncrieff's and Slagle's work—and Herb's subjective observations—support the idea that what I have long taken to be conventional wisdom about MC cartridge loading—described by Moncrieff in the paragraph above—is wrong.

It is true that, when viewed as purely electrical circuits, MC cartridges have a resonant peak. Physics provides a simple formula for predicting the resonance of an electrical system. By plugging in typical values of inductance and capacitance for an MC cartridge, we find a resonant frequency not in the treble but in the low MHz range. (A mechanical resonance in the near-ultrasonic range has been hypothesized for a stylus moving in a record groove—the so-called stylus groove resonance—but assuming it exists, loading down the cartridge with resistance will not dampen it.)

Such values are so far above the audible range that even if they were real, damping them is unlikely to have any direct effect on the audible-range frequency response. But it doesn't matter: A phono cartridge is an electromechanical system, not just electrical. It isn't possible to vibrate a phono cartridge at MHz frequencies. Any electrical resonance will be mechanically damped.

The behavior in the audible range is easily tested. I had an Ortofon Windfeld Ti MC cartridge set up on my turntable. The Windfeld is a low-output MC cartridge with a low source resistance: 7 ohms. In my system, the Windfeld was feeding a Pass Laboratories XP-27 phono preamplifier, which has a range of passive loading from 30 ohms to 47k ohms. I connected the phono preamp's output to my Focusrite Scarlett 2i2 audio interface and connected that by USB to my Macintosh laptop. I played the pink noise track from the Analogue Productions Ultimate Analogue Test LP and recorded it at a sampling rate of 96kHz using Adobe Audition. I did this twice, first with the load resistance set to 47k ohms, then with it set to 30 ohms. I repeated the measurement with a few different values of capacitive load to confirm that this wouldn't affect the result. (It didn't.)

You can see the results in the figure at the top of this page. In the audible range, the only effect of heavier loading (lower R; red trace) is a slight reduction in cartridge output. Over the range measured, the frequency response is unchanged.

Yet anyone with ears can hear that loading down an MC cartridge does something. If it doesn't affect the frequency response, what does it do? Moncrieff suggested that proper loading reduces distortion—especially intermodulation distortion—perhaps by reducing the tendency of the stylus to flit about inside the record groove. Makes sense to me.

Where did the conventional wisdom come from? Perhaps it was helped along by the fact that it's qualitatively correct: There is indeed a high-frequency electrical resonance, and a resistive load can dampen it. It's just that the peak is too high in frequency to matter. A second source of confusion may be the use of step-up transformers (SUTs). Just as a transformer reduces the effective load resistance by an amount equal to the turns ratio squared, a SUT multiplies cartridge capacitance by the same amount. That's enough to bring the resonant frequency much closer to the audible range.

I find results like this exciting—it's fun to be freed up from convention. If one accepts this result, then it's time to rethink the whole subject of MC cartridge loading. Does an optimal value of the load resistance exist at all? Moncrieff says no—that there is no lower limit: The lower the load resistance, the better.

I was unaware of any of this until a few weeks ago—yet Moncrieff did his work 43 years ago. Has anyone since then solved the mystery of MC cartridge loading? If you know the answer, drop me a line at stletters@stereophile.com.

COMMENTS
MontyM's picture
Submitted by MontyM on December 12, 2023 - 9:39am

Jim, thanks for this interesting article. Without giving it the scientific attention you have, I have experienced the phenomenon you explore. I use a Zesto Allasso SUT that has flexible gain and loading settings. I adjust the settings so that the MC cartridge I'm using sounds best to me. I can't think of an example (admitting to a small sample size of maybe a dozen) where the cartridge has sounded best by simply following the 'factor of 10 rule of thumb.' This hobby is constant discovery.

-- Monty

JRT's picture
Submitted by JRT on December 12, 2023 - 10:16am

At his Hagerman Audio Labs website, Jim Hagerman has posted some worthwhile information on cartridge loading, and other related subject matter.

Article on cartridge loading:
https://www.hagtech.com/loading.html

More articles at the following link:
https://hagerman-audio-labs.myshopify.com/pages/frontpage

Jim Hagerman's picture
Submitted by Jim Hagerman on December 18, 2023 - 5:54pm

Thanks for the shoutout! I've done a bunch of revisions and updates to that page over the past few days so it better reflects recent information.

Jim Hagerman's picture
Submitted by Jim Hagerman on December 30, 2023 - 2:49pm

I've added a new section regarding transimpedance phonostages (those zero ohm input types) where cartridge is used in current mode. Turns out it was easy to calculate bandwidth of the cartridge, which occurs when the reactive (imaginary) impedance (L) equals the real impedance (R) of the coil. This happens at:

f = R / (2*pi*L);

JRT's picture
Submitted by JRT on December 12, 2023 - 10:26am

I think that if you were to pass the signals associated with both raw curves through a buffer and any sort of reasonably good RIAA filter (analog or digital) to flatten the curves with the same forcing function, and then subtract one from the other, a difference curve might be a better visual reference rather than trying to discern the varying vertical distance between the two raw response curves which are sloped varying curves.

John Atkinson's picture
Submitted by John Atkinson on December 12, 2023 - 3:46pm
JRT wrote:
I think that if you were to pass the signals associated with both raw curves through a buffer and any sort of reasonably good RIAA filter (analog or digital) to flatten the curves

The spectra are derived from the RIAA-equalized output of the cartridge. Pink noise has decreasing energy as the frequency increases. I suspect you are thinking the test signal was white noise, which has equal energy per frequency.

JRT wrote:
a difference curve might be a better visual reference rather than trying to discern the varying vertical distance between the two raw response curves which are sloped varying curves.

I still have Jim's WAV files, from which I generated the graph, so I could try that.

John Atkinson
Technical Editor, Stereophile

Jim Austin's picture
Submitted by Jim Austin on December 12, 2023 - 3:55pm

Looks like we were both answering this at the same time, though you beat me!

Thanks John.

Jim

Jim Austin's picture
Submitted by Jim Austin on December 12, 2023 - 3:53pm

JRT, what you're seeing in the graph is RIAA-corrected, from an RIAA-encoded test record. Perhaps that's unclear because it's pink noise in an FFT analysis, whereas in audio most are used to seeing 1/3-octave spectra. In a 1/3-octave spectrum, pink noise appears as a flat, straight line--but in an FFT analysis, which has the same frequency range in each channel (instead of a third of an octave), it's a straight line with a negative--downward--slope, 3dB/octave. It isn't perfectly flat because the cartridge's response isn't perfectly flat. There could be some RIAA error, but I'd expect the Pass Labs phono pre to be pretty good. If there were a high-end peak, I'd expect to see more of a rise at higher frequencies. The main result here, though, is that the two curves look so similar.

It is true however that there are better ways to display this. This is admittedly a quick-and-dirty attempt to do a quick check to corroborate the Moncrief result with as little work as possible. It's also important that this is just one cartridge; other cartridges may behave differently. I would not be shocked to learn that the Ortofon is especially well-behaved.

Jim Austin, Editor
Stereophile

eatapc's picture
Submitted by eatapc on December 12, 2023 - 10:29am

Thanks for investigating this. It makes me wonder if all the frequency vs. loading charts I've seen over the years — showing rising frequency response in MC cartridges, tamed with correct loading — were based on real measurements. Is it possible that the plots were simply drawings based on assumptions or theories that were not tested?

As to the reduced IM distortion, I note that Rob Robinson of Channel D has done measurements showing improved tracking ability with his transimpedance phono stages, which load cartridges down to a virtual short. https://www.channld.com/seta/linoC2.html

Bonsai's picture
Submitted by Bonsai on December 12, 2023 - 3:58pm

There are quite a few MC EQ preamplifiers out there now that feature 0 Ohms input impedance by treating a MC as a current generator and feeding the cart into a virtual earth. These designs appear to offer very good performance and even a 40 Ohm DL103 suffers no deleterious effects.

See www.hifisonix.com and the X-Altra MC/MM phono preamp for an example.

Ortofan's picture
Submitted by Ortofan on December 12, 2023 - 5:23pm

... variations in capacitive loading on the frequency response of moving-coil type cartridges, as discussed by FM Acoustics in their Newsletters 9+10, dated Autumn 2000, pages 4-5:

https://www.fmacoustics.com/download-1/

Jim Austin's picture
Submitted by Jim Austin on December 12, 2023 - 6:00pm

For this cartridge, I tried the maximum and minimum capacitive-loading settings on the Pass Labs phono preamp and there was no measurable difference in the frequency response.

I repeated the measurement with a few different values of capacitive load to confirm that this wouldn't affect the result. (It didn't.)

Jim Austin, Editor
Stereophile

Ortofan's picture
Submitted by Ortofan on December 12, 2023 - 6:47pm

... available from the Pass Labs XP-27 is only 750pF.

Note that the values shown in the FM Acoustics document are 0.68uF and 1.5uF - several orders of magnitude greater than those in the XP-27.
In their particular example, a load of 1.5uF in parallel with 68Ω yields the flattest frequency repsonse.

dave slagle's picture
Submitted by dave slagle on December 13, 2023 - 7:39am

It would be nice if the associated test conditions for the above referenced plots (from FM Acoustics) showing varying frequency response for different loads were referenced. This would help us better understand how we have seemingly opposite behaviors from similar measured results.

dave

Michael Fremer's picture
Submitted by Michael Fremer on December 14, 2023 - 2:24pm

Having told readers for decades that with MCs capacitive loading isn't really an issue....but everything else is now floating on open waters....very confusing...

Ortofan's picture
Submitted by Ortofan on December 14, 2023 - 5:40pm

... use to reach that conclusion?

Did you try, for example, only hundreds of picofarads, or did you go up as high as a few microfarads?

Jim Hagerman's picture
Submitted by Jim Hagerman on December 18, 2023 - 5:58pm

You are correct, it isn't an issue. Unless an SUT is being employed, whereby the capacitance seen by cartridge is reflected through SUT multiplied by turns-ratio squared.

mosfet50's picture
Submitted by mosfet50 on December 13, 2023 - 3:21pm

That curve is bode plot not an FFT.

Fo=1/(2piLC); resistors only effect freq. in high freq. circuits (RF frequencies)

Jim Austin's picture
Submitted by Jim Austin on December 15, 2023 - 6:11am

I believe that "Bode Plot" refers to a mans of presentation, with logarithmic data on both axes. FFT, on the other hand, is a specific calculation computational technique. An FFT could be presented on a Bode Plot or with linear response on either or both axes.

Regardless of how it's presented, this graph was calculated (in Adobe Audition) as an FFT, with the same frequency range in each frequency "bin." Adobe provides little technical information about its calculations, but that much at least is clear.

Jim Austin, Editor
Stereophile

mosfet50's picture
Submitted by mosfet50 on December 15, 2023 - 7:18am

A bode plot is a magnitude/phase representation as a function of frequency. A Fast Fourier Transform is a representation of a signals analysis as a function of frequency as opposed to time:

"Fourier analysis converts a signal from its original domain to a representation in the frequency domain and vice versa."
source: Wikipedia

For example, if we wanted to see the odd order harmonics of a square wave, we would us an FFT. A bode plot graph would not show that.

If we wanted to see the variation in gain of an amplifier over a specific frequency range and its phase we would use a bode plot, an FFT graph would not show that.

In this instance we're looking for the gain of a cartridge relative to a specific frequency range an FFT graph would not show that, a bode plot graph would.

Regardless of the math used to calculate the graph the representation is a bode plot - magnitude/freq.

Jim Austin's picture
Submitted by Jim Austin on December 15, 2023 - 7:30am

A bode plot is a magnitude/phase representation as a function of frequency.

I see that now. But there is no phase information in this plot--only magnitude vs frequency. So technically that is a "Bode magnitude plot." But I can assure you--having used FFTs extensively during my PhD research--that this is an FFT plot. Perhaps confusion arises from the fact that the number of samples here is very large, resulting in a very fine resolution. Perhaps you were expecting to see discrete frequency components? You would not see that here, because the frequency constituents of pink noise are (quasi-) continuous and the resolution of the FFT is rather fine.

I've got editing to do; no time to continue this conversation.

Best Wishes,

Jim Austin

mosfet50's picture
Submitted by mosfet50 on December 15, 2023 - 8:16am

Jim,
The mathematics used to generate a graph is a non sequitur. We're looking at a specific analysis of magnitude/freq. not analysis of a signal/freq. The graphs are specific to their intended results and observations. We simply don't call a bode plot and FFT, entirely different things and if a plot leaves off one aspect like phase it still gives us the information the bode plot is generated to do, we do this all the time with graphs. Calling it a "bode magnitude graph" is redundant just is calling it a 'bode magnitude only graph' would be. It's just a bode plot!

teched58's picture
Submitted by teched58 on December 15, 2023 - 11:22am

Bode plots are a time honored technique which IIRC are introduced pretty early on in the sequence of undergraduate classes EEs take. (In my memory, both Bode plots and Kirchhoff's law were big things in first-year EE. Neither are difficult, which I guess is why they appear early in the curriculum.)

OTOH, I suspect while most physicists are well familiar with FFTs, they aren't exposed to Bode plots.

cgh's picture
Submitted by cgh on December 15, 2023 - 11:51am

Now I am confused as to how to interpret the graph. I assumed it wasn't labeled entirely correctly re 1/f noise and that it was just a graph of dB versus a sweep in frequency under two different loads to show that the traces follow the same path with shift (what the ee people are calling a Bode plot).... not that there was a number of time base obs FT'd.... but maybe that's what was meant by pink noise? It's the FFT of a pink noise source under two different loads. If it was an FFT of pink noise it wouldn't be decreasing like that... hence why I assumed it was what the ee guys are calling a Bode plot.

The plot tells the story regardless... there's no dependence.

Jim Austin's picture
Submitted by Jim Austin on December 17, 2023 - 6:32am

ugh, read my comment--above--with subject line "JRT, what you're seeing in". It answers your question about the graph's negative slope.

Jim Austin, Editor
Stereophile

mosfet50's picture
Submitted by mosfet50 on December 15, 2023 - 12:07pm

So is ohms law, what's your point? Maybe it was introduced early on in the curriculum because it's germane to electronic engineering and will be used often.

What physicists are exposed to is subjective, we don't know what most physicists are exposed to.

Geeze, we're evaluating the frequency response of a cartridge for Pete's sake, hence the bode plot which ideally suits that application.

PeterPani's picture
Submitted by PeterPani on December 13, 2023 - 11:52pm

Shouldn't the producer of a cartridge be able to give an optimal value for the cartridge loading?

mosfet50's picture
Submitted by mosfet50 on December 14, 2023 - 1:31am

All you have to know is the inductance of the cartridge.

The XL(inductive reactance) is relative to frequency of course (XL=2piFL) but you can pick the resonant freq. the cartridge works best at by doing the math with different caps and even plot different graphs using different capacitors.

The plot above is exactly right as you decrease the resistance you load the cartridge more but let's remember this is basic ohm's law E=IR, the resistance doesn't affect Fo (resonant freq.)

Look at it this way, the cartridge is a generator, it generates a voltage because of the winding resistance the current is limited, according to E=IR the load resistance drops and the current rises but it can only go so high before the voltage drops and the output of the cartridge drops.

This has nothing to do with the freq. There is no R element in Fo=1/2piLC so the output drops as the resistance drops BUT the freq. curve stays the same as the above plot shows.

Do the bode plot with different capacitors loading the output and you will see the plot change - pick your favorite flavor.

I'm not sure I understand what the "myth" is or what the person found in the 80's but this is basic AC circuit analysis.

Jim Austin's picture
Submitted by Jim Austin on December 15, 2023 - 6:37am

It damps--broadens and lowers--the peak; for a series resonant circuit, Q is proportional to 1/. That's the theory, which is undoubtedly correct. The issue here seems to be that the resonant frequency is well above the range where it's relevant in an electromechanical system like this.

Jim Austin, Editor
Stereophile

mosfet50's picture
Submitted by mosfet50 on December 15, 2023 - 7:36am

Which is what the bode plot shows, there's no change in freq. relative to resistance.

Jim Austin's picture
Submitted by Jim Austin on December 15, 2023 - 7:38am

... and sufficient resistance was added, you would see the peak decline in magnitude and broaden.

Jim Austin, Editor
Stereophile

mosfet50's picture
Submitted by mosfet50 on December 15, 2023 - 2:04pm

Pure resistance can not change frequency, It can not create peaks, dips, rolls offs, etc. It's not possible! Does anyone see resistance in any of these formulas:

Fo=1/2piFL: Xc=1/2piFC ; XL=2piFL

No, so how can resistance effect a frequency solely by itself? It can't.

The only thing a resistor on the output of a cartridge can do is load the circuit and drop the gain. Again, that's why the bode plot doesn't show any frequency change.

So if anyone wants to change the frequency profile of a cartridge they better have an inductor or capacitor in their circuit because, as far as I know, cartridges do not have enough inductance or capacitance to change a frequency in the audio spectrum with just a resistor on the output. Again, again and again as the bode plot above clearly shows.

Anton's picture
Submitted by Anton on December 14, 2023 - 7:55am

Thank you for teaching me cool new stuff!

Ed Oz's picture
Submitted by Ed Oz on December 14, 2023 - 1:10pm

Shannon Parks has championed this take on MC loading with his products, saying that - based on his research and experimentation - what we're hearing with loading changes are shifts in gain, not frequency response. Thus, his Puffin and Waxwing phono stages offer just two load choices and continuously variable gain and volume. (Thanks for this article, Jim!)

mc1z's picture
Submitted by mc1z on December 14, 2023 - 1:38pm

Years ago, I had an all c-j system with the Premier 15 phono stage & got a SoundSmith Norma MI cartridge to replace a Van den Hul DDT-II that had too low an output to be well matched with the c-j phono stage. After being without a decent system for almost 10 years, I recently starting putting a budget system together, but I still had the cartridges. I tried a couple phono preamps, both with 100 ohm loading - the Van den Hul sounded fine, but the SoundSmith sounded so bad, I contacted Peter Lederman to discuss returning it for repair. He reminded me that the documentation for the Norma states that it requires a load of 1500 ohms minimum (in capital letters, in fact), so I started looking for a low-cost phono stage that would work. That led me to Jim Hagerman's page mentioned above, then to his products for sale. In contacting him, he knew exactly why the SoundSmitih MI cartridge needed the higher load and offered to substitute the 470 ohm spot in his Bugle MC with a 2.1k ohm load. Listening to the Norma while switching between 100, 220, 2.1k & 47k was enlightening. At 100, it was unlistenable, mostly so at 220, beautiful at 2.1k, and a bit "screechy" at 47k. I haven't yet tried the experiment with the Van den Hul MC, but plan to soon.

mosfet50's picture
Submitted by mosfet50 on December 14, 2023 - 2:02pm

Remember you're not changing the freq. response of the cartridge with resistors as the bode plot above clearly shows. However it's possible to overload the cartridge with low value resistors.
A cartridge can be affected by resistors this is not what the science says, it only says the freq. won't be changed. If there's a capacitor in the cartridge circuit you might create oscillations at certain frequencies that could very well interfere with the sound negatively.

mc1z's picture
Submitted by mc1z on December 15, 2023 - 5:16am

If my post appeared to disagree with the overall topic of the initial post, I apologize for the confusion. i certainly didn't mean to - not least because I don't know enough about the topic specifically to have an opinion, which, in turn is because I don't care. And I sincerely don't mean that sentence to be belittling or quarrelsome in any way - it's just the simplest way i know to describe the nature of my interest in the topic. Not being a component designer, there are a lot of things I'm way more interested in delving into the theory of than cartridge impedance. I only meant to say that my experience (and the knowledge of SoundSmith's founder) indicates that cartridge loading is so important, with some cartridges, at least, that it's almost a necessity to know what your cartridge (in your system) "likes", and supply that load - otherwise, you're not getting what you paid for. (And, even more off-topic, my old fingers sure appreciate the ability to play with the loading via a switch on the phono stage, rather than having to deal with a lot of fiddly screws & DIP switches.) I suspect that everything I said in my original post was essentially a statement of the obvious to this audience - if so, as I said, I really didn't mean to confuse the topic...

haroon's picture
Submitted by haroon on December 18, 2023 - 7:01am

Cartridge and Phono designers don't agree with you Jim! They blame the phono preamplifier's inability to deal with ultrasonic or low radio-frequency energy.

Jonathan Carr, designer of Lyra cartridges and Ralph Karsten, designer of Atma-Sphere have a very different explanation for loading making cartridge sound different.

Jonathan Carr
https://forum.audiogon.com/posts/2590914

Ralph Karsten
https://forum.audiogon.com/posts/2588565

Audiophile and industry are stuck on traditions/rituals/marketing without giving a thought to reasons behind them. For example, we see a separate ground wire on balanced XLR tonearm cables and ground lug for a balanced XLR input on many phono preamplifiers. It is not needed with three pin XLR connector, which already has Pin 1 as ground. But separate ground was needed with two pin RCA connector where ground pin was taken up by -ve/ inverted signal.

Charlie-S's picture
Submitted by Charlie-S on January 27, 2024 - 10:04am

Late to this games, but I’ll offer this observation. As recognized, unlike most MM and MI cartridge designs, the low source impedance of most MC cartridge places any electrical resonances far above the audio range. However, all have MECHANICAL stylus resonances within an octave or so of the highest audio frequencies. One is the resonance formed by the effective tip mass acting against the compliance of the vinyl material. Another is formed by cantilever compliance connecting the tip mass and generator (magnet or coil) mass. The relative importance of these resonances depends on specifics of the cartridge design. The very best cartridges have reduced masses and stiffened cantilevers such as to place these mechanical resonances above 20 kHz.

Shunting a MC’s low generator resistance with a low resistance load produces some electrodynamic damping of the mechanical resonances. Lower shunt resistance increases damping. (Folks of my age may remember Simpson analog multimeters having the word “TRANSIT” next to one mode switch position. This mode places the lowest resistance meter shunt across the meter coils, serving to reduce meter swings while carrying the instrument.)

Are these mechanical resonances audible? Yes, directly so if well within the audio signal range. But high-energy stylus motion such as record ticks and pops also excite the resonances. And if not sufficiently damped will produce ultrasonic energy possibly overloading the phono preamplifier, thus indirectly degrading audio quality.

(Incidentally, the much higher coil resistance typical of MI/MM cartridges precludes this type of mechanical resonance damping. But the RLC electrical filtering of its electrical output helps protect preamp input from ultrasonic overload.)

Finally, why did Jim Austins frequency response measurements show no differences with loading? Possible because the Ultimate Analogue Test LP’s pink noise track (which is band-limited to 20 kHz) has insufficient energy to excite his Winfeld Ti’s MC mechanical resonances. Being a very high-end cartridge, I’d assume any mechanical resonances lie above 20 kHz and are well controlled.

X