What Makes a Good RIAA or Line Stage?

Thom:

Regarding playback eq deviations, a small width one may indeed be rarely noticed. Wider band deviations will almost certainly be noticeable if you have a more accurate playback curve at hand for comparison (or have experienced one recently), but if said wider band deviation is the best that you have experienced (or some time has passed since you listened to a more accurate network), maybe you won't mind (or notice). However, although I haven't measured the Lamms, I know from my own work that what we perceive as measureable frequency deviations (as would be the case with an improperly designed RIAA network) may not always be so. Component choices, HF bleedthrough via the power supplies, resonances in the RF range all play a role in the perceived frequency balance. For example, although it appears to be accepted knowledge now that different capacitors (or resistors) have their characteristic signatures, the same also applies for active devices (even if they conform to the same nominal spec). If I don't like the perceived frequency balance that I am getting, it is therefore not a problem to change that while keeping the measured frequency response in the audible band the same. The process may involve some trial and error, and it may take me a few tries to get where I want, but it certainly can be done.

Regarding the analogy with concert halls, I get your point, but I am not sure if it is on target. Many instruments have very different tonal balances depending on the angle and distance that you listen to them from, and you need to physically put your ears where the microphones are to verify whether what you think you should be getting is really what is inscribed on the LP or not (and don't forget that mikes have different frequency responses from our ears). I am fortunate enough to have friends who are recording engineers and have allowed me to sit by the microphones (sometimes on a ladder!), tap into the mike feed, go back to a normal seat in the audience, listen to the analog tape master on the same day, and then a few days later, listen to the lacquer masters. Very, very educational. I encourage you to search out opportunities to experience this.

I very much agree with Ralph's comments on the desireability for high overload margin, and I will add that this is needed at ultrasonic frequencies as well as audible ones. Groove dirt and damage played through the cartridge manifest themselves as transient impulses (very high amplitude, very high frequency content) that at least the front end of the phono stage needs to deal with. If the phono stage doesn't have good overload margins and recovery, pops and ticks will be emphasized, so will record noise in general, and this can also shift the perceived tonal balance upwards so everything sounds brighter than it should.

To add another point, good behaviour in the RF region is also desireable, because there is enough energy (particularly in the 500kHz~ 2MHz range) normally reaching the phono stage that, if the phono stage has problems in this range, IMD can result in inharmonic distortions subheterodyned down into the audible range. Obviously, AM radio stations broadcast in this band, and need to be dealt with. However, phono cartridge loading can also generate resonances in this same region. The inductance of the cartridge's signal coils will react with the capacitance of the interconnet cable to create a resonance in the RF range. Let's take a Denon DL-103. Measuring, I get 40.5uH coil inductance. phono cable capacitance 150pF, resonant frequency 1.94MHz. Now let's see what happens to the measured frequency response when we vary the input load resistance of the phono stage. With a load of 47kohm, the electrical response is flat out to 100kHz but starts to rise, and by 1.94Mhz it is about 7dB up. If we say that the correct load resistance is sq.rt. (L divided by C), we get 500 ohms, and while the frequency measurement looks the same as with the 47kohm load, it stays more or less flat out to a -3dB point of 1.77Mhz. Even if we load at 270ohms, approximately half of the optimal 500ohms, the frequency response still stays flat out to 100kHz, and at 1MHz, we are only down by 2dB.

So, even when you give a low-medium input impedance MC various loads, the audible frequencies are not directly affected. The measureable frequency variations are occuring at ultrasonic frequencies. So why do people report major difference in sound when the input loading is altered? IME, HF behaviour of the phono stage and IMD is the answer. IOW, if the phono stage has exemplary behaviour at RF frequencies, whether the triggering source is a radio station or a resonance between the coil inductance and cable capacitance, that stuff will remain at RF frequencies and you won't hear it (at least not easily - grin). But if a sensitive part of the phono stage has performance issues at those same RF frequencies, IMD will make it far more likely that, for example, changes to cartridge loading result in big changes to the sound. And listening while altering the input loading of phono stages with high HF overload and good RF behaviour as compared to those that do not, bears this out (at least in my experience).

Do note, however, that since coil inductance and cable capacitance determine the resonant frequency, with enough coil inductance and capable capacitance, the resonant frequency can drop to within or close to the audible range, and the likelihood of hearing the effects becomes far higher, regardless of how well the phono stage may do at RF frequencies.

If the designer has taken this sort of stuff into account as well as obvious things like an accurate RIAA network and low noise, the greater the chances are that all of your LP collection (or at least more of it - grin) will sound good.

Again, I agree with Ralph that nasty recordings are often a better guide to the real worth of a phono stage than kind ones. Usually, when I am testing or auditioning equipment, I prefer to put on "system-breakers" - recordings that I know from experience have a good chance of throwing a system into fits. None of that sissy audiophile stuff! (^o^).

regards, jonathan carr

Regarding the rationale for the non-standard 3.18uS (50kHz) turnover, first it needs to be established that all cutting lathes have their HF resonances in this same region. Is this true for each and every one of the Neumann models, and what about JVC, Sculley, Westrex et al and their respective models? Next, it should be pointed out that half-speed mastered LPs will have this HF resonance shifted by one octave (100kHz instead of 50kHz).

regards, jonathan carr

The magnitude of RIAA error is not particularly useful unless we also consider the range of frequencies that are affected by the error. in practice, a 1dB deviation that only hits one note is not going to be very noticeable, but a 0.1dB error that spans an octave or more can be quite noticeable. In fact, the manner of musical presentation changes when this happens. That said, I do think that the less total deviation there is from the RIAA curve, the better. True, LP recordings and mastering systems have their own deviations, but they can deviate in any direction, and as long as we keep as close to the standard curve as we can, the frequency deviations in one's LP collection should average out. At least we won't be favoring certain recordings over others, which would certainly be the case with an RIAA playback network that wasn't right.

I don't think that speaker colorations are an acceptable excuse to tolerate RIAA deviations. Admittedly it is next to impossible to exactly duplicate electronic colorations in the speaker and vice versa so that they can be truly compared, but at least in my experience, it has seemed that electronic colorations are much more noticeable and less forgivable than speaker colorations. I think that this is because acoustic colorations in the environment are part of everyday life, and compensating for this is a constant, subconscious process.

Regarding when balanced phono amps were introduced, I have on my bench a schematic for a discrete FET balanced phono amp from the Japanese audio magazine M&J which is dated January 1985, and I am pretty sure that there are earlier examples (especially from the tube guys).

I don't think that complexity in a design is necessarily a bad thing, because a major goal of this approach should be to get smaller "modules" with more well-defined tasks/behaviour/environment. This makes it easier to design, understand and debug the functions and can lead to better performance, even if the overall complexity becomes greater. A simple circuit can lead to a wider, less clearly defined range of responsibilities being assigned to fewer parts, and this can result in lower performance.

IME, NFB is just another tool, neither good nor bad by itself. The results of using NFB have a lot more to do with the capabilities and sensibiities of the designer than NFB per se. I usually dial in the amount of NFB by ear as well as by measurement, and sometimes I'm at 0dB of NFB, sometimes 50dB.

Although I fully agree with the "holistic" approach, I think it is possible to achieve a good-sounding line or phono amp using a variety of technologies, circuits and approaches. However, topologies and components by themselves don't know what they are supposed to sound like. Good sound, bad sound, they don't know any better. The most important component of all is the designer, and the final sound extracted from the topologies and components is only as good as the designer allows.

regards, jonathan carr

Thom, a replay curve that deviates more from the RIAA standard can sound better on select recordings, but it will almost certainly also sound worse on other recordings, compared to a replay curve that deviates less. My experience is that a flat RIAA curve is likely to allow you to enjoy more of your LP collection, not less. If 0.1dB or better is possible, go for it, as I think that on the whole, you will be ahead. The one categorical exception is when an LP contains sampling noise (sometimes CRT monitor noise). There are albums by Kraftwerk and Lorie Anderson that have this which I find painful. Everything sounds find until the sampler kicks in, and then I look for a wad of cotton, or wish that I'd designed an RIAA playback network that shelved down the top end (grin).

As an aside, the other possible solution for an RIAA playback curve would be to implement an EQ trim control, like with the FM Acoustics designs, or maybe a Cello/Viola Pallette. I've listened to and played with both, and yes, I can see their point.

I also find that a theoretically "better" solution - better power supply, better regulators, better amplification circuitry etc. will nearly always improve the sound of nearly every LP that you own. I don't find that "more accurate" means that you become more picky about the LPs that you can find enjoyable. Yes, you may become more aware of recording, EQ or mastering issues, but the music and performance comes through even more strongly, more than enough to overwhelm trivial concerns about the recording. The better my designs become, the more I appreciate a greater number of musicians.

I do find, however, that when it comes to component selection, you have to use your ears and subjective taste, in addition to your head. I've picked components that on paper should have been the cat's meow, but in listening turned out to be a pig's kiss instead. The designer cannot know each minute particular of every component that he chooses, and as they say, the devil is often in the details. So unlike the case with overall topology or circuitry or layout, with components I find it necessary to have a "range of candidates" on hand and go with whatever sounds the best - in the context of the circuit being tested. Engaging in this is more like cooking or choosing clothes than it is intellectual design, and is the phase where the more artistic types can strut their stuff, and pull level with or even ahead of other designers who may be their intellectual superiors. That's the fun part about audio (designing it as well as using it) - there is a place for the sensibilities as well as the intellect.

regards, jonathan carr

Quite so, Mothra. I've been at recording sessions where I was able to physically put my ears where the microphones were, and then listen to the electrical feed from microphones. In the majority of cases, there is a substantial difference between what the ear and microphone hears, even from the same location. And in most modern recordings, a variety of microphones are used, and each modifies the sound in distinctive ways.

Also, the location and angle of the microphones will in most cases be quite different from what you would hear if you were at a performance of the same event. Most microphones are located far closer to the instrument than any audience seat, and the angles will be quite different, too.

If you have a friend who plays the violin (for example), it is very instructive to listen to it being played at a distance (like you would hear from an audience seat), then listen to it from a distance of under one meter to get the microphones' perspective, and also listen to the instrument from above (again to get the microphones' perspective).

Now put all of the above together and think about the implications for a home audio reproduction system. Since the recording likely does not sound like what you would have heard live from a seat in the audience, if you have set up your audio system to sound like what you'd hear live, it is almost certain that your audio system is modifying what's on the recording, and not in a small way, either!

However, there are recordings that include a list of the equipment used, and also microphone placement drawings. If you know what the recording gear sounds like, and also study the placement drawings, you can form a closer guesstimate of what these recordings should probably sound like, and this can be a somewhat better guide to setting up your system (although you still won't know what the mixing contributed, as Mothra pointed out).

And my hat goes off to you, and all recording engineers who capture those great performances which justify our having audio systems!