Simple reasoning and practical tests, against the standard practice.
Preface
The life of an engineer is not simple; when it comes to build and sell High-End Audio, you have additional obstacles, namely misconceptions and myths of the Audio world. To make matter worse those gross misconceptions are widely implemented in many popular products, not because are of any practical benefit, but because the market wants them and have them implemented gives a kind of entitlement of the product to be high-end.
The High-End Audio world is impregnated of paranoia, on forums you find fierce debates about matters that are wrong, irrelevant or of no consequence. You can read pages and pages of passionate discussion to establish which brand of resistors sound better. People are swearing on the own child’s head that ‘that’ resistor sound much better than any other, of course, none of those philosophers has the system professionally tested and tuned on the listening room, and ⅔ of the bandwidth they listen to is way out of phase… however, who’s care? The sound of THAT resistor is more important.
However, I’m not here to give vent at my rants, let’s go to the heart of the matter. Today I want to discuss one feature that is present in many commercial SUT (Step Up Transformer) for MC cartridges.
It’s not a secret that we produce a line of High-End, Hand-Made SUTs; one of the questions we receive about our SUTs is: Why your SUT haven’t a selectable impedance?. At this question, we answer: Because it is useless!
Now, such answer is not really what the discerning audio enthusiast want to ear, firstly is because is against the general belief that is very important to have a selectable input impedance, secondly is because the explanation about WHY it is useless need a bit more time and efforts than just a brief statement. If you want to know the reasons because that magic selector is just smoke on the eyes, then you must arm yourself of patience and read this article.
What the impedance selector does and how it works
To understand how the impedance selector works we must understand how the SUT impedance works.
A SUT has not an impedance of its own, at the primary (where you connect your MC cartridge) the impedance the pick-up see is a ‘reflection’ of the load applied on the secondary (where your phono stage is attached), the formula to calculate the primary’s impedance is: (Secondary load/SUT’s ratio^2); 99.9% of the phono stages have a load of 47 KΩ, for a transformer with a ratio of 1:10 the formula is: (47000/(10^2))= 470Ω, for a 1:20 SUT is: (47000/(20^2))= 117.5Ω.
It’s clear that, if we want an impedance higher than that calculated above, we need to increase the phono stage load, and this means to modify the device (and you need to know how to do it and void your warranty), if we desire a lower impedance we need to lower the phono stage’s load. For doing so, we must put a resistor in parallel with the phono stage. Two resistors in parallel (one is the load resistor inside your phono stage, 47KΩ), results in a load following the formula: 1/Rt=1/R1+1/R2. If you want to lower your impedance to 100Ω with a 1:10 SUT then you put in parallel with the 47KΩ phono load a 13KΩ resistor, the resulting load is: 10183 Ω, we now calculate the impedance at the SUT’s primary: (10183/(10^2))= 101.83Ω, simple, isn’t it?
So that’s what that ‘magic’ selector do, put in parallel with your phono stage resistors of different values. As a result, the impedance at the SUT’s primary change.
Do I need that selector?
The short answer is: NO!
Almost all MC cartridges require an impedance from, roughly, 100Ω to 1000Ω with an internal cartridge resistance of few Ohms. If you are between the min and max impedance value that the cartridge manufacturer specifies for the product, then you are fine. As a matter of facts, impedance bridging shouldn’t be a problem with almost all SUTs, of course, there are exceptions and owners of very exotic cartridges might not fall into this statement, but for the rest of us mortals, the statement is correct 99% of the cases. Of course, you can find on the net some ‘guru’ that's keeping telling you that the cartridge impedance must be matched perfectly! Probably who say so have read some manual of high-frequency/radio tech where, in facts, the impedance of two devices (typically a transmitter and cable/antenna combo) must be matched almost perfectly, penalty loss of signal and, in extreme cases, damages to the transmitter. In audio this is NOT true, the golden rule is that you build a system where the impedance is bridged. This means that the impedance of the driving device (that who generate the voltage) must have an impedance 1/10 or lower than the device that must be driven, and this is true with MC cartridges as well; main reason for this is that most audio devices have an impedance that change with the frequency, therefore exact impedance matching IS NOT possible!
However, I read on that forum that changes the impedance helped in to improve the sound of the cartridge!
So in most cases who said so is right! However, the problem is NOT matching the cartridge impedance but curing a SUT’s that is ringing!
Notes about SUT’s behaviour at higher frequencies.
If you test a SUT with a sinusoidal signal you will be sleeping sweet dreams for all your life; matters get more complicated when you feed your SUT with a square wave. Why a square wave? The audio signal is NOT square! Indeed, but an audio signal is not a sinusoid either.
The audio signal is very complex, the wave you have to reproduce is an intricate mess where you have a fundamental, several harmonics, and continuously varying in amplitude, sometimes in matters of milliseconds, or less. The square wave is like a sinusoid with an infinite number of harmonics and with 2 steep sides that give us clues about how the amplifier (or the SUT) behave in fast-changing transients. A good square wave reproduction give us clues about the behaviour of the amplifier in difficult situations, please take note that not always an amplifier can reproduce square waves, coupling capacitors, RIAA networks and other passive components will degenerate the square and generate other kinds of traces, this will not hamper the ability to reproduce sound with a high fidelity.
Increasing the frequency the square wave becomes more and more ‘demanding’ for our SUT, the time the cores and windings must react to the changing signal is sorter and shorter, the magnetic fields inside our transformer are frenetically working trying to cope with the steep sides of the square wave. Moreover, then the SUT start to ‘ring’.
Ringing and transient’s reaction
In the following picture, we have the trace of a 1KHz square wave reproduced by one of our MC01/10 1:10 SUT.
The trace is almost perfect, just a minimal overshoot at the extremes of the 2 sides of the wave but of no concern, the overshoot is less than 0.02 ms in duration. This kind of phenomena is common with SUTs because of the hysteresis of the system.
Now we increase the frequency to 10 KHz
Here we have a severe ringing problem. Ringing is common to all SUTs, is the incapability of the system (windings and cores) to stabilise after a fast transient happened at high frequency, it might affect at different extends, sometimes very moderate, sometimes quite severe like here. In this case, we have ringing phenomena that last nearly 40 ms with an amplitude of 30 mV in the relation of a signal of 120 mV ( ~25% ). In this case, a complex audio signal at higher frequencies (above 3-4 KHz) and with fast transients will be distorted audibly.
Now, in a commercial SUT with the selectable load, we can try to lower the load to 20KΩ (200Ω cartridge impedance in a 1:10 SUT), we simulate this setting in the next trace.
The new trace shows that the ringing is very much diminished, now the amplitude is only 8 mV in comparison to a signal of 80 mV (10% instead of 30%). Wait a moment. Did we say signal 80 mV?
The signal we fed into the system was unchanged; it means that reducing the load changed the ratio of the SUT from 1:10 to 1:7.5. That’s the reason why I say that the ‘magic’ selectors that change your impedance are good for nothing. Maybe will help to reduce ringing, but at the cost of amplification, if your cartridge has, say, 0.4 mV output with a 1:10 SUT you get 4 mV, and that’s a decent value to feed any phono stage, but now the same cartridge will give an output of only 3 mV, insufficient for many phono stages.
So what? We must keep this nasty ringing then!
A far better cure to ringing is to increase the capacitance of the secondary winding.
In this trace, we see the same SUT with the standard load of 47 KΩ but a 300 pF capacitor connected in parallel with the load. The output level remained the same (120 mV), and now the ringing is almost gone with a value of only 5.5mV to 120 mV (circa 4%) adding a further 60 pF proved to lower this ringing value even more (2.5%). The value of 300 pF was chosen to take into consideration that a cable with medium capacitance (like our litz 0.5 mt long interconnect) will be added between the SUT and the phono stage.
Let’s replace the ‘magic’ resistive selector with a capacitive one then!
Not really!
The need for a capacitor can vary from few pF (just the cable capacitance) to several hundred, and the effects must be verified in lab conditions with a trial and error method. Leave this to the user will mean either a selector with tens of possibilities and high chances to choose a wrong value. If the value is too low then the effect will be minimal, if too high we will get a square wave that will look like this:
That’s why we test and tune all our SUTs, who produce step-ups on industrial scale cannot do that and give you the ‘magic’ selector that’s is half of a cure for ringing, making you believe that you are ‘matching’ the cartridge, but at the end of the day you have a system that is not optimised. In the following image, The Vinyl Source SUT with a calibration capacitor soldered in place.
What about the frequency response?
Some ‘guru’ out there on the net will probably be horrified; capacitors are evil and, if used, only audio grade 650V metallic film wounded by Tibetan monks at 20000 feet above sea level are acceptable, because anything else is rubbish!
A capacitor in THAT position (a parallel RC network) won’t have any effect on the sound quality, so don’t worry if it is a tiny one that isn’t built in a former USRR nuclear bunker! Apart from that, the frequency response won’t be affected at all because the parallel RC network doesn’t act as a filter. Here is the frequency response after the tuning:
Here is the frequency response, linear within +- 0.125 dB from 12 Hz up 1.5 KHz and the next the rest of the audio band with the same linearity till 39 KHz (we use to split into 2 traces, so the low frequency is more detailed).
The punch line
The high audio world can be a very confusing place; we lost the capability to discern what is real tech and what is ‘snake oil’, customers spend fortunes buying more and more gear but is not able to make all the components work well together. How many of you spend a considerable amount of money for silly accessories that promise to improve your audio massively? Also, how many of you invest in professional support to make your system at the optimum? Many of the former, very few of the latter. It’s easy to spend money in a new toy and get bored at it after some time, in the quest of the holy grail of ‘perfect’ sound we lost the ability to discern how the real sound looks like. The industry wants all of us unsatisfied; the unhappy and unsatisfied human being is the perfect customer.
Keep sane and enjoy the music.