Stepped attenuators
Join Our Community Subscribe to Paul's PostsControlling the volume level in your stereo system is an obvious and important function. The element used to turn the level up and down is important as well.
If you look inside a preamplifier you’d discover it’s one of the simpler devices in your chain. Most consist of only three blocks in the signal path: an input selector, a volume/balance control and a gain stage.
The input selector is nothing more than a switch that connects the input you wish to listen to into the signal chain. This switch can be a mechanical one requiring you to turn the knob to choose or a series of relays that act like switches. Relays are simple switches that use a small electro magnet to open and close their switch contacts. This allows designers to send a tiny electrical signal to control the relay, rather than relying on mechanical force to move the switch from one position to another. There are even electronic versions of switches that use a transistor as the switch (usually a MOSFET).
Once selected, the source you want to listen to is connected to the preamplifier’s volume control; the subject of this post. Most preamplifiers use what is known as a pot to control the volume. “Pot” is short for potentiometer and is a simple device consisting of a resistive element and a contact point (known as a wiper). The resistive element sounds fancy but it too is simple and made from a material that resists electricity like carbon or certain types of metals that don’t conduct too well; also known as a resistor (because it resists electricity). The contact point is a simple piece of good conducting metal that touches the resistive element (like silver). When you turn the pot up or down, you are moving the contact point (wiper) along the resistive element in a wiping motion.
The sound quality of this device is dependent entirely on the quality of the resistive element and the contact material of the wiper, once implemented properly. A good pot can be extremely expensive.
If you look closely at the picture you’ll note that the wiper is dividing the resistive element in two. In the example I show, A is a smaller area than B as divided by the wiper. This means that A has a smaller resistor value than B. If we were to actually measure these values we’d see that A is probably 25% of the whole and B the remaining 75%. If the pot is a 10k value (10,000 Ohms) then we could say it can be modeled with two resistors, the A resistor measuring 2,500 Ohms and the B resistor measuring 7,500 Ohms (the two resistors will always add up to the 10,000 value of the pot).
I mentioned the sonic performance of this device depends on the quality of that resistive element and the contact material of the wiper. There’s no doubt the best resistive elements can be found in individual resistors and the best contact materials can be found on expensive switches – none of which is true of potentiometers. Thus, if we decide to use discrete resistors and a quality switch to select those resistors, instead of a pot, we would have a better sounding volume control. This arrangement is known as a stepped attenuator.
You can see in the picture such a device. The switch is nothing more than a series of contacts and, as you can see, there are four sets of resistors: two to make up the A and B combinations of our pot for one channel and another set for the other.
The advantages of a stepped attenuator are the possibility of better sound because of the parts quality. The downside of this arrangement is the volume is controlled in a series of fixed steps and you may or may not like the step level.
There are pots that can come close to the performance of a stepped attenuator, but rarely do they.
Most high end preamps of today use an electronic version of the stepped attenuator, replacing the mechanical contacts of the pictured switch with either relays or electronic relays.
I dislike stepped attenuators
For the very reason i like to listen to music at a volume… I … want to hear it at
Not what the attenuator tells me to, in stepped volume
The tail wagging the dog
A spare Audio Research SP4 preamp i have has a stepped volume
Good sound, but no deal in my system
But i take your point about a better ‘sound’… technically speaking
Oh well…
Don;’t feel alone, I am not a big fan either. If you use CMOS switches and lots of resistors you can get close and do it in 1/2dB steps which isn’t bad. There’s always a trade off between the best sound and the best functionality.
Sub-groups of the Connecticut Audio Society did preamp comparisons including one session to explore volume control technology. I brought Bryston and VTL models for active potentiometer designs and a Creek passive volume control which were ascending in that order. We then tried a stepped attenuator which was yet more refined – but the champion was a stepped transformer.
The system output was refurbished ESL63’s driven by Charlie King’s sublime amplifier. It is a Dyna Stereo 70 chassis with the input and driver stage replaced by a cascode so there is a single capacitor in the signal path: a PentaCap oil filled, skived Teflon, non-inductively wound structure in a military grade case which is likely the best sounding cap available.
My current system is an Ayre AX-7 which is FET switched resistors integrated with the power amplifier, likewise a cascode voltage gain stage driving FET outputs for a solid state replica of the King topology.
Paul, please pardon my ignorance, but I don’t understand how varying the relative distances between the wiper and the two ends of the conductive material controls volume. Wouldn’t the signal just favor the path of least resistance, since both paths are not very long?
Good question Mike. It happens because one end of the conductive material is at ground (and therefore zero signal) and the other is connected directly to the music (and therefore 100%). So, forgetting the middle wiper for a moment, the resistive element has two extremes, zero and everything and in between, there’s every state between those two extremes.
The output of the pot is always the wiper so it is just picking off the signal at a point that’s 100% or 0% or somewhere in between.
Paul, I’ll show my electronics ignorance by adding to Mike A’s question. Doesn’t the signal pass from A throughout to W, therefore the value from wiper (W) to B is irrelevant? The total value of the pot, 10K in your example, then determining the value of R by the percentage of that value for the position of the wiper (W).
I thought only one resistor was needed for each stepped position per channel.
Your answer to Mike just explained the first part of my question, but I still don’t understand why two resistors are necessary for each selected position?
Acuvox submitted a correct answer, as did coppy, but I fear it still may not be clear to those struggling to get this.
Think of the pot as a single resistor going from A to B on the drawing. A has the signal in and B is going to ground. Ground is zero – nada, nothing. If you then move W to the very middle of the pot, halfway between A and B, you get 50% of the level. This works because you’ve divided our single resistor in half.
So if the pot “resistor” is 10,000 Ohms, when you place W halfway down this resistor, it is the same thing as having two resistors in series: a 5,000 Ohm from A to W and another 5,000 Ohms between W and B.
If we ignore that it’s a pot with a wiper, then we could easily model this with two resistors: both 5K. At the junction between the two resistors, you only have half the volume – this is because we’re going from 100% volume to 0% volume at the top of the two resistors and the bottom of the two resistors. In between these two resistors you must have half.
Think of this instead of music as light. At the top of the string we have a bright flashlight pointing down. At the bottom we have a black hole with zero light. As the light travels from its brightest to the black hole, it is getting progressively closer to zero. Half way there’s only 50% of the light. If you were to tap off of the middle point you’d only get half the light.
Dioes that get our understanding closer or am I still not hitting the mark as to why we need two elements?
Since you are discussing preamplifiers, any chance PS Audio is looking into producing another one? If so, how about one of a digital nature?
BTW–How is the power amp development going? Do you have an update?
Yes, I am considering a preamplifier but it would be all analog if so. Power amp creeps along – made some excellent progress so far.
You still need a consistent path to ground for the signal. All of the signal may not exit on the wiper, the the preceding circuit needs to see a constant impedance. So any spot on the POT is really two resistors, before and after the wiper. The resistor ladders simply duplicate that at fixed levels so the signal will see a constant impedance to ground. Think that’s correct.
Resistance is the ratio of voltage to current for a two terminal device. If you have two resistors in series as a load for signal, the input resistance is the sum of the two, In the case of a potentiometer this is constant. ignoring the loading of the following stage. With opamps or normal discrete circuitry this is a good approximation.
The output impedance is equal to the resistance between the wiper and ground, and so varies with the volume setting. Further, all impedance includes a variable depending on frequency which may account for the difference in sound between types of analog volume controls. (Note: I count continuous time as analog, as this is the difference between digital and analog in my model of hearing).
A true stepped attenuator switches both resistors to maintain constant input and output impedance. This could be done with a dual concentric potentiometer, but would take perfect tracking, most likkely with hand trimming via knife or laser. Tracking is also a major issue in stereo potentiometers. The left and right vary relatively, shifting the balance as you change volume. Stepped attenuators obviate tracking errors.
Another sonic differentiator is the distortion of resistors. These factors are normally measured as voltage sensitivity (the change in resistance with applied voltage), frequency sensitivity and non-linear dielectric absorbtion. The latter factor is subtle in resistors, but much more obvious in capacitors and cables.
The distortion of resistive elements with long life under a wiper means that potentiometers are always a compromise. You could use ultra-low distortion material but it would wear out too fast.
For many years almost all broadcast audio mixing boards used stepped attenuators for reasons of reliability. It was not until the 1970s that conductive plastic resistive tracks became durable enough for constant operation (with, say, 4 audio sources, 2-3 minute singles and 30-60 second commercials, each pot might get turned up & down 20 times an hour or more.) Intermittent signal loss and crackly noises are a definite no-no!
Many of these were true works of art. Big as a navel orange, contact studs arranged in concentric rows under a gently spring-loaded rotor, and highly polished shaft bushings. In operation they were quite smooth, with only a light feel of the contacts gliding across the studs. 2 dB steps were most common. A web search for “Daven attenuator” will bring up some pictures, as well as a few used units offered for sale (not cheap!)
Even so, these required a regular clean & lube maintenance schedule. They almost never failed though.
Best,
Mike