Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Michael Lawrence]

Hi y'all-

As an offshoot of this discussion, I wanted to come up with a way to actually measure and plot the fader laws of mixing consoles to see how much variation there is.

In discussing this on the SynAudCon forums, Langston Holland and I conducted the following test:

• Measure length of fader track (usually 100mm on pro desks)
• Divide into 11 segments (every 10mm, 0 through 100mm inclusive) and label a piece of tape
• Stick tape to console, send 1kHz sine wave through channel and record the level at each of the steps

Langston said "I'd use a 1kHz sine wave and a multimeter, bypass EQ, etc. from the measurement path, start out with the fader at max, adjust the source voltage and mic preamp so you get 10V on the consoles output, which is very near clipping so you might have to use a lower max voltage. 10 is an easy number to deal with and you want to be close to clipping so you get the most stable readings (S/N) when measuring the lowest fader positions. "
For those who are members, the thread is here.

So we did that. He used an APB Dynasonics H1020 and I used an X32. Charts below in both linear and log formats. The method and therefore the data are inexact but still interesting.

X axis is fader position in mm, from 0 at the bottom to 100 at the top. Y axis is Volts. I didn't label them because I'm tired and I hate Excel.

Both are pretty close to true log (linear-in-dB) for most of the throw. I'd be interested in seeing results from more desks.

If anyone wants to try this out, maybe we can get a dataset together.

[Rob Spence]

Are not faders on digital desks simply digital encoders?

Linear or not would be in software?


Sent from my iPad using Tapatalk Pro

[Michael Lawrence]

Are not faders on digital desks simply digital encoders?

Linear or not would be in software?

Exactly. Usually 10-bit (1024 position values) for digital. But the curve would be up to whoever writes the code
It's much harder to do in analog.

[Justice C. Bigler]

What's the point?

What exactly are you trying to measure/prove?

[Michael Lawrence]

What's the point?

What exactly are you trying to measure/prove?

It's relatively challenging to make an analog pot or fader have a linear-in-db or "audio" taper. It requires extra components to tap and load the fader strip and wiper. So analog desks will have different fader laws depending on how the faders are designed. By contrast, the digital fader is an encoder and so any law can be accomplished in software.
British circuit designer Doug Self, who designed large-format consoles for Neve and Soundcraft, has an entire chapter in one of his books about incrementally developing circuitry to accomplish more useful fader laws.
For example, the attached image is from a Yamaha MG series mixer, and as you can see it's NON-linear-in-dB around unity in a very interesting way. I don't think anyone would design this taper (to cram the 'sweet spot' into a tiny region) for its own sake. I'm betting this was because the curve ended up a certain way, and it was cheaper to silkscreen the scale in a weird way than to add more components to each fader circuit block to change the law.

I find it interesting :)

[John Roberts {JR}]

It's relatively challenging to make an analog pot or fader have a linear-in-db or "audio" taper. It requires extra components to tap and load the fader strip and wiper.

Bzzzt wrongo...  The vast majority of (analog) faders are simple screened resistive ink cured onto a non-conductive substrate, so manufacturing a good audio taper pattern is a step and repeat process. At most there may be 2 hits of conductive resist, one to insure a decent taper down near the bottom (hop off region), then the rest.

The most difficult analog taper I ever had to deal with was mic preamp gain pots, that required good adjustability down at the single digit ohm (high gain) end, while providing 10-20k total resistance for the low gain end. These required something like 4 resist screen overlay steps in manufacturing.   

So analog desks will have different fader laws depending on how the faders are designed. By contrast,

As designed by the pot manufacturer and specified by their customers. Most follow an A20 (i.e. audio taper down -20 dB at 50% travel) some are A10 (only down -10dB at mid travel).  It looks like both the APB (analog) and x-32 (digital) target the -20dB nominal standard. The analog fader appears piecewise linear, because it probably is (I don't remember which brand faders JP used, I recall he changed a couple times over the years. Customers care more about how faders feel than precise tapers  ). The digital law appears smoother, despite being actually quantized, but a fader law executed by a 16+ bit multiplier will have very tiny steps. 

the digital fader is an encoder and so any law can be accomplished in software.

indeed...

British circuit designer Doug Self, who designed large-format consoles for Neve and Soundcraft, has an entire chapter in one of his books about incrementally developing circuitry to accomplish more useful fader laws.
For example, the attached image is from a Yamaha MG series mixer, and as you can see it's NON-linear-in-dB around unity in a very interesting way. I don't think anyone would design this taper (to cram the 'sweet spot' into a tiny region) for its own sake. I'm betting this was because the curve ended up a certain way, and it was cheaper to silkscreen the scale in a weird way than to add more components to each fader circuit block to change the law.

thats what we call ergonomics or human factors engineering... got to keep the paying customers happy or they won't buy your desk. 

When you have punters running around measuring stuff with bench test equipment, they expect their mixer/console to deliver unity gain, when the meter, fader, or both, say 0dB, that is not trivial in analog with production tolerances, so an expanded range around 0dB will make it a little easier to hit in production. Old school analog VU meters have a ton of resolution around 0VU so this 0dB problem is not exactly a new issue, but digital meters are likewise not very fuzzy about where 0dB is.

I expect the modern digital platforms will be able to deliver this apparent concise gain/level performance more easily than even high quality analog (like APB), giving the customers a false sense of serenity, having nothing to do with sound quality.

I find it interesting

I found this interesting back in the 70s...

JR

[Michael Lawrence]

Bzzzt wrongo...  The vast majority of (analog) faders are simple screened resistive ink cured onto a non-conductive substrate, so manufacturing a good audio taper pattern is a step and repeat process. At most there may be 2 hits of conductive resist, one to insure a decent taper down near the bottom (hop off region), then the rest.

Hi John!

I have before me Doug Self's Small Signal Audio Design, 2nd edition. Ch 13 is a quite in-depth investigation into this issue. Quoting from page 353:

Inexpensive faders are usually made using the same two-slope carbon-film construction as are rotary log volume controls, but the more expensive and sophisticated types use a conductive-plastic track with multiple taps connected to a resistor ladder, as described in the previous section. This allows much better control over the fader law.

High-quality faders typically have a conductive-plastic track, contacted by multiple gold-plate metal fingers to reduce noise during movement.

He goes to show a fader control law of the A20 variety and says Note that a fader does not attempt to implement a linear-in-dB log scale; the attenuation is spread out over the top part of the travel and much compressed at the bottom. This puts the greatest ease of control in the range of most frequent use."

So are you saying there's no resistor tapping involved, or that it's less frequently used due to modern manufacturing techniques?

Self also spends a lot of time on using tapping and loading to make a linear pot behave in a logarithmic fashion because he states that log pots are not generally very accurate.

Although he designed large format consoles for Soundcraft and Neve, most of Self's work is with home audio equipment, amplifiers, and preamps, etc. So it's possible that the manufacturing techniques are different due to the economies of scale?

a fader law executed by a 16+ bit multiplier will have very tiny steps.

I believe many of the current digitals on the market use a 10-bit system.

an expanded range around 0dB will make it a little easier to hit in production.

That's what's interesting about the Yamaha example I provided- the 0dB region is compressed rather than expanded. Which seems odd.

To be clear, I am not making any inferences about sound quality or mixer operation with regards to fader law. I'm just interested in the topic. :)

[Frank Koenig]

What JR says. So far as I know (analog) audio is the only application where non-linear pots are common. In the instrumentation and control system worlds pretty much all pots are linear as that's usually what you want and only they can be manufactured with decent, say better than 2%, repeatability.

And while I agree that exact calibration of faders is unnecessary for most mixing, I rather like that my modern digital mixer has the facility to calibrate faders. This not only appeals to my inner German but there are occasions when knowing with confidence that two faders match or that I can remember and reuse a gain difference between two busses makes for one less thing to worry about.

--Frank

PS: I'll confess that in my audio hobbying back in the day I did use the loaded linear pot approach, mostly as good audio taper pots were unavailable to me. It does a lousy approximation of audio taper and forces other compromises. -F

[John Roberts {JR}]

What JR says. So far as I know (analog) audio is the only application where non-linear pots are common. In the instrumentation and control system worlds pretty much all pots are linear as that's usually what you want and only they can be manufactured with decent, say better than 2%, repeatability.

And while I agree that exact calibration of faders is unnecessary for most mixing, I rather like that my modern digital mixer has the facility to calibrate faders. This not only appeals to my inner German but there are occasions when knowing with confidence that two faders match or that I can remember and reuse a gain difference between two busses makes for one less thing to worry about.

--Frank

PS: I'll confess that in my audio hobbying back in the day I did use the loaded linear pot approach, mostly as good audio taper pots were unavailable to me. It does a lousy approximation of audio taper and forces other compromises. -F

"Slugging" pots (using a resistor load to pull a taper from linear) is a trick done sometimes out of necessity (small companies can not afford custom tapers...been there), but the bulk resistance of screened resistive elements is something like 20% so repeatability of slugged pots is too variable for mass production (or not so mass production). Most pot manufacturers offer a decent range of standard tapers.

I have slugged (actually my technician did the work) individual sections of a 4 ganged pot back last century to make an analog 4 pole L-R crossover that didn't suck. Slugged the individual pot sections to roughly match at 50% rotation. But that was low volume and decades before digital technology was remotely practical.

JR 

[Riley Casey]

Ah for the days of hand matching 1% MF resistors for setting filters in the BSS360 crossovers. That and totally screwing up the roll your own EQ filters from the R / C tables. 

...
I have slugged (actually my technician did the work) individual sections of a 4 ganged pot back last century to make an analog 4 pole L-R crossover that didn't suck. Slugged the individual pot sections to roughly match at 50% rotation. ...

JR

[Riley Casey]

dupes R us

[John Roberts {JR}]

Hi John!

I have before me Doug Self's Small Signal Audio Design, 2nd edition. Ch 13 is a quite in-depth investigation into this issue. Quoting from page 353:

congrats

Inexpensive faders are usually made using the same two-slope carbon-film construction as are rotary log volume controls, but the more expensive and sophisticated types use a conductive-plastic track with multiple taps connected to a resistor ladder, as described in the previous section. This allows much better control over the fader law.

High-quality faders typically have a conductive-plastic track, contacted by multiple gold-plate metal fingers to reduce noise during movement.

He goes to show a fader control law of the A20 variety and says Note that a fader does not attempt to implement a linear-in-dB log scale; the attenuation is spread out over the top part of the travel and much compressed at the bottom. This puts the greatest ease of control in the range of most frequent use."

the range of most frequent use is the top 10-15 dB, but customers will notice if the kill at full down is incomplete, and if hop off region is jumpy or not smooth.

So are you saying there's no resistor tapping involved, or that it's less frequently used due to modern manufacturing techniques?

Used by who? I haven't designed a console this century, and my last friend still doing consoles (at APB) died a couple years ago so I can't even ask him (if I cared).

There are some high end premium fader that use multiple screens, taps or whatever... do not look for expensive parts in any reasonable scale manufacturing. I would expect trick faders to become scare as analog consoles become scarce. The economics for full custom parts does not favor low volume SKUs.

Self also spends a lot of time on using tapping and loading to make a linear pot behave in a logarithmic fashion because he states that log pots are not generally very accurate.

but slugging pots is even less accurate. As I have already posted the bulk resistance of screened resistive elements is 20% (because they need that much). The final pot element resistance depends on being cured in an oven after screening. I've seen pot manufacturers screw this up too.

Although he designed large format consoles for Soundcraft and Neve, most of Self's work is with home audio equipment, amplifiers, and preamps, etc. So it's possible that the manufacturing techniques are different due to the economies of scale?

no comment. I've heard about his book(s?), but never read one. Those books did not exist when I was learning and still don't tell the whole story (probably?). There was a good series about console design written by Steve Dove and published in Studio Sound decades ago (or you could read the article I wrote for RE/P back in 1980). I think Steve's series was reprinted in some audio encyclopedia (that also didn't exist when I needed to read one.) Of course don't take my word for anything. 

I believe many of the current digitals on the market use a 10-bit system.

Don't confuse the encoder resolution with the digital domain multiplier resolution. I do not know for a fact but expect fader changes to be smoothed and using full internal 16+ bit multiplier resolution (it would be kind of crazy not to). Of course static gain settings will be limited to interface encoder resolution, but moving between static settings can be smoother. 

That's what's interesting about the Yamaha example I provided- the 0dB region is compressed rather than expanded. Which seems odd.

1 dB level change is barely audible without A/B comparisons. Odb Fader markings are mainly used for lining up channel strips for unity. Other fader changes will be done by ear so don't need clutter in the markings. 

To be clear, I am not making any inferences about sound quality or mixer operation with regards to fader law. I'm just interested in the topic.

Audio fader laws are based on human loudness change perception.

If you want something really interesting look at control laws for everything else inside consoles. Customers make ASSumptions about quality all the time from minor things as unrelated to signal quality as EQ boost-cut speed around flat, etc....  lots of secret sauce under that hood. Glad I am no longer dealing with the phoolish mojo that keeps console customers happy.

JR 

[Frank Koenig]

The other thing I remember doing (we're talking high school here) to get around all this, and the poor gain matching between stereo channels using dual pots, was building an 11 step stereo step attenuator using 1% resistors. 11 step rotary switches were all I could find and I made the steps 3 dB -- too course of course. The last step from -27 to -infinity was a doozy. I got around that with an ~20 dB hi/lo range switch. So it was kind of an 11 speed transmission with a 2 speed transfer case. It had a really cool knob I got off a WW2 radio though. Things got better...

--Frank

[John Roberts {JR}]

The other thing I remember doing (we're talking high school here) to get around all this, and the poor gain matching between stereo channels using dual pots, was building an 11 step stereo step attenuator using 1% resistors. 11 step rotary switches were all I could find and I made the steps 3 dB -- too course of course. The last step from -27 to -infinity was a doozy. I got around that with an ~20 dB hi/lo range switch. So it was kind of an 11 speed transmission with a 2 speed transfer case. It had a really cool knob I got off a WW2 radio though. Things got better...

--Frank

Alps actually sold a stepped attenuator that was a bunch of SMD resistors forming a precise voltage divider, back before SMD were common. Don't know if they still make them... Kind of esoteric and academic for digital audio paths.

JR

[Ron Hebbard]

Alps actually sold a stepped attenuator that was a bunch of SMD resistors forming a precise voltage divider, back before SMD were common. Don't know if they still make them... Kind of esoteric and academic for digital audio paths.

JR

  Mr. Roberts Sir!  While were chatting about attenuators and curves, would you care to say a few words about the 20 Hz. to 20 KHz. range of the rotary control on your Loft tech TS-1 sweepable sine generator and db meter? 
Toodleoo!
Ron Hebbard

[John Roberts {JR}]

  Mr. Roberts Sir!  While were chatting about attenuators and curves, would you care to say a few words about the 20 Hz. to 20 KHz. range of the rotary control on your Loft tech TS-1 sweepable sine generator and db meter? 
Toodleoo!
Ron Hebbard

Actually it was safely more than 20-20k so something like 15Hz to 30kHz... so it could cover 20-20k for measurements. (it actually has an internal trimpot to set the 15Hz endpoint because analog circuitry is not very stable over that much range and without the trim might not start that low, or might be so low it stops oscillating.) I think it worked pretty damn good for a cost effective (cheap for what it was) analog test equipment.

For years (actually decades) I have resisted the temptation to upgrade that design with modern digital technology. Using a rotary digital encoder I could cover the frequency range with one knob easily, even providing adaptive rate of change. Fast when first turned and slower every time you reverse direction. The major problem caused by combining that much adjustment range with a 1Hz resolution frequency counter attached meant that many customers to try to hit 1.000kHz exactly (not trivial with one knob), this is despite the low likelihood of typical stand alone analog sine wave generators putting out exactly 1kHz when they say they do.  This is an example of (failed) human factors engineering in a nutshell.   

I would love to do a digital TS-1 that could be better and cheaper, but I also promised myself to never compete with software smart phone applications that can be cheaply copied (even before there were smart phones). So not gonna happen... 

Toodleoo Ron

JR   

[Nathan Riddle]

Hi y'all-

As an offshoot of this discussion, I wanted to come up with a way to actually measure and plot the fader laws of mixing consoles to see how much variation there is.

In discussing this on the SynAudCon forums, Langston Holland and I conducted the following test:

• Measure length of fader track (usually 100mm on pro desks)
• Divide into 11 segments (every 10mm, 0 through 100mm inclusive) and label a piece of tape
• Stick tape to console, send 1kHz sine wave through channel and record the level at each of the steps

Langston said "I'd use a 1kHz sine wave and a multimeter, bypass EQ, etc. from the measurement path, start out with the fader at max, adjust the source voltage and mic preamp so you get 10V on the consoles output, which is very near clipping so you might have to use a lower max voltage. 10 is an easy number to deal with and you want to be close to clipping so you get the most stable readings (S/N) when measuring the lowest fader positions. "
For those who are members, the thread is here.

So we did that. He used an APB Dynasonics H1020 and I used an X32. Charts below in both linear and log formats. The method and therefore the data are inexact but still interesting.

X axis is fader position in mm, from 0 at the bottom to 100 at the top. Y axis is Volts. I didn't label them because I'm tired and I hate Excel.

Both are pretty close to true log (linear-in-dB) for most of the throw. I'd be interested in seeing results from more desks.

If anyone wants to try this out, maybe we can get a dataset together.

When I get some time next week I'll test my SQ6, QU24, Zed10FX, & small Behringer whatever mixers.

[ProSoundWeb Community]