The pots I measured are regular consumer-grade, carbon track types from well-known manufacturers: CTS, Radiohm, Piher, MBLE,.....
Something important to be noted, in my test setup, only the contact-resistance is measured, in order to magnify any possible problem.
In a normal utilization, the figures would be down by 40~60dB, but if they are used in "creative" circuits heavily loading the wiper, the reduction could be only 20dB for instance.
I concentrated on the wiper to track resistance, because it is a relatively low-hanging fruit, but the track resistance itself had already been identified in several studies as significantly non-linear, because the track is in essence a thick-film resistor.
For me, this means that in order to match (and preferably exceed) the sub-ppm performance of modern amplifiers, you need to use pots in a purely potentiometric way (or turn to another solution)
I don´t focus on contact resistance, distortion, etc. because the main problem is another: over loaded linear pots are not and will bever be log pots, period, theyare at best a crude approximation and in only a couple cases, and at worst they are a mess.:
1) they "work" only in voltage divider circuits ... i.e. "passive volume pots"
2) they heavily load the driving circuit, by showing 5X to 10X *lower* impedance than nominal, a BIG consideration
3) said impedance varies all the time with pot rotation.
4) similar to a broken clock which gives the right time twice a day, a a simulated Log pot matches the equivalent Log attenuation at only *one* setting (besides tivial 0 and 10)
5) of course they do NOT work properly as 2 terminal gain controls in an NFB loop.
NONE of the above problems is shown by true Log pots, or even the Commercial version made by using two 5:1 to 10:1 linear tracks inside the pot.
Difference in performance is so large that I always walk the extra mile and get the proper Log pots, which to boot cost the same as Linears or cents more.
1) they "work" only in voltage divider circuits ... i.e. "passive volume pots"
2) they heavily load the driving circuit, by showing 5X to 10X *lower* impedance than nominal, a BIG consideration
3) said impedance varies all the time with pot rotation.
4) similar to a broken clock which gives the right time twice a day, a a simulated Log pot matches the equivalent Log attenuation at only *one* setting (besides tivial 0 and 10)
5) of course they do NOT work properly as 2 terminal gain controls in an NFB loop.
NONE of the above problems is shown by true Log pots, or even the Commercial version made by using two 5:1 to 10:1 linear tracks inside the pot.
Difference in performance is so large that I always walk the extra mile and get the proper Log pots, which to boot cost the same as Linears or cents more.
Mr Fahey
Please enlighten us where you can get accurately logarithmic pots for "pennies more" than linear ones.
The (slightly) variable loading of a padded linear pot has no appreciable effect when used in a correctly designed circuit. Sadly, the lack of basic electronic knowledge evinced by many of the posters to this forum suggests that there's probably little chance of good quality design around here.....
I have designed many commercial products - some of which are "industry standards" - and many of them use padded linear potentiometers in an effort to get closer to the ideal of dB-calibrated attenuation for faders and other controls.
Some circuit configurations require other pot laws - reciprocal, square, and sometimes just plain eccentric - and all of these (and more) can be achieved by loading a linear pot.
The reality of commercial electronic design demands that the cheapest components that we can get away with are used in manufacture. On many occasions - just for amusement (though it has been demanded by a couple of clients at times) - I, or one of my team, have assembled one of our products using "audio grade" components. This unit is then rigorously tested against a production-line "standard" product. In no instance have we found the "special" unit to perform significantly better than the standard production unit, despite it costing up to 10 times as much in component parts! This shows that correctly designed electronics - taking into account things like component tolerances and batch differences - doesn't require "special" components.
Still - we're trying to deal with people who believe that the colour of the sleeving on their speaker cables affects the way their system "sounds".....
Please enlighten us where you can get accurately logarithmic pots for "pennies more" than linear ones.
The (slightly) variable loading of a padded linear pot has no appreciable effect when used in a correctly designed circuit. Sadly, the lack of basic electronic knowledge evinced by many of the posters to this forum suggests that there's probably little chance of good quality design around here.....
I have designed many commercial products - some of which are "industry standards" - and many of them use padded linear potentiometers in an effort to get closer to the ideal of dB-calibrated attenuation for faders and other controls.
Some circuit configurations require other pot laws - reciprocal, square, and sometimes just plain eccentric - and all of these (and more) can be achieved by loading a linear pot.
The reality of commercial electronic design demands that the cheapest components that we can get away with are used in manufacture. On many occasions - just for amusement (though it has been demanded by a couple of clients at times) - I, or one of my team, have assembled one of our products using "audio grade" components. This unit is then rigorously tested against a production-line "standard" product. In no instance have we found the "special" unit to perform significantly better than the standard production unit, despite it costing up to 10 times as much in component parts! This shows that correctly designed electronics - taking into account things like component tolerances and batch differences - doesn't require "special" components.
Still - we're trying to deal with people who believe that the colour of the sleeving on their speaker cables affects the way their system "sounds".....
Don´t add straw man arguments, please, I am talking *commercial grade* Log pots as (hate to self quote, but if you selectively read
)>>"the Commercial version made by using two 5:1 to 10:1 linear tracks inside the pot."<< which is what any mainstream Supplier (think Mouser, Farnell, Digikey, etc.) will supply you, what major Commercial Manufacturers use, etc.
I can buy "just now" linear pots from, say, 100 suppliers, while maybe 10 or 20 of them might have Log ones *in stock* (in any case all others can order them for me, specially since I am a bulk quantity buyer).
Basic pot price is the same, Log or Lin, but since Log ones are less popular my shopping selection is somewhat more restricted so I often end up paying a few cents more.
Next?
You are completely off track here and I wonder how anybody with the most basic knowledge of Ohm´s Law or how resistors combine might say that.
Let´s compare side by side 2 pots, a commercial Log one and a commercial Linear one.
Please don´t try to kick the ball outside the field by referring to Lab equipment or "ideal" pots and such .... conditions the padded linear does not meet anyway
Let´s take 2 100k pots, one of each type, set each to, say, 0, 5 and 10 on a 0-10 scale, and compare attenuation and load seen by driving stage.
* 100k Log
setting....0.........5.........10
output....0........-20dB....0dB
load.......100k.....100k....100k <--- pay attention to this
* 100k LIN with a 5k6 resistor from wiper to Ground.
And why 5k6 and not other value?
Because it can provide classic "10% Log attenuation" when set to 5, like any plain vanilla Log pot can.
A little Math, to save you the trouble of checking my choice:
a) a 100k Lin pot set to 5 will have both sections (ground to wiper and wiper to hot) same resistance: 50k each.
b) if I want to have 10% of the original signal at that point, lower section must show 10% resistance of upper section, hence, around 5k.
5k6 in parallel with 50k is 5.04k , so a 5k6 resistor meets the classic Log attenuation value.
So:
setting...0.........5..........10
output...0.......-20dB.......0dB
load......100k....55k.........5.3k <--- pay attention to this
If you think loading previous stage with 5k3 instead of 100k means
I guess you are set in for a rude awakening.
And if you choose a higher loading resistor value to minimize the heavy loading effect, then you are NOT meeting the most basic Log pot spec.
So basic that it´s usually shown in datasheet or description (i.e.10% Log, 30%Log, etc.)
Speak about yourself and don´t insult qualified Forum Members just to try to prop yourself higher ... the exact opposite happens.
A padded Lin pot meets proper Log curve only at *one* point, pick one, I made it meet on 5 simply because it´s a popular spec, choose another if you will, it won´t change the reality that it will not anywhere else ... except at trivial solutions 0 and 10 , of course, where *all* pots meet, no matter what curve they follow.
Reminds me of the broken clock, which "gives you the correct time twice a day"
No, you can´t turn one curve into another, all you can do is meet it at *one* point.
Oh, now you are showing your true colours
WHO mentioned "Audio Grade" components?
I am a long standing Manufacturer, for almost 50 years now, as pennywise as anybody or wouldn´t have lasted that long, and since true Log pots (I am talking Commercial grade, as always) cost same or pennies more than poorly performing FAKE ones, why bother?
It ends up costing *less* , go figure.
And as mentioned before, can used anywhere without special precautions, while FAKE Log ones can not, unless you drive them from a low source impedance , sometimes requiring a dedicated buffer, and you can not use them in a "floating" application, such as in an Op Amp NFB net.
Which I use a lot, specially in variable gain Preamp and Mixer input stages, Mixer channel mixing Master stages, Frequency setting in Equalizers, etc.
So why settle for the inferior or plain unusable kludge when I can use the real one?.
WHO mentioned "Audio Grade" components? [2]
WHO mentioned "Audio Grade" components? [3]
As a side note, I have personally manufactured Potentiometers at the SEIWA Factory, trying prototypes until I found what I needed, including harder to find "S" curve ones (Log from 0-5 , Anti Log from 5-10) , different % Log ones, of course anti-Log ones, the works.
They let me experiment using their originally Philips Eindhoven equipment, dies, etc. , as well as original Philips resistive paint (which when oven cured leaves behind the carbon track).
Doubt anybody here can match that actual, hands on experience.
Funnily enough, I worked for Philips at Eindhoven for a couple of years when I left University in the 1970s.
A reciprocal law pot (a padded linear one) is significantly closer to true logarithmic than the usual two-taper "logarithmic" pots found from most manufacturers. There is sometimes a requirement to provide precise dB attenuation calibrations on a control - the only real way to do that is to use a rotary switch connecting accurate resistive pads!
You sneer at my "penny pinching" approach to design. This is usually what the customer demands, irrespective of the "quality" of the product! The Music Industry is the worst for this approach. I'm glad that most of my design effort goes into high quality broadcast equipment these days.
A reciprocal law pot (a padded linear one) is significantly closer to true logarithmic than the usual two-taper "logarithmic" pots found from most manufacturers. There is sometimes a requirement to provide precise dB attenuation calibrations on a control - the only real way to do that is to use a rotary switch connecting accurate resistive pads!
You sneer at my "penny pinching" approach to design. This is usually what the customer demands, irrespective of the "quality" of the product! The Music Industry is the worst for this approach. I'm glad that most of my design effort goes into high quality broadcast equipment these days.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.