The log faking resistor does not help solve the very low attenuation problem.
The log faking creates an S shaped control law and the very bottom of the S becomes quite steep. The opposite to what is wanted.
The log faking ONLY gives an improvement over part of the control rotation range. That range does NOT include the bottom of the rotation.
The log faking creates an S shaped control law and the very bottom of the S becomes quite steep. The opposite to what is wanted.
The log faking ONLY gives an improvement over part of the control rotation range. That range does NOT include the bottom of the rotation.
I don't follow.
10dgrees of rotation at the top end might reduce volume by 3dB. It would be nice that 10° at the middle would also reduce volume by 3dB.
And great if 10° near the bottom would also reduce volume by 3dB.
That would be constant dB for equal rotation.
For a full 270° of rotation one would ideally have 81dB of attenuation, i.e. ~1db/3° of rotation.
NFB vol pot, log faking and conventional never get close to that ideal.
10dgrees of rotation at the top end might reduce volume by 3dB. It would be nice that 10° at the middle would also reduce volume by 3dB.
And great if 10° near the bottom would also reduce volume by 3dB.
That would be constant dB for equal rotation.
For a full 270° of rotation one would ideally have 81dB of attenuation, i.e. ~1db/3° of rotation.
NFB vol pot, log faking and conventional never get close to that ideal.
Not dismissing the whole project.
Just the range of attenuation/rate.
If I listen at -30dB most of the time and use -50db at many other times, then +95db is not in my range and -122db is also not in my range.
If what I would "just accept" compresses all my listening into 4° of rotation, then the pre-selected range is not acceptable.
It is wise to check before one builds.
It is a nonsense to build without making any informed estimates and then come back saying "my vol pot law does not work. What have I done wrong?"
Just the range of attenuation/rate.
If I listen at -30dB most of the time and use -50db at many other times, then +95db is not in my range and -122db is also not in my range.
If what I would "just accept" compresses all my listening into 4° of rotation, then the pre-selected range is not acceptable.
It is wise to check before one builds.
It is a nonsense to build without making any informed estimates and then come back saying "my vol pot law does not work. What have I done wrong?"
B.P says "no added resistors" they ruin the vol pot.
Last edited:
OK since nobody stepped up, I did a quick one to get some ballpark numbers. I assumed a dual linear pot with 5% mismatch between track resistance, I think that's realistic looking at some datasheets. Let's say one track is 10,000 ohms, the other is 10,500 ohms.
As I understood the suggestion, you would use say 500 ohms in series with the higher track for a total of 11,000 ohms, and 1000 ohms in series with the lower valued one for also 11,000 total.
Now if you turn it all the way towards the series R's, you have an att ratio of 500/10,500 and 1,000/10,000 which my Radio Shack calculator tells me is -26.85dB versus -20dB. So your channel balance is shot.
Now lets go to mid point rotation of the pot. Att values now are 5,250/11,000 and 5,000/11,000 or respectively -6.42dB and -6.85dB. Better, but not great.
So I don't see the advantage here, quite the opposite.
But maybe I misunderstood, and if so I'd look forward to a corrected calculation.
Jan
to maintain a 10:1 ratio the 10k needs a 1k0 added resistor.
The 10k5 needs a 1050r resistor.
At the top end stop the NFB gain becomes 10000/1000 for one channel and 10500/1050
Both maximum gains are identical @ +20dB
Now move the VR down to 95% of track value. One will measure 9500r and the other will measure 9975 ±5% .
The gains will be 9500/(1000+500) = 6.3333times (+16.0326dB) and 9975/(1050+525) = 6.3333times (+16.0326dB) and all the other for the ±5% potential difference between the tracks. If the tracks follow each other and still maintain the 5% difference then the balance between the channels is maintained.
Without the added resistor, the 95% gains will be 9500/500 = 19times (+25.5751dB) and 9975/525 = 19times (+25.5751dB)
This time the exact tracking maintaining the 5% difference results in exactly the same balance as with the added resistor.
The clever bit that might show up in excel is when you add on the new gains for the limits @ +5% and +4% and compare the two limits for the added resistor and without added resistor.
At the top end of the gain range the added resistor will make the balance error smaller for the small difference in tracking of the two parallel tracks.
The 10k5 needs a 1050r resistor.
At the top end stop the NFB gain becomes 10000/1000 for one channel and 10500/1050
Both maximum gains are identical @ +20dB
Now move the VR down to 95% of track value. One will measure 9500r and the other will measure 9975 ±5% .
The gains will be 9500/(1000+500) = 6.3333times (+16.0326dB) and 9975/(1050+525) = 6.3333times (+16.0326dB) and all the other for the ±5% potential difference between the tracks. If the tracks follow each other and still maintain the 5% difference then the balance between the channels is maintained.
Without the added resistor, the 95% gains will be 9500/500 = 19times (+25.5751dB) and 9975/525 = 19times (+25.5751dB)
This time the exact tracking maintaining the 5% difference results in exactly the same balance as with the added resistor.
The clever bit that might show up in excel is when you add on the new gains for the limits @ +5% and +4% and compare the two limits for the added resistor and without added resistor.
At the top end of the gain range the added resistor will make the balance error smaller for the small difference in tracking of the two parallel tracks.
It's just maths.
As long as the pot is linear, then the differential of the change in ratio between each side of the wiper will be a constant (i.e. change will be linear).
Start adding resistors here and there and you move away from simple maths and into the realms of reality and ohms law with the actual resistance values then coming in to play with the ratio. Begin adding +n to one side of the ratio equation and again you've lost the linearity of the differential.
That's how I immediately see it/feel it anyhow... but I in no way have any EE knowledge, I just follow instructions, a kit builder
For those who want precise gain control and tracking (L gain = R gain), would it not be easier to have a staged set of pots / gains controlling the feedback? By that, I mean two or three sets of pots which switch from one to another via some good relays where each attenuation configuration is used at its optimum sweet spot in terms of control and linearity?
By that I mean having one circuit for very low volumes / high level of attenuation which is designed to give you a smooth, linear increase and not an S-curve that seems to plague the "one pot suits all levels" scenario? Then switch to a more generic circuit from say -40 to -10 so you have good control at moderate volume levels and then maybe a third which controls the -10 to +whatever limit you want to set yourself?
The trick then would be controlling the switchover points and trying to get them to match so you don't have volume jumps as you go from one to another. But if that's done at an infrequently used volume level, that's not so critical?
From my experience with old amps with volume attenuation, I've frequently heard mismatched gains between left and right at low volumes. Often, one speaker will be clearly audible and nothing from the other until you get to a point when the latter will "catch up".
I do also think that having precise tracking at moderate to high volume levels is less important as you simply don't notice it as much.
I've set myself an extra challenge of seeing how Bruno's design will work with an LDR-based attenuation control. I'm probably unrealistic as they tend to be super-log rather than linear, and reading Bruno's posts, his measurements were actually with an incorrectly inserted log pot, not a linear one so linear results should be even better...
By that I mean having one circuit for very low volumes / high level of attenuation which is designed to give you a smooth, linear increase and not an S-curve that seems to plague the "one pot suits all levels" scenario? Then switch to a more generic circuit from say -40 to -10 so you have good control at moderate volume levels and then maybe a third which controls the -10 to +whatever limit you want to set yourself?
The trick then would be controlling the switchover points and trying to get them to match so you don't have volume jumps as you go from one to another. But if that's done at an infrequently used volume level, that's not so critical?
From my experience with old amps with volume attenuation, I've frequently heard mismatched gains between left and right at low volumes. Often, one speaker will be clearly audible and nothing from the other until you get to a point when the latter will "catch up".
I do also think that having precise tracking at moderate to high volume levels is less important as you simply don't notice it as much.
I've set myself an extra challenge of seeing how Bruno's design will work with an LDR-based attenuation control. I'm probably unrealistic as they tend to be super-log rather than linear, and reading Bruno's posts, his measurements were actually with an incorrectly inserted log pot, not a linear one so linear results should be even better...
I want to build this preamp but I cannot find sufficient performance data. I would like to ask you if you guys know:
- The input / output impedance of this preamp?
- The input / output maksimum signal RMS voltage levels?
- Input noise equivalent / output noise level for specified gain or input voltage?
- Power transformer windings voltage (I know last question is silly but I really have a concerns here)
I need this preamp as a linking stage between unbuffered UCD400OEM amplifiers with 9V rms input / 1.8kOhm input impedance and Soekris R-2R DAC with single-ended 1.4V rms output / 625 Ohms output impedance.
Thank you very much in advance.
- The input / output impedance of this preamp?
- The input / output maksimum signal RMS voltage levels?
- Input noise equivalent / output noise level for specified gain or input voltage?
- Power transformer windings voltage (I know last question is silly but I really have a concerns here)
I need this preamp as a linking stage between unbuffered UCD400OEM amplifiers with 9V rms input / 1.8kOhm input impedance and Soekris R-2R DAC with single-ended 1.4V rms output / 625 Ohms output impedance.
Thank you very much in advance.
Yes I have read them all but I cannoct deduct from it if the preamp can drive a load with higher than 6,9 Vrms signal level. It is said that differential amplifier section is clipping at this level.
There is also no information if it can comfortably provide +20dB voltage gain into difficult load of UCD amplifier core.