I recently bought an EAR 834p clone. Sounds quite nice nice, but I am going to upgrade the coupling caps.
I have a bit of an issue with subsonics making their way to my speakers, causing some low frequency woofer movement.
Can I adjust the size of the coupling caps after either the first tube (a 0.1uf instead of a .15uf) or the 3rd tube (.47uf instead of a 1uf) to effectively create a high pass filter on subsonics? Which is the better place to do so?
https://www.vinylengine.com/turntable_forum/download/file.php?id=95993&mode=view
I tried these filters on my interconnects: https://store.acousticsounds.com/d/60042/DB_Systems-DB_Systems_Subsonic_Filter-Turntable_Accessories
They worked to filter the subsonics, but I did not care for how they colored / confised the sound, and they rolled off too much audible bass.
I have a bit of an issue with subsonics making their way to my speakers, causing some low frequency woofer movement.
Can I adjust the size of the coupling caps after either the first tube (a 0.1uf instead of a .15uf) or the 3rd tube (.47uf instead of a 1uf) to effectively create a high pass filter on subsonics? Which is the better place to do so?
https://www.vinylengine.com/turntable_forum/download/file.php?id=95993&mode=view
I tried these filters on my interconnects: https://store.acousticsounds.com/d/60042/DB_Systems-DB_Systems_Subsonic_Filter-Turntable_Accessories
They worked to filter the subsonics, but I did not care for how they colored / confised the sound, and they rolled off too much audible bass.
Last edited:
The first coupling cap is definitely a high pass pole and will effect sub-sonics. The second is more of a shallow shelf, with only a very small effect. I'd worry more that the RIAA 50Hz pole is not defined, except by running out of open loop gain (there's no resistor across the 330pF feedback capacitor). This may or may not be OK, and too hard to guess without a modern LTspice simulation, but a possible issue.
System sub-sonic response depends critically on the cartridge compliance x tonearm effective mass resonance being placed within a narrow range above the warp spectrum and below the music spectrum. It's very, very commonly placed way below optimum (8 to 12 Hz). Electrical filtering can remove the amplitude of the warp signals but cannot remove the time modulation, FM and PM. Probably still worth some effort, but you may want something steeper than a single pole rolloff.
All good fortune,
Chris
System sub-sonic response depends critically on the cartridge compliance x tonearm effective mass resonance being placed within a narrow range above the warp spectrum and below the music spectrum. It's very, very commonly placed way below optimum (8 to 12 Hz). Electrical filtering can remove the amplitude of the warp signals but cannot remove the time modulation, FM and PM. Probably still worth some effort, but you may want something steeper than a single pole rolloff.
All good fortune,
Chris
Changing the first wont work well as it's inside a feedback loop. Changing the second will work well
Depending on the input impedance of the power amp different values will be needed.
The ending resistor ( 220k ) parallell with 220k in the power amp and a 0.1uF will give
a corner of 14hz, that may be what yoy want.
Depending on the input impedance of the power amp different values will be needed.
The ending resistor ( 220k ) parallell with 220k in the power amp and a 0.1uF will give
a corner of 14hz, that may be what yoy want.
Cris read the schematics correct and corrected my erroneous statement 
I am corrected.
The first cap however has a less well defined R part. ( 2M parallell with the NFB )
Maybe adding a cap and resistor after the 1uF is the way to go ? Or a RC network before the
first stage , althogh that would be dependent on cartridge properties.
I am corrected.
The first cap however has a less well defined R part. ( 2M parallell with the NFB )
Maybe adding a cap and resistor after the 1uF is the way to go ? Or a RC network before the
first stage , althogh that would be dependent on cartridge properties.
You're right that the first cap's rolloff frequency is hard to define (without an exact LTspice model of the whole project) because it depends, as you say, critically on two somewhat "soft" resistances, the source impedance from the first stage (resistively degenerated by un-bypassed cathode resistor) and the summing junction's impedance (poorly defined because it depends critically on open loop gain within the feedback loop which is, at best, only slightly higher than closed loop gain). And, there's no well defined 50Hz pole, to make it even more nebulous. If all ECC83 valves were entirely predictable forever it would only be a little easier. But.
So, how could we guesstimate? The first stage's output resistance could be sloppily guessed to be the anode load of 330K in parallel with valve anode "resistance" of 80K plus mu of 100 times cathode resistor of 2K2, so 300K (very rough estimates, but probably good enough for what we're doing. Let's call the resulting source resistance 160K.
On the other side of the coupling cap is the 2M grid leak resistor in parallel with a complex impedance formed by the amplified current loading from the 330pF RIAA feedback cap. How to interpret this? The designer apparently believed that the fed-back stage(s) would run out of open loop gain at 50 HZ, because they included no resistor across the 330pF cap to set the 50Hz RIAA pole (and have also pushed the 500Hz and 2122Hz pole components to compensate).
If the open loop gain were infinite (and therefore there would be a resistor across the 330pF cap to set the 50Hz pole) the impedance at the second stage grid would be zero, and the RC time constant / high pass pole would be determined simply by the coupling cap and the source resistance of maybe 160K Ohms. But it's not even close to infinite; rather it's somewhere much nearer unity, and very dependent on the gain of the particular valve in use. Without feedback, the loading side of the coupling cap sees only the 2M grid leak. Our desired pole is set by the product of the coupling cap with the sum of source (maybe 160K) plus load "resistance" (2M grid leak in parallel with summing junction loading due to feedback). So, again, how do we guess the load resistance without knowing exactly (because details matter here) the fed-back stage(s) gain? Let's take a WAG and call the total (sum of source and load) R 1M Ohm. No good reason to choose this number other than convenience. We don't have any design frequency in mind anyway, other than 25Hz seemed to work OK.
If we believe all of these WAG assumptions, we calculate C=1/2pi(25Hz)(1M Ohm) = 6n36F give or take A LOT. If we don't believe the assumptions, we'd need to either measure accurately or prepare to experiment. A decent starting point might be a common value like .01uF, then twice and then half that. But solving a mechanical problem in the mechanical domain is still the best route, if possible.
All god fortune,
Chris
So, how could we guesstimate? The first stage's output resistance could be sloppily guessed to be the anode load of 330K in parallel with valve anode "resistance" of 80K plus mu of 100 times cathode resistor of 2K2, so 300K (very rough estimates, but probably good enough for what we're doing. Let's call the resulting source resistance 160K.
On the other side of the coupling cap is the 2M grid leak resistor in parallel with a complex impedance formed by the amplified current loading from the 330pF RIAA feedback cap. How to interpret this? The designer apparently believed that the fed-back stage(s) would run out of open loop gain at 50 HZ, because they included no resistor across the 330pF cap to set the 50Hz RIAA pole (and have also pushed the 500Hz and 2122Hz pole components to compensate).
If the open loop gain were infinite (and therefore there would be a resistor across the 330pF cap to set the 50Hz pole) the impedance at the second stage grid would be zero, and the RC time constant / high pass pole would be determined simply by the coupling cap and the source resistance of maybe 160K Ohms. But it's not even close to infinite; rather it's somewhere much nearer unity, and very dependent on the gain of the particular valve in use. Without feedback, the loading side of the coupling cap sees only the 2M grid leak. Our desired pole is set by the product of the coupling cap with the sum of source (maybe 160K) plus load "resistance" (2M grid leak in parallel with summing junction loading due to feedback). So, again, how do we guess the load resistance without knowing exactly (because details matter here) the fed-back stage(s) gain? Let's take a WAG and call the total (sum of source and load) R 1M Ohm. No good reason to choose this number other than convenience. We don't have any design frequency in mind anyway, other than 25Hz seemed to work OK.
If we believe all of these WAG assumptions, we calculate C=1/2pi(25Hz)(1M Ohm) = 6n36F give or take A LOT. If we don't believe the assumptions, we'd need to either measure accurately or prepare to experiment. A decent starting point might be a common value like .01uF, then twice and then half that. But solving a mechanical problem in the mechanical domain is still the best route, if possible.
All god fortune,
Chris
Thank you Chris! Quite insightful, especially your observation around the 50Hz pole and the feedback loop. I am too far removed from my engineering college days to remember how to properly evaluate these circuits across the freq range.
I will try a smaller cap on the first stage, but I likely have to figure out a different solution on how to apply a high pass filter on the turntable output.
Now on to my next part - figuring out where they installed the noble volume pot in this circuit. I believe, based on a first look at the board, that it is in place of the 220k resistor at the output. If I test the output impedance in ohms with my multimeter, I get anywhere from single digit ohms at one end of the pot range, up to 47k ohms at the other end. Adjusting it down and turning up the volume on my preamp, resulted it pretty similar sound, and did not affect the mechanical woofer low frequency oscillation.
I will try a smaller cap on the first stage, but I likely have to figure out a different solution on how to apply a high pass filter on the turntable output.
Now on to my next part - figuring out where they installed the noble volume pot in this circuit. I believe, based on a first look at the board, that it is in place of the 220k resistor at the output. If I test the output impedance in ohms with my multimeter, I get anywhere from single digit ohms at one end of the pot range, up to 47k ohms at the other end. Adjusting it down and turning up the volume on my preamp, resulted it pretty similar sound, and did not affect the mechanical woofer low frequency oscillation.
You might be better off by adding the highpass after the volume pot, this is (almost)outside the NFB thus
unaffected by tube/component properties and only depending on the input impedance of the following stage.
A better way would be to slightly change the topology :
Decouple the highpass from NFB, move the NFB connection from the cathode
to the grid using a somewhat smaller cap ( 0.22 ?) . That way the 1 uF cap could be replaced without
affecting the NFB. This would make NFB unaffected by changes at the cathode followers output.
unaffected by tube/component properties and only depending on the input impedance of the following stage.
A better way would be to slightly change the topology :
Decouple the highpass from NFB, move the NFB connection from the cathode
to the grid using a somewhat smaller cap ( 0.22 ?) . That way the 1 uF cap could be replaced without
affecting the NFB. This would make NFB unaffected by changes at the cathode followers output.
Peter - That is interesting. I experimented today. Removed the small 1uf caps and replaced with Obbligao .47uf caps. Nice sound, but too rolled off at the bottom end without fixing the subsonic issues that pop up on certain records. I then tried wiring the .47uf caps in line in some cheap interconnects as outs from the phone to the preamp. That seemed to work to reduce the subsonics. Better sound than the DB Filters, but not quite there yet
I like your idea of moving the nfb from the cathode to the grid and reducing the cap size for the filter. i will review the board topology and see if I can do that easily.
I like your idea of moving the nfb from the cathode to the grid and reducing the cap size for the filter. i will review the board topology and see if I can do that easily.