I have a Sanusi BA-2000 that I like a lot that I have recapped and refurbished a bit but to me there is a perceivable lack high frequency content compared to other amps. It seems like this is a consensus among other owners. The stated specs say otherwise and that it's response is pretty much flat in the audio band. I was hoping to dive into the circuit a bit to try and find what part of the design could be affecting this. My initial thought is that there are some frequency / phase compensation circuits affecting this and perhaps they could be modified.
On the driver board, I'm looking at (guessing!) C03 is (33pf) is part of a low pass filter (maybe this value could be lowered to 22pf). C29 and 31 as miller compensation and C33 and C19 as low pass filter and compensation. Can adjusting the miller compensation caps have an effect in the audible spectrum? Also, could I lower the value of c33 and bleed off less high end? There also is a capacitor C55 that appears to be hanging off the feedback path as well.
I'm just taking wild guesses here so if there's any way to get to the bottom of why this amp seems to sound a bit dull and ways to change that, I'm all ears.
On the driver board, I'm looking at (guessing!) C03 is (33pf) is part of a low pass filter (maybe this value could be lowered to 22pf). C29 and 31 as miller compensation and C33 and C19 as low pass filter and compensation. Can adjusting the miller compensation caps have an effect in the audible spectrum? Also, could I lower the value of c33 and bleed off less high end? There also is a capacitor C55 that appears to be hanging off the feedback path as well.
I'm just taking wild guesses here so if there's any way to get to the bottom of why this amp seems to sound a bit dull and ways to change that, I'm all ears.
There are three caps that you could play with.
I'd start by reducing the C35 from 22pF to 10pF. That should do it.
There's also C45 (3.9pF)... but it is so low in value that you could only remove it... which could cause instability at very high frequencies.... maybe... maybe not.
And then there's that C33 (220pF)... ouch. See if you can find 180pF or 150pF caps... and try.
I'd use the oscilloscope at the output and a bit of a capacitive load (together with a resistive load that mimics your speaker impedance) at the end of the actual speaker cables you intend to use; look for no oscillations at the output with a shorted input.
The caps should be polystyrene. Do not use disk ceramics and do not use silvered mica.
I think a combo of 10pF for C35 and 180pF for C33 should do the trick, while still preserving very good stability (which does not mean that you should not check for oscillations with an oscilloscope). If that is not enough high frequencies, sparkle and space at the top end... try 150pF for C33.
The above changes will definitely affect the phase margin... but I do not see any other solution if you want to bring some top end into play.
Actually, there is another one... you could try with much better quality power rails' decoupling caps (C41 & C43, C51 & C53, and to some extent the C13)
I'd start by reducing the C35 from 22pF to 10pF. That should do it.
There's also C45 (3.9pF)... but it is so low in value that you could only remove it... which could cause instability at very high frequencies.... maybe... maybe not.
And then there's that C33 (220pF)... ouch. See if you can find 180pF or 150pF caps... and try.
I'd use the oscilloscope at the output and a bit of a capacitive load (together with a resistive load that mimics your speaker impedance) at the end of the actual speaker cables you intend to use; look for no oscillations at the output with a shorted input.
The caps should be polystyrene. Do not use disk ceramics and do not use silvered mica.
I think a combo of 10pF for C35 and 180pF for C33 should do the trick, while still preserving very good stability (which does not mean that you should not check for oscillations with an oscilloscope). If that is not enough high frequencies, sparkle and space at the top end... try 150pF for C33.
The above changes will definitely affect the phase margin... but I do not see any other solution if you want to bring some top end into play.
Actually, there is another one... you could try with much better quality power rails' decoupling caps (C41 & C43, C51 & C53, and to some extent the C13)
I recommend you test frequency response before changing components. Perhaps there's a failed part that needs replacement. If the frequency response is flat, I'd leave the amp alone and use tone controls to adjust sound to your preference. If you want to experiment, just beware Boky's cautions.
Good luck!
I agree with avoiding use of ceramic caps anywhere in the signal path except for COG/NPO dielectric. This class of ceramic caps offers superlative performance and are used as the tuning cap of preference in extremely low distortion oscillators.
Good luck!
Thanks all, I'll report back as I test or make any changes. Electrolytic decoupling caps have already been upgrades to high quality, low esr types with slightly increased values. I'm not sure why polystyrene would be necessary over c0g as those capacitors are not in the audio path directly.
Last edited by a moderator:
Many vintage Sansui amps, including BA-2000, used fusible resistors in many locations. These fusible resistors are a known issue. Often they will increase significantly in resistance value. I removed fusible resistors from an AU-317 that were originally 150 ohms, but their values had all risen, a couple of them to over 2k ohms, and one was over 80k ohms!
https://audiokarma.org/forums/index.php?threads/fuse-resistors-the-wheres-whys-and-hows.430439/
This might be something to consider.
Pic of BA-2000 fusible resistors:
https://audiokarma.org/forums/index.php?threads/ba-2000-on-the-bench.694143/#post-9279788
https://audiokarma.org/forums/index.php?threads/fuse-resistors-the-wheres-whys-and-hows.430439/
This might be something to consider.
Pic of BA-2000 fusible resistors:
https://audiokarma.org/forums/index.php?threads/ba-2000-on-the-bench.694143/#post-9279788
Last edited:
Polystyrene won’t be THAT inductive in these cap values. They are fine, but you have to be careful because they are fragile. Too much heat from the iron and they short. Leads tend to break off if the cap isn’t properly supported. Silver micas can practically be beat on with a hammer and nothing happens to them. Hold the soldering iron right up to the coating on the cap and it doesn’t melt.
Silver micas have high ESR compared to other comparable types. For compensation caps in audio that normally does not matter. Look at the impedances in the circuit. They are are high. When looking at RF circuits in the 50 ohm range (plus or minus a decade, maybe) the Q can be severely limited. If a cap needs to get down in the tenths of ohms at high frequency then they are not good. But even in the Miller cap position in most audio amps the ESR doesn’t hurt a thing. Distortion from dielectric absorption is not zero, but the small amount is swamped up by the nonlinear Cob of the transistor it’s hanging off of. Nowhere near the massive distortion produced by class 2 ceramics with high voltage swings.
Silver micas have high ESR compared to other comparable types. For compensation caps in audio that normally does not matter. Look at the impedances in the circuit. They are are high. When looking at RF circuits in the 50 ohm range (plus or minus a decade, maybe) the Q can be severely limited. If a cap needs to get down in the tenths of ohms at high frequency then they are not good. But even in the Miller cap position in most audio amps the ESR doesn’t hurt a thing. Distortion from dielectric absorption is not zero, but the small amount is swamped up by the nonlinear Cob of the transistor it’s hanging off of. Nowhere near the massive distortion produced by class 2 ceramics with high voltage swings.
I don’t see anything in the circuit that would obviously limit the audio frequency bandwidth. Perhaps you could check the frequency response of the amp it is being compared to. It may exhibit ultrasonic peaking making it sound “airy” or “extremely detailed”. Just a simple 1kHz small signal square wave is very telling - if there is overshoot or ringing at the leading/trailing edges is indicative of peaking. Even if the peak is waaay above 20kHz it can subjectively affect the sound (It will affect the IM distortion spectrum, and that is measurable). Also, if the Sansui has severely rounded edges to its square wave response it probably is rolling off too soon. If that happens I’d be looking for resistors that shifted or electrolytics that have gone bad. From what I see in the schematic that shouldn’t be happening.
https://www.audiovintage.fr/2014/08/12/sansui-ba-2000/
An obviously gray and flat, powerless, disembodied sounding design.
An obviously gray and flat, powerless, disembodied sounding design.
C0G types have very little voltage coefficient of capacitance (translation: low distortion) and can be very low loss. Better in many respects than silver mica. Capacitance density isn’t as high - getting 200+ volt parts or >100 pF can be like pulling teeth. 500 volt micas up to 10,000 pF are easy (maybe not cheap, but easy). The ceramics are also more piezoelectric than micas - external mechanical stresses have more effect electrically. Polystyrenes are the best of the bunch - just damn easy to break. I tend to use silver micas for my “anti-oscillation-caps” (I’m often running 200+ volt supplies, even with solid state amps and always with tubes) and polystyrene where it just performs filtering. Stuff that’s not critical gets ceramic throughout.
Thank you @wg_ski
A follow-up question if I may. I restore vintage amps where typically ceramics are used as Miller caps on the VAS, NFB and elsewhere on the amp board in (what I believe to be) "anti-oscillation" roles. So I am happy replacing those original ceramics with modern C0G types.
But my question is regarding how high to go with the voltage ratings. As an example, lets say an amp has rails of +50V and -50V. What minumum voltage rating(s) would you suggest for the VAS Miller cap and NFB (cap across the larger resistor) for modern C0G replacements in that amp?
A follow-up question if I may. I restore vintage amps where typically ceramics are used as Miller caps on the VAS, NFB and elsewhere on the amp board in (what I believe to be) "anti-oscillation" roles. So I am happy replacing those original ceramics with modern C0G types.
But my question is regarding how high to go with the voltage ratings. As an example, lets say an amp has rails of +50V and -50V. What minumum voltage rating(s) would you suggest for the VAS Miller cap and NFB (cap across the larger resistor) for modern C0G replacements in that amp?
From the voltages shown on p13 of the manual, I surmise that it runs quite low bias, if any, on the output stage.
Cranking that up a little might be worth trying.
Cranking that up a little might be worth trying.
The difference of the rails is 100V, so use that as the minimum voltage rating.
It may be related to input drive impedance.
It should be tested with a sine wave at 20khz
With a low Z source and then put a 15k resistor is series with the signal source. If it drops off it is input Z not being consistent across the audio band.
It should be tested with a sine wave at 20khz
With a low Z source and then put a 15k resistor is series with the signal source. If it drops off it is input Z not being consistent across the audio band.