I could always detect the HF roll-off with true NOS DAC and 44.1kHz sampled material. The measured performance was some 3.5 dB down at 20kHz. But, this should not cause the reported lifelessness.
I think that the lifelessness comes from overprocessed gain structure and complex compensation to provide stability. Also, that bootstrap execution looks like positive feedback to me, which in turn demands even higher stability measures.... just to prevent oscillations. But, only one half of the sinewave is bootstrapped..?? No? It would be interesting to load the amplifier with various types of load (R/L/C) and look/check for the symmetrical or non-symmetrical clipping/behaviour... These loads should be connected with the intended speaker cables in place... to count for the cables' capacitance/inductance.
I think that the lifelessness comes from overprocessed gain structure and complex compensation to provide stability. Also, that bootstrap execution looks like positive feedback to me, which in turn demands even higher stability measures.... just to prevent oscillations. But, only one half of the sinewave is bootstrapped..?? No? It would be interesting to load the amplifier with various types of load (R/L/C) and look/check for the symmetrical or non-symmetrical clipping/behaviour... These loads should be connected with the intended speaker cables in place... to count for the cables' capacitance/inductance.
I'm going to assume the frequency plot is an electrical measurement and address the droop in gain at 20kHz--- and remain without opinion re cause of lifelessness.
I recommend some probing with a 'scope--- gain loss of 3dB should be easily observable. I share suspicion about the absence of a resistor shunting L3 but a check with the scope should reveal if that's an issue. Likewise, checking flatness of response at C31 should resolve that possibility. Be sure to use a X10 probe to ensure minor added capacitive loading when measuring.
I believe this amp has "composite" architecture: note that from Q10 base to amp output at R45 and R46, the amp has nominal gain of (1+R17/R16) or about 31.2 V/V. This sets low frequency gain of the inner amp. The global gain is unity at DC and (1+R24/R25) or about 47.4 V/V. [I don't understand the purpose of R18 but it's so large re (R25+R16) that I think it has small effect.]. So I agree that there's DC servo behavior but also believe opamp Z2 performs distortion reduction at audio frequencies. C10 ensures global stability around the composite amplifier and its connection at Q12 is conservative with respect to stability; I suspect distortion performance might be improved if it were connected across R24, but stability caution is advised.
If measurements indicate the composite amp compensation is at fault, I can offer thoughts about how to tweak cap values.
I recommend some probing with a 'scope--- gain loss of 3dB should be easily observable. I share suspicion about the absence of a resistor shunting L3 but a check with the scope should reveal if that's an issue. Likewise, checking flatness of response at C31 should resolve that possibility. Be sure to use a X10 probe to ensure minor added capacitive loading when measuring.
I believe this amp has "composite" architecture: note that from Q10 base to amp output at R45 and R46, the amp has nominal gain of (1+R17/R16) or about 31.2 V/V. This sets low frequency gain of the inner amp. The global gain is unity at DC and (1+R24/R25) or about 47.4 V/V. [I don't understand the purpose of R18 but it's so large re (R25+R16) that I think it has small effect.]. So I agree that there's DC servo behavior but also believe opamp Z2 performs distortion reduction at audio frequencies. C10 ensures global stability around the composite amplifier and its connection at Q12 is conservative with respect to stability; I suspect distortion performance might be improved if it were connected across R24, but stability caution is advised.
If measurements indicate the composite amp compensation is at fault, I can offer thoughts about how to tweak cap values.
Last edited:
Ok, the way I test is confusing, as it's pink noise and 48th octave analysis of the 8 Ohm loaded amplifier output. I wish my speakers made a plot like that with a mic...
Ok, I wonder what aspect it is then. It's a 100WPC audio amp and presents a night and day sound difference, compared to another 100WPC Kenwood stereo amp - the one that measures flat to 20k.
This is my garage system. There's a pair of JBL HLS 610s hanging from the rafters, which are normally nice and bright and clear as a bell. One of my "DAC"s is an old M-Audio NRV10 mixer connected to the PC via firewire. For whatever reason I chose channels 7-8 for the PC to output to. First thing I did was grab a handful of 10K eq for those channels; brightened, but didnt clear the fog and still sounds nothing clarity wise like the Kenwwod and most other amps through the HLS if I recall correctly. Tried other DACs via USB (M-Audio, Focusrite Scarlet solo) no difference in character.
So, what else is there about a big SS amp that makes it as dull (i.e. markedly unenjoyable but not distorted) sounding, compared to many other nothing special units to the same speakers? I did think about my hearing and was quite surprised that I actually "could hear the measurement". I'll agree it could be not that at all. So, what then? It's a 20+ year old unit, with the signal going though at least 7 sets of caps that have some effect on how it sounds.
Took a moderna boost yesterday and it's just moppin the floor with me today, after a restless night. Pardon my lack of enthusiasm at the moment - I am interested in learning why it's the way it is (Amplifier 101) and fixing it, if possible. If not, it's my last ebay boat anchor which is just sooo too bad. Seller will get + feedback as this amp is what it is and everything works right, except how it sounds to me. It's something anyone could hear IMHO, not just a fellow audiophile.
You can perform one test . Perform frequency response test at different volume potentiometer positions . Worst case is when potentiometer is electrically in the middle (because logarithmic, it will be not in the middle position visually ),same resistance from output to input and to ground .
In every position you test , set volume pot on source (PC or tone generator ) to such level ,that you get identical output to speaker at 1kHz .
If you get different frequency responses - its input filters fault , need to adjust values , if response same - then maybe other parts related, maybe by inductor not being shunted with resistor in parallel .So more tests !
In every position you test , set volume pot on source (PC or tone generator ) to such level ,that you get identical output to speaker at 1kHz .
If you get different frequency responses - its input filters fault , need to adjust values , if response same - then maybe other parts related, maybe by inductor not being shunted with resistor in parallel .So more tests !
I would try to get the output response as flat as possible with an external EQ. I don't believe the -3db decrease can change the sound that much but I can't hear the amp so I can't make a strong argument 😀. If the FR is flat on the output and it still sounds dull compared to another amplifier that's equally flat, I don't think changing the internals to get it flat will change the sound either.
Hope you get better soon,
Hope you get better soon,
I believe its C31 being the main culprit. C16 and C17 are also pretty big. Only maybe 22pf needed for stability.
I'd just replace C31 with 33pf and call it a day.
-3 dB @ 20k is very noticeable further down the path and it imparts a significant phase shift. That is audible to my ears and it appears to be audible by yourself as well being you already verified your suspicions.
I dont understand this day and age we live in where everything in the pro sound business is "just good enough". What's wrong with putting in the effort to make it the way it should be if it doesn't involve a ton of money and work? I get criticized for being too picky because most people settle for slightly better than average. You guys can have your MP3s and "remastered" vinyl that sounds worse than the original cassette version of the same album, compressed to death and loud. I'm honestly sick of the pro audio world nowadays, mainly becuase its cheap, disposable, non-serviceable plastic junk. As long as its loud, light and cheap, that appears all which matters to most. Sorry, not for me. - end of rant
I'd just replace C31 with 33pf and call it a day.
-3 dB @ 20k is very noticeable further down the path and it imparts a significant phase shift. That is audible to my ears and it appears to be audible by yourself as well being you already verified your suspicions.
I dont understand this day and age we live in where everything in the pro sound business is "just good enough". What's wrong with putting in the effort to make it the way it should be if it doesn't involve a ton of money and work? I get criticized for being too picky because most people settle for slightly better than average. You guys can have your MP3s and "remastered" vinyl that sounds worse than the original cassette version of the same album, compressed to death and loud. I'm honestly sick of the pro audio world nowadays, mainly becuase its cheap, disposable, non-serviceable plastic junk. As long as its loud, light and cheap, that appears all which matters to most. Sorry, not for me. - end of rant
Small offtopic : in cassette tape times ,i had portable radio ,and removed some capacitors from amplifiers input and that portable radio began to sound much better at hf range . I have asked friend ,who made speakers and so ,worked in audio shop , why manufacturers do put parts in devices ,that makes sound worse that it can be ?? He said - if it will sound good ,you will not buy more expensive device , you will be happy with that cheap and they have no business .Looks like it dividing by class , or cheapness . Usually home theater multi channel amplifiers are worse specs as i see than just stereo , especially older .
Just tested input filter alone in sim - looks like it has smaller impact , not 3db . So drop at hf must be then complex ,not only C31 related , probably after line inputs jacks filters present too ,before volume control .
Just tested input filter alone in sim - looks like it has smaller impact , not 3db . So drop at hf must be then complex ,not only C31 related , probably after line inputs jacks filters present too ,before volume control .
Well yeah, although my room is flat to 20Hz ;-), but this plot isn't flat anywhere, which seems very suspicious. All those little jumps are electrically impossible. I've looked at hundreds of these and they are all straight lines until the corner frequencies.
It always looks like that when the measurements are done at discrete freq steps, unless the software interpolates and/or smoothes it.
Jan
Jan
True, but then IF
I´d love to see that same plot but sweeping a sinewave.
Also: maybe there is not such a 3.5dB drop at all.
From first post I am bewildered by the brickwall "impossible" drop just above 20kHz ... wtf?
Where does that come from? 😱
So the shown drop might easily come from that filter, not one in the amp but in the measuring instrument or system.
Back to the original complaint: "lifelessness" might easily come from other factors, such as different damping factor, different current source/sink capability, available energy stored in min filter caps, even residual DCR in terminal posts or wiring.
that to certain degree amounts to "cheating", doesn´t it?
I´d love to see that same plot but sweeping a sinewave.
Also: maybe there is not such a 3.5dB drop at all.
From first post I am bewildered by the brickwall "impossible" drop just above 20kHz ... wtf?
Where does that come from? 😱
So the shown drop might easily come from that filter, not one in the amp but in the measuring instrument or system.
Back to the original complaint: "lifelessness" might easily come from other factors, such as different damping factor, different current source/sink capability, available energy stored in min filter caps, even residual DCR in terminal posts or wiring.
You forget then volume potentiometer's added 5k , if in middle position , 10k /2 .There's no buffer after potentiometer .
So what numbers do we have now for 1khz , 10khz , 20 khz drop ? I don't see 20k point in picture .
But anyway its higher drop than just input filter in my sim test ,probably 33pf feedback capacitor has negative effect on frequency response .
But anyway its higher drop than just input filter in my sim test ,probably 33pf feedback capacitor has negative effect on frequency response .