Attaching to the thread - this also is contentious for me. I have honestly not yet got my 3875s but I've tried this on TDA 2030 and STK 4191, and I got similar types of results - different because of the chips, but similar.
It's just a question of removing a couple of caps and resoldering one elsewhere. I have done extensive A/Bs - over weeks, and I have reached definite conclusions. Would love to know your results...
It's just a question of removing a couple of caps and resoldering one elsewhere. I have done extensive A/Bs - over weeks, and I have reached definite conclusions. Would love to know your results...
Just a thought - if we could get a dozen or so people involved we could do this vaguely scientifically
If over a week or so people were to build a an inverting and normal version using just one or two of the chips with a definative schematic and email the results to me I could then draw up the results and see how members opinions differ or coincide. Obviously this is open to abuse, so the results should not be taken as fact, but it might be a bit of fun!
Anyone up for this?
If over a week or so people were to build a an inverting and normal version using just one or two of the chips with a definative schematic and email the results to me I could then draw up the results and see how members opinions differ or coincide. Obviously this is open to abuse, so the results should not be taken as fact, but it might be a bit of fun!
Anyone up for this?
OK, I'll bite.
I do not have a schematic editor so I can't post a schematic. All constructions have been done using off-the-shelf (~RS) parts and 1% MFRs.
Circuits are as per datasheet recommendations. The STKs need some additonal biasing circuitry along with feedback resistors, so this was implemented as per MFR recommendations. The TDA 2030 does not need anything except for feedback circuitry and PS. All configs used a 220pF cap at input, and a 100 nF/4.7 ohm Zobel.
All rails were decoupled with 100uF electro and 0.1 uF film caps. Supplies were as recommended - the 2030 were run off a 19V rail supply and the STK off 35 volt rails.
The non-inverting configs used 22K/1K with impedance matching resistor of 22K between input and ground. The DC blocking cap was a 2.2. uF film and tantalum cap - used alternately to hone in on sonic abilities. The Dc blocker for the feedback path was a 100uF/63V cap.
The inverting configs were the same as above except for the exclusion of the 100uF from the feedback pin which became now the input. The resistor was left in place. offset voltage at output was 4 mV with the STK and 25 mV with the 2030. Input Caps were left in place.
Results are not between the chips, but between the connections. This is after switching and checking out the input cap options.
I find that the non-inverting config sounds acceptable - lots of meat, body and slam to the sound. Very upfront qualities - almost like an agressive sound, very good to listen to initially. Gets fatiguing specially when coupled with large speakers. Sounds good on small speakers - the extra punch fills in a missing dimension to the lower bass level output.
With the inverting config, extensive problems listening on small speakers. Sound becomes brighter, airier, and the bass moves lower and becomes more well-defined. The midbass slam almost vanishes, to be replaced by smooth, rounded midrange that never irritates or compes you to lower the volume. Best coupled with large speakers that can bring out detail.
The differences in sound are too compelling to ignore. I find the latter better, some may find the earlier better. Definitely the non-inverting may appeal to those who listen to a lot of rock/pop. The inverting sounds superb on just about everything - especially acoustic guitar recordings.
The inverting config has massive problems driving long speaker lengths. Does not like more than 2m or so between itself and the speakers, or maybe it is just my chips.
On my Worldspace reciever inverting brings out the digital artefacting at the higher frequencies and MP3 you cannot even listen to, they sound absolutely horrible. It only sounds good when connected to my CD player and my nice big Jamos, and nothing else in between. That is how it is being used. I would not un-invert them for the world.
Different qualities, for different applications. All in all I would sum up by saying that the inverting is a revealing kind of application, and the non-inverting is for acceptable listening.
Now, what's the theory behind this? Or am I way out of whack?
I do not have a schematic editor so I can't post a schematic. All constructions have been done using off-the-shelf (~RS) parts and 1% MFRs.
Circuits are as per datasheet recommendations. The STKs need some additonal biasing circuitry along with feedback resistors, so this was implemented as per MFR recommendations. The TDA 2030 does not need anything except for feedback circuitry and PS. All configs used a 220pF cap at input, and a 100 nF/4.7 ohm Zobel.
All rails were decoupled with 100uF electro and 0.1 uF film caps. Supplies were as recommended - the 2030 were run off a 19V rail supply and the STK off 35 volt rails.
The non-inverting configs used 22K/1K with impedance matching resistor of 22K between input and ground. The DC blocking cap was a 2.2. uF film and tantalum cap - used alternately to hone in on sonic abilities. The Dc blocker for the feedback path was a 100uF/63V cap.
The inverting configs were the same as above except for the exclusion of the 100uF from the feedback pin which became now the input. The resistor was left in place. offset voltage at output was 4 mV with the STK and 25 mV with the 2030. Input Caps were left in place.
Results are not between the chips, but between the connections. This is after switching and checking out the input cap options.
I find that the non-inverting config sounds acceptable - lots of meat, body and slam to the sound. Very upfront qualities - almost like an agressive sound, very good to listen to initially. Gets fatiguing specially when coupled with large speakers. Sounds good on small speakers - the extra punch fills in a missing dimension to the lower bass level output.
With the inverting config, extensive problems listening on small speakers. Sound becomes brighter, airier, and the bass moves lower and becomes more well-defined. The midbass slam almost vanishes, to be replaced by smooth, rounded midrange that never irritates or compes you to lower the volume. Best coupled with large speakers that can bring out detail.
The differences in sound are too compelling to ignore. I find the latter better, some may find the earlier better. Definitely the non-inverting may appeal to those who listen to a lot of rock/pop. The inverting sounds superb on just about everything - especially acoustic guitar recordings.
The inverting config has massive problems driving long speaker lengths. Does not like more than 2m or so between itself and the speakers, or maybe it is just my chips.
On my Worldspace reciever inverting brings out the digital artefacting at the higher frequencies and MP3 you cannot even listen to, they sound absolutely horrible. It only sounds good when connected to my CD player and my nice big Jamos, and nothing else in between. That is how it is being used. I would not un-invert them for the world.
Different qualities, for different applications. All in all I would sum up by saying that the inverting is a revealing kind of application, and the non-inverting is for acceptable listening.
Now, what's the theory behind this? Or am I way out of whack?
Yes, Eric. Absolute phase was maintained throughout. I did experiment with reversing speaker connection but since my system is not a very high-res one I did not notice many differences. Maybe since I listen to lots of rock I could not discern much difference there, but also not with my collection of nice Windham hill/Taylor CDs.
I have heard comments that the sound is a bit 'dry' 'analytical' and 'revealing'. I do not know the meaning of those words so much but I love a listen on a good CD - I like 'The Ragpickers dream' and 'Sailing to Philadelphia' (both, Mark knopfler) as good examples of range and diversity in music. Also tested were 'Friday night in San Fransisco', 'Shakti' (John Maclaughlin), 'Best of friends' (John Lee Hooker), and 'Crash' (Dave matthews band), 'Drops of Jupiter' (Train), 'Leaning into the night' (Ottmar Liebert), 'Tried and True' (Suzanne Vega), and the two Coldplay albums (Parachutes and A rush of Blood). My music may not be representative of those that require absolute phase, so the jury may be out on this for a while yet - I've heard sax is an instrument that requires absolute phase to sound 'correct', and I don't listen to Sax - except for the occasional Miles Davis tape (not CD)
Also some of my vintage and compilation CDs are no longer listenable. I have to resort to my computer system and MP3 collection to listen to it - mastering defects become so obvious.
I have heard comments that the sound is a bit 'dry' 'analytical' and 'revealing'. I do not know the meaning of those words so much but I love a listen on a good CD - I like 'The Ragpickers dream' and 'Sailing to Philadelphia' (both, Mark knopfler) as good examples of range and diversity in music. Also tested were 'Friday night in San Fransisco', 'Shakti' (John Maclaughlin), 'Best of friends' (John Lee Hooker), and 'Crash' (Dave matthews band), 'Drops of Jupiter' (Train), 'Leaning into the night' (Ottmar Liebert), 'Tried and True' (Suzanne Vega), and the two Coldplay albums (Parachutes and A rush of Blood). My music may not be representative of those that require absolute phase, so the jury may be out on this for a while yet - I've heard sax is an instrument that requires absolute phase to sound 'correct', and I don't listen to Sax - except for the occasional Miles Davis tape (not CD)
Also some of my vintage and compilation CDs are no longer listenable. I have to resort to my computer system and MP3 collection to listen to it - mastering defects become so obvious.
Hi Sangram, I ask because some of your comments could apply to AP sonics differences in my experience.
I find that with good 2 ways that AP is clearly aparrent.
Also, source impedence is important, more particularly for the inverting configuration, and this may be a factor in your experience of sonics differences.
Eric.
I find that with good 2 ways that AP is clearly aparrent.
Also, source impedence is important, more particularly for the inverting configuration, and this may be a factor in your experience of sonics differences.
Eric.
Koinichiwa,
And here the reason where the first layer of sonic problems may come from. With only 1k input impedance you are really pushing things in the source. You cannot simply dothings the way you did. You cause more problems than you solve.
I think you may wish to repeat the comparison using the latest inverting schematic, with 10k/22k/220k resistors around the negative input and 18k to ground on the positive input.
Also, if you use the same configuration in the output but have problems with long speaker cable length then there is something fundamentally wrong. I suspect somewhere, somehow you have build a few bugs into your inverting circuit in addition to excessively loading the source.
Sayonara
sangram said:
And here the reason where the first layer of sonic problems may come from. With only 1k input impedance you are really pushing things in the source. You cannot simply dothings the way you did. You cause more problems than you solve.
I think you may wish to repeat the comparison using the latest inverting schematic, with 10k/22k/220k resistors around the negative input and 18k to ground on the positive input.
Also, if you use the same configuration in the output but have problems with long speaker cable length then there is something fundamentally wrong. I suspect somewhere, somehow you have build a few bugs into your inverting circuit in addition to excessively loading the source.
Sayonara
Kuei Yang Wang said:
OK, I'll try 220K but last I remember anything above 100K was creating too much noise. I'll try it out anyway.
Also the source may be not as loaded as there is a volume pot before the input cap. I rarely push it above 50% mark, so I assume the source is looking into an impedance ~50K or so at 50%. Not a problem as such.
I forgot to mention. I am using a 18K resistor to ground in the TDA 2030. The STK with the calculated value was giving offset ~1V. It needs the value to be equal to the feedback resistor, to ground the + input. I will try a few more values and report back.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.