I use much smaller caps next to the chips. I have a 100uF Panasonic FM and a X7R ceramic directly on the supply pins. The caps decouple the inductance of the supply leads, and provide a small reservoir at a very small distance. I like to equate it to the L1 and L2 cache on a computer CPU. Basically the aim is to get a supply with a very low impedance, and enough current reserve for short-term peaks. the former is achieved by large caps and the latter by caps with low ESR. Caps that deliver both are usually outrageously expensive, so cap mixing is a way to achieve both objectives. I still think a pair of M-caps would be good, but spending hundreds of dollars on caps is not my thing.
I think Andrew had earlier posted a benchmark of 1000uF for every ampere of output current. I run twice that, per rail, so I end up with 40,000uF for the supposed 10A capability of the 3886. I have a strong feeling that the rail capacitors are in series as far as the load is concerned and actually the effective capacitance is only half of what we think, but I'm not really sure about that.
I think Andrew had earlier posted a benchmark of 1000uF for every ampere of output current. I run twice that, per rail, so I end up with 40,000uF for the supposed 10A capability of the 3886. I have a strong feeling that the rail capacitors are in series as far as the load is concerned and actually the effective capacitance is only half of what we think, but I'm not really sure about that.
Zoodle said:
According to some (including myself), you lose a bit of the "magic" in the mids and highs with a big cap power supply. That said, I feel the trade off is worth it, as I want some bass with my music. How much capacitance you need will depend on the amp and the speakers (and somewhat the transformer too).
Interesting read on the subject:
http://www.diyaudio.com/forums/showthread.php?postid=581230#post581230
limits
I have a 3886 straight chipamp and a 4780 "parallel," both with big power supplies.
I started out on the "you don't need that much power" bandwagon and "wow the chip circ is so simple but so good sounding" club but after a few years and trying different amps and speakers I am becoming convinced that the smooth tonal qualities of a chip do not make up for its lack of power delivery.
Even with Snell Type E's (efficient and an easy load) the chipamps struggle to play at medium levels in a mid size room. And they certainly do warm up at the effort.
I know a lot of people make really nice chipamps and there are commercial variations with strong followings but do no be surprised when the fancy chips with fancy capacitors run out of steam.
That said, if you don't need current or high volumes a chipamp can sound very good.
I have a 3886 straight chipamp and a 4780 "parallel," both with big power supplies.
I started out on the "you don't need that much power" bandwagon and "wow the chip circ is so simple but so good sounding" club but after a few years and trying different amps and speakers I am becoming convinced that the smooth tonal qualities of a chip do not make up for its lack of power delivery.
Even with Snell Type E's (efficient and an easy load) the chipamps struggle to play at medium levels in a mid size room. And they certainly do warm up at the effort.
I know a lot of people make really nice chipamps and there are commercial variations with strong followings but do no be surprised when the fancy chips with fancy capacitors run out of steam.
That said, if you don't need current or high volumes a chipamp can sound very good.
Sound of a GainClone
Hi guys!
Recently I build a ChipAmp with LM3875 in conbination with a NOS DAC (see photo gallery, page 44). After a proper break-in period I decided to upgrade the power supply and I added another small pcb with two 4700/50 caps and a rc-snubber of 1R & 100nF on top of the rectifier pcb. This does quite a lot to the performance of the chipamp! Everything (not only bass) has become tighter and more integrated, better defined and better refined!
But even though, after a few days of serious listening I came to the following conclusion about the sound: A well implemented chipamp has a quite different sound signature than conventional solid state amps. It is capable of delivering a very sweet, open and detailed sound, without being harsh or stressed at all. Really, soundwise this amp reminds me of a really good tube amp, but without the SOUL of a good tube amp!
And indeed, this is something I am missing: The soul, the heart and the spirit which a very good amp should be able to transport to the listeners mind. Despite the fact that this GainClone sounds more refined and open than my previous Naim Nap 140 (of the early 90's), it isn't able to deliver this feeling of really played music by real people. It has a more soft and slightly rounded presentation, and a slight grayness overall. All this leads to the conclusion, that this amp doesn't involve me as much as a really musical solid state or a good tube amp could do. Bigger caps in the ps or not, this doesn't make a difference here.
My speakers are a pair of Reference 3a MM (early version of the current model DeCapo; crossoverless 8" carbon driver, 1" Vifa tweeter with nothing else than a 4,7uF cap in series; efficiency is about 90db, impedance never goes below 6 ohms).
Well, just my two cents and of course, a very subjective view. But this is my experience, and beside all the technical issues I think we should discuss the sound of our builds a bit more here in this forum, because this is what all this is about: Nice sound for happy listening!
Regards - martin
Hi guys!
Recently I build a ChipAmp with LM3875 in conbination with a NOS DAC (see photo gallery, page 44). After a proper break-in period I decided to upgrade the power supply and I added another small pcb with two 4700/50 caps and a rc-snubber of 1R & 100nF on top of the rectifier pcb. This does quite a lot to the performance of the chipamp! Everything (not only bass) has become tighter and more integrated, better defined and better refined!
But even though, after a few days of serious listening I came to the following conclusion about the sound: A well implemented chipamp has a quite different sound signature than conventional solid state amps. It is capable of delivering a very sweet, open and detailed sound, without being harsh or stressed at all. Really, soundwise this amp reminds me of a really good tube amp, but without the SOUL of a good tube amp!
And indeed, this is something I am missing: The soul, the heart and the spirit which a very good amp should be able to transport to the listeners mind. Despite the fact that this GainClone sounds more refined and open than my previous Naim Nap 140 (of the early 90's), it isn't able to deliver this feeling of really played music by real people. It has a more soft and slightly rounded presentation, and a slight grayness overall. All this leads to the conclusion, that this amp doesn't involve me as much as a really musical solid state or a good tube amp could do. Bigger caps in the ps or not, this doesn't make a difference here.
My speakers are a pair of Reference 3a MM (early version of the current model DeCapo; crossoverless 8" carbon driver, 1" Vifa tweeter with nothing else than a 4,7uF cap in series; efficiency is about 90db, impedance never goes below 6 ohms).
Well, just my two cents and of course, a very subjective view. But this is my experience, and beside all the technical issues I think we should discuss the sound of our builds a bit more here in this forum, because this is what all this is about: Nice sound for happy listening!
Regards - martin
If you read through the forum, you will gain the impression that chipamps can sound very different although they are all based on the same IC. You could of course start a thread, where people describe their setup and sonic signature. Maybe that turns up correlations, which can help people tweak their amp to a certain character.
Impedance and sensitivity are only two aspects of a speaker load. A speaker that remains above a certain impedance and has a reasonable sensitivity can still be a difficult load. Big woofers in small bass-reflex enclosures are often difficult loads, as are speakers with high inductance, too simple cross-overs, too complex cross-overs or bumpy impedance curves.
Impedance and sensitivity are only two aspects of a speaker load. A speaker that remains above a certain impedance and has a reasonable sensitivity can still be a difficult load. Big woofers in small bass-reflex enclosures are often difficult loads, as are speakers with high inductance, too simple cross-overs, too complex cross-overs or bumpy impedance curves.
Hi,
I just report my experience and impressions.
CD player: Meridian 508-20 (commercial of course)
speakers: full range Jordan MLTL-48 86db sensitivity (self-made)
Amplifiers:
YBA Integre solid state integrated commercial amp, 50W 8ohm, 90W 4Ohm, dual mono, current price 2500euro (built in France)
Yarland M84 commercial integrated tube amp 15W 8/4Ohm push-pull, EL84 tubes, current price 310euro (built in China)
LM1875 integrated chipamp, kit from chipamp.com, 1 trafo 225VA 2x18VAC (self-made).
2 LM3875 monoblocks, premium kit from audiosector.com, trafo 300VA 2x25VAC, Panosonic caps, + passive pre with stepped attenuator from diyfidelity.com.au (self made).
I am not an expert and I cannot report about technique, but just about sound impression.
I did not believe I could get significant improvement by changing the amplification; I built the LM3875 just for fun and the satisfaction to have a complete system built by myself, next step the dac of course.
My prejudice was confirmed by the comparison among the YBA, the chinese tube amp and the LM1875 little chipamp.
A little bit of preference for the YBA, but not to justify the factor 10 increase in cost and I do not think I could hear differences in a blind test.
But I had to change the idea, when I connected the speakers to the LM3875 monoblocks. The sound dramatically changed, a lot more of detail, a lot of information unveiled. For sure a step towards the simulation of live experience.
I think I could easily identify the improvement even in a blind test.
In conclusion, to me the LM3875 chipamp I have now is clearly better than the YBA, a respected and costly amp rated in the past class A on Stereophile and I have the feelling that a builder more clever than me by proper selection of components can get even more.
While I am now completely satisfied with Jazz and small ensemble music, with my system I still miss something with most of classical orchestral records I have, but I still do not know if this is really a lack of the system, especially of the speakers, or because most of my classical CDs are of the 80s and the system now shows the poor quality of the recording and mastering.
Renato
P.S. If I take into account for the time I spent building the chipamps, the YBA amp is much cheaper
I just report my experience and impressions.
CD player: Meridian 508-20 (commercial of course)
speakers: full range Jordan MLTL-48 86db sensitivity (self-made)
Amplifiers:
YBA Integre solid state integrated commercial amp, 50W 8ohm, 90W 4Ohm, dual mono, current price 2500euro (built in France)
Yarland M84 commercial integrated tube amp 15W 8/4Ohm push-pull, EL84 tubes, current price 310euro (built in China)
LM1875 integrated chipamp, kit from chipamp.com, 1 trafo 225VA 2x18VAC (self-made).
2 LM3875 monoblocks, premium kit from audiosector.com, trafo 300VA 2x25VAC, Panosonic caps, + passive pre with stepped attenuator from diyfidelity.com.au (self made).
I am not an expert and I cannot report about technique, but just about sound impression.
I did not believe I could get significant improvement by changing the amplification; I built the LM3875 just for fun and the satisfaction to have a complete system built by myself, next step the dac of course.
My prejudice was confirmed by the comparison among the YBA, the chinese tube amp and the LM1875 little chipamp.
A little bit of preference for the YBA, but not to justify the factor 10 increase in cost and I do not think I could hear differences in a blind test.
But I had to change the idea, when I connected the speakers to the LM3875 monoblocks. The sound dramatically changed, a lot more of detail, a lot of information unveiled. For sure a step towards the simulation of live experience.
I think I could easily identify the improvement even in a blind test.
In conclusion, to me the LM3875 chipamp I have now is clearly better than the YBA, a respected and costly amp rated in the past class A on Stereophile and I have the feelling that a builder more clever than me by proper selection of components can get even more.
While I am now completely satisfied with Jazz and small ensemble music, with my system I still miss something with most of classical orchestral records I have, but I still do not know if this is really a lack of the system, especially of the speakers, or because most of my classical CDs are of the 80s and the system now shows the poor quality of the recording and mastering.
Renato
P.S. If I take into account for the time I spent building the chipamps, the YBA amp is much cheaper
Redshift187 said:
Redshift187,
I suspected something like that might happen. Can you parallel two types of cap - a monster for that 'bottomless well' effect, plus a smaller cap? Can we have the best of both worlds, or doesn't it work that way?
I'm pretty sure you always want to put one or more small value caps in parallel with a large cap. That statement still applies to that situation. Something else people do, that I haven't done, is parallel say 10x 1000uF instead of using 1x 10,000uF.
Redshift187 said:
The smaller caps definitely work well -- and sometimes you can find good quality ones for very low price. the 10u and 100n caps adjacent to the chip seem to help with the thermal protection for some reason -- you can get the protection to cycle under certain conditions without them.
I use a large reservoir of 40-80,000uF near the transformer, with a 100uF and 100nF right next to the chip. My supply leads are about 6", if I was using a PCB or had shorter runs I could get by with only a 100nF I guess.
I like mF at the rectifiers. +-20mF to +-75mF does for 4 to 8ohm and for ClassAB and ClassA.
I also like 00s uF at the amplifier. 220uF to 2200uF seems to do.
00s nF seems to suit many amps right at the the highest current consumers.
Note, this applies equally to all SS power amps.
I also like 00s uF at the amplifier. 220uF to 2200uF seems to do.
00s nF seems to suit many amps right at the the highest current consumers.
Note, this applies equally to all SS power amps.
Hey you guys,
you see what happens? This thread is getting right into the technical aspect of amp building again!
If we want to discuss performance here (and I would be happy if we want to do so), we should concentrate on what really matters. See, I don't want to hijack this thread, nor do I want to lead the discussion. But everytime before I start building something, I try to inform myself as much as possible, because to build an amp does not only consume time and money, it also could lead you to a more or less undesired result, and this automatically leads to less fun when listening to music ;-)
So here it goes:
In the past I owned a Naim Nap 140 and Nap 160 (sometimes used with a NAC 62), some Rotels (best one was the RA-931mkII), build two amps using a single TDA7294 per channel and another one unsing TDA7293. And recently I build a GainClone with LM3875, with a fully encapsulated high quality 250VA toroid (2x24V sec.), first with the small caps on board only, later with additional large caps on the rectifier board.
My findings are, in short:
- The old Naim amps do have a very special character, very musical, with lot of drive and verve, best suited for some old style rock music, but not that kind of speedy and agile when it comes to more modern electronic music. Sometimes they can sound a bit grainy in the upper mids, and bass is strong but not that thight and controlled than with a well implemeted TDA7293.
- Rotel is nice mid-range hifi, but really, the RA-931mkII is a different beast: the most musical, open and real sounding Rotel I ever listened to. Only the lack of refinement and smoothness in mids and highs leaded me to build something more convincing.
- The TDA7293/94 are some special amps. I think Linn uses these chips in their LK-series (85 and 140). They really grab the speakers and hold them, playing every tune carefully articulated, but with some great musical sense. They have dynamics, speed and timing, and some sort of realism, which makes you think of listening to live music.
- The LM3875 delivers a very open, sweet and clean sound, with a large soundstage. But there is little control on the speakers and a tendency to some boomy kind of upper bass. There is also a definite lack of realism to the music. Every of the amps mentioned above involved me more into the music, told me more about the message the artist wanted to send. Even if the sound is sweet and very detailed, there is some artificial aspect on the whole performance. For example, a guitar has strings, made of metal. When a guitar is played, I should hear something which sounds like metal. But with the GainClone it simply doesn't sound like metal, it just sounds like a sweet sounding something, which has something to do with a guitar, but isn't one in reality. There seems to be no real connection to the ground, no EARTH-feeling, if you knwo what I mean.
Sorry if I bother you, but I'm always happy about statements which help me to decide wheather I should build something or not. So this might be helpful to those who are unsure about what project to start. But again, this is only MY oppinion!
Regards
martin
P.S.: pacificblue, if you search the web about Reference 3a speakers, you can read nearly everywhere (be it a commercial magazin or some forum threads of owners) that these speakers are a very easy load to nearly all amps, small valve amps included. Further more, I measured the impedance curve by myself, and I can confirm these claims. So, if a well build GainClone is not able to drive these speakers with moderate volume levels and some rather less complex music (like Tracy Chapman or Mark Knopfler or something like that), than I really don't know what these amps are supposed to drive. Or is this ChipAmp intended for the use with very sensitive single driver fullrange horn loaded speakers ONLY?
Don't get me wrong pacificblue, I really don't want to seem impolite here, I mean this as a serious question!
you see what happens? This thread is getting right into the technical aspect of amp building again!
If we want to discuss performance here (and I would be happy if we want to do so), we should concentrate on what really matters. See, I don't want to hijack this thread, nor do I want to lead the discussion. But everytime before I start building something, I try to inform myself as much as possible, because to build an amp does not only consume time and money, it also could lead you to a more or less undesired result, and this automatically leads to less fun when listening to music ;-)
So here it goes:
In the past I owned a Naim Nap 140 and Nap 160 (sometimes used with a NAC 62), some Rotels (best one was the RA-931mkII), build two amps using a single TDA7294 per channel and another one unsing TDA7293. And recently I build a GainClone with LM3875, with a fully encapsulated high quality 250VA toroid (2x24V sec.), first with the small caps on board only, later with additional large caps on the rectifier board.
My findings are, in short:
- The old Naim amps do have a very special character, very musical, with lot of drive and verve, best suited for some old style rock music, but not that kind of speedy and agile when it comes to more modern electronic music. Sometimes they can sound a bit grainy in the upper mids, and bass is strong but not that thight and controlled than with a well implemeted TDA7293.
- Rotel is nice mid-range hifi, but really, the RA-931mkII is a different beast: the most musical, open and real sounding Rotel I ever listened to. Only the lack of refinement and smoothness in mids and highs leaded me to build something more convincing.
- The TDA7293/94 are some special amps. I think Linn uses these chips in their LK-series (85 and 140). They really grab the speakers and hold them, playing every tune carefully articulated, but with some great musical sense. They have dynamics, speed and timing, and some sort of realism, which makes you think of listening to live music.
- The LM3875 delivers a very open, sweet and clean sound, with a large soundstage. But there is little control on the speakers and a tendency to some boomy kind of upper bass. There is also a definite lack of realism to the music. Every of the amps mentioned above involved me more into the music, told me more about the message the artist wanted to send. Even if the sound is sweet and very detailed, there is some artificial aspect on the whole performance. For example, a guitar has strings, made of metal. When a guitar is played, I should hear something which sounds like metal. But with the GainClone it simply doesn't sound like metal, it just sounds like a sweet sounding something, which has something to do with a guitar, but isn't one in reality. There seems to be no real connection to the ground, no EARTH-feeling, if you knwo what I mean.
Sorry if I bother you, but I'm always happy about statements which help me to decide wheather I should build something or not. So this might be helpful to those who are unsure about what project to start. But again, this is only MY oppinion!
Regards
martin
P.S.: pacificblue, if you search the web about Reference 3a speakers, you can read nearly everywhere (be it a commercial magazin or some forum threads of owners) that these speakers are a very easy load to nearly all amps, small valve amps included. Further more, I measured the impedance curve by myself, and I can confirm these claims. So, if a well build GainClone is not able to drive these speakers with moderate volume levels and some rather less complex music (like Tracy Chapman or Mark Knopfler or something like that), than I really don't know what these amps are supposed to drive. Or is this ChipAmp intended for the use with very sensitive single driver fullrange horn loaded speakers ONLY?
Don't get me wrong pacificblue, I really don't want to seem impolite here, I mean this as a serious question!
Martin,
Thanks for an informative read.
I've been looking for a chip amp builder familiar with the Linn/Naim sound - and here you are! Have you heard the LM3886? Do you know any chips used in other Linn amps?
In any case, thanks for suggesting the TDA7293/94.
Thanks for an informative read.
I've been looking for a chip amp builder familiar with the Linn/Naim sound - and here you are! Have you heard the LM3886? Do you know any chips used in other Linn amps?
In any case, thanks for suggesting the TDA7293/94.
Hi Zoodle,
thanks for your comment. I would not go that far to claim that a TDA7293 is familiar with the Naim sound, but these little amps are quite an alternative. No, I didn't listen to a LM3886, but the comparison of LM3875 and TDA7293 convinced me to stay with the TDA's. AFAIK, the TDA'S have a Mos-Fet output stage. I don't know exactly if this is the key here, but I like the Mos-Fet sound for some reason.
Regards
martin
thanks for your comment. I would not go that far to claim that a TDA7293 is familiar with the Naim sound, but these little amps are quite an alternative. No, I didn't listen to a LM3886, but the comparison of LM3875 and TDA7293 convinced me to stay with the TDA's. AFAIK, the TDA'S have a Mos-Fet output stage. I don't know exactly if this is the key here, but I like the Mos-Fet sound for some reason.
Regards
martin
You cannot separate those two aspects, because a lot of the sonic signature and performance depends on the technical implementation. A bad power supply will always lead to a bad amplifier, whether too small or too big or simply made from the wrong components. Wrong choice of filters at the input as well. Some people claim that the sound even changes with the gain setting.martinbls said:
±32 V and a speaker that dips down to 6 Ohm, hence is no nominal 8 Ohm speaker anymore, combined with the TF package, could bring the IC into a temperature region, where the protection system starts to degrade the sound. Try the same amplifier with a 2x18 V transformer and a heatsink that remains cool enough to touch during operation.
This is, what the datasheet says:martinbls said:
- Component stereo
- Compact stereo
- Self-powered speakers
- Surround-sound amplifiers
- High-end stereo TV's
From this you can derive that it was originally designed as a low-cost low- to medium-power mid-fi solution. It turned out to be above average in its class and much better than the price would let you expect, which started the Gainclone hype off. It certainly depends on your definition of moderate levels, but I would estimate that a well-made LM3886 based amplifier will serve well for more than 90 % of listening situations. The remaining less than 10 % are, what tweaking and the Solid-State Forum (and mabye PA-Forum) are about.
TI has some PurePath chipsets that go as high as 300w per channel. You'll need to solder SMD and use a microcontroller to initialize the modulator, though.
Not sure what this other stuff means...
I like my Linn 85 i got of ebay in perfect condition for £180, not sure its value, but i had a good feeling about it so i bought it anyway.
I dont want to get to technical so i will sum it up in baby language
I like my amplifier, it is a good amplifier, better than my old amplifier
Ok i could get abit tecnical, It seems to produce the waveform more precicely....
I like my Linn 85 i got of ebay in perfect condition for £180, not sure its value, but i had a good feeling about it so i bought it anyway.
I dont want to get to technical so i will sum it up in baby language
I like my amplifier, it is a good amplifier, better than my old amplifier
Ok i could get abit tecnical, It seems to produce the waveform more precicely....
One thing i would like to know is, how it seems to not get hot, and have such a small skimpy circit board, and weigh half a gram, and yet still sound fantatsic! defies the audiotry equivolant to pysics.
If some one could shine a light on this i would be interested, as i am thinking of buying another!
If some one could shine a light on this i would be interested, as i am thinking of buying another!
Well, if there would be a universal understanding of terms like "musical, open, real, punchy etc." that wouldn't be necessary but since you really can't comrehend the fine aspects of another persons verbalized perception this means nearly nothing.
Just consider that we have a system with an enormous amount of variables that influence sound, and besides those we can measure that influence sound, there are still quite a lot of things going on that we can't get a grasp on yet but make a difference in hearing nonetheless.
While I am sure that we all are compassionate about our musical equiment and like to describe it as colourful as possible I don't think discussing chipamp performance with vague terms will help. Also I think a few more blind tests would help everyone, so why not build a little box that switches amplifieres/parts randomly (or not) and lets you write down your guess and afterwards you can look how good you guessed.
I don't want to go all objectivist here, but reading some descriptions seem like watching a magic trick, where no one knows what is going on but something amazing is obviously happening...
cheers,
hurtz
Yes, besides the obvious problems with the undefined meanings of such descriptions, and even if those problems went away completely, describing the "sound" of different chips is completely meaningless without the complete technical details of the entire build. And even then, most people would have little basis for interpreting such descriptions in any meaningful way, much less transforming them into more-calibrated comparisons.
Or maybe you don't realize how much the details of the implementation can affect the results. I'm not saying that you don't. But there are many significant factors, not in the schematics, that most DIY chipamp builders are either unaware of or ignore.
All of the types of sonic attributes that you have described can be drastically affected by implementation details OTHER THAN the chip that is used.
Or maybe you don't realize how much the details of the implementation can affect the results. I'm not saying that you don't. But there are many significant factors, not in the schematics, that most DIY chipamp builders are either unaware of or ignore.
All of the types of sonic attributes that you have described can be drastically affected by implementation details OTHER THAN the chip that is used.
- Status
- Not open for further replies.