Chip Amps - Anyone using or building them

[tuyen]

I sure hope so.  I would be surprised if a $50 chip amp from eBay sounds better than all valve amps out there!

[Lansche plasma guy]

Hi Guys.

The chip amp wasnt a chip amp.

It was a small digitsl amp which had a bit of a cult following (and may still do)?? Called the Virtue one it is a digital amp.

It uses the Tripath 2050 module which msny people thought was the best sounding small power diisl module.

Stock it was okay. I changed all the wiring to 1940's very thin solid core copper eire sheathed in silk. I chaged the resistorsand changed the coupling caps from the Auri caps to some vintage stacked paper and oil types.

This was all done to give the Virtue amp a little more body and soul. It was a lovely little amp and great value for money.

But it will never compete on mega efficient horns with a single ended amp using a 6gw8 amp i also heavilly modified. The lile 6gw8 is just jaw dropping if it can drive the speakers. The valve has a driver tube and power tube in the same glasse nvelope. The circuit has almost nothing in it. This makes for. Crazy short signal path along with the very best of sublime valve sounds. Amazing tone and really vivid. Much less lush than the typical big bottle single ended amp like the300b, type 50 etc. but it only puts out sbout 1.5 watts.

Ive also had great experiences with gainclones. But not all Gainclones are created ewuaand i found that small changes had big effects on the sound. Depending on your dydtem and other components you can strip them so there isnt much left. Just be careful of oscillation and stability.

But eit the right speaker the best of these can still have lots of magic. Im not saying they are going to compete with the very best valve smps but in the right system they can sound amazing providing veryhigh level sound and the same sort of signature as the more modern SET sound. They are a great match for speakers that are a little deada nd dark and slow (think Audionote speakers for which they can be amazing match).

May be amazing on really efficient horns but the horns but maybe not. The horns may reveal the truth too much whch may expose the gainclones. On the flip side you could say that the gainclones may be a little too clear and clean for really efficient horns (Avantgardes etc) and highlight the impossibility of making a multi way horn speaker with horns different lengths and depths sound coherent. A bit of valve colour is nice to vaseline coat the horns so to speak and being out the magic of the speakers.

Back on topic with chip amps: i had some deeply involving times (sometimes) with a sympathetic system powered by the chip amps and enjoyed a few years of modifying them.

Theres more thsn one way to Rome (but one of the fastest ways is with a vslve direct drive electrostatic speaker which can do 25hz-50khz at 110db) :-)

LPG.

[Newman]

but a valve amp will do things a chip amp can only dream about

 

And vice versa, with interest  

[Newman]

After a bit more research and I find that a few chip amp recipe models out there

E.G. Gainclone - Tripath T Amps & Nelson Pass

What are the main design differences and does one type trump the others ?

I read with interest some posts that use battery powered chip amps

What advantage does this give it over direct AC power ?

 

Gaincard = class AB power amp on a chip

 

Tripath = class D variant power amp on a chip

 

Pass = no chip amp I am aware of?-- although he as a circuit or 2 that uses only one active component (a MOSFET), but that is not an 'amp on a chip'. I'll exclude his designs pending new information.

 

Comparing, AFAIK the Gaincard-type amps have better performance and power than the class D (including T) ICs that are out there (so far?) - and they can be ganged for higher voltage or current ratings and hence power. Very nice. Class D chip amps of course have the upper hand for efficiency and hence small and light completed amps, usually with no heatsink.

 

Batteries are not suited to power amps unless they are particularly puny power amps -- too much drain on the battery. [edit: show me a few links? - interesting]

Edited November 4, 2013 by Newman

[ehtcom]

No, they don't actually have those components, as I see it, that's just an equivalent schematic. 

 

My point being, it isn't just "one" magical amplifying device. Many "components" are contained within the integrated circuit, or "chip".

In fact, probably a lot more active devices than shown on the equivalent circuit.

[Zaphod Beeblebrox]

My point being, it isn't just "one" magical amplifying device. Many "components" are contained within the integrated circuit, or "chip".

In fact, probably a lot more active devices than shown on the equivalent circuit.

 

Correct. So-called 'chip amps' are EXTREMELY complex devices. Not that that fact makes them bad. It's just that when a manufacturer using them makes claim to simplicity, they are telling lies.

[Zaphod Beeblebrox]

After a bit more research and I find that a few chip amp recipe models out there

E.G. Gainclone - Tripath T Amps & Nelson Pass

What are the main design differences and does one type trump the others ?

I read with interest some posts that use battery powered chip amps

What advantage does this give it over direct AC power ?

 

First off:

 

Forget so-called 'Gainclones'. All rely in the LM3886 and derivative chips. A FAR better choice is the TDA1514A. The TDA1514A is far more sophisticated and better sounding. It is full complementary symmetry. The LM series are quasi-complementary symmetry. Quasi comp SS amps have, for all intents, not been used by any amplifier manufacturer since the mid 1970s.

 

Second off:

 

Class T amplifiers are a very different product. Class T is a Class D variant. Some listeners like them. I don't. They are compact, efficient and reasonably priced. If you don't mind a less efficient, larger amplifier, but at comparable cost, the a TDA1514A (or even an LM series chip) will, IMO, provide superior sound, at a similar cost.

 

Thirdly:

 

I doubt Nelson Pass would be caught dead stooping to use a chip amp. I will remind you that chip amps are:

 

* Designed to be cheap.

* Designed to be convenient.

* Designed to be used in TV sets, surround sound amps and other small appliances.

* Designed to be reliable.

* Designed to be easy to use.

 

None have been designed to push the limits of high end sound reproduction. They do a very good job at a reasonable price though.

 

Lastly:

 

Forget batteries. All chip amps possess excellent PSRR (Power Supply Rejection Ratio) and will perform very well on a properly designed and sized mains power supply.

Edited November 4, 2013 by Zaphod Beeblebrox

[Lansche plasma guy]

I think one of the reasons chip amps can sound so good for the money isnt the number of components but the length of the signal.

I've never heard ultimate soind from a chip amp. But i have heard superb sound and for very very little money.

I remember with a friend comparing a lm3875t chip amp to the Audionote Conquest (or conqueror? It was the dearer mono block amps) and also against the Wavelenth Audio Napoleon amps (the ones with the silver output transformers which ten years ago where $20k in Australia).

Driving some Audionote Anj with out oard power supplies. In that system on theday the most sublime match was with the little lm3875t amp. The chip amp sounded very similar to the Gordon Rankins Wavelength Audio Napoleon.

The Audionote amps had beautiful tone and sounded totally different to the other amps. It sounded gorgeous but was much too dark and slow sounding on the Audionotes.

But it would be wrong to infer from the above that im saying the lm3875t gainclone amp was better or id amazing. But that particular chip amp in that particular system on the day was. The chip amps have a family sounding but there isa huge difference between the really good sounding ones and the not so good ones. The chips themselves are very complex but the amps themselves only use a handful of parts. Each one of those parta make a huge difference to the sound of the amp.

But for no money as a kit they can be stupid good sounding.

LPG

[Zaphod Beeblebrox]

I think one of the reasons chip amps can sound so good for the money isnt the number of components but the length of the signal.

 

 

No. I've heard that same hypothesis many times. It does not make any rational sense. The time taken for electricity to travel the length of 30cm of wire is approximately 2 nano seconds. That's 2 X 10^-9 seconds. Give or take. The human ear can, at best, resolve a time difference in the order of 5 X 10^-5 seconds. That equates to around 7.5 km of wire.

 

The human ear is exquisitely sensitive to many facets of sound reproduction, but it cannot resolve things in the micro or nano second region. It ain't physically possible.

 

The reality is far more prosaic:

 

Chip amps ARE very good for a number of reasons. They are well designed and, given their intended market, they must be. They are designed to sell in million off quantities. Designers ensure that their product is reliable, cost-effective and decent sounding.

 

Their tiny size also creates a 'wow, look at that' feeling in the user.

Edited November 4, 2013 by Zaphod Beeblebrox

[Addicted to music]

No. I've heard that same hypothesis many times. It does not make any rational sense. The time taken for electricity to travel the length of 30cm of wire is approximately 2 nano seconds. That's 2 X 10^-9 seconds. Give or take. The human ear can, at best, resolve a time difference in the order of 5 X 10^-5 seconds. That equates to around 7.5 km of wire.

 

The human ear is exquisitely sensitive to many facets of sound reproduction, but it cannot resolve things in the micro or nano second region. It ain't physically possible.

 

The reality is far more prosaic:

 

Chip amps ARE very good for a number of reasons. They are well designed and, given their intended market, they must be. They are designed to sell in million off quantities. Designers ensure that their product is reliable, cost-effective and decent sounding.

 

Their tiny size also creates a 'wow, look at that' feeling in the user.

Not sure whether I agree with this sound explanation,

In the mid 80s Naim produced 2 version of there top of the range preamp, the 1st had typical track layouts to ensure the components on in a neat fashion. The 2nd was CAD designed to eliminate and shortened the paths to each component. You can see that the tracks where done so the electrons naturally had less to travel and less corners to go around. The board used identical parts and was compatible to the PSU they had. Nothing else was changed. Doing a shootout between the versions definitely displayed an inherent difference.

[Zaphod Beeblebrox]

Not sure whether I agree with this sound explanation,

In the mid 80s Naim produced 2 version of there top of the range preamp, the 1st had typical track layouts to ensure the components on in a neat fashion. The 2nd was CAD designed to eliminate and shortened the paths to each component. You can see that the tracks where done so the electrons naturally had less to travel and less corners to go around. The board used identical parts and was compatible to the PSU they had. Nothing else was changed. Doing a shootout between the versions definitely displayed an inherent difference.

 

It's not the DISTANCE that matters, but the layout. Shorter may not, necessarily be better, though shorter may be easier to manage difficult problems. I have been involved with products where altering the layout (but making cabling any shorter) elicited huge differences in an audio product.

 

About 20 years ago I was involved in the design of a device which operated at audio frequencies (though it was not an audio device). At the final hurdle, I found that increasing the thickness of the PCB traces enabled me to obtain the performance demanded by my client.

[Newman]

First off:

 

Forget so-called 'Gainclones'. All rely in the LM3886 and derivative chips. A FAR better choice is the TDA1514A. The TDA1514A is far more sophisticated and better sounding. It is full complementary symmetry. The LM series are quasi-complementary symmetry. Quasi comp SS amps have, for all intents, not been used by any amplifier manufacturer since the mid 1970s.

 

Oh my goodness. Where does all this prejudice come from? I remember it from previous discussions of chip amps. Is it because of this line of thought that KenTripp asked if the Gaincard uses the TDA1514A? i.e. because the LM series can't be any good?

 

At least now I know where it is coming from: 'Quasi complementary symmetry' prejudice. 

 

The original reason for the rise of quasi comp was the cost and difficulty in making PNP transistors to complement NPN transistors. A quasi comp amp is all-NPN. Once affordable PNPs came along, designers started using full complementary designs, including ME I suppose. However, some of the most exotic amp builders stuck with quasi comp. Naim comes to mind. The reason being that even today PNP transistors are more difficult to make than NPN, so complementary pairs have different behaviour even if matched. 

 

The best designers freely chose one or the other configuration and made great amps with them. The problems of quasi comp and the problems of full comp can both be dealt with by good design. There is no need to be dismissive of one or the other, except to serve personal agendas.

 

Looking at the quasi comp equivalent circuit in the LM series ICs, the use of on-chip fabrication allows superb matching of output devices, better than any discrete design irrespective of brand. I find it fascinating that the Nat Semi designers chose this output design. There was obviously no cost saving by going quasi comp on an IC, which was the reason for quasi comp in the 70s (and the reason for your prejudice), so I think the designers have been innovative. I would love to know their thoughts. I reckon Nat Semi see their output stage as cleverly using LSI techniques and solving some issues that plague a conventional output stage when used on a power IC, like the conventional fully comp output on the TDA series. The LM output stage is probably not reproducible (at least its performance is not reproducible) using discrete components. It is the LM that in fact is probably more sophisticated, not the TDA.

[Zaphod Beeblebrox]

Oh my goodness. Where does all this prejudice come from? I remember it from previous discussions of chip amps. Is it because of this line of thought that KenTripp asked if the Gaincard uses the TDA1514A? i.e. because the LM series can't be any good?

 

At least now I know where it is coming from: 'Quasi complementary symmetry' prejudice. 

 

The original reason for the rise of quasi comp was the cost and difficulty in making PNP transistors to complement NPN transistors. A quasi comp amp is all-NPN. Once affordable PNPs came along, designers started using full complementary designs, including ME I suppose. However, some of the most exotic amp builders stuck with quasi comp. Naim comes to mind. The reason being that even today PNP transistors are more difficult to make than NPN, so complementary pairs have different behaviour even if matched. 

 

The best designers freely chose one or the other configuration and made great amps with them. The problems of quasi comp and the problems of full comp can both be dealt with by good design. There is no need to be dismissive of one or the other, except to serve personal agendas.

 

Looking at the quasi comp equivalent circuit in the LM series ICs, the use of on-chip fabrication allows superb matching of output devices, better than any discrete design irrespective of brand. I find it fascinating that the Nat Semi designers chose this output design. There was obviously no cost saving by going quasi comp on an IC, which was the reason for quasi comp in the 70s (and the reason for your prejudice), so I think the designers have been innovative. I would love to know their thoughts. I reckon Nat Semi see their output stage as cleverly using LSI techniques and solving some issues that plague a conventional output stage when used on a power IC, like the conventional fully comp output on the TDA series. The LM output stage is probably not reproducible (at least its performance is not reproducible) using discrete components. It is the LM that in fact is probably more sophisticated, not the TDA.

 

The reason I suggested the TDA1514A over any of the LM series chips, is because I have listened to them and measured them. The TDA1514A is clearly audibly superior. By a long margin. The use of complementary symmetry is just one, major, advantage of the TDA1514A. The TDA1514A is an old, but superior, design.

 

Further points:

 

* The cost advantage of using quasi-complementary symmetry over full complementary symmetry is approximately zero. PNP devices are marginally more expensive to manufacture. Just.

 

* Modern PNP devices are functionally identical to modern NPN devices from the same line. Early PNP and NPN devices were quite different. It is not the case today.

 

* I never suggested that the LM series were crap. They are impressive devices, given their size and cost. It's just that the TDA1514A is superior.

[Newman]

Forget so-called 'Gainclones'. All rely in the LM3886 and derivative chips....The LM series are quasi-complementary symmetry. Quasi comp SS amps have, for all intents, not been used by any amplifier manufacturer since the mid 1970s.

* I never suggested that the LM series were crap. They are impressive devices, given their size and cost.

 

 no comment

[Steve M]

Hi Guys.

The chip amp wasnt a chip amp.

It was a small digitsl amp which had a bit of a cult following (and may still do)?? Called the Virtue one it is a digital amp.

It uses the Tripath 2050 module which msny people thought was the best sounding small power diisl module.

Stock it was okay. I changed all the wiring to 1940's very thin solid core copper eire sheathed in silk. I chaged the resistorsand changed the coupling caps from the Auri caps to some vintage stacked paper and oil types.

This was all done to give the Virtue amp a little more body and soul. It was a lovely little amp and great value for money.

But it will never compete on mega efficient horns with a single ended amp using a 6gw8 amp i also heavilly modified. The lile 6gw8 is just jaw dropping if it can drive the speakers. The valve has a driver tube and power tube in the same glasse nvelope. The circuit has almost nothing in it. This makes for. Crazy short signal path along with the very best of sublime valve sounds. Amazing tone and really vivid. Much less lush than the typical big bottle single ended amp like the300b, type 50 etc. but it only puts out sbout 1.5 watts.

Ive also had great experiences with gainclones. But not all Gainclones are created ewuaand i found that small changes had big effects on the sound. Depending on your dydtem and other components you can strip them so there isnt much left. Just be careful of oscillation and stability.

But eit the right speaker the best of these can still have lots of magic. Im not saying they are going to compete with the very best valve smps but in the right system they can sound amazing providing veryhigh level sound and the same sort of signature as the more modern SET sound. They are a great match for speakers that are a little deada nd dark and slow (think Audionote speakers for which they can be amazing match).

May be amazing on really efficient horns but the horns but maybe not. The horns may reveal the truth too much whch may expose the gainclones. On the flip side you could say that the gainclones may be a little too clear and clean for really efficient horns (Avantgardes etc) and highlight the impossibility of making a multi way horn speaker with horns different lengths and depths sound coherent. A bit of valve colour is nice to vaseline coat the horns so to speak and being out the magic of the speakers.

Back on topic with chip amps: i had some deeply involving times (sometimes) with a sympathetic system powered by the chip amps and enjoyed a few years of modifying them.

Theres more thsn one way to Rome (but one of the fastest ways is with a vslve direct drive electrostatic speaker which can do 25hz-50khz at 110db) :-)

LPG.

 

 

I think one of the reasons chip amps can sound so good for the money isnt the number of components but the length of the signal.

I've never heard ultimate soind from a chip amp. But i have heard superb sound and for very very little money.

I remember with a friend comparing a lm3875t chip amp to the Audionote Conquest (or conqueror? It was the dearer mono block amps) and also against the Wavelenth Audio Napoleon amps (the ones with the silver output transformers which ten years ago where $20k in Australia).

Driving some Audionote Anj with out oard power supplies. In that system on theday the most sublime match was with the little lm3875t amp. The chip amp sounded very similar to the Gordon Rankins Wavelength Audio Napoleon.

The Audionote amps had beautiful tone and sounded totally different to the other amps. It sounded gorgeous but was much too dark and slow sounding on the Audionotes.

But it would be wrong to infer from the above that im saying the lm3875t gainclone amp was better or id amazing. But that particular chip amp in that particular system on the day was. The chip amps have a family sounding but there isa huge difference between the really good sounding ones and the not so good ones. The chips themselves are very complex but the amps themselves only use a handful of parts. Each one of those parta make a huge difference to the sound of the amp.

But for no money as a kit they can be stupid good sounding.

LPG

 

Good posts Bryan and some honest insight into the potential of chip amps ...

 

You and I share more than a few things in common with audio, including hearing something special in chip amps. I agree they are not necessarily world beaters compared to VERY good valve and Class A or AB solid state devices, but into the right loudspeaker - they almost can be.

 

The best chip amp that I have ever heard belongs to Thomo/Jon on SNA, this thing is almost SET like in its openess and really vivid sounding. It is a cobbled together little amp made by Sonic Art here in Australia using the LM3886T chip, which I have heard described as being the same as the LM3875, but with a mute switch built into the chip. The LM3875 is the chip used by Sakura Systems 47 Labs Japan in their famous Gaincard amplifier with RRP $4K, which started the whole trend/fad with audiophile marketed chipamps worldwide. The LM3875 & LM 3886T essentially sound the same to me.

 

I currently own four chip amps a LM3875 (34w/ch), LM4870 (60w/ch) and 2 x LM3886T(34w/ch). The LM3886T amps I built from an Altronics kit, which when I examined it closely was exactly the same as that used in Jon's Sonic Art chipamp. Mine doesn't quite sound as good as Jon's for some reason (being maybe a few percent down in SQ)? My DIY versions use better quality parts being uprated 300VA trafos (160VA required), Sanyo Oscon and Panasonic FC caps and an Alps Blue potentiometer whereas Jon's is a bog standard board and parts, with a cheap plastic $2 volume pot from RS Components ...this frustrates me a bit?? It could be the pot, so will have to swap it out at some stage to see the difference.

 

Anyway, here's a pic of my DIY LM3886T chipamp, which I use for midrange and treble duties on my Edgar Horns. It does a great job because the speaker has 109dB efficiency and therefore does not stress out the chip, keeping distortion in the amp low.

 

 

Steve.

 

Edited November 5, 2013 by Steve M

[Newman]

Nice post Steve. I'm definitely not showing you a photo of my ancient LM1875 effort, not now!

[Full Range]

@ Steve M

In my research I came across you blog and build of your amp

Nice build

[tuyen]

Have a pair of MyRef_C (Ultimate BOM) monoblocks that use the LM3886 chip.  Quite an impressive sounding implementation of the chip, for the DIY enthusiasts out there. 

 

http://www.diyaudio.com/forums/chip-amps/103119-various-questions-diy-noob-regarding-myrefc.html

http://www.diyaudio.com/forums/group-buys/160768-myref_c-ultimate-bom.html

http://www.diyaudio.com/forums/chip-amps/167458-myrefc-build-guide.html

http://www.diyaudio.com/forums/swap-meet/71304-everything-required-assemble-myrefc-kookabura.html

 

Some pics as everyone loves pics:

 

Edited November 5, 2013 by tuyen

[Full Range]

Had a PM conversation by a member that coincidentally lives just a few kilometres away

He is offering a loan of chip amp unit to test it out some time this weekend or next week

Bring it on I say

[hochopeper]

tuyen,

 

the mains voltage at the rear of the IEC socket is exposed ... be careful if you've got them plugged in with the chassis open! Is the chip mounted to the underside of the chassis? The thing that's concerned me about those extruded chassis is that you have zero ventilation so I wonder how things go on the inside of that during summer ... just a thought.

 

 

Cheers,

Chris

[tuyen]

Hi Chris,

 

Thanks for the warnings.   The LM3886T chip is mounted to the right side panel.  Hardly gets warm enough to worry about ventilation.

[A J]

Hi Chris,

Thanks for the warnings. The LM3886T chip is mounted to the right side panel. Hardly gets warm enough to worry about ventilation.

I wouldn't have thought that it would overheat inside as the only real heat generator is the chip which dissipates it to the case. ?

They look pretty nice build hey

[hochopeper]

Hi Chris,

 

Thanks for the warnings.   The LM3886T chip is mounted to the right side panel.  Hardly gets warm enough to worry about ventilation.

 

 

I wouldn't have thought that it would overheat inside as the only real heat generator is the chip which dissipates it to the case. ? They look pretty nice build hey

 

 

tuyen, off the top of my head, yeah that should be okay, the thermal concern is not just for the chip but also for diodes and capacitor temps because the heat inside the enclosure would have a chance of being quite a bit more than what you feel when you touch the outside of the enclosure and on a warm day in Aus it could get reasonably warm. Haven't looked at the calcs for this scenario just mentioning the limitation for a sealed enclosure for this application.

 

Cheers,

Chris

[ArthurDent]

No cooling problems here, just swinging in the breeze

 

 

[THOMO]

Good posts Bryan and some honest insight into the potential of chip amps ...

 

You and I share more than a few things in common with audio, including hearing something special in chip amps. I agree they are not necessarily world beaters compared to VERY good valve and Class A or AB solid state devices, but into the right loudspeaker - they almost can be.

 

The best chip amp that I have ever heard belongs to Thomo/Jon on SNA, this thing is almost SET like in its openess and really vivid sounding. It is a cobbled together little amp made by Sonic Art here in Australia using the LM3886T chip, which I have heard described as being the same as the LM3875, but with a mute switch built into the chip. The LM3875 is the chip used by Sakura Systems 47 Labs Japan in their famous Gaincard amplifier with RRP $4K, which started the whole trend/fad with audiophile marketed chipamps worldwide. The LM3875 & LM 3886T essentially sound the same to me.

 

I currently own four chip amps a LM3875 (34w/ch), LM4870 (60w/ch) and 2 x LM3886T(34w/ch). The LM3886T amps I built from an Altronics kit, which when I examined it closely was exactly the same as that used in Jon's Sonic Art chipamp. Mine doesn't quite sound as good as Jon's for some reason (being maybe a few percent down in SQ)? My DIY versions use better quality parts being uprated 300VA trafos (160VA required), Sanyo Oscon and Panasonic FC caps and an Alps Blue potentiometer whereas Jon's is a bog standard board and parts, with a cheap plastic $2 volume pot from RS Components ...this frustrates me a bit?? It could be the pot, so will have to swap it out at some stage to see the difference.

 

Anyway, here's a pic of my DIY LM3886T chipamp, which I use for midrange and treble duties on my Edgar Horns. It does a great job because the speaker has 109dB efficiency and therefore does not stress out the chip, keeping distortion in the amp low.

 

 

Steve.

 

My Sonic Art chip amp uses the 3876T chip not the 3886T.

I agree the chip amps sound great but mine has always sounded best when used with my Supratek valve preamp which fleshes out the tonal balance nicely.

Used like this it sounds very much like a good modern SET.My preamp uses 300Bs so to me this indicates the chip amp is very tranparent.More so than the vast majority of transistor amps I have tried.

You need reasonably efficient speakers with an 8 ohm load.Mine drives an old pair of B&W DM2As very well.