grubyhalo
New Member
A better option would be to source the PCB's locally from suds. Also, he has got the approval from Pass for his PCB's:
http://www.hifivision.com/diy/4506-working-nelson-pass-first-watt-6.html#post64126
Gainclone vs Tripath T Amps Vs Nelson Pass F5
- Thread starter Hiten
- Start date
A better option would be to source the PCB's locally from suds. Also, he has got the approval from Pass for his PCB's:
http://www.hifivision.com/diy/4506-working-nelson-pass-first-watt-6.html#post64126
grubyhalo
New Member
Hmmm, interesting...
soundsgreat
New Member
Hi,
Hey Gruby,how's you ? you are most welcome my friend
!
One more interesting thing that some may have missed out is,Now you can get the best of both the worlds i,e chips and discreet by using both!!
The latest series of LM chips are now becoming very popular in highend circles as frontend drivers ! you can use LM49810 and such as frontend driver to drive the discreet deviceshyeah: !!
One such small 70W compact Amp is being discussed here : Compact Sized LME49810/11 +ThermalTrak Amp
If you are one of those who want to move your neighbors out and shake the foundation of your house there are versions which pump 300+Watts with multiple discreet devices driven by the said LM chips !
Regards.
Hey Gruby,how's you ? you are most welcome my friend
!One more interesting thing that some may have missed out is,Now you can get the best of both the worlds i,e chips and discreet by using both
!! The latest series of LM chips are now becoming very popular in highend circles as frontend drivers ! you can use LM49810 and such as frontend driver to drive the discreet devices
hyeah: !!One such small 70W compact Amp is being discussed here : Compact Sized LME49810/11 +ThermalTrak Amp
If you are one of those who want to move your neighbors out and shake the foundation of your house there are versions which pump 300+Watts with multiple discreet devices driven by the said LM chips !
Regards.
The nested topology has been attributed to Walt Jung,
and may have come up earlier too. You should be
able to find the original paper here -
Services
As mentioned by Soundsgreat, I did build the
Mauro Penasa gainclone amp. It is a very satisfying and
musical amp. The gainclone is excellent with the
smaller recession buster speakers. I definitely
prefer this one to the Tripath 2024 amp I tried out
for sometime.
The new DIYA group buy Mauro Amp should be even better,
as it is supposed to come with the choicest premium
parts based on experiments by DIYA members
to determine the best caps and resistors etc.
FWIW, I find that my main speakers (Sachiko
Single Driver) are best matched with my KT77 PP tube amp.
and may have come up earlier too. You should be
able to find the original paper here -
Services
As mentioned by Soundsgreat, I did build the
Mauro Penasa gainclone amp. It is a very satisfying and
musical amp. The gainclone is excellent with the
smaller recession buster speakers. I definitely
prefer this one to the Tripath 2024 amp I tried out
for sometime.
The new DIYA group buy Mauro Amp should be even better,
as it is supposed to come with the choicest premium
parts based on experiments by DIYA members
to determine the best caps and resistors etc.
FWIW, I find that my main speakers (Sachiko
Single Driver) are best matched with my KT77 PP tube amp.
Thanks everyone. Searched net to understand the technical terms (feedback loop etc.) Benefits of Negative Feedback
So basically the concept was implemented to increase bandwidth and reduce THD. But gainclone already has low THD and good frequency response values.=> ADX Electronics - Why Gainclones work
Normaly THD of <1% and fq. response 20 hz to 20 khz are accepted, so why go beyond that ? why improve things which we will not be able to hear ?
Edit : Or am I missing something here ?
(feedback loop etc.) Benefits of Negative Feedback So basically the concept was implemented to increase bandwidth and reduce THD. But gainclone already has low THD and good frequency response values.=> ADX Electronics - Why Gainclones work
Normaly THD of <1% and fq. response 20 hz to 20 khz are accepted, so why go beyond that ? why improve things which we will not be able to hear ?
Edit : Or am I missing something here ?
Last edited:
The human ear/brain can tolerate and ignore fairly high distortion as long as it is even-harmonic, and in many cases even-harmonics of sine waves are perceived to be pleasant sounding. However, the ear/brain can perceive very small amounts of odd-harmonic distortion - as little as 0.01% of some odd harmonics will be perceived as fatiguing when heard over a period of hours. It therefore helps to reduce odd-harmonic distortion to very low values - 0.003% or lower is desirable.
The second aspect is that we need to pass at least the 5th harmonic of the highest audible frequency to be able to reconstruct the audio waveform accurately. This is somewhere between 60 kHz and 100 kHz - the entire signal path has to be able to reproduce this accurately.
The third aspect is minimization of phase distortion - the phase of all frequency components up to (say) 100 kHz needs to be relatively constant. The flatter the phase response is, the better. This contributes a great deal to the perception of imaging and soundstage - the ability to locate individual musical instruments precisely with your eyes closed just by listening to the music. The instruments must be perceived as having the same physical location, as they actually had at the concert hall or studio where the recording took place. This usually a much larger (virtual) area than the physical area where the speakers/headphones are actually located during playback.
The fourth factor is that most real speakers/x-overs do not present a constant resistive load to the output stage of the amplifier - instead, there is a frequency-dependent complex impedance presented to the output of an amplifier, and this varies depending on the combination of drivers, crossovers and speaker cabinet design. The output stage has to be able to accurately and stably drive a wide variety of different speaker loads, including difficult loads with large capacitive components.
Getting all these properties (and some more) exactly right is pretty difficult, even with discrete designs. The miracle of modern chipamps is that they do a pretty good job on almost every one of the important properties, and these can even be improved further by choosing the right topology. It's possible to make chipamp circuits that go right into high-end audiophile territory at relatively low costs - the only area where they might find it difficult to challenge discretes is in power-handling at levels above 100W/ch.
The second aspect is that we need to pass at least the 5th harmonic of the highest audible frequency to be able to reconstruct the audio waveform accurately. This is somewhere between 60 kHz and 100 kHz - the entire signal path has to be able to reproduce this accurately.
The third aspect is minimization of phase distortion - the phase of all frequency components up to (say) 100 kHz needs to be relatively constant. The flatter the phase response is, the better. This contributes a great deal to the perception of imaging and soundstage - the ability to locate individual musical instruments precisely with your eyes closed just by listening to the music. The instruments must be perceived as having the same physical location, as they actually had at the concert hall or studio where the recording took place. This usually a much larger (virtual) area than the physical area where the speakers/headphones are actually located during playback.
The fourth factor is that most real speakers/x-overs do not present a constant resistive load to the output stage of the amplifier - instead, there is a frequency-dependent complex impedance presented to the output of an amplifier, and this varies depending on the combination of drivers, crossovers and speaker cabinet design. The output stage has to be able to accurately and stably drive a wide variety of different speaker loads, including difficult loads with large capacitive components.
Getting all these properties (and some more) exactly right is pretty difficult, even with discrete designs. The miracle of modern chipamps is that they do a pretty good job on almost every one of the important properties, and these can even be improved further by choosing the right topology. It's possible to make chipamp circuits that go right into high-end audiophile territory at relatively low costs - the only area where they might find it difficult to challenge discretes is in power-handling at levels above 100W/ch.
Cranky
Regarding Mauros Ref. Gainclone :
Is LM318 used to increase gain, because using nested feedback loop lowers gain of an amp ? If yes, would it introduce its own, shall I say sound characteristics ? Is that why you recommended OPA 548/549 opamps ?
Thanks
Of the 5 nested/multiloop topologies that I simulated extensively (Quad/Current Dumping, Sandman, Technics, Jung and Penasa), the Penasa is the most stable or easiest to stabilize, but the Jung can be optimized for lower distortion. However, it's harder to stabilize than the Penasa - the compensation has to be tuned for the specific opamp/chipamp combination.
I tried plugging in an OPA541 model into the simulation of a Penasa-type gainclone with an NE5532 as the front end, and it was disappointing in comparison to the LM1875. The OPA541 gets to about -66 dB THD20 (~0.05 % distortion), while the LM1875 can get
below -100 dB THD20 (~0.001% distortion), almost entirely even-harmonic. The OPA541 does have the advantage of being unity-gain stable, but that comes at a price: an over-stabilized dominant pole that contributes to poor highs and generally dull sonics.
Basically, the sonics of the LM1875 in a Penasa topology are close to magical - but it's limited to about 20W power into 8 ohms. Using 2 x LM1875 in parallel increases this a bit - to about 30 - 40W at 0.001%, which is great; but the power and dynamic range are still below the LM3886/LM318 combo in the Penasa Rev C.
If you look at the economics, the LM3886 or TDA7294 win outright - they get to the 60W/channel territory at Rs.200 or below. The next possibility is a paralleled 2 x LM3886 (or a single LM4780) per channel, which gets to the 100W/channel territory at reasonable cost. This is possibly the way to go for a super-premium gainclone at a BOM of less than Rs.1k/channel. That's what I'm trying to investigate next, but I can't find a reliable LM3886 SPICE model. It will probably work without major changes within the existing 2 x LM1875 compensation schema which I've simulated, but it nevertheless has to be prototyped, auditioned and compared to the LM3886/LM318 Penasa Rev C - which is the reference gainclone of choice today.
I tried plugging in an OPA541 model into the simulation of a Penasa-type gainclone with an NE5532 as the front end, and it was disappointing in comparison to the LM1875. The OPA541 gets to about -66 dB THD20 (~0.05 % distortion), while the LM1875 can get
below -100 dB THD20 (~0.001% distortion), almost entirely even-harmonic. The OPA541 does have the advantage of being unity-gain stable, but that comes at a price: an over-stabilized dominant pole that contributes to poor highs and generally dull sonics.
Basically, the sonics of the LM1875 in a Penasa topology are close to magical - but it's limited to about 20W power into 8 ohms. Using 2 x LM1875 in parallel increases this a bit - to about 30 - 40W at 0.001%, which is great; but the power and dynamic range are still below the LM3886/LM318 combo in the Penasa Rev C.
If you look at the economics, the LM3886 or TDA7294 win outright - they get to the 60W/channel territory at Rs.200 or below. The next possibility is a paralleled 2 x LM3886 (or a single LM4780) per channel, which gets to the 100W/channel territory at reasonable cost. This is possibly the way to go for a super-premium gainclone at a BOM of less than Rs.1k/channel. That's what I'm trying to investigate next, but I can't find a reliable LM3886 SPICE model. It will probably work without major changes within the existing 2 x LM1875 compensation schema which I've simulated, but it nevertheless has to be prototyped, auditioned and compared to the LM3886/LM318 Penasa Rev C - which is the reference gainclone of choice today.
grubyhalo
New Member
Too bad! Did you order the UcD models from Hypex or are these something else?
~ Linuxguru
Regards
As I have small room Power rating of 30-40W will be sufficient for me. Lets see how much the group buy cost turns out. also have seen one LM3886/LM318 kit on ebay for about Rs. 3000. But I think components will not be as good as 'Group Buy' kit.
I inquired through a friend at Lamington Road. LM3886 was for about Rs. 350 and LM3875 for about Rs. 325.
Regards
Vaguely related - there are a large number of "YuanJing" buffered gainclone PCBs, kits as well as assembled boards available on EBay from various sellers in China and Hong Kong. It has some design issues, some of which can be worked around to make a decent-sounding (if not audiophile grade) gainclone at prices below Rs.1k for a stereo board.
Here's my post at diyaudio listing some of the problems and fixes:
"Yuanjing" Gainclone 3886 - eBay amazing value ? - Page 3 - diyAudio
Note that this is not a nested gainclone, but only a buffered gainclone. However, it is a reasonable "starter" kit for a first-time DIY kit-builder, which will give acceptable results at first shot for most builders, and can be significantly improved as detailed above for those who want better sound.
Here's my post at diyaudio listing some of the problems and fixes:
"Yuanjing" Gainclone 3886 - eBay amazing value ? - Page 3 - diyAudio
Note that this is not a nested gainclone, but only a buffered gainclone. However, it is a reasonable "starter" kit for a first-time DIY kit-builder, which will give acceptable results at first shot for most builders, and can be significantly improved as detailed above for those who want better sound.
Linuxguru thanks for sharing the risk involved in buying such kits.
I was refering to this ..
Classic lines LM3886 dual-channel +LM318 AMP/ DIY - eBay (item 320478341831 end time Feb-21-10 03:45:26 PST)
I dont know if it is nested or buffered. The seller has not given any details and seems to be selling from China so very much doubtful about quality of components.
One should rather have plain GC kit from Peter Daniel or group buy you mentioned.
I was refering to this ..
Classic lines LM3886 dual-channel +LM318 AMP/ DIY - eBay (item 320478341831 end time Feb-21-10 03:45:26 PST)
I dont know if it is nested or buffered. The seller has not given any details and seems to be selling from China so very much doubtful about quality of components.
One should rather have plain GC kit from Peter Daniel or group buy you mentioned.
I haven't seen that one first-hand, but it's possibly an unauthorised knock-off of the Mauro Penasa Rev A stereo board. The Rev A was fairly good, but the Rev C monoblocks are much better overall - the layout as well as the compensation schema is different on the Rev C boards, and sonically it's a fairly major improvement over the Rev A.
However, it is possible to build a Rev C on a Rev A bare board with some kludged wiring modifications, so you can always get a Rev A bare PCB and assemble a Rev C with it. It won't hurt to try it out if it costs less than say $20 or so.
However, it is possible to build a Rev C on a Rev A bare board with some kludged wiring modifications, so you can always get a Rev A bare PCB and assemble a Rev C with it. It won't hurt to try it out if it costs less than say $20 or so.
Here we have a better pricing LM3886 are 170/- each..
rocket_sled
New Member
Would these be good for subwoofer duty then ?
There are a few variants listed, which one did you get and what was the approx landed price.
grubyhalo
New Member
Would this work for you?
Class D Audio Class D Power Amplifier, 250W X 2, 500W X 1 - Home and Pro Audio - Amps
rocket_sled
New Member
Looks good, i'll check it out !
For excellent sound that won't break the bank, the 5 Star Award Winning Wharfedale Diamond 12.1 Bookshelf Speakers is the one to consider!
Similar threads
