I hope that someday, the MyRef line of amplifiers gets a Driver Chip Amp like that one.
I wonder why IRFPs seemed better to my ears than 2SJ201/2SK1530. I am not sure what kind of power topic started wants to acquire, but if it is about sweetness and simplicity, why not to make some SE class A amp. Or crazy regulator chipamp for fun. There are endless options, and still only you can determine which way you want to go. I think every DIYer have experienced some kind of addiction of some specific type of amps. For some these are chips, for some hybrids, some SE. And your mind will just fool your ears if you will believe that the one you chose is the best. At least that is what I've experienced, subjectively sounds better, but I didn't have any proof. Sorry about ofttopicy monologue.
The point is, that if an amp, or system is really accurate then the sweetness or otherwise is totally dependent on the recording, has nothing to do with the playback system. For example, I should be able to put on a recent, driving blues recording, and it should have tremendous crunch, bite, visceral kick; then a string serenade immediately after and I start floating off with fairies ... that's how audio should work ...
Frank
Frank
For me the sweetness of an amp comes down more to its PSU and decoupling, than the chip used. I've had sweet sound with LM3886, TDA7265, TDA7293/4, TDA2030A, LM4766 but only by ensuring the grounds aren't contaminated by noise from the PSU. Once that's fixed then the next stage is getting sweetness with depth and that seems to require better filtering of HF ripple on the PSU as no chipamp I've seen has good PSRR at higher frequencies - the best peg out below 50dB @ 20kHz.
Absolutely true, thanks to sum it up.
A Entirely divergent view??
After 40+ years of 'dicking' with Audio I've learned the hard way (invariablythat ones' Speakers are 'at least' 85/90% of the sound produced.
Seriously: Invest in Speakers that do ..Everything.. you want before wandering off on equipment tangents.
I've been involved /responsible for a 'lot' of bespoke hi end AV setups in Clients' homes, typically north of $250K worth, serious coin.
I've learned /observed that electronics ie: Amplification.. beyond a certain quality threshold point
Can only suggest reordering ones' priorities for best result for $$ /time expended. ..as the First step Buy the best speakers that you can't afford
Poor? Save up for years if needs be, It's Basic and worth it.
PS a 6.5 " driver is Not even close to providing accewptable Bass.
Last edited:
Actually a double 6.5" or double 7" with a parallel chip amp can be quite amazing on output power (when used in normal living rooms). That's just barely enough for trouser flapping bass. But, the cost of speaker+amp is less than $100 per channel.
The Duntech Soverign is not especially economical, but fortunately most normal size rooms don't need that much speaker.
Exclusivity market pricing is not the only path to high performance. Fair pricing is according to the size area to fill with audio.
The double-woofer speaker is actually the economy performance champ since the reduced crossover price, reduced crossover noise and double cone area move compromise out of the speaker and into the amplifier where the compromise can be eventually defeated by bracing up the amplifier to handle the higher current.
I think we'll see a bit better show with the MyRef EVO2 120w (and my new TDA7294 80w) after they have completed testing and become more widely available to drive these double-woofer speakers.
The Duntech Soverign is not especially economical, but fortunately most normal size rooms don't need that much speaker.
Exclusivity market pricing is not the only path to high performance. Fair pricing is according to the size area to fill with audio.
The double-woofer speaker is actually the economy performance champ since the reduced crossover price, reduced crossover noise and double cone area move compromise out of the speaker and into the amplifier where the compromise can be eventually defeated by bracing up the amplifier to handle the higher current.
I think we'll see a bit better show with the MyRef EVO2 120w (and my new TDA7294 80w) after they have completed testing and become more widely available to drive these double-woofer speakers.
Last edited:
Daniel, I'd be interested in your view on how the parallel LM3886 amps will change the performance of the Sunflowers. They have a single 8 ohm woofer but 4 ohm for the other two legs. Matched with the current MyRefs (FE) the performance is outstanding. I have not had to use my sub since they were built.
Not quite true, the stereo sounds as good as good playing it's worst component.
I agreed that speakers and room acoustics is what most home stereos need to have improved first. But once you have decent speakers you need a decent amp and preamp to reveal the quality and good DAC as well + good recordings.
Usually person with good (absolute hearing) ends up with LP (class A DAC) and class A horns/line arrays or planar/ESL.
Subs and conventional consumer speakers deliver terrible booming bass. Best bass that preservers timber I heard so far was from Magnepans. That speakers let you enjoy very realist pipe organ recordings reproduction at home once you find amps that can let them sing.
The Sunflower has a light load. The MyRef-FE will drive it easily. With the Sunflower, 4 ohms for midrange and treble is not a problem because the signal isn't very tall, and so the demand isn't tall either. The chip is not strained. The speakers are suitable for that amp.
Let's take a look at what a MyRef-Evo2 can do. It will send some spike circuit noise (that hallmark midrange tone of the Overture series) into the ballast resistors where you can't hear the noise. It will also send some of the crossing noise into the ballast resistors where you can't hear the noise. Any distortion that no two chips do at precisely the same moment in time, in precisely the same way, will be sent to the ballast resistors. The ballast resistors themselves will help give the chip amp the more lavish sound of a better discrete amp (with similar build involving parallel output devices). What you get, (in comparison to the MyRef-FE) is the darker voice treatment and the mids a little more laid back and the treble a bit more refined. The parallel amplifier's more ear friendly arrangement will be more fun during loud playback.
The nested and current pump designs with their nested feedback loops already send some of this noise away by same proportion as the global loop(s). The parallel outputs will only help improve performance further. But, the tone will be different and thus your ideal component selections will be different. At the power circuit, you might not need 1000u amp decoupling caps to get rid of the overture series loud mids, since some of that noise is caught in the ballast resistors. Perhaps 680u or 470u (low esr compact) will be needed to perk up the Evo2.
To level out the situation, you use bigger amp board decoupling caps for shouty amps, smaller amp board caps for laid back amps. That's the major step to level out the harmonic response. Really good power filtering can allow the use of even smaller caps in a sort of have your cake and eat it too--clearer and more open without being more forwards. A more elegant power circuit could be important with the Evo2 since that amp doesn't need as much power side dampening/dulling to overcome its tamer mids. National Semiconductor recommends you try 470u caps for amp board decoupling and maybe that's large enough for the parallel amp. Hot tip: The low bass and bari of the LM3886 is gorgeous if we can just get the mids under control so we can utilize the big clear bass feature. Parallel amp is a step in the right direction.
After you have the two amplifiers in the comparison playing with a really similar frequency response, any remaining difference is that the current headroom of the parallel amp can sound effortless and lavish as if you spent a lot more for the amp.
Being naturally less forwards in tone, the MyRef-Evo2 is perhaps less suited for extremely quiet playback; however, it is less likely to be exhausting for longer and/or louder listening sessions.
An additional difference, according to AN1192.pdf from National Semiconductor, is that the spike system limiter may or may not begin making noises for all excursion greater than 45 watts. This is postponed to 80 watts with the parallel amp. So, we should have tamer mids. . . which can allow for better quality tuning of the power circuit for clearer audio with more lavish bass bari and treble, if you choose to utilize the tolerances that way.
Last edited:
My apologies Daniel, I saw this when you posted it and never got back till now.
You are getting me very excited about the EVO2. I talked to Siva this morning on another matter and he is planning to solidify the MiniRef first.
I had been thinking about doing a bi-amp with the FE on top, but from your information, it appears that won't be necessary. Can't wait till the amp shows and has it's own dedicated thread.
To shamelessly display my lack of knowledge to the world, with your usual clarity could you give a layman's description of "decoupling"? It's one of those things that comes up so often but I haven't asked my friend Mr. Google about it yet.
You are getting me very excited about the EVO2. I talked to Siva this morning on another matter and he is planning to solidify the MiniRef first.
I had been thinking about doing a bi-amp with the FE on top, but from your information, it appears that won't be necessary. Can't wait till the amp shows and has it's own dedicated thread.
To shamelessly display my lack of knowledge to the world, with your usual clarity could you give a layman's description of "decoupling"? It's one of those things that comes up so often but I haven't asked my friend Mr. Google about it yet.
Neither I nor Google would make any sense of it.
So, here's Tom on power decoupling:
This example infers that there may be an umbilical cable, diode drop, or resistor in-between the 10,000u tank and 1000u decoupling caps on the MyRef board, so that large signal demand comes from the 10,000u and small signal demand comes from the 1000u, but that is not the case if the umbilical cable has been replaced with a 1cm trace. On these integrated amp+power boards, the amp decoupling caps lose charge every time the tank caps need charge. This may be pretty for tone but it is terrible for clarity. I would try a resistor or diode drop instead of shorting out the decoupling with a short length trace. That applies to EVO2, since it is less needy on tone but could use a boost in clarity. To further level out the differences between the FE versus the EVO2, it is highly likely that the amp power decoupling caps need stepped down to 680u or in any case somewhat smaller size than the single chip amp.
Let's look at power supply design for a moment.
Transformer, Bridge Rectifier, Smoothing caps, loss, Tank caps, loss, Decoupling caps, trace, bypass caps, amp. The above mentioned loss may be a cable, light bulb, fuse, transistor, long trace, diode, or resistor. They're all different, but the most important difference is that a ballast loss exists. For each loss that is shorted/omitted, then TWO power supply features are hindered.
So, when I look at the MyRef boards, I think that something needs installed series between the 10,000u and 1000u. On first guess, I'd try a fast silicon diode. Now, the power won't get stolen from the 1000u until the rails droop 1/2v. A 0.22R resistor may also work since the power would come out of the transformer and into the 10,000u more readily than coming out of the 1000u through 0.22R and into the 10,000u. SO, there are some compact options that don't have to involve an umbilical cable.
Crazy trick: A 1 ohm 1/2w resistor parallel to a fast silicon diode will give you a tiny amount of inexpensive reverb and a lot of noise filtering capacity. This exchange of decreasing disharmonic noise for increasing harmonic noise is generally attractive and worth an interview because it is also low loss. However, I wanted to mention it isn't totally noise free--just a lot more likely to be attractive. The resistor part gives the feature of sending noise away from the amp--a feature you wouldn't get with the diode alone (except circuitously); however, the speed of the diode is not the same as the speed of the resistor and you may get a tiny dose of reverb. I'd be sorely tempted to try that on a MyRef, since it is quite likely to outperform an umbilical cable, and the total amount of noise does not increase, and disharmonic noise will decrease. That's a decent feature.
Edit: The D//R unit goes between the 10,000u and 1000u, instead of an umbilical cable.
Edit2: That method is unorthodox--The orthodox solution for clarifying integrated power+amp boards is to install regulators.
Unfortunately, the answer I have to give for the MyRef board, on the topic of power decoupling, is that you didn't install it yet. EDIT3: To install, you'd put in something that looks like CRC.
A possible exception (to missing power decoupling) is if using Standard 10,000u in competition for charge with Low ESR 1000u and then possibly, the more demanding Low ESR 1000u can hold some charge whilst the 10,000u are charging. But, CSD (component specific design) is time consuming and it is problematic if a specified part is discontinued or otherwise hindered in availability. A bit more comprehensive design could make component selection much easier.
Last edited:
Meanwhile, back to Tom's quote above and also referencing National Semiconductor AN1192 specs: 470u decoupling with 100n bypass. They also list a power supply board, meaning that an umbilical cable is used. If it had been all in one board, it would be 470u//100n bypass since the decoupling and the cable that does it have come up absent. The sound is far different.
Probably the solo chip LM3886's benefit from the dark voice and noise of the integrated power+amp combination board so as to tame the mids for which that chip has a surplus. Maybe all of this typing would have been better if applied to a different amplifier, such as the TDA7294 with 220u decoupling and 4.7n bypass? I do not know.
The only thing I actually know is that you do have options to explore and that power circuit is of equal importance to amplifier circuitry because they are totally interrelated. Any audio feature or effect that you can think of can be installed at either a power board or an amplifier board, because whatever you do to the reference is also done to the signal.
Probably the solo chip LM3886's benefit from the dark voice and noise of the integrated power+amp combination board so as to tame the mids for which that chip has a surplus. Maybe all of this typing would have been better if applied to a different amplifier, such as the TDA7294 with 220u decoupling and 4.7n bypass? I do not know.
The only thing I actually know is that you do have options to explore and that power circuit is of equal importance to amplifier circuitry because they are totally interrelated. Any audio feature or effect that you can think of can be installed at either a power board or an amplifier board, because whatever you do to the reference is also done to the signal.
Hi All,
Four weeks to ship and a few hours to build and I finally can hear a TDA7294. The kit is quite simple compered to the LM3886 MyRefs I'm used to. However, I am very impressed with the sound it produces. A short description would be warm, powerful (in my build), accurate and a bit relaxed. The bottom is cleaner and better defined at this "stock" level than the standard LM builds I have done, but not as powerful. I can clearly understand why so many like the character of this chip.
So let me do a few "IMHO" comparisons with the MyRefs and in particular thr Fremen Edition. What stands out the most is the difference in headroom. Below a particular volume level the TDA really shines. Anything above that shows a lot of inter-modulation distortion, harshness and strain. The last few versions of the MyRef stay clean all the way to max. I'm sure there is a lot more sophistication in the power supply technology for the MR than what this TDA build has now. It would be interesting to hear this amp with something specifically designed to power it.
Unfortunately, there are no props for the TDA's top end when compared to the MyRefs. Again, that may not be a fair comparison as the kit I built is bare bones from my understanding. The MR FE is silky, smooth, defined, airy and shimmering on all types of music. If you are familiar with it, Ravel's Daphnis & Chloe has a wide range of dynamics, instrument combinations and volume levels. As long as everything is soft to moderate the TDA sounds beautiful. It just doesn't take long/much for louder and higher range portions to get muddy and a touch blaring. Bells and chimes break through brightly during crescendos and loud finales, but just below that it gets pretty mushy. Some of that may be resolved with higher grade components on the build, which I hope to do soon.
What I hear in the dynamics in percussion and general mid-rang regesters is very impressive. Without any specific knowledge or technical background, I would suspect the TDA is simply "faster" than the LM3886. If anyone has that knowledge please pass it along. There is a real element of genuine excitement produced by this chip, particularly with wild classical tracks.
So the few hours I've owned a TDA7294 have been very enjoyable to say the least. I did run several passes of the Isotek burn-in cd that cleaned away some fuzz that was there at first power-up. This kit is capable of a BTL configuration and I am very interested in trying that as well as some of the other dual chip setups Daniel W. has suggested. At $30 for a two board kit with all the necessary do-dads, one can't go wrong and a second set won't break the budget.
Thumbs Up for the TDA7294
And....Happy Holidays Y'all.
Four weeks to ship and a few hours to build and I finally can hear a TDA7294. The kit is quite simple compered to the LM3886 MyRefs I'm used to. However, I am very impressed with the sound it produces. A short description would be warm, powerful (in my build), accurate and a bit relaxed. The bottom is cleaner and better defined at this "stock" level than the standard LM builds I have done, but not as powerful. I can clearly understand why so many like the character of this chip.
So let me do a few "IMHO" comparisons with the MyRefs and in particular thr Fremen Edition. What stands out the most is the difference in headroom. Below a particular volume level the TDA really shines. Anything above that shows a lot of inter-modulation distortion, harshness and strain. The last few versions of the MyRef stay clean all the way to max. I'm sure there is a lot more sophistication in the power supply technology for the MR than what this TDA build has now. It would be interesting to hear this amp with something specifically designed to power it.
Unfortunately, there are no props for the TDA's top end when compared to the MyRefs. Again, that may not be a fair comparison as the kit I built is bare bones from my understanding. The MR FE is silky, smooth, defined, airy and shimmering on all types of music. If you are familiar with it, Ravel's Daphnis & Chloe has a wide range of dynamics, instrument combinations and volume levels. As long as everything is soft to moderate the TDA sounds beautiful. It just doesn't take long/much for louder and higher range portions to get muddy and a touch blaring. Bells and chimes break through brightly during crescendos and loud finales, but just below that it gets pretty mushy. Some of that may be resolved with higher grade components on the build, which I hope to do soon.
What I hear in the dynamics in percussion and general mid-rang regesters is very impressive. Without any specific knowledge or technical background, I would suspect the TDA is simply "faster" than the LM3886. If anyone has that knowledge please pass it along. There is a real element of genuine excitement produced by this chip, particularly with wild classical tracks.
So the few hours I've owned a TDA7294 have been very enjoyable to say the least. I did run several passes of the Isotek burn-in cd that cleaned away some fuzz that was there at first power-up. This kit is capable of a BTL configuration and I am very interested in trying that as well as some of the other dual chip setups Daniel W. has suggested. At $30 for a two board kit with all the necessary do-dads, one can't go wrong and a second set won't break the budget.
Thumbs Up for the TDA7294
And....Happy Holidays Y'all.
Last edited:
Well, that's actually a really good result for a basic kit in stock condition. Both the gain and the feedback current may be a bit high and the NFB cap and bootstrap may be a bit small. That combination causes it to miss the highest highs and lowest lows, but that is what the datasheet example schematic does.
Try the bootstrap cap directly at 47u for slightly more low bass.
(some newer schematics show 47u)
And, try the NFB cap at least as large as 220u (or bigger)--Also try some tiny value polyester caps and/or electrolytic of 0.47uf or less, as bypass caps parallel with your NFB cap.
Then the lowest lows and highest highs should be working.
(the schematic shows 22u, which is a bass blocker)
Also for low bass this amplifier appreciates a large power supply reservoir capacitance, not located on same board as amplifier.
If it still does some midfi behavior, the gain might be set too high, since it doesn't really have enough gain capacity for a 70w amplifier. The quality drops if the gain is too high. If you want hi-fi, I think you want low gain power amp pushed by a preamp.
The feedback current might be too high due to feedback-shunt and feedback resistor values too small, which also causes fragility and current noise of overloading the small signal section beyond linearity.
To make a reasonable compromise, you could set the feedback-shunt resistor to 2k2 and the feedback resistor to 56k (or even 68k if going without a preamp). Nearby values make really similar results. This ballpark is a good spot to start exploring. That example is off a bit, but not nearly as far as the datasheet schematic.
Before using the amplifier in bridge mode, I'd like to see it running quite happily clear and definitely running cool temps. The operating voltage for bridge mode is with a 20+20vac transformer (max). So, if you're going higher than that, do consider output caps or speaker protector kit. I did test those baby latfets to 127 watts before I ran out of transformer, but I was using speaker protection, just in case.
Try the bootstrap cap directly at 47u for slightly more low bass.
(some newer schematics show 47u)
And, try the NFB cap at least as large as 220u (or bigger)--Also try some tiny value polyester caps and/or electrolytic of 0.47uf or less, as bypass caps parallel with your NFB cap.
Then the lowest lows and highest highs should be working.
(the schematic shows 22u, which is a bass blocker)
Also for low bass this amplifier appreciates a large power supply reservoir capacitance, not located on same board as amplifier.
If it still does some midfi behavior, the gain might be set too high, since it doesn't really have enough gain capacity for a 70w amplifier. The quality drops if the gain is too high. If you want hi-fi, I think you want low gain power amp pushed by a preamp.
The feedback current might be too high due to feedback-shunt and feedback resistor values too small, which also causes fragility and current noise of overloading the small signal section beyond linearity.
To make a reasonable compromise, you could set the feedback-shunt resistor to 2k2 and the feedback resistor to 56k (or even 68k if going without a preamp). Nearby values make really similar results. This ballpark is a good spot to start exploring. That example is off a bit, but not nearly as far as the datasheet schematic.
Before using the amplifier in bridge mode, I'd like to see it running quite happily clear and definitely running cool temps. The operating voltage for bridge mode is with a 20+20vac transformer (max). So, if you're going higher than that, do consider output caps or speaker protector kit. I did test those baby latfets to 127 watts before I ran out of transformer, but I was using speaker protection, just in case.
Last edited:
Thanks Daniel, this chip contains a lot of potential if what I'm hearing is truly stock performance. It definitely fits well within the definition stated in the thread title. Even with the shortcomings I perceived, I was enticed to listen to some major productions all the way through with its warm and dynamic presentation.
For the benefit of myself and others, hopefully we can go through the adjustments you list one by one on the thread after the first of the year. Though I am gaining a degree of understanding of terminology and function, I still need a little "paint by the numbers" direction to clarify the process.
I did purchase a pair of the Peter Daniels Universal Power Supply boards but won't use them for what I had planed. Don't know if they are capable of what you propose for this chip, but they are available.
Looking forward to drilling down on this project.
For the benefit of myself and others, hopefully we can go through the adjustments you list one by one on the thread after the first of the year. Though I am gaining a degree of understanding of terminology and function, I still need a little "paint by the numbers" direction to clarify the process.
I did purchase a pair of the Peter Daniels Universal Power Supply boards but won't use them for what I had planed. Don't know if they are capable of what you propose for this chip, but they are available.
Looking forward to drilling down on this project.
Attachments
Sure, you can give it its own thread, and introduce it, along with the schematic. I've attached the schematic so that it doesn't disappear. Four of the caps are undersized and two of the resistors have too much current, and the funny thing is that the caps are too small to control where the current is supposed to go. There's a little traffic jam, which is the source of the "muddy" that happens when you try to play loud. It just needs some really basic/generic audiometric tuning to get it in the ballpark (of playing both loud and well) before you begin other fine tuning.
For power supply, we could do the "ladderback" style which is both basic and excellent. That is built with multiple parallel 3300u (at least 8 of them--used as 4 per each rail) on ordinary copperless phenolic perfboard. It would be a great introduction and works well with that amp. This is just recifier+tank, a simple power supply, but these are not dull. Well, it is a good place to start.
Attached is your amp schematic in "stock condition":
Attachments
![!Bz9HT2Q!2k~$(KGrHqJ,!h4Ew5Hd,rhsBMYPP-RcRQ~~_3[1].JPG](/community/data/attachments/295/295189-071f06a504d8d08081c42e940d833ddc.jpg)
Last edited:
Hmmm... I'm not so sure, but I guess LM1875 sound is terribly wrong. I like TDA2030 more. It is much more accurate imo. They do sound different, but I think preferring LM1875 is wrong
Just kidding 