Idiot... nah, you are hard with yourself! Wouldn't say that... hey, and at least you were right on something: they probably had a much lower ESL per definition ;-)
I am sorry about that expense: it is easy to get confused after a long day at work when ordering, happens to the best. Whereas 0.01uF can be useful in some audio applications, as you have guessed for the kind of frequencies we are dealing with this is too low and impact is probably minor.
Now, IMHO it doesn't really matter what make you go for as long as close to 0.1uF and please C0G and not X7R etc.
Murata turned out to be fine as I know they work great from past experience plus they are very small and affordable. But you can go for whatever equivalent, should do the job here.
The only part where I can't stand 100% for other components is the OPA PS bypass where we really ended up cooking strangely with different spices (read components) to tune to our ears, making the most of what seemed to be a quite defficient PS to start with.
Good luck and again the positive is you spotted it, so you have even more room for sonic improvements 🙂
Keep the good work and keep us posted, you ARE DOING great!
Claude
I am sorry about that expense: it is easy to get confused after a long day at work when ordering, happens to the best. Whereas 0.01uF can be useful in some audio applications, as you have guessed for the kind of frequencies we are dealing with this is too low and impact is probably minor.
Now, IMHO it doesn't really matter what make you go for as long as close to 0.1uF and please C0G and not X7R etc.
Murata turned out to be fine as I know they work great from past experience plus they are very small and affordable. But you can go for whatever equivalent, should do the job here.
The only part where I can't stand 100% for other components is the OPA PS bypass where we really ended up cooking strangely with different spices (read components) to tune to our ears, making the most of what seemed to be a quite defficient PS to start with.
Good luck and again the positive is you spotted it, so you have even more room for sonic improvements 🙂
Keep the good work and keep us posted, you ARE DOING great!
Claude
Last edited:
I ended up just getting the TDK .1uf C0G. I also got a bunch of Nichicon UFG caps to replace all the remaining electrolytic s in this amp.
More info to follow........
More info to follow........
A link to this TDKs perhaps? Always interesting to have alternative products when some are on back orders, so if you are happy with them worth for us noting them down 🙂
Replacing all caps, well, why not, entertaining project. Just a few things:
- with our 2 units, desoldering some components turned out to be VERY difficult despite top equipment. In fact, we never succeeded re volume board. Plus the board doesn't like abuse at all
- We adressed nearly all caps that were significant (IMHO), and some were worth adressing, others less so. Perhaps a good starting point to avoid diminishing returns is looking at those we dimmed interesting to tackle?
- Bypassing and adding capacity without removing the existing caps gave us probably 90% of the achievable result. The place where this sentence would have been the most challenged would have been of course the caps in the signal path, but for whatever reason and in an unexpected manner the amps didn't react a lot to that... including taking them completely out (best cap is no cap LOL)
These were our findings but it shall be interesting to read yours. Do you have perhaps a spare unit to allow direct sonic comparisons? Desoldering can be long (those output filter caps) and the best way to hear if there is an improvement you like is comparing 2 otherwise identical units. Also some parts require some burn-in.
Last but not least, I would make sure whatever mod I do is reversible, in case of. Oh, and while taking caps out and before soldering the new one, check carefully the tiny tracks... they really don't like any abuse when desoldering (or was in un-gluing in some case LOL).
Great project, keep us posted
Claude
Replacing all caps, well, why not, entertaining project. Just a few things:
- with our 2 units, desoldering some components turned out to be VERY difficult despite top equipment. In fact, we never succeeded re volume board. Plus the board doesn't like abuse at all
- We adressed nearly all caps that were significant (IMHO), and some were worth adressing, others less so. Perhaps a good starting point to avoid diminishing returns is looking at those we dimmed interesting to tackle?
- Bypassing and adding capacity without removing the existing caps gave us probably 90% of the achievable result. The place where this sentence would have been the most challenged would have been of course the caps in the signal path, but for whatever reason and in an unexpected manner the amps didn't react a lot to that... including taking them completely out (best cap is no cap LOL)
These were our findings but it shall be interesting to read yours. Do you have perhaps a spare unit to allow direct sonic comparisons? Desoldering can be long (those output filter caps) and the best way to hear if there is an improvement you like is comparing 2 otherwise identical units. Also some parts require some burn-in.
Last but not least, I would make sure whatever mod I do is reversible, in case of. Oh, and while taking caps out and before soldering the new one, check carefully the tiny tracks... they really don't like any abuse when desoldering (or was in un-gluing in some case LOL).
Great project, keep us posted
Claude
Here are the TDK's I ordered......
TDK FA16C0G1H104JNU00
Here are the very nice (and very expensive) KEMET low ESL ones in case someone wants to try them
KEMET C350C104J1G5TA
And for sure were rapidly getting to diminishing returns here. I will submit the 2 most important changes if one does nothing else will to be to replace the main caps with high quality, low ESR ones af AT LEAST 3000uf capacity and changing the op-amps to whatever flavor you like. The bypasses definitely help too, but for me, soldering in such small areas with such tiny components was more difficult than just replacing the main caps and it took way longer.
Like I said when I started this, I have zero electronics experience. I havent touched a soldering iron in 20+ years and that was only for things like speaker wires, not delicate PCB's so if I can pull this off, anyone with the motivation can. With this lack of experience though it means I dont really even understand the layout of an amp PCB or what all the components do or are for. My working theory is all the caps (likely all the components) are fake and will be improved in whatever job they have by replacing them with genuine, high quality ones. So thile the remaining 10-12 electrolytic caps might not improve the sound much, there is a chance it will. Yes, I read all your struggles with the volume pot....Ill pass on messing with that at this point. As well as changing out the inductors and related components. The cost benefit of those is low compared to the cost of the amp, I think. It makes me wonder if the TPA3251 chip itself is fake? I would imagine so, but who knows. Im guessing changing the IC would be very difficult.
Everything Im doing is as much about getting an understanding of all of this as much as it is about better audio. I can simply buy better audio. I want to get a grasp on how amp circuits work so I can build a much better (and more $) later myself. I would imagine a $300 DIY amp would be exceptionally good. But alas, there is so much to learn before then. Power supplies, filtering caps, board layouts, component interaction and intended purpose for each, etc.
TDK FA16C0G1H104JNU00
Here are the very nice (and very expensive) KEMET low ESL ones in case someone wants to try them
KEMET C350C104J1G5TA
And for sure were rapidly getting to diminishing returns here. I will submit the 2 most important changes if one does nothing else will to be to replace the main caps with high quality, low ESR ones af AT LEAST 3000uf capacity and changing the op-amps to whatever flavor you like. The bypasses definitely help too, but for me, soldering in such small areas with such tiny components was more difficult than just replacing the main caps and it took way longer.
Like I said when I started this, I have zero electronics experience. I havent touched a soldering iron in 20+ years and that was only for things like speaker wires, not delicate PCB's so if I can pull this off, anyone with the motivation can. With this lack of experience though it means I dont really even understand the layout of an amp PCB or what all the components do or are for. My working theory is all the caps (likely all the components) are fake and will be improved in whatever job they have by replacing them with genuine, high quality ones. So thile the remaining 10-12 electrolytic caps might not improve the sound much, there is a chance it will. Yes, I read all your struggles with the volume pot....Ill pass on messing with that at this point. As well as changing out the inductors and related components. The cost benefit of those is low compared to the cost of the amp, I think. It makes me wonder if the TPA3251 chip itself is fake? I would imagine so, but who knows. Im guessing changing the IC would be very difficult.
Everything Im doing is as much about getting an understanding of all of this as much as it is about better audio. I can simply buy better audio. I want to get a grasp on how amp circuits work so I can build a much better (and more $) later myself. I would imagine a $300 DIY amp would be exceptionally good. But alas, there is so much to learn before then. Power supplies, filtering caps, board layouts, component interaction and intended purpose for each, etc.
As anyone knows, the main issue with those Power caps is the height (26mm max). Rolling new OP amps is easy )
The ELBK350ELL452AM25S is nice and rated @35V / 4700uf Low ESR (18x25mm) but not so easy to find.(about 3.5 euros each)
Another cheap alternative is the JCCON /Sanyo Low ESR 35V@4700Uf has been measured between 4400uf and 5000uf (16x25mm) see users feedback in Aliexpress. (about 0.5 euros each)
The stock ones in the A04 with (black PCB) are the ELNA RJD 35V / 3000UF can be found here (18x20mm) : 2 pcs 3000 uF 35 V Japon ELNA RJD Serie 18x20mm Faible Impedance Haute Fiabilite 35V3000uF Condensateur pour Audio | AliExpress
Last edited:
order place at Digikey for power supply and bypass caps. thanks for the info. Starting to et pricey! I received 2 pairs of opamp LM4562NA/NOPA, and RC4560IP late yesterday. Will wait to swap them out on Saturday.
I admire your determination and willingness to learn!
Yes, as stated, anyone can midly tune this amp - Gilles started also from scratch during the lockdown.
The chip itself is genuine, would be too difficult to fake at that stage. The other parts could be fake, or not, but fake doesn't necessary mean bad or having a negative impact on sound - depends what they are and what function they have. At the end, whatever Aiyima did, they did not too bad because the unit sounds quite OK once your remove the 1 or 2 majour bottlenecks. Somewhere they stand out of the Chinese competitors that did similar, so lucky guess with fake parts or well arranged cooking, the end result is fine.
Of course, the day you will build you own amp, you will enter an entire new territory with full control on components and assembly quality (yours, LOL). Given you only get started, I would though advise you look for kits or well documented builts: designing an amp looks easy, but so many things can go wrong and the devil is in the details... so many things to sort.
Good luck for the tweaks to come... and the future amps
Claude
Yes, as stated, anyone can midly tune this amp - Gilles started also from scratch during the lockdown.
The chip itself is genuine, would be too difficult to fake at that stage. The other parts could be fake, or not, but fake doesn't necessary mean bad or having a negative impact on sound - depends what they are and what function they have. At the end, whatever Aiyima did, they did not too bad because the unit sounds quite OK once your remove the 1 or 2 majour bottlenecks. Somewhere they stand out of the Chinese competitors that did similar, so lucky guess with fake parts or well arranged cooking, the end result is fine.
Of course, the day you will build you own amp, you will enter an entire new territory with full control on components and assembly quality (yours, LOL). Given you only get started, I would though advise you look for kits or well documented builts: designing an amp looks easy, but so many things can go wrong and the devil is in the details... so many things to sort.
Good luck for the tweaks to come... and the future amps
Claude
The Kemet are indeed fine (used similar ones myself in another project)... but quite pricey. With all the C0G we needed for this project, they would have surpassed the amp's price LOL. They are also quite big so not sure they fit easily under the board without going the risk to short them when putting it back into the casing.
The TDK you went for are very good aswell and IMHO better suited for this project - regearding their size and price anyway. I doubt one can hear any real difference in this project between these and the Kemet, so you went for the right choice. Note the Muratas would have been fine aswell...
Claude
The TDK you went for are very good aswell and IMHO better suited for this project - regearding their size and price anyway. I doubt one can hear any real difference in this project between these and the Kemet, so you went for the right choice. Note the Muratas would have been fine aswell...
Claude
You can almost guarantee any components youre getting from AliExpress will not be genuine. I would strongly recommend only using Mouser or Digikey for components. Yes, size is a trick for sure. Nichicon UBY series as Claude is using are also very nice. Theyre sold out here which is why I went another direction.
We probably differ in opinion here. Obviously you know way more about this than I do but you cant replicate equivalent quality components with inferior parts and mfg practices. Almost all the quality electronics components come from Japan who are known for mfg superiority in basically everything. I would be extremely surprised if many other places can replicate the machine they have for quality production. There are also other countries known for good mfg practices (US, Germany, etc) but those arent where fake components are coming from.
I do agree Aiyima has done a good job with this amp, especially considering the price but I would be willing to bet anyone if they took this exact same PCB and chose the exact same components used but from suppliers known to sell genuine parts the difference in sound quality would be noticeable. There is simply no way a company like Aiyima (or whoever builds these) can be using quality components, assemble, and sell an amp like this for this price, even with the economy of scale and the low cost of Chinese labor. The parts list alone from Mouser/Digikey even if bought in high volumes would almost certainly exceed the $50 cost of this.
My $0.02 on this topic anyway
I guess there is some missinterpretation from what I tried to say...
1- Never ever buy single components from Ali... & Co, always from trustfull sources. That's the only way to know what you really have and as DIYer you can afford to pay the few components wereas you can't afford to pay for someone testing your fakes for you or an insurance reimbursing you if all goes wrong
2- ONCE YOU HAVE GENUINE COMPONENTS, uttermost quality is not necessarly always required. Overkill, diminishing returns, you have to be clever there. Depending on the function of the part and in what circuit it is used, also how it is used, you have to determine the required specs that match with the quality level you intend to reach. That requires some experience and understanding. It is the only way to keep things reasonable and avoid paying for 1000$ caps that may well indeed be key in some exceptional circumstances. Also, best is the good's enemy: sometimes, having some parts with lesser specs (some specific specs) can be better. Example, you don't need... and even don'twant to quick caps / low ESR caps in the the first C of a CRC filter (but note that ripple would be key there on the other hand). On some occasions you can do with very average caps or components because they aren't really critical and also tolerances can be high (ex. 1M resistor to ground to avoid switching pop). And yes, you would want low ESR caps in teh second C and ripple who cares really...
Many many other examples, but few times... it is always best to understand roughly what is going on and then shoot for VG quality parts or the step above, not necessarly the ultimate
A champion is using normal parts (and a few exceptional parts), mounting all of them in unexpected ways, is IMHO Nelso Pass. Genius... in simplicity. Master Chief in cooking.
All IMHO
Claude
1- Never ever buy single components from Ali... & Co, always from trustfull sources. That's the only way to know what you really have and as DIYer you can afford to pay the few components wereas you can't afford to pay for someone testing your fakes for you or an insurance reimbursing you if all goes wrong
2- ONCE YOU HAVE GENUINE COMPONENTS, uttermost quality is not necessarly always required. Overkill, diminishing returns, you have to be clever there. Depending on the function of the part and in what circuit it is used, also how it is used, you have to determine the required specs that match with the quality level you intend to reach. That requires some experience and understanding. It is the only way to keep things reasonable and avoid paying for 1000$ caps that may well indeed be key in some exceptional circumstances. Also, best is the good's enemy: sometimes, having some parts with lesser specs (some specific specs) can be better. Example, you don't need... and even don'twant to quick caps / low ESR caps in the the first C of a CRC filter (but note that ripple would be key there on the other hand). On some occasions you can do with very average caps or components because they aren't really critical and also tolerances can be high (ex. 1M resistor to ground to avoid switching pop). And yes, you would want low ESR caps in teh second C and ripple who cares really...
Many many other examples, but few times... it is always best to understand roughly what is going on and then shoot for VG quality parts or the step above, not necessarly the ultimate
A champion is using normal parts (and a few exceptional parts), mounting all of them in unexpected ways, is IMHO Nelso Pass. Genius... in simplicity. Master Chief in cooking.
All IMHO
Claude
Last edited:
Ok, thats fair and I totally agree with that. Diminishing returns in audio exists everywhere. And 100% you need to know what specs are best for any given component, which I still have a LONG way to go on. I have to rely on others for now and some trial and error.
Everything in audio is about removing the bottlenecks. You can have the very best components money can buy but if you add a low quality component to one place in the signal chain that will determine the overall quality. People argue all the time about the difference cables (speaker or signal) make. Putting $1000/ft speaker cable coming out of a low quality source wont fix anything, however, if the source components are being held back by poor quaity cables they will 100% make a difference.
If I knew more about electronic circuitry I would be able to make better decisions about the choices I made so far. If it wasnt for the work you and others put in here I would never have had a starting point. I am however a quick study and someone who likes to understand concepts, not just follow steps. So again, HUGE thank you to you, others here, and this entire DIY community!
Nah, you are learning fast, and soon we will all be learning from you
Meanwhile, don't hesitate to ask any question regarding whatever topic/project: I can't garantee you I know the answer, but you are very welcome anytime.
Take care
Claude
Meanwhile, don't hesitate to ask any question regarding whatever topic/project: I can't garantee you I know the answer, but you are very welcome anytime.
Take care
Claude
I appreciate that. I do have a question then. Do you know top of your head what the role of the 9 capacitors surrounding the opa's are? What about the 3 470uf ones? Is it as simple as following the traces on the board to know where these components fit and what theyre for?
That was an example ) ELBK350ELL452AM25S come from Mouser or TME 🙂
Audiophonics in Europe is also a trusty site.
Quote: I do have a question then. Do you know top of your head what the role of the 9 capacitors surrounding the opa's are?
=> Yes. In fact you should consider they are 4 + 4 + 1. The "1 is the one in the middle of the raw of 5 and turns out to be VERY sensitive and completely different from the others re function. The others are 4 per channel, at various stages of the OPA stage...S.
Let's tackle the 4 + 4. They are basicaly DC decoupling caps, all directly in the signal path. You can see the OPAs as preamp function, as they turn out to have some gain and are not though as anticipated just OPAs to symmetrize (= make a balanced signal out of a usual RCA signal) the input signal. But converting RCA to balanced has though to be considered their main function somewhere as that HAS to be... as the TI chip wouldn't accept any RCA signal directly otherwise. Let's say Aiyima used the opportunity "while at it" to dial some gain and not just symmetrize and buffer the input signal but also adding some gain to make sure you don't run out of steam. Low gain though...
If you go through my various posts you will find it is a 2 stage affair per channel and for various reasons explained you need to make sure you don't have any DC and so between input and inside and output of this OPA "preamp stage" you end up needing 8 DC decoupling caps (unless you deal with the first input DC decoupling caps as we did elsewhere, hence jumpers/shorts in our case).
Now the last cap, the one in the middle, is PS supply related. Here Aiyima didn't go the trouble to feed the OPA with +V and -V, but went instead with 0V and +V, meaning the OPAs needed some virtual middle voltage reference/level connected to one of their entry pin. You can read all that in my post and find more details on internet, quite common layout. In short, that cap somewhere is more PS related and decoupling it turned out to be key. I didn't know what to expect re its sensitivity, was a question mark hence us addressing it separately to evaluate it closely. I did though expect much more sensitivity from the caps that are directly in the signal path, the 8 others, and at the end surprisingly their senstitivity was quite low. I guess we are hitting a limit there, regardless their location and gain stages. On top equipment they are nearly always a bottleneck... Chip amp Class D probably give us the illusion of being very transparent, but it is not as much as high end stuff, let's be honest.
The 470uF caps are also PS related, but then they are various PS inse the unit: some for analogue, some for digital, some for a mix of both and we have various voltage levels (hence also voltage regulation stages) going around. Some are for logical operations and I expected their sensitivity to be low: after all it is just logic and not signal. But they did affect things, and as I don't have the schematic it is difficult to guess why A affects B. Hence us just trying with some common sense while though remaining open to every experiment, including the ones that shouldn't bring mass benefits.
Quote "Is it as simple as following the traces on the board to know where these components fit and what theyre for?"
That would be great and works quite well on usual old school boards. But not here. Reason being you have a multilayer board and that means that some traces are burried inside the PCB. That's quite a problem, because some traces are indeed obvious as being thick and on one of either PCB side, but some are just hidden inside. And sadly hidden traces can change a lot of things or making tweaking hazardous. At least all you can be sure of is that the traces you see and follow on both sides are there. AT LEAST you know part of the schematic that way. Question is: what are you missing. And if many options, as here... what would be the most logical completion of the missing puzzle and is there a way to check that without destroying the board? We spent some time on this, did it quite in depth for most signal / sound traces and discovered 1 or 2 things as reported. The Aiyima is very close to the TI eval board schematic... but not exactly it and of course there are lots of additions to make it a complete amp with power switch and led LOL.
We didn't though bother tracing everything outside the signal path or additional PS as up to a point it made little sense re time spent vs potential added value. There is still a grey zone but I am 90% confident there is nothing worthwhile inside. Could be wrong though, still 10% LOL
I hope ths helps
Claude
=> Yes. In fact you should consider they are 4 + 4 + 1. The "1 is the one in the middle of the raw of 5 and turns out to be VERY sensitive and completely different from the others re function. The others are 4 per channel, at various stages of the OPA stage...S.
Let's tackle the 4 + 4. They are basicaly DC decoupling caps, all directly in the signal path. You can see the OPAs as preamp function, as they turn out to have some gain and are not though as anticipated just OPAs to symmetrize (= make a balanced signal out of a usual RCA signal) the input signal. But converting RCA to balanced has though to be considered their main function somewhere as that HAS to be... as the TI chip wouldn't accept any RCA signal directly otherwise. Let's say Aiyima used the opportunity "while at it" to dial some gain and not just symmetrize and buffer the input signal but also adding some gain to make sure you don't run out of steam. Low gain though...
If you go through my various posts you will find it is a 2 stage affair per channel and for various reasons explained you need to make sure you don't have any DC and so between input and inside and output of this OPA "preamp stage" you end up needing 8 DC decoupling caps (unless you deal with the first input DC decoupling caps as we did elsewhere, hence jumpers/shorts in our case).
Now the last cap, the one in the middle, is PS supply related. Here Aiyima didn't go the trouble to feed the OPA with +V and -V, but went instead with 0V and +V, meaning the OPAs needed some virtual middle voltage reference/level connected to one of their entry pin. You can read all that in my post and find more details on internet, quite common layout. In short, that cap somewhere is more PS related and decoupling it turned out to be key. I didn't know what to expect re its sensitivity, was a question mark hence us addressing it separately to evaluate it closely. I did though expect much more sensitivity from the caps that are directly in the signal path, the 8 others, and at the end surprisingly their senstitivity was quite low. I guess we are hitting a limit there, regardless their location and gain stages. On top equipment they are nearly always a bottleneck... Chip amp Class D probably give us the illusion of being very transparent, but it is not as much as high end stuff, let's be honest.
The 470uF caps are also PS related, but then they are various PS inse the unit: some for analogue, some for digital, some for a mix of both and we have various voltage levels (hence also voltage regulation stages) going around. Some are for logical operations and I expected their sensitivity to be low: after all it is just logic and not signal. But they did affect things, and as I don't have the schematic it is difficult to guess why A affects B. Hence us just trying with some common sense while though remaining open to every experiment, including the ones that shouldn't bring mass benefits.
Quote "Is it as simple as following the traces on the board to know where these components fit and what theyre for?"
That would be great and works quite well on usual old school boards. But not here. Reason being you have a multilayer board and that means that some traces are burried inside the PCB. That's quite a problem, because some traces are indeed obvious as being thick and on one of either PCB side, but some are just hidden inside. And sadly hidden traces can change a lot of things or making tweaking hazardous. At least all you can be sure of is that the traces you see and follow on both sides are there. AT LEAST you know part of the schematic that way. Question is: what are you missing. And if many options, as here... what would be the most logical completion of the missing puzzle and is there a way to check that without destroying the board? We spent some time on this, did it quite in depth for most signal / sound traces and discovered 1 or 2 things as reported. The Aiyima is very close to the TI eval board schematic... but not exactly it and of course there are lots of additions to make it a complete amp with power switch and led LOL.
We didn't though bother tracing everything outside the signal path or additional PS as up to a point it made little sense re time spent vs potential added value. There is still a grey zone but I am 90% confident there is nothing worthwhile inside. Could be wrong though, still 10% LOL
I hope ths helps
Claude
Wow, I REALLY appreciate the detailed answer. Some of that is going over my head but thats ok. I get the gist of what youre saying.
So, if the 8 (2x4) caps are for the opa's how come they are completely different? I think theyre all 10uf (I think) but the voltage is different and the physical size of course. Based on post #1 in this thread the new ones I ordered are 22uf so Ill be curious as to what this does. In theory good quality caps should help in a similar (maybe??) way as to how you bypassed the opa's. Sounds like replacing the 470uf's wont likely do much but theyre very cheap and I had an order placed anyway,
As far as comparisons to the TI eval board it looks like this entire section (the 9 opa caps) isnt present. I know TI isnt using the same type of opamp circuit (or at all) that Aiyima is but its an interestingly different layout, at least to my very untrained eyes. I wish eval boards weret so expensive.
Total aside and not related to this thread........
I just updated the caps in my NHT SuperZero's crossover. The original were electrolytics and I changed to Clarity CX film caps of the same capacitance. Still have a long way to go breaking them in, but man what a difference vs 20 year old electrolytics. Very similar to what upgrading the amp did. Its like a veil was lifted off the sound.
So, if the 8 (2x4) caps are for the opa's how come they are completely different? I think theyre all 10uf (I think) but the voltage is different and the physical size of course. Based on post #1 in this thread the new ones I ordered are 22uf so Ill be curious as to what this does. In theory good quality caps should help in a similar (maybe??) way as to how you bypassed the opa's. Sounds like replacing the 470uf's wont likely do much but theyre very cheap and I had an order placed anyway,
As far as comparisons to the TI eval board it looks like this entire section (the 9 opa caps) isnt present. I know TI isnt using the same type of opamp circuit (or at all) that Aiyima is but its an interestingly different layout, at least to my very untrained eyes. I wish eval boards weret so expensive.
Total aside and not related to this thread........
I just updated the caps in my NHT SuperZero's crossover. The original were electrolytics and I changed to Clarity CX film caps of the same capacitance. Still have a long way to go breaking them in, but man what a difference vs 20 year old electrolytics. Very similar to what upgrading the amp did. Its like a veil was lifted off the sound.
Last edited:
Quote: "So, if the 8 (2x4) caps are for the opa's how come they are completely different? I think theyre all 10uf (I think) but the voltage is different and the physical size of course.
=> I posted all the details already in this thread. That's probably the section most people didn't understand what it was really about as a tad different from what TI gave the Chinese.
To start with the caps aren't really that different, they are all 10uF. TBH, don't care about the voltage rating, that came with the caps and the rest is pure marketing and cosmetic.
Now, the 10uF value was also chosen as it is s std value, even TI didn't bother calculating what is required- Pass seldom does either and uses 10uF and opens door to tweaking for those that like that. The caps form a low pass filter with the impedance (resistance mainly) of the neighbour stage that is associated so you have to make sure you don't cut off low frequencies going too low. My posts in the B1 Korg thread detailled how to chose / calculate these caps and their values for a different gear but at a similar function. Key word is RC low pass...
Short answer: they didn't bother calculating and took std failsafe value, same everywhere and then marketing. Calculation even gave me that I could go high quality 1uF (only) caps for the 4 shiny bits closest to the chip amp radiator. I wanted to go WIMA MKS as that would just fit. And didn't bother because of the low sensitivity to tweaking when I was able to fit the best ever cap (read NO cap) on one location. Note some reported 1uF might be a tad too small despite the entry impedance of the TI chip. Try and learn...
22uF aren't going to help you really, but could sound different... because they are different caps. The value increase is absolutely not relevant here.
Quote: "In theory good quality caps should help in a similar (maybe??) way as to how you bypassed the opa's."
=> Nope, this are DC blocking caps, completely different function than the PS caps. Being caps and bypassed, sometimes with similar parts as I did was just pure coincidence. It turned out it made sense in that case, but I could have done things very differently in another gear (and did).
Quote: Sounds like replacing the 470uf's wont likely do much but theyre very cheap and I had an order placed anyway,
=> Keep us posted!
Quote: As far as comparisons to the TI eval board it looks like this entire section (the 9 opa caps) isnt present.
=> The entry and output caps are present. The caps between stages aren't. That's where the Chinese went very differently. Normaly, due to that simplified PS with virtual middle voltage reference, you would absolutely need DC decoupling caps between the OPA stages. Otherwise you kind of saturate with your PS overruling the signal. BUT there is one exception to it: if your gain is 1 (eg no gain). TI went also for that simplified PS... but for unity gain and though could do without 1 cap per channel.
Quote: I know TI isnt using the same type of opamp circuit (or at all) that Aiyima is
=> Nope, very very similar topology in fact there, only the resistor value differ and lead to a non unity gain sides Aiyima. All posted in the thread, don't have the values in mind, but was a small gain around of 2 or so...
Good you ask the right questions, that was quite a finding on our side and I wondered why no one really bounced back on this... well you do now LOL
Quote: Total aside and not related to this thread........
I just updated the caps in my NHT SuperZero's crossover. The original were electrolytics and I changed to Clarity CX film caps of the same capacitance.
=> This is in fact a tricky part: caps in LS Xover.
OK, no doubt your film caps sound ways better than the old (or new BTW) 'lytics. Thats the positive. I guess small value so possibly for the tweater or mids, caps filtering frequencies (highpass, midpass, depending where the caps are and what schematic)
Now to the negative. Unless you know what you are doing or want to address some defficiencies re sound with cheap LS, you can't just replace 'lytics with films. It could work, but that's luck.
Short explanation: when manufacturer design Xover they do carefully chose the right components. Indeed, in your case lytics where chosen probabaly because of cost and/or size. BUT when going for a cap, they also took into account its imperfection. As you noticed caps have an ESR. In fact, that "parasitic" resistance is not a constant, by far not: it varies with frequency. Often replacement parts of good hifi LS state not only the cap value but the cap technology and the ESR or TANd at a given frequency (which ends up being the same provide you know what we are talking about).
Let's take an example. Often you will find a cap in thesignal path of the tweater, to form a high pass filter and protecting it from low frequencies. Now, you also often find a resistor in this signal path to the tweater: it is there to adjust the sound volume of the tweater and make it sound even with the rest of the speakers. That R value is very important. When people design that king of Xover, they do though also take the ESR of the cap into account, as that is also a R in the signal path that will tob some dynamic. Somewhere, roughly explained (it is a tad more complicated), if they found out that with perfect components they needed a resistor of value R in front of the tweater, and then a perfect C, they will now that in fact the tweater will be fed though the resistor of their choice (say R1) and the ESR of the cap (say R2). Because of that they will have to chose R1 such that at the end, in the corresponding frequency band, R1+R2 = R they wanted. In fact, they will fit a lower resistor value than calculation would give, to take into account your lytic imperfection (having a quite high R).
Now, film caps of the same capacity than your lytics will have an ESR that is several orders of magnitude smaller than the OEM lytic. Make that R=0. In reality it is that close to that for our purpose. Now guess what is happening? When replacing the lytic with films one would made R2=0 and now that means that the tweater sees more energy than it should!
Bottom line: in that particular case, to get the benefits of the film caps WHILE not changing the tonal balance of the LS you need to introduce in serie with your new film cap a small resistor to compensate for the lower ESR. If one doesn't do that, the treble will sound louder. Clearer for sure due to teh better caps, but sadly also out of the intented proportions. A resistor gives you back balance while enjoying the quality increase of the film caps. And then you need to do that all again for the other speakers, well noting that some don't benefit from films really (coudl even be worst some time in the bass, depends), and that sometimes, rarely, on some locations on multiway LS you can do without a resistor to compensate because teh complex filter doesn't really bother on some specific locations. Simulations and computers to the rescue for complex filters.
I hope I haven't spoiled the party, and all this in order to help you completing your LS tweak... supposed you don't know all that already
Claude
=> I posted all the details already in this thread. That's probably the section most people didn't understand what it was really about as a tad different from what TI gave the Chinese.
To start with the caps aren't really that different, they are all 10uF. TBH, don't care about the voltage rating, that came with the caps and the rest is pure marketing and cosmetic.
Now, the 10uF value was also chosen as it is s std value, even TI didn't bother calculating what is required- Pass seldom does either and uses 10uF and opens door to tweaking for those that like that. The caps form a low pass filter with the impedance (resistance mainly) of the neighbour stage that is associated so you have to make sure you don't cut off low frequencies going too low. My posts in the B1 Korg thread detailled how to chose / calculate these caps and their values for a different gear but at a similar function. Key word is RC low pass...
Short answer: they didn't bother calculating and took std failsafe value, same everywhere and then marketing. Calculation even gave me that I could go high quality 1uF (only) caps for the 4 shiny bits closest to the chip amp radiator. I wanted to go WIMA MKS as that would just fit. And didn't bother because of the low sensitivity to tweaking when I was able to fit the best ever cap (read NO cap) on one location. Note some reported 1uF might be a tad too small despite the entry impedance of the TI chip. Try and learn...
22uF aren't going to help you really, but could sound different... because they are different caps. The value increase is absolutely not relevant here.
Quote: "In theory good quality caps should help in a similar (maybe??) way as to how you bypassed the opa's."
=> Nope, this are DC blocking caps, completely different function than the PS caps. Being caps and bypassed, sometimes with similar parts as I did was just pure coincidence. It turned out it made sense in that case, but I could have done things very differently in another gear (and did).
Quote: Sounds like replacing the 470uf's wont likely do much but theyre very cheap and I had an order placed anyway,
=> Keep us posted!
Quote: As far as comparisons to the TI eval board it looks like this entire section (the 9 opa caps) isnt present.
=> The entry and output caps are present. The caps between stages aren't. That's where the Chinese went very differently. Normaly, due to that simplified PS with virtual middle voltage reference, you would absolutely need DC decoupling caps between the OPA stages. Otherwise you kind of saturate with your PS overruling the signal. BUT there is one exception to it: if your gain is 1 (eg no gain). TI went also for that simplified PS... but for unity gain and though could do without 1 cap per channel.
Quote: I know TI isnt using the same type of opamp circuit (or at all) that Aiyima is
=> Nope, very very similar topology in fact there, only the resistor value differ and lead to a non unity gain sides Aiyima. All posted in the thread, don't have the values in mind, but was a small gain around of 2 or so...
Good you ask the right questions, that was quite a finding on our side and I wondered why no one really bounced back on this... well you do now LOL
Quote: Total aside and not related to this thread........
I just updated the caps in my NHT SuperZero's crossover. The original were electrolytics and I changed to Clarity CX film caps of the same capacitance.
=> This is in fact a tricky part: caps in LS Xover.
OK, no doubt your film caps sound ways better than the old (or new BTW) 'lytics. Thats the positive. I guess small value so possibly for the tweater or mids, caps filtering frequencies (highpass, midpass, depending where the caps are and what schematic)
Now to the negative. Unless you know what you are doing or want to address some defficiencies re sound with cheap LS, you can't just replace 'lytics with films. It could work, but that's luck.
Short explanation: when manufacturer design Xover they do carefully chose the right components. Indeed, in your case lytics where chosen probabaly because of cost and/or size. BUT when going for a cap, they also took into account its imperfection. As you noticed caps have an ESR. In fact, that "parasitic" resistance is not a constant, by far not: it varies with frequency. Often replacement parts of good hifi LS state not only the cap value but the cap technology and the ESR or TANd at a given frequency (which ends up being the same provide you know what we are talking about).
Let's take an example. Often you will find a cap in thesignal path of the tweater, to form a high pass filter and protecting it from low frequencies. Now, you also often find a resistor in this signal path to the tweater: it is there to adjust the sound volume of the tweater and make it sound even with the rest of the speakers. That R value is very important. When people design that king of Xover, they do though also take the ESR of the cap into account, as that is also a R in the signal path that will tob some dynamic. Somewhere, roughly explained (it is a tad more complicated), if they found out that with perfect components they needed a resistor of value R in front of the tweater, and then a perfect C, they will now that in fact the tweater will be fed though the resistor of their choice (say R1) and the ESR of the cap (say R2). Because of that they will have to chose R1 such that at the end, in the corresponding frequency band, R1+R2 = R they wanted. In fact, they will fit a lower resistor value than calculation would give, to take into account your lytic imperfection (having a quite high R).
Now, film caps of the same capacity than your lytics will have an ESR that is several orders of magnitude smaller than the OEM lytic. Make that R=0. In reality it is that close to that for our purpose. Now guess what is happening? When replacing the lytic with films one would made R2=0 and now that means that the tweater sees more energy than it should!
Bottom line: in that particular case, to get the benefits of the film caps WHILE not changing the tonal balance of the LS you need to introduce in serie with your new film cap a small resistor to compensate for the lower ESR. If one doesn't do that, the treble will sound louder. Clearer for sure due to teh better caps, but sadly also out of the intented proportions. A resistor gives you back balance while enjoying the quality increase of the film caps. And then you need to do that all again for the other speakers, well noting that some don't benefit from films really (coudl even be worst some time in the bass, depends), and that sometimes, rarely, on some locations on multiway LS you can do without a resistor to compensate because teh complex filter doesn't really bother on some specific locations. Simulations and computers to the rescue for complex filters.
I hope I haven't spoiled the party, and all this in order to help you completing your LS tweak... supposed you don't know all that already
Claude
Another fantastic, incredibly informative response.....thanks again.
I knew about the difference in film vs electrolytic and what to expect. Clarity are some of the "warmest" sounding film caps so my hope was a bit of a compromise. For sure Ken Kantor went with these for size. These are tiny speakers with a PCB style crossover and the original caps are very small to fit on the board. If I end up not liking the results after they break in Ill definitely try both a small (any idea how small is good??) capacitor or just changing back to electrolytic caps which I may try either way.
What I didnt know was all the why's apart from the lower resistance of film caps vs lytics. So far though they sound really good. The highs arent boosted much if at all, which is interesting as thats what I was expecting. Treble is overall smoother and more resolving.
This was as much (if not more) me just playing as it was anything else. I didnt get the big $$ Clarity's, just the entry level ones which didnt cost too much....$25 for all 4 or so.
I knew about the difference in film vs electrolytic and what to expect. Clarity are some of the "warmest" sounding film caps so my hope was a bit of a compromise. For sure Ken Kantor went with these for size. These are tiny speakers with a PCB style crossover and the original caps are very small to fit on the board. If I end up not liking the results after they break in Ill definitely try both a small (any idea how small is good??) capacitor or just changing back to electrolytic caps which I may try either way.
What I didnt know was all the why's apart from the lower resistance of film caps vs lytics. So far though they sound really good. The highs arent boosted much if at all, which is interesting as thats what I was expecting. Treble is overall smoother and more resolving.
This was as much (if not more) me just playing as it was anything else. I didnt get the big $$ Clarity's, just the entry level ones which didnt cost too much....$25 for all 4 or so.
Well, see how it goes then, sounds promising 🙂
I don't know these speakers, their schematics nor the price range and what the designer did when he designed them... and not even what mods you did and 'where". BUT I am sure you will find nice folks here on DIYaudio in the LS speaker section willing to help much better than I can, possibly knowing already most answers.
Important is you went for exactly the same value re capacity, that's the starting point. Of course depending on value some like tweaking or bypassing, but you better know then exactly what you ar doing.
It is tricky without all the inputs I listed. Take very cheap 200$ speakers. Even some of the best make went in the past the mistake of going for simple schematics without proper listening and assuming simple perfect parts for the price range (read no ESR takien into account...). In that case juste replacing lytics with films caps alone can work, because films are closer in fact to what was intented LOL. That's though for low fi usualy, or very old speakers not also designed by ear / maker. I modified at least one of these speakers and never bothered about the the extra resistor as the balance was probably retablished as it should. Overkill though to mod that kind of speakers usualy, unless for a friend...
Take on the other hand a properly designed speaker where all this was taken into account... and far more (cabinet reflections, mechanical "gains") and proper listening in various conditions, systems, real live listening room etc.
Suppose you know what you are doing and are the king of Xsim & Co. You wisely replace that lytic with the appropriate film + resistor. But the owner bought these speakers used, old, worn. He bought them because of their good old sound. Yet after decades the highs are muddy and muffled, ESR went sky high with age etc. You restore the right balance and give them a new life, sounding in fact as new but even with extra resolution and transparency. And now the owner says they sound agressive and he doesn't like them because they indeed sound as intented but not his idea of these speakers.
Or you fit that extra resistors as you should and the owner is past 60 and in fact he prefers the sound without extra R... because he has hearing probs and more treble helps restauring his listening experience.
Not easy, uh? That's why I can't tell you what R value to go for. Note it is R and not a small C.
What I could perhaps do (but for sure in the LS section people here would be delighted to do the same) is if you tell me what specs were the initial lytics (I mean what TANd at what frequency or what ESR spec the cap had, that is very important), where that cap was located on the schematic, what cap you went for as replacement... then I could possibly calculate a R value to compensate for the lower ESR and get things back to "as intented but with better components".
But honestly, this is not the thread, and some love to spend time on all that and have marvelous tools for that (I have a pocket calculator and my poor brain), and would love to help I am sure... if you need some advice IMHO try the Speaker section and explain your case with all the data you can gather.
That's an entirely different playfield: in this hobby I found that designing fom scratch (high quality) RIAA amps and LS are probably the most complicated things...
Have fun
Claude
I don't know these speakers, their schematics nor the price range and what the designer did when he designed them... and not even what mods you did and 'where". BUT I am sure you will find nice folks here on DIYaudio in the LS speaker section willing to help much better than I can, possibly knowing already most answers.
Important is you went for exactly the same value re capacity, that's the starting point. Of course depending on value some like tweaking or bypassing, but you better know then exactly what you ar doing.
It is tricky without all the inputs I listed. Take very cheap 200$ speakers. Even some of the best make went in the past the mistake of going for simple schematics without proper listening and assuming simple perfect parts for the price range (read no ESR takien into account...). In that case juste replacing lytics with films caps alone can work, because films are closer in fact to what was intented LOL. That's though for low fi usualy, or very old speakers not also designed by ear / maker. I modified at least one of these speakers and never bothered about the the extra resistor as the balance was probably retablished as it should. Overkill though to mod that kind of speakers usualy, unless for a friend...
Take on the other hand a properly designed speaker where all this was taken into account... and far more (cabinet reflections, mechanical "gains") and proper listening in various conditions, systems, real live listening room etc.
Suppose you know what you are doing and are the king of Xsim & Co. You wisely replace that lytic with the appropriate film + resistor. But the owner bought these speakers used, old, worn. He bought them because of their good old sound. Yet after decades the highs are muddy and muffled, ESR went sky high with age etc. You restore the right balance and give them a new life, sounding in fact as new but even with extra resolution and transparency. And now the owner says they sound agressive and he doesn't like them because they indeed sound as intented but not his idea of these speakers.
Or you fit that extra resistors as you should and the owner is past 60 and in fact he prefers the sound without extra R... because he has hearing probs and more treble helps restauring his listening experience.
Not easy, uh? That's why I can't tell you what R value to go for. Note it is R and not a small C.
What I could perhaps do (but for sure in the LS section people here would be delighted to do the same) is if you tell me what specs were the initial lytics (I mean what TANd at what frequency or what ESR spec the cap had, that is very important), where that cap was located on the schematic, what cap you went for as replacement... then I could possibly calculate a R value to compensate for the lower ESR and get things back to "as intented but with better components".
But honestly, this is not the thread, and some love to spend time on all that and have marvelous tools for that (I have a pocket calculator and my poor brain), and would love to help I am sure... if you need some advice IMHO try the Speaker section and explain your case with all the data you can gather.
That's an entirely different playfield: in this hobby I found that designing fom scratch (high quality) RIAA amps and LS are probably the most complicated things...
Have fun
Claude