Work in progress ) wiating for AC/DC filter to finish
here goes again
Stock version
A small 110*70*35MM TPA3255 module with :
SAGAMI 7W14A inductors
Full Nichicon caps, as below :
Nichicon 2200UF/63V VZ serie power caps
Nichicon FW serie Buffer caps (Audio Grade)
DIP8 OP amps socket
My minor Updates :
For TPA3255 module : Remove 2X Nichicon VZ serie 2200UF/63V > Install 2X Nichicon 4700UF KW serie (low ESR Audio grade)
For LLC SMPS PSU : Remove Output Caps 2X Nichicon PM serie 50V/1800UF > Install 2X Nichicon 2200UF KW serie
OP amps : Remove NES5532 Op amps > Install OPA1656 gold pins.
Last edited:
No worries, it will come with exprience and it is not like there is a simple answer or solution for all fits, otherwise all would go for it.
There are models, there is physics, that's the baseline. And then there is also some cooking some people don't want to believe in. We are far away from having discovered everything re psychoacoustic, that's quite new in fact and we know most interesting works 20y after, oncethey are kind of public. Applies to some physics aswell BTW. But I disgress...
So try to understand the models, most importantly try to understand what alidity field they are related to and hence where the limits are from the models, and then cook as you fancy unless you want to start R&D...
Have fun
Claude
There are models, there is physics, that's the baseline. And then there is also some cooking some people don't want to believe in. We are far away from having discovered everything re psychoacoustic, that's quite new in fact and we know most interesting works 20y after, oncethey are kind of public. Applies to some physics aswell BTW. But I disgress...
So try to understand the models, most importantly try to understand what alidity field they are related to and hence where the limits are from the models, and then cook as you fancy unless you want to start R&D...
Have fun
Claude
So do you think I could get away with a 2.2uf film cap for the 4 OPA output caps? IE, is that enough capacitance?
Regarding your specific questions...
First off, please eventhough the treatment might for whatever coincidence (as in our case) ending up being the same, please don't talkabout the 9 caps. There aren't really a bunch of 9 caps as explained. There is on single outstanding cap, that on that special occasion feeds one of the inputs of the OPA but that should IMHO be regarded as a PS caps. And then there are the 8 remaining caps, caps in the the signal path. DC blocking caps. These see different impedances and are located at different "stages", so that for each sêcific case one could calculate the optimal capacity value to go for, and that wouldn't be 10uF. But sorry, I can't get any deeper into this: key words are again high pass filter RC mainly...
Now, regarding the 8 caps, IMHO all this is just preferences. I decided to leave the existing caps there and just go for 0.1uF + C0G bypasses.
In details
a) Electrolytics (22uf) + .1uf C0G bypass
=> Why not, but I doubt the increased electrolytic will change anything once bypassed, other than colouring perhaps.
b) Small film (MKS2B051001N00JSSD) for example or (ECQ-E1106JF)
=> same comments... usualy working on the signal path caps is one of the priorities, here IMHO it isn't. I commented already why. No ROI, waste of time... little gain.
c) Even smaller capacity (1-5uf) film (ECW-FD2W225J4) or similar
=> That depends on which cap / its location and what low frequency you will cut off... to be sure you could start with the 4 caps located physicaly near the heatsink. You to try and to tell us, no idea about the outcome vs a)
d) Stick with the 10uf ceramics and bypass them with .1uf C0G
=> Was my choice. Little to be gained anyway, so why bother? Plus everytime you remove a cap you take the risk to wreck the tracks and go into big trouble. Here it is risk x 9.
Again, we were able n our VERY SPECIFIC set up to take 2 caps completely out and hence to evaluate the gain of a perfect cap (= no cap). We were disapointed to hear very little result. Whatever combo is chosen, regardless the $$ of the caps (but admitelly the colouration one might like), at least 50% of the gains achievable are obtained in our opinions with the simple .1uF C0G bypass. And 100% would be very little gain anyway, low magnitude...
e) Other?
=> Don't bother, the truth is elsewhere. And if you are really after the ultimate, start another project. This is just a class D amp that as it stands struggles to compete vs modern 500E amps. Once tuned its sonic play perhaps in the 1000-2000E arena. TBD, we don't know yet, we haven't tried vs modern competitor in that bracket. We don't own such hifi gear (either amps below 1000E... or much more in fact). But don't expect miracles or believe people telling you this is the last say... It is though good enough once tweaked to be happy with re power amp and excellent VFM. All IMHO
Also since I'm thinking of it..... Are film or ceramic better for bypass caps?
=> Generaly speaking FILMS. Ceramic are non linear, piezoelectric etc. and I avoid them usualy in the signal path. And everywhere they would probably be inferior to film caps, unless considering non audio applications. Having said that, sometimes you just have to accept them as there is no alternative. And once you consider PS bypass caps, perhaps the question becomes academic as then the difference matters less re distorsions etc.
And to make it more complex, there are VG ceramics and average films... they might be a no man's land and no clear cut.
I can get small .1uf film caps or C0G ceramics
=> C0G is not a technology per se but a spec re results / delivery. If they are C0G then they are probably VG as bypass caps in the PS. Good enough for this project. Generaly speaking the best would be PPP, but given my experience with the C0G Muratas I am not sure these are not on a par with most MKS in this application... OK, a step under PPP, but then still on a par with many "films" (a word that englobes many things) and probably no audible difference when it comes to PS bypasses for a project as this one.
As of me? Regarding the high end project I am on, space and cost no object, I will go PPP and C0G where I can't re PS bypass. Because I have the choice and it is a proven path, doesn't mean the best path. As for the signal path, I will go $$, that is reasonable PPP caps with no bypass. Here on this Aiyima? No way to do the same, as that doesn't make any sense to me - To be provocative (no comments please) "get rid of Class D to get proper soundstage depth first, LOL" (sorry, harsh I know 🙂
I stop here: rest is non relevant to this project and might end up in bad debates like "what is better than what"...
Why don't you give all that a try and make you own mind? Nobody can predict the outcome and further the result might depend on your taste and ears, also on your system. Some people like it bass heavy to compensate for smaller LS, just as an example. That's one of the reasons why we called it temporarly a day with Gilles (see posts!), because we understood these caps bypasses would be (very) fine tuning, more spicing the end result. Because of that we though wiser to concentrate on the main things first, to have a more mature baseline. What if you say bypass X sounds better then Y, but that's only true because your baseline is wrong and you haven't adddressed many other bottlenecks and X is just correcting a bad sound and because of that standing out. And then, once your unit evolves and gets mature, you realise that in fact Y would fit better, because the main concern X was covering doesn't need to be addressed anymore. Just an example we faced...
My advice: address other bits than these 8 caps (just do the basics like easy bypass), starting with OPA, pot, PS etc. And once done, spice your 8 caps to adjust your sound as this unit will never be perfect, but to get it closer to what you like for YOUR system and ears.
Just my 2 ps
Claude
First off, please eventhough the treatment might for whatever coincidence (as in our case) ending up being the same, please don't talkabout the 9 caps. There aren't really a bunch of 9 caps as explained. There is on single outstanding cap, that on that special occasion feeds one of the inputs of the OPA but that should IMHO be regarded as a PS caps. And then there are the 8 remaining caps, caps in the the signal path. DC blocking caps. These see different impedances and are located at different "stages", so that for each sêcific case one could calculate the optimal capacity value to go for, and that wouldn't be 10uF. But sorry, I can't get any deeper into this: key words are again high pass filter RC mainly...
Now, regarding the 8 caps, IMHO all this is just preferences. I decided to leave the existing caps there and just go for 0.1uF + C0G bypasses.
In details
a) Electrolytics (22uf) + .1uf C0G bypass
=> Why not, but I doubt the increased electrolytic will change anything once bypassed, other than colouring perhaps.
b) Small film (MKS2B051001N00JSSD) for example or (ECQ-E1106JF)
=> same comments... usualy working on the signal path caps is one of the priorities, here IMHO it isn't. I commented already why. No ROI, waste of time... little gain.
c) Even smaller capacity (1-5uf) film (ECW-FD2W225J4) or similar
=> That depends on which cap / its location and what low frequency you will cut off... to be sure you could start with the 4 caps located physicaly near the heatsink. You to try and to tell us, no idea about the outcome vs a)
d) Stick with the 10uf ceramics and bypass them with .1uf C0G
=> Was my choice. Little to be gained anyway, so why bother? Plus everytime you remove a cap you take the risk to wreck the tracks and go into big trouble. Here it is risk x 9.
Again, we were able n our VERY SPECIFIC set up to take 2 caps completely out and hence to evaluate the gain of a perfect cap (= no cap). We were disapointed to hear very little result. Whatever combo is chosen, regardless the $$ of the caps (but admitelly the colouration one might like), at least 50% of the gains achievable are obtained in our opinions with the simple .1uF C0G bypass. And 100% would be very little gain anyway, low magnitude...
e) Other?
=> Don't bother, the truth is elsewhere. And if you are really after the ultimate, start another project. This is just a class D amp that as it stands struggles to compete vs modern 500E amps. Once tuned its sonic play perhaps in the 1000-2000E arena. TBD, we don't know yet, we haven't tried vs modern competitor in that bracket. We don't own such hifi gear (either amps below 1000E... or much more in fact). But don't expect miracles or believe people telling you this is the last say... It is though good enough once tweaked to be happy with re power amp and excellent VFM. All IMHO
Also since I'm thinking of it..... Are film or ceramic better for bypass caps?
=> Generaly speaking FILMS. Ceramic are non linear, piezoelectric etc. and I avoid them usualy in the signal path. And everywhere they would probably be inferior to film caps, unless considering non audio applications. Having said that, sometimes you just have to accept them as there is no alternative. And once you consider PS bypass caps, perhaps the question becomes academic as then the difference matters less re distorsions etc.
And to make it more complex, there are VG ceramics and average films... they might be a no man's land and no clear cut.
I can get small .1uf film caps or C0G ceramics
=> C0G is not a technology per se but a spec re results / delivery. If they are C0G then they are probably VG as bypass caps in the PS. Good enough for this project. Generaly speaking the best would be PPP, but given my experience with the C0G Muratas I am not sure these are not on a par with most MKS in this application... OK, a step under PPP, but then still on a par with many "films" (a word that englobes many things) and probably no audible difference when it comes to PS bypasses for a project as this one.
As of me? Regarding the high end project I am on, space and cost no object, I will go PPP and C0G where I can't re PS bypass. Because I have the choice and it is a proven path, doesn't mean the best path. As for the signal path, I will go $$, that is reasonable PPP caps with no bypass. Here on this Aiyima? No way to do the same, as that doesn't make any sense to me - To be provocative (no comments please) "get rid of Class D to get proper soundstage depth first, LOL" (sorry, harsh I know 🙂
I stop here: rest is non relevant to this project and might end up in bad debates like "what is better than what"...
Why don't you give all that a try and make you own mind? Nobody can predict the outcome and further the result might depend on your taste and ears, also on your system. Some people like it bass heavy to compensate for smaller LS, just as an example. That's one of the reasons why we called it temporarly a day with Gilles (see posts!), because we understood these caps bypasses would be (very) fine tuning, more spicing the end result. Because of that we though wiser to concentrate on the main things first, to have a more mature baseline. What if you say bypass X sounds better then Y, but that's only true because your baseline is wrong and you haven't adddressed many other bottlenecks and X is just correcting a bad sound and because of that standing out. And then, once your unit evolves and gets mature, you realise that in fact Y would fit better, because the main concern X was covering doesn't need to be addressed anymore. Just an example we faced...
My advice: address other bits than these 8 caps (just do the basics like easy bypass), starting with OPA, pot, PS etc. And once done, spice your 8 caps to adjust your sound as this unit will never be perfect, but to get it closer to what you like for YOUR system and ears.
Just my 2 ps
Claude
Quote: So do you think I could get away with a 2.2uf film cap for the 4 OPA output caps? IE, is that enough capacitance?
=> Calculation and specsheets said yes - it should me more than enough FOR THESE CAPS.
I haven't done it though, as explained. And Dr Mordor (I favour experience over theory in many touchy cases) said that despite calculations it was a tad short.
Bottom line: you run a 30Hz sinus tone (or whatever your LS are capable of) at a given pot position and record with an oscilloscope or worst case an application the LS sonic output level. You don't touch the pot and try your 2uF caps and record the output level of said signal. If the same, you win and I was right.
Then you step back from theory, let it all burn-in and after a day you listen to it and decide if you LIKE the sound better. Then you will get the best of every world, provided you have a good memory for sonic qualities.
All IMHO
Claude
=> Calculation and specsheets said yes - it should me more than enough FOR THESE CAPS.
I haven't done it though, as explained. And Dr Mordor (I favour experience over theory in many touchy cases) said that despite calculations it was a tad short.
Bottom line: you run a 30Hz sinus tone (or whatever your LS are capable of) at a given pot position and record with an oscilloscope or worst case an application the LS sonic output level. You don't touch the pot and try your 2uF caps and record the output level of said signal. If the same, you win and I was right.
Then you step back from theory, let it all burn-in and after a day you listen to it and decide if you LIKE the sound better. Then you will get the best of every world, provided you have a good memory for sonic qualities.
All IMHO
Claude
I received my OP1656 on Brown Dog adapters today. I would say this... anyone in the USA that does not plan on soldering that tiny 1656 op amp on an adapter, then Cimarron Technology (Eagle, Idaho) is the place to buy from ($13.60 each is very fair - $31.15 shipped). Besides the quality, you are guaranteed a genuine 1656 op amp.
I am not sure if Cimarron Technology would ship out of the USA; may be worth contacting them and asking if interested.
I am not sure if Cimarron Technology would ship out of the USA; may be worth contacting them and asking if interested.
Attachments
That's a lot to digest. I think I now feel more over my head than I did at the beginning....lol. I'll keep you posted as to what I decide but I think I need a better understanding of what I'm doing and why I'm doing it vs just trying stuff out. My guess is I'll end up with a second amp to start clean at some point. I'm so far away from the baseline now it's very hard to tell what's improvement and what's regression, or what's in my head.
I probably should have just stick with the mods listed here and been content.
I probably should have just stick with the mods listed here and been content.
Very nice. These might be worth a second try of OPA1656 for me. I'm not 100% sure the ones I ordered were genuine, though I got them from a US seller. The pins on it were on the verge of too big for these sockets so I had to do some coaxing just to get them seated enough to work which may have also altered their sound. Plus the soldering job was suspect at best, nothing like these look.
They are my go to company for soic or vssop to dip8 adapters. I have ordered some adapters with op-amps pre-soldered as well.
Cheers & happy building,
Pete
I think it was daniboun in a post somewhere that I read where you need round pins not square on the Dip-8 adapter for the A04 amp. So, if you have square, that may be the issue, and you risk the possibility of damaging the sockets.
I have been reading and absorbing, taking notes on this 3251 (A04); so much information. One of which was the Brown Dog adapers. After I purchased the heavily fluxed 1656 on adapters from eBay, I read about the Brown Dog's and noted that. I just searched for and found that info... turion64 posted that info in post #1824 (pg 183) on the lengthy TPA3251d2 thread. I seriously had a hard time finding the 1656 op amps mounted on adapters here in the USA, until turion64's post.
So, much great thanks to turion64 for that information!
Also, I greatly appreciate the contributions of Rob43 for this thread, rhing, daniboun, and ClaudeG, plus many others. Seriously, so much information passed on. Think of it, 30+ years ago, you'd have to go to the library and research and research, then join some clubs and drive to the meetings and learn with a very slow process of information acquisition.
I have been reading and absorbing, taking notes on this 3251 (A04); so much information. One of which was the Brown Dog adapers. After I purchased the heavily fluxed 1656 on adapters from eBay, I read about the Brown Dog's and noted that. I just searched for and found that info... turion64 posted that info in post #1824 (pg 183) on the lengthy TPA3251d2 thread. I seriously had a hard time finding the 1656 op amps mounted on adapters here in the USA, until turion64's post.
So, much great thanks to turion64 for that information!
Also, I greatly appreciate the contributions of Rob43 for this thread, rhing, daniboun, and ClaudeG, plus many others. Seriously, so much information passed on. Think of it, 30+ years ago, you'd have to go to the library and research and research, then join some clubs and drive to the meetings and learn with a very slow process of information acquisition.
Quote: I probably should have just stick with the mods listed here and been content.
Nah, this is for fun, and experimenting + learning is part of it.
Tell you what: why not just indeed going for the mods listed here FIRST, to sharpen your ears and knwoledge... and to have at the end a baseline, thous also something others might refer to. And THEN start your own kitchen, your own mods, one by one, evaluating each one separately vs the listed mods?
Based on the sensitivity of some changes, based on the sonic results, you can start digging on how it worked / what changed and why. And if you have brand new ideas, you can still ask questions (to yourself or here) before going for it, making it muche easier than tackling a lot of options in one go or reinventing the wheel... Just the tyre is already a lot of work ;-)
Well, that's what I did in the past and still do. For example when I built the B1 Korg I went exactly by the book (who am I to deviate from Papa!?), to have a real Pass reference / sonic signature, and then, once I had played with the settings within recommendations... and away from that once I understood it could do a lot more, I started modding step by step, posting each time or asking questions to make sure.
I know have what I consider (and many since...) an enhenced B1 Korg - still what Papa designed, better to my ears in some areas he BTW left open probably on purpose for us curious DIYer, no doubt given the responses he gave in person. And the community helped me progressing very quickly on some areas whete I had doubts, explaining things in depth...
Worked for me - but that's just me and the way I like working
Claude
Nah, this is for fun, and experimenting + learning is part of it.
Tell you what: why not just indeed going for the mods listed here FIRST, to sharpen your ears and knwoledge... and to have at the end a baseline, thous also something others might refer to. And THEN start your own kitchen, your own mods, one by one, evaluating each one separately vs the listed mods?
Based on the sensitivity of some changes, based on the sonic results, you can start digging on how it worked / what changed and why. And if you have brand new ideas, you can still ask questions (to yourself or here) before going for it, making it muche easier than tackling a lot of options in one go or reinventing the wheel... Just the tyre is already a lot of work ;-)
Well, that's what I did in the past and still do. For example when I built the B1 Korg I went exactly by the book (who am I to deviate from Papa!?), to have a real Pass reference / sonic signature, and then, once I had played with the settings within recommendations... and away from that once I understood it could do a lot more, I started modding step by step, posting each time or asking questions to make sure.
I know have what I consider (and many since...) an enhenced B1 Korg - still what Papa designed, better to my ears in some areas he BTW left open probably on purpose for us curious DIYer, no doubt given the responses he gave in person. And the community helped me progressing very quickly on some areas whete I had doubts, explaining things in depth...
Worked for me - but that's just me and the way I like working
Claude
+1 on kempral81 (and thanks for the compliment BTW)
Ouh, if these OPA1656 were indeed mounted on square pins adapters, unless that was a special order per request, it doesn't sound good at all. Cheap way to avoid the expense of a couple of cents, means often possibly fake chip and/or bad solder job due to lack of skill / knowledge and defo very bad for the existing socket (damaging it if it ever fits).
Run away from that, just get the genuine part and a SMD training kit and get started with SMDs before soldering your first SMD OPA (what I did after 20 years). Or get a student to do it for you (what I did 20y ago LOL). Or get the good from a reputable shop, defo no Ebay & co for that, even worst than single components
All IMHO
Claude
Ouh, if these OPA1656 were indeed mounted on square pins adapters, unless that was a special order per request, it doesn't sound good at all. Cheap way to avoid the expense of a couple of cents, means often possibly fake chip and/or bad solder job due to lack of skill / knowledge and defo very bad for the existing socket (damaging it if it ever fits).
Run away from that, just get the genuine part and a SMD training kit and get started with SMDs before soldering your first SMD OPA (what I did after 20 years). Or get a student to do it for you (what I did 20y ago LOL). Or get the good from a reputable shop, defo no Ebay & co for that, even worst than single components
All IMHO
Claude
I ordered a pair of brown dogs to give the 1656's a fair shake. Hopefully I'm as impressed with them as everyone else. If not, no worries.
Don't forget also to give it a clean PS. They are excellent OPAs, but perhaps more demanding re PS than old average ones...
Sorry... PS = power supply
Meant don't forget OPA PS bypass caps, they do make a lot of difference on this Aiyima as we found out.
It could well be that " if just amp rolling " you prefer OPA X , because it is less sensitive to current feed, whereas you later find out OPA Y is way better once the current feed to the OPA is really good and in our case that means quite a few bypass caps closest to each OPA. Seems to be another bottleneck of this unit.
It is a bit like say a race car that can't make do with an average tyres or geo, whereas a daily shed is far more compliant and often even happier with less sharp settings to mask its limits.
Perhaps a share of the minority that prefered here another OPA than OPA1656 might have gone for say a bass heavier or less treble sharp op amp... because the OPA current feed wasn't optimal. Or because it addressed other sonic balance problems (small LS without first ocatve etc.). Just a guess... I don't see many bypass caps on pix in this thread re OPA... but a lot of amp rolling is reported.
To even things, you need say at least kind of a ".1uF C0G + 10uF+10uF possibly of mixed tech" (say 'lytic + X7R). Of course 2x 10uF of 22uF PPP would be much better, but that's out of scope, hence making do...
All IMHO
Claude
Meant don't forget OPA PS bypass caps, they do make a lot of difference on this Aiyima as we found out.
It could well be that " if just amp rolling " you prefer OPA X , because it is less sensitive to current feed, whereas you later find out OPA Y is way better once the current feed to the OPA is really good and in our case that means quite a few bypass caps closest to each OPA. Seems to be another bottleneck of this unit.
It is a bit like say a race car that can't make do with an average tyres or geo, whereas a daily shed is far more compliant and often even happier with less sharp settings to mask its limits.
Perhaps a share of the minority that prefered here another OPA than OPA1656 might have gone for say a bass heavier or less treble sharp op amp... because the OPA current feed wasn't optimal. Or because it addressed other sonic balance problems (small LS without first ocatve etc.). Just a guess... I don't see many bypass caps on pix in this thread re OPA... but a lot of amp rolling is reported.
To even things, you need say at least kind of a ".1uF C0G + 10uF+10uF possibly of mixed tech" (say 'lytic + X7R). Of course 2x 10uF of 22uF PPP would be much better, but that's out of scope, hence making do...
All IMHO
Claude
Gotcha, yeah, its a shame so much his required to make the 1656's "sound good". Arent the 4 10uf OPA input caps responsible for that? Would using something larger and/or with lower ESR accomplish similar results? That would also potentially improve any opamp used too.
It actually looks like (to my inexperienced eyes) that the LM4562's require more input current than the OPA1656's (10mA x 2 vs 3.9mA x 2)