Linuxfan - I found a suggestion from you in another thread
Just to make sure the IC is turned the right way - is this correct for connecting balanced xlr to the dip 8 adaptor ?
+ve XLR-F pin 2 (red) to DIP8 pin 1
-ve XLR-F pin 3 (white) to DIP8 pin 7
AGND XLR-F pin 1 (braid) to DIP8 pin 4
Attachments
As per previous post and pix, the 0.1uF C0G Murata to bypass the OPA’s PS was a clear winner with +2. At this stage we fitted the caps under the unit’s board to release space on the top of the OPA DIP8 board, but once finalised it all belongs closest to the OPA, so on the top of the little boards.
We brought both our amps to the same tuning state, both with our latest 0.1uF mod, and we added 1x 10uF C7R Murata cap per OPA on one of the two amps to carry on our bypass test. We did so under the board for the same reason : testing and keeping free space at the top in case off…
That additional move was mixed blessing. It did add some kick in the upper bass region, more energy and more prominent upper bass/low mids, all the rest remaining unchanged. It was fun and those who find the OPA1656 too neutral, or Class D too neutral and who prefer warming up the sound with op amps will love that bypass ! As of us, we found it gave a nice kick on some pop and rock music, but it also brought that lower range forward on the scene while not bringing real lower bass to balance things. To us, it brings the upper bass (say 100-150Hz region) forward, project sounds, gives kick but also blurs the message. On complexe music it was too much and to forward, and clearly not linear, bringing a « upper bass hump » in the frequency response. Perhaps somewhat seductive on average pop records missing bass and that don’t have a great soundstage spacialisation, +1 or +2 then like a loudness effect, but clearly a damage on good recorded music with real soundstage, -1 or worst, and that recurrent feeling that the amp isn’t neutral anymore, cheats with a upper bass hump. We decided it sounded best… WITHOUT the 10uF cap.
However, we must say there is something to the energy it brought. We also mentioned that so far most tweaks enhenced all the frequency range, but the bass less so than the rest, so we may need adjust the balance with the appropriate tweaks at the end to get a perfect balance. In fact we would like that extra kick to be everywhere and not frequency biased. Or if a hump is there, then it has to be way way smaller to be acceptable (eg linear response) and also mandatory shifted as low as possible, in the deep bass, where it does less damage or might even prove a plus to get the balance right. We need to dig.. Rob as a lot fo work to do playing with OPA PS bypasses !
At that stage I draw a first conclusion. To address the DC blocking caps in the signal path, we purchased Elnas, Wimas and other main caps, many of them, and also some additional bypass caps for those in form of Murata 0.1uF. We found out that the original caps bypassed brought an up to +0.5 gain. Read small gain, but easy to do. Taking a cap completely out, which is the maximum sonic gain you can achieve, brought us only +1. That is probably not achievable with DC blocking caps, or say perhaps the best PPP can achieve that. So we are set to play with lot of variations around the 2 remaining DC blocking caps per channel, involving a lot of money and time and risk… for an extra remaining potential of maximal 2x0.5 =+1. That’s a lot of trouble for nearly naught ! Hence me posting the unit wasn’t very responsive in terms of pure gains to DC blocking caps tuning, but it would be a nice option to spice/adjust the sound at the end, should we need it to. Things as adjusting the tonal balance as final touch.
And now we just found out that OPA PS bypassing had far more sonic potential gains while also influencing/adjusting the sonic balance far more ! For this reason, we will skip further DC blocking caps tweaks – they aren’t worth it- and move on like Rob to OPA PS bypass caps tuning… using the Elna, Wimas and Murata we deserved for the signal caps tuning.
Once we are done with OPA PS tuning, we will also dig further into other low voltage PS bypasses. Despite having fit 2.2uF « a bit everywhere » on other PS caps, it appears that X7Rs are likely not to sound as great as C0Gs, even as PS bypasses for pure audio PS. And we do know that there are similar low voltage PS as for the OPAs, adressing directly the sound, starting with « somewhere » a +12V feed for the TI chip amp’s internal 1st amplification stage, fed separately and acting as an internal OPA probably. We need to find that one and try Murata 0.1s on it, and also on that other analogue +12V feed that serves the TI’s internal analogue and digital regulators. The other (digital section or not) low voltage feeds, be it 12 or 5V, are IMHO less important and already partly addressed.
So more OPA PS bypass tweaks to come… and now additional bypasses added on our TODO list, those that could address any +12V analogue PS in direct relation to sound. Oh, and while at it, we will also give these Muratas a go on the main PS 3000uF caps - these have already benefited from 2x2.2uF X7Rs… who knows how they like extra 0.1uF C0Gs
Enjoy music
Claude… and Gilles
We brought both our amps to the same tuning state, both with our latest 0.1uF mod, and we added 1x 10uF C7R Murata cap per OPA on one of the two amps to carry on our bypass test. We did so under the board for the same reason : testing and keeping free space at the top in case off…
That additional move was mixed blessing. It did add some kick in the upper bass region, more energy and more prominent upper bass/low mids, all the rest remaining unchanged. It was fun and those who find the OPA1656 too neutral, or Class D too neutral and who prefer warming up the sound with op amps will love that bypass ! As of us, we found it gave a nice kick on some pop and rock music, but it also brought that lower range forward on the scene while not bringing real lower bass to balance things. To us, it brings the upper bass (say 100-150Hz region) forward, project sounds, gives kick but also blurs the message. On complexe music it was too much and to forward, and clearly not linear, bringing a « upper bass hump » in the frequency response. Perhaps somewhat seductive on average pop records missing bass and that don’t have a great soundstage spacialisation, +1 or +2 then like a loudness effect, but clearly a damage on good recorded music with real soundstage, -1 or worst, and that recurrent feeling that the amp isn’t neutral anymore, cheats with a upper bass hump. We decided it sounded best… WITHOUT the 10uF cap.
However, we must say there is something to the energy it brought. We also mentioned that so far most tweaks enhenced all the frequency range, but the bass less so than the rest, so we may need adjust the balance with the appropriate tweaks at the end to get a perfect balance. In fact we would like that extra kick to be everywhere and not frequency biased. Or if a hump is there, then it has to be way way smaller to be acceptable (eg linear response) and also mandatory shifted as low as possible, in the deep bass, where it does less damage or might even prove a plus to get the balance right. We need to dig.. Rob as a lot fo work to do playing with OPA PS bypasses !
At that stage I draw a first conclusion. To address the DC blocking caps in the signal path, we purchased Elnas, Wimas and other main caps, many of them, and also some additional bypass caps for those in form of Murata 0.1uF. We found out that the original caps bypassed brought an up to +0.5 gain. Read small gain, but easy to do. Taking a cap completely out, which is the maximum sonic gain you can achieve, brought us only +1. That is probably not achievable with DC blocking caps, or say perhaps the best PPP can achieve that. So we are set to play with lot of variations around the 2 remaining DC blocking caps per channel, involving a lot of money and time and risk… for an extra remaining potential of maximal 2x0.5 =+1. That’s a lot of trouble for nearly naught ! Hence me posting the unit wasn’t very responsive in terms of pure gains to DC blocking caps tuning, but it would be a nice option to spice/adjust the sound at the end, should we need it to. Things as adjusting the tonal balance as final touch.
And now we just found out that OPA PS bypassing had far more sonic potential gains while also influencing/adjusting the sonic balance far more ! For this reason, we will skip further DC blocking caps tweaks – they aren’t worth it- and move on like Rob to OPA PS bypass caps tuning… using the Elna, Wimas and Murata we deserved for the signal caps tuning.
Once we are done with OPA PS tuning, we will also dig further into other low voltage PS bypasses. Despite having fit 2.2uF « a bit everywhere » on other PS caps, it appears that X7Rs are likely not to sound as great as C0Gs, even as PS bypasses for pure audio PS. And we do know that there are similar low voltage PS as for the OPAs, adressing directly the sound, starting with « somewhere » a +12V feed for the TI chip amp’s internal 1st amplification stage, fed separately and acting as an internal OPA probably. We need to find that one and try Murata 0.1s on it, and also on that other analogue +12V feed that serves the TI’s internal analogue and digital regulators. The other (digital section or not) low voltage feeds, be it 12 or 5V, are IMHO less important and already partly addressed.
So more OPA PS bypass tweaks to come… and now additional bypasses added on our TODO list, those that could address any +12V analogue PS in direct relation to sound. Oh, and while at it, we will also give these Muratas a go on the main PS 3000uF caps - these have already benefited from 2x2.2uF X7Rs… who knows how they like extra 0.1uF C0Gs
Enjoy music
Claude… and Gilles
Attachments
So, let's carry on with OPA PS tweaks.
At that stage we have the 0.1uF C0G Muratas in both amps and they will stay - perhaps we will at the end just change their location to bring them closer to the OPA’s legs. We won't come back to them, they are done and confirmed, what follows included them always.
The question is : what can we do about much larger energy feeds and filterings, read bigger cap's values as PS bypasses ? Rob will explore all that in depth, but while at it, let's see what we can do... while probabaly shamelessly copying Rob's final findings at the end.
We need to understand why the additional 10uF X7Rs were a negative, why they shifted the register bringing a hump while indeed bringing more welcome energy in play. Is it due to the capacity value, or to the technology, or both ?
TBH, I have a clue on that and there would be in an ideal world a perfect solution regarding the PS caps : fit 30uF or so of paralleled PPP caps, job done. Perfect feed, makes a nearly perfectly quite and powerfull battery feed but with extremly low ESR across all frequencies. ESR is also probably the cause of our problem, as we want it low but ESR on our caps is also very frequency related – read some of our caps have an OK quite low ESR but that turns to bad highish ESR a(nd probably other probs) at lower frequencies… those frequencies where we are experiencing problems.
We can’t fit a lot of large PPP caps, that would be cost overkill and not in line with the form factor. What we can do though, is trying to understand and be clever, playing with cap technology and quality bypasses.
To explore things, we will confront various bypass configurations . Our baseline is « bypass only 0.1uF C0G Murata », and all what follows will be additions to that configuration :
- 1x 10uF X7R (turns out not good)
- 1x 10uF X7R + 10uF Elna Simic II (Audio electrolytics)
- 2x 10uF Elna Simic II (Audio electrolytics)
- Possibly 30uF, mix of caps TBD, playing with tech and paralleling caps to bring the ESR down while maxing value to see where the energy gain stops
Once done and depending on the result, we will have the option to add on top a 1uF Wima MKS – OK, no PPP, but close to it fo PS purposes.
We fitted the caps quick and dirty for test purposes only, leaving the housing open and having each time one amp in one configuration, the other amp in the other configuration and being as ever able to compare nearly on the fly.
Evaluations have already started, brun-in is important to avoid mistakes, and we will of course report on the end results once done…
Claude and Gilles
At that stage we have the 0.1uF C0G Muratas in both amps and they will stay - perhaps we will at the end just change their location to bring them closer to the OPA’s legs. We won't come back to them, they are done and confirmed, what follows included them always.
The question is : what can we do about much larger energy feeds and filterings, read bigger cap's values as PS bypasses ? Rob will explore all that in depth, but while at it, let's see what we can do... while probabaly shamelessly copying Rob's final findings at the end.
We need to understand why the additional 10uF X7Rs were a negative, why they shifted the register bringing a hump while indeed bringing more welcome energy in play. Is it due to the capacity value, or to the technology, or both ?
TBH, I have a clue on that and there would be in an ideal world a perfect solution regarding the PS caps : fit 30uF or so of paralleled PPP caps, job done. Perfect feed, makes a nearly perfectly quite and powerfull battery feed but with extremly low ESR across all frequencies. ESR is also probably the cause of our problem, as we want it low but ESR on our caps is also very frequency related – read some of our caps have an OK quite low ESR but that turns to bad highish ESR a(nd probably other probs) at lower frequencies… those frequencies where we are experiencing problems.
We can’t fit a lot of large PPP caps, that would be cost overkill and not in line with the form factor. What we can do though, is trying to understand and be clever, playing with cap technology and quality bypasses.
To explore things, we will confront various bypass configurations . Our baseline is « bypass only 0.1uF C0G Murata », and all what follows will be additions to that configuration :
- 1x 10uF X7R (turns out not good)
- 1x 10uF X7R + 10uF Elna Simic II (Audio electrolytics)
- 2x 10uF Elna Simic II (Audio electrolytics)
- Possibly 30uF, mix of caps TBD, playing with tech and paralleling caps to bring the ESR down while maxing value to see where the energy gain stops
Once done and depending on the result, we will have the option to add on top a 1uF Wima MKS – OK, no PPP, but close to it fo PS purposes.
We fitted the caps quick and dirty for test purposes only, leaving the housing open and having each time one amp in one configuration, the other amp in the other configuration and being as ever able to compare nearly on the fly.
Evaluations have already started, brun-in is important to avoid mistakes, and we will of course report on the end results once done…
Claude and Gilles
Attachments
Now I have made some comparition between OPA1642 and 1654. On the 1654 I did put a bypass cap (10uF Silmic and 0.1uF ceramic).
At first I liked the sound from 1654 but this morning I was listening to some barock music and the violin sounded very odd, to much of the harshness from the strings. I then switched back to 1642 and the music came back!
Conclution: I do prefer OPA1642 with a great margin.
/S
At first I liked the sound from 1654 but this morning I was listening to some barock music and the violin sounded very odd, to much of the harshness from the strings. I then switched back to 1642 and the music came back!
Conclution: I do prefer OPA1642 with a great margin.
/S
Ahem... OPA1642 is a fairly good DUAL op amp indeed... while OPA1654 is more average QUAD amp... that won't fit into our units unless you want to join both channels into one op amp....
Give OPA1656 a go, that one should outperform slightly for our audio application OPA1642, which is nevertheless very good in other applications.
Last but not least, none of these OPA should sound odd, but having oscillations or band favorisation in the PS with a wrong PS caps layout (be it values, or caps choice, or wires) can induce indeed very very odd things... and make even good op amps sound odd, as you will soon read here. Some OPA are then more resistant to others to these PS probs...
All IMHO and important is you are happy
Claude
Give OPA1656 a go, that one should outperform slightly for our audio application OPA1642, which is nevertheless very good in other applications.
Last but not least, none of these OPA should sound odd, but having oscillations or band favorisation in the PS with a wrong PS caps layout (be it values, or caps choice, or wires) can induce indeed very very odd things... and make even good op amps sound odd, as you will soon read here. Some OPA are then more resistant to others to these PS probs...
All IMHO and important is you are happy
Claude
The OPA PS bypasses have turned out to be quite interesting. As mentioned, our baseline was « bypass only 0.1uF C0G Murata », and all what follows is in addition to that configuration.
First we went for 1x 10uF X7R… and as posted we didn’t like the result, but it did make quite a difference and all wasn’t negative.
We went for 10uF/35V Elna Simic II, and that worked much better… but in fact not really better than nothing ! The improvements if any were tiny and we didn’t find any of the positive we had with the X7R… and also none of its negative.
We thought perhaps doubling the capacity and dividing by 2 the ESR, going for 2x10uF Elna would bring a change. Not really, sounded nearly identical to 1 Elna, that is close to nothing and not worth the hassle. At that stage we felt that going for any total bypass value much higher than 20uF wouldn’t be efficient, which matches also my experience on previous OPA PS. So we discarted the 3x Elna bypass try.
Finaly we decided to go for a « 1x 10uF X7R + 10uF Elna Simic II » bypass. We hoped that the caps would combine and perhaps get the best of both world. And it worked ! In conjunction with the 0.1uF C0G, the combo sounded clearly best and suprisingly good. The blurred sound of the « X7R only » is gone, its low mid hump is gone, but the extra speed, the kick and a new bass extension are there. The sound is now very flesh on the bone and very consistant, airy, 3D, precise but never agressive, authentic neutral to slightly enjoyable euphonic. Gilles said the X7R was like a portion of butter you throw on your bred, the Elna acted then as a knife to pread it evently, everywhere. Gone is the hump, instead we have bass extension and if ever (?) a very negligeable low bass hump… which is a bit seductive anyway with the Klipsch. Butter on the bred… A clear difference, that combo earned a +2 gain.
Now this is getting really interesting: if you combine the 3 bypass caps, you get +4 gain to our ears. To put things in perspective, the move to an OPA1656, which is considered « a majour move a deaf could hear », was +5 on our scale. Of course the bypasses don’t have the same sonic effects as swapping OPAs, but to our ears the sonic impact of our PS bypass has a magnitude of 80% of the OPA mod. That is quite a lot a lot of impact. And at the end, when you mod the OPA and its PS you get +9… that isn’t the same Aiyima anymore, it is getting very enjoyable !
So we highly recommend that mod. If you look at the pix, you will see our final layout in quite a few of them, which differs from the test layout on previous pix. The 10uF X7R cap stayed under the amp’s board. The 0.1uF cap went on the DIP8 board, closest as possible to the OPA +V and –V legs, as it should. The 10uF Elna went on top of it, because that’s where we have space for it. Last but not least, we fitted the caps on top of the DIP8 board in such a manner that we could… still add a small last cap, to give it a try : Wima MKS 1uF. Not that we feel we really need to, but as I have them lying around it would be stupid not to try them... when I can access the unit again.
Triggered by all this, we will carry on bypasing all the other low voltage PS as discussed… but that will be only after we gave the main PS a quick try with an additional 0.1uF C0G bypass, our great winner here.
Enjoy music
Claude… and Gilles
First we went for 1x 10uF X7R… and as posted we didn’t like the result, but it did make quite a difference and all wasn’t negative.
We went for 10uF/35V Elna Simic II, and that worked much better… but in fact not really better than nothing ! The improvements if any were tiny and we didn’t find any of the positive we had with the X7R… and also none of its negative.
We thought perhaps doubling the capacity and dividing by 2 the ESR, going for 2x10uF Elna would bring a change. Not really, sounded nearly identical to 1 Elna, that is close to nothing and not worth the hassle. At that stage we felt that going for any total bypass value much higher than 20uF wouldn’t be efficient, which matches also my experience on previous OPA PS. So we discarted the 3x Elna bypass try.
Finaly we decided to go for a « 1x 10uF X7R + 10uF Elna Simic II » bypass. We hoped that the caps would combine and perhaps get the best of both world. And it worked ! In conjunction with the 0.1uF C0G, the combo sounded clearly best and suprisingly good. The blurred sound of the « X7R only » is gone, its low mid hump is gone, but the extra speed, the kick and a new bass extension are there. The sound is now very flesh on the bone and very consistant, airy, 3D, precise but never agressive, authentic neutral to slightly enjoyable euphonic. Gilles said the X7R was like a portion of butter you throw on your bred, the Elna acted then as a knife to pread it evently, everywhere. Gone is the hump, instead we have bass extension and if ever (?) a very negligeable low bass hump… which is a bit seductive anyway with the Klipsch. Butter on the bred… A clear difference, that combo earned a +2 gain.
Now this is getting really interesting: if you combine the 3 bypass caps, you get +4 gain to our ears. To put things in perspective, the move to an OPA1656, which is considered « a majour move a deaf could hear », was +5 on our scale. Of course the bypasses don’t have the same sonic effects as swapping OPAs, but to our ears the sonic impact of our PS bypass has a magnitude of 80% of the OPA mod. That is quite a lot a lot of impact. And at the end, when you mod the OPA and its PS you get +9… that isn’t the same Aiyima anymore, it is getting very enjoyable !
So we highly recommend that mod. If you look at the pix, you will see our final layout in quite a few of them, which differs from the test layout on previous pix. The 10uF X7R cap stayed under the amp’s board. The 0.1uF cap went on the DIP8 board, closest as possible to the OPA +V and –V legs, as it should. The 10uF Elna went on top of it, because that’s where we have space for it. Last but not least, we fitted the caps on top of the DIP8 board in such a manner that we could… still add a small last cap, to give it a try : Wima MKS 1uF. Not that we feel we really need to, but as I have them lying around it would be stupid not to try them... when I can access the unit again.
Triggered by all this, we will carry on bypasing all the other low voltage PS as discussed… but that will be only after we gave the main PS a quick try with an additional 0.1uF C0G bypass, our great winner here.
Enjoy music
Claude… and Gilles
Attachments
Hi, Prometheus here:
I haven't paid much attention to the TPA32XX boards in a while but permit me to inquire as to why the large PS caps haven't been replaced? I see bypass caps experiments, but none with the large bulk caps themselves.
In my previous experience with TPA boards these caps *do* significantly impact sound quality, certainly with changes in value if not brand (Muse vs. Panasonic KA, etc etc).
It's possible that with the 325X this is no longer the case, but if one is willing to go to such lengths to mod all the other components then replacing the main PS caps should have been one of the first mods.
I haven't paid much attention to the TPA32XX boards in a while but permit me to inquire as to why the large PS caps haven't been replaced? I see bypass caps experiments, but none with the large bulk caps themselves.
In my previous experience with TPA boards these caps *do* significantly impact sound quality, certainly with changes in value if not brand (Muse vs. Panasonic KA, etc etc).
It's possible that with the 325X this is no longer the case, but if one is willing to go to such lengths to mod all the other components then replacing the main PS caps should have been one of the first mods.
Attachments
Last edited:
Hi,
We have plan for this, as for the coils (details earlier in this thread). Sadly Claude and I can't meet currently due to lockdown here in France. I have the amps and some bits (mainly small caps) with me, Claude has the rest, including these big caps... We have to wait !...
Gilles... and Claude
Err... You may want to read the posts from the beginning. This cap swap mod is of course on the top of the list, as the output coils/caps and volume potentiometer BTW, and the required parts already here since 2 months... but we are downlocked due to Corona and Gilles hasn't got the hardware to carry out these mods... sadly. Desoldering machine and parts are with me, 250km away from Gilles, and the day we intended to get started together at mine was precisely the day the president decided to lockdown the population and Gilles left back home in a hurry. So only add-on mods or fitting the pre-assembled modules I did that could reach him - or based on new orders at Mouser- are feasible... for now. We couldn't follow the order of TODO mods you see in the first page, due to the virus, and that despite Gilles putting a second order at Mouser. But we wouldn't want to order twice the big or expensive bits I have lying here...
Back to your point, we already mentioned the Nichicon we intend to use instead of the existing presumably fake Elna. We chose them carefully with very low ESR and very high ripple, suited for high and quick / high frequency transfers. They have been discussed in the previous thread. BTW, as mentioned we used them aswell in our SMPS filter and that was the first mod we did... goes between SMPS and the red caps, report on first page or so.
We don't plan though to go inboard for more than (the existing) 2x 3000uF and are not convinced either higher values for this amp are needed or suitable. Having said, that we do indeed have planned on our TODO list from day 1 to try to double the existing 6000uF... and have already added offboard 3000uF (the filter), with little effect though. TI got away with 4400uF in total on its eval board... and we are currently at 9000uF for each amp (and have 2 amps as going bi-amping). Not sure additional 3-6000uF help, or won't trigger the SMPS security at start up (LOL), but we shall see.
Last but not least, bypassing the existing PS caps "whatever they are", the way we did it already... and will do it as next mod tomorrow, is also a way to enhence the quality of those thanks to hopefully clever bypasses (and that worked so far as reported, that was 2nd mod or so). Not best, but as we can't yet replace the original caps as long as locked at home... what else can we do?
WE WANT OUT !!!
Claude
Back to your point, we already mentioned the Nichicon we intend to use instead of the existing presumably fake Elna. We chose them carefully with very low ESR and very high ripple, suited for high and quick / high frequency transfers. They have been discussed in the previous thread. BTW, as mentioned we used them aswell in our SMPS filter and that was the first mod we did... goes between SMPS and the red caps, report on first page or so.
We don't plan though to go inboard for more than (the existing) 2x 3000uF and are not convinced either higher values for this amp are needed or suitable. Having said, that we do indeed have planned on our TODO list from day 1 to try to double the existing 6000uF... and have already added offboard 3000uF (the filter), with little effect though. TI got away with 4400uF in total on its eval board... and we are currently at 9000uF for each amp (and have 2 amps as going bi-amping). Not sure additional 3-6000uF help, or won't trigger the SMPS security at start up (LOL), but we shall see.
Last but not least, bypassing the existing PS caps "whatever they are", the way we did it already... and will do it as next mod tomorrow, is also a way to enhence the quality of those thanks to hopefully clever bypasses (and that worked so far as reported, that was 2nd mod or so). Not best, but as we can't yet replace the original caps as long as locked at home... what else can we do?
WE WANT OUT !!!
Claude
Last edited:
With the lockdown and a couple of changes in our tweaks, Gilles has now « only » 5 0.1uF bypass cap locations left to try. The problem is he has only 4 caps left… for 2 amps, 6 potentialy missing if all should be fruitfull (unlikely but well). That’s why we decided to do quick and dirty tests, not permanent installations, to see what bites and what not.
While doing this, we thought again about one area we don’t master : the 6 caps behind the facade were we randomly put bypass caps with admitelly very little benefits, 3 of them with 0.1uF bypasses (6 for 2 amps). For sure some must be useless and could be taken out, to use them elsewhere now we know the amp a bit better ! I was at least quite convinced the 6.5V cap could, per definition, not really be directly signal amplification PS related, more a logic or digital affair with 5V or 3.3V, something that wouldn’t have any sonic added value. Remember Gilles can only measure power down and has just a multimeter, so we can’t be sure (and won’t waste time for such a minor area). Anyway, we decided to take the caps out again on these 3 smaller value caps (120uF, 100uF, 22uF). It couldn’t really harm we thought.
But it did. As ever, humans are more perceptive when it comes to « retiring a given bonus », than when « adding a plus ». This was small, in the 0.5 league, but retiring the bypass caps caps did affect the sound. Gilles ran several tests, putting some caps back in, making sure it wasn’t his mind and also checking the 2 amps he has are really « still » identicaly sounding. They are absolutely to Gilles ears, which is quite a testament to the Aiyima and to the TI chip in particular. We wasted a lot of time playing with that and yet the end, for a fraction of second, on very complex music (real operas), when the singer goes for a high pitch note, there is this distinctive « lack of confidence in the tone he hears» with all bypasses bar the 6.5V one, and yet it sounds so much more consistant with this insignifiant very low voltage section bypass. Subtle, but there, eventough we tried hard to convince ourselves it simply couldn’t be, it wouldn’t make sense, we should listen again next morning… and eventhough we would have liked to take that cap out ato use it elsewhere, it soundd better with it.
Bottom line : all we did has indeed an impact, sometimes very small, but for the sake of 50 cents we shouldn’t reverse to less quality now. All we did stays, the 2 amps are back to identical configuration (full tuning as of today’s stand) and… sound indeed identical.
We will try our bypasses with the 4 caps left we have, rolling locations each by each, and see what works and not. Should all work, then I will post the required numbers of caps to Gilles’ so he can equip both amps. I have just enough left and they are very small so an enveloppe should do it for these and another tweak. If the post works. Or fingers X there is no more lockdown (ahem) and I just bring the caps… in person with all the hardware that waits here since 2 months !
Enjoy music
Claude… and Gilles
While doing this, we thought again about one area we don’t master : the 6 caps behind the facade were we randomly put bypass caps with admitelly very little benefits, 3 of them with 0.1uF bypasses (6 for 2 amps). For sure some must be useless and could be taken out, to use them elsewhere now we know the amp a bit better ! I was at least quite convinced the 6.5V cap could, per definition, not really be directly signal amplification PS related, more a logic or digital affair with 5V or 3.3V, something that wouldn’t have any sonic added value. Remember Gilles can only measure power down and has just a multimeter, so we can’t be sure (and won’t waste time for such a minor area). Anyway, we decided to take the caps out again on these 3 smaller value caps (120uF, 100uF, 22uF). It couldn’t really harm we thought.
But it did. As ever, humans are more perceptive when it comes to « retiring a given bonus », than when « adding a plus ». This was small, in the 0.5 league, but retiring the bypass caps caps did affect the sound. Gilles ran several tests, putting some caps back in, making sure it wasn’t his mind and also checking the 2 amps he has are really « still » identicaly sounding. They are absolutely to Gilles ears, which is quite a testament to the Aiyima and to the TI chip in particular. We wasted a lot of time playing with that and yet the end, for a fraction of second, on very complex music (real operas), when the singer goes for a high pitch note, there is this distinctive « lack of confidence in the tone he hears» with all bypasses bar the 6.5V one, and yet it sounds so much more consistant with this insignifiant very low voltage section bypass. Subtle, but there, eventough we tried hard to convince ourselves it simply couldn’t be, it wouldn’t make sense, we should listen again next morning… and eventhough we would have liked to take that cap out ato use it elsewhere, it soundd better with it.
Bottom line : all we did has indeed an impact, sometimes very small, but for the sake of 50 cents we shouldn’t reverse to less quality now. All we did stays, the 2 amps are back to identical configuration (full tuning as of today’s stand) and… sound indeed identical.
We will try our bypasses with the 4 caps left we have, rolling locations each by each, and see what works and not. Should all work, then I will post the required numbers of caps to Gilles’ so he can equip both amps. I have just enough left and they are very small so an enveloppe should do it for these and another tweak. If the post works. Or fingers X there is no more lockdown (ahem) and I just bring the caps… in person with all the hardware that waits here since 2 months !
Enjoy music
Claude… and Gilles
GLOL
I am though still suprised such a ridiculously small cap, admittedly in a chosen key position, can have so much effect.
Must confess though I am delighted about it and bypassing remains in some cases a very cost and formfactor effective way of "replacing with litlle trade offs" bigger $$ caps
We would probably never have pushed that "bypass quest" if we weren't locked down ;-)
Gilles had nothing else to play with re amp tuning, let's hope the serious stuff à la output filter and main caps gives more benefits LOL!
Claude
I am though still suprised such a ridiculously small cap, admittedly in a chosen key position, can have so much effect.
Must confess though I am delighted about it and bypassing remains in some cases a very cost and formfactor effective way of "replacing with litlle trade offs" bigger $$ caps
We would probably never have pushed that "bypass quest" if we weren't locked down ;-)
Gilles had nothing else to play with re amp tuning, let's hope the serious stuff à la output filter and main caps gives more benefits LOL!
Claude
Hi,
Why don't you wait for a couple of days: we will by then be done with ALL the 0.1uF bypasses one can think of with this amp, and able to tell where it is the most efficient.
As of pix, they are already in the posts and all bypasses have already be shown.
As of today, re 0.1uF bypasses, I would tend to say that you could use 2 for the op amps PS (one per OPA) and 1 for the VMID bypass (that's the 10uF cap in the middle of the row of 5 caps just under the OPA).
Then there is still one spare to use and we are currently investigating 4 different locations... but TBH you really want more of these caps to put them where they have an added value. Even if it is less efficient than in the locations I just mentioned, they are no brainer considering the price and the gains.
I hope this helps
Claude
Why don't you wait for a couple of days: we will by then be done with ALL the 0.1uF bypasses one can think of with this amp, and able to tell where it is the most efficient.
As of pix, they are already in the posts and all bypasses have already be shown.
As of today, re 0.1uF bypasses, I would tend to say that you could use 2 for the op amps PS (one per OPA) and 1 for the VMID bypass (that's the 10uF cap in the middle of the row of 5 caps just under the OPA).
Then there is still one spare to use and we are currently investigating 4 different locations... but TBH you really want more of these caps to put them where they have an added value. Even if it is less efficient than in the locations I just mentioned, they are no brainer considering the price and the gains.
I hope this helps
Claude