[dboy]

And if it was me, I would at least stick the surface mount caps by the 3116 back in.  I would guess the MR doesn't like a lot of capacitance close by, but a little sm cap with some trace in between should be OK.

Thanks Randy. The app notes for MR state the following, so if the SMD caps are as the TPA3116 datasheet circuit then I guess I could have left them in:

"Ensure that no output cap is present. If this is unavoidable, try to make the cap value less than 0.1uF or greater than 20uF. It is possible to go much higher than this, into the thousands of microfarads, if certain precautions are taken..."

With my Charlize I was not sure what the effect of the MR would be so I just soldered it in with all the caps still present to see what happened. The result was such a knockout that I never got around to trying without them - I just didn't want to mess with that sound at all. This time around I wish I had tried that first for comparison. I am too clumsy at soldering to reinstall them. I do have another Audiobah board to play with so may have to try that - it just means removing the input caps to get a like for like comparison as the dedicated amp for the Buffalo.

[randytsuch]

Thanks Randy. The app notes for MR state the following, so if the SMD caps are as the TPA3116 datasheet circuit then I guess I could have left them in:

"Ensure that no output cap is present. If this is unavoidable, try to make the cap value less than 0.1uF or greater than 20uF. It is possible to go much higher than this, into the thousands of microfarads, if certain precautions are taken..."

With my Charlize I was not sure what the effect of the MR would be so I just soldered it in with all the caps still present to see what happened. The result was such a knockout that I never got around to trying without them - I just didn't want to mess with that sound at all. This time around I wish I had tried that first for comparison. I am too clumsy at soldering to reinstall them. I do have another Audiobah board to play with so may have to try that - it just means removing the input caps to get a like for like comparison as the dedicated amp for the Buffalo.

So little SM caps are not fun, but I'd give it a shot before you give up on that board.  It really helps (at least it helps me) to have a pair of tweezers to hold the cap, and a magnifying glass to see the part.  I have a visor type magnifying glass so I can just wear it when I work. 

Also make sure there is no short after you solder the cap in place.

[dboy]

Correction - just reread the cap values on the datasheet circuit and I think I was right to remove them if Audiobah board has same values of 1nF & 100nF (x2). 100nF is 0.1uF right? I'm a bit tired and got confused before. I gotta crash now - big day tomorrow. Cheers (& thanks for advice - may try it this one day soon for the experience)! 

[Markvdv]

What makes you think any of the capacitors you removed are output capacitors? If you want to see output capacitors I can advise you to look at tpa3123 datasheet, page 1 simple schematic show 470uF output capacitors.

http://www.ti.com/lit/ds/symlink/tpa3123d2.pdf

[Folsom]

They're filter capacitors on the output, not actually output capacitors. They're part of the filter to remove as much of the RF left overs from the switching (amplification) that comes along with the signal straight out of the chip. The 3110 doesn't have them and it works, but it's low power and doesn't do well next to some other electronics (or rather the speaker wires and speakers may not).

[dboy]

What makes you think any of the capacitors you removed are output capacitors? If you want to see output capacitors I can advise you to look at tpa3123 datasheet, page 1 simple schematic show 470uF output capacitors.

http://www.ti.com/lit/ds/symlink/tpa3123d2.pdf

Hi Mark - As I understand it the capacitors I removed are decoupling capacitors connected to pins 17,18,19,31 & 32 of the TPA3116D2. The reference to output capacitors was from the app notes for the Music Rail - i.e. caps on the power supply output from the Music Rail. Since the decoupling caps on the TPA board are linked directly to this power supply I interpreted the app notes as advising me to remove them. I may be wrong and am open to any advice, but I suspect you have misinterpreted what I wrote. 

http://www.bybeelabs.com/uploads/J0409_Bypass_Capacitors.pdf

[dboy]

OK - it was foolish of me to post negative listening impressions before I was confident the sound had settled down properly. Listening for a few more hours this evening I feel the sound has evolved significantly and I am much less concerned that removing those caps may have been a mistake.

Musicality has improved and dynamics are now very impressive (the deep bass has arrived in a big way). The sound still has a little less romance about it than my previous setup, but I believe I am hearing deeper into the recording, eg acoustics of recording spaces. With one CD (yes, CD!) that I know very well I suddenly became aware (or thought I did) of a change of vocal microphone between tracks. Never noticed that sort of thing before.

Feeling more confident now that I am on the right track...   

[OzarkTom]

My buddy Rex received his cryo'd power cord and all he can say is "wow, it is icing on the cake." Rex has two systems, the one he is using the Astron-Itube with this power cord is a NAD 3020B running the Pioneer SL22's. Rex agrees, the power cord makes a significant improvement, bigger soundstage, smoother, and more musical.

[Markvdv]

Hint: The Music Rail, like the TPA3116, needs decoupling capacitor soldered directly to pins to function properly.

[dboy]

Hint: The Music Rail, like the TPA3116, needs decoupling capacitor soldered directly to pins to function properly.

Are you referring to the following section from the app note?

"At least 10uF should be strapped from input to ground. Better to use 33uF or higher and place it close to the module until stability is confirmed. You can use film or electrolytic, but if electrolytic is used, you should use 33uF minimum. If the module is within a few inches of the stock cap bank this part may be omitted. "

If this is what you mean then the 10,000uF cap is about 1-2 inches from the MR - on the underside of the perf-board you can see in the picture I posted. Judging by the current performance I believe the Music Rail is functioning properly.

If you think it will really make a difference then I'll perhaps try a smaller cap on the pins when I am sure things are fully burned in. There is no point now as the sound is developing constantly.

[dboy]

[randytsuch]

I like your box.   Nice wood. 

[dboy]

Cheers! As you can probably tell, my carpentry skills are no more advanced than my electronics skills. But I try to reap the benefit of choosing quality materials to work with. 

[acdiy]

hi
how much these mod would cost ?
thank you
regards

Coilcraft SER2915L 10uh inductors
TDK .22uf 250v X7R caps (bootstraps)
330uf 25v Panasonic hybrid electrolytics (PS caps)
Wima FKP, MKP, and MKS for .68uf,10nf, and 1nf output filter caps
36db gain resistor setting
Dayton MKP 2.2uf input caps
Jantzen Superior 2.2uf input caps

Amp sounds *excellent*. Clean and clear, no grain or edge. *Pure*.Best build yet. All done now w/ the TPA. Ready to move on.

I do want to get one of those power supplies that I think someone keeps mentioning. What are they again?

[dboy]

Hi acdiy - you are probably best to do your own research on parts costs as a significant portion of the cost will probably be postage and that will depend where you live and where you buy from. You might or might not be able to get all those parts from the same place. Somebody else might chime in but basically, trust your friend google.

[Markvdv]

Outputcapacitors like inputcapacitors I thought were always coupling capacitors, not bypasses/decoupling to ground. Nevermind. Bybee means decoupling caps so it seems, maybe because they are similar to MR and like just adding a random mkp bypass to electrolytic capacitor (especially ones that already have ceramic bypasses) often deteriorates impedance and noise filtering (like outputfilter there can be huge peaking). All decoupling caps removed do seem to be (just) outside dangerzone for MR, maybe 0.1uF was a bit close but I guess if it really was, Bybee would have mentioned 0.05uF or 0.01uF so probably 0.1uF is fine, but you can use smaller close to 0.1uF like 0.082uF. The position of the original probably 0.1uF and probably 1nF on that green board already made them kind of bad. I can't look under that black heatsink but maybe there are decoupling caps closer to chip, they would easily fit under heatsink, and they really should be closer to chip to work properly.

If there are no decoupling ceramics under the heatsink you can remove heatsink and add two 0.082uF chippin to ground easily (1 left, 1 right). If MR doesn't get hot you maybe could attach it to same heatsink, then MR output pins are a lot closer to the pcbtraces going to original positions where you removed the ceramics, you could use tracepads leading to chip there, 1 for left one for right. Next I would cut power pcb-trace other side of pcb twice so there wouldn't be power crossing inputsignal just above, that creates noise too. (do look for mutepower in the corner of that trace, you probably want muting to keep working so cut after that via and other side past inputs above like in other corner too)  You add electrolytics outside MR dangerzone to improve bass, you will probably notice that with 0.082uF directly on chippins and MR mounted close, those electrolytics don't influence highfrequencies anymore like befor. (Mounted like this original powerinput on pcb is not used, electrolytics get power from traces fed by MR, so powertraces are on chipside of pcb now, meaning no electrolytics for bass mounted means no mutecircuitry functioning if pcbtrace on bottomside is cut twice!!!) My 2 cents.

I thought Sagami inductors used on green are rather good. But somewhere I think somebody said they are not made for this purpose. So there probably is a number on them identifying them, you can look if they are not suitable then on Sagami website, there certainly are usable Sagami double inductors there that look similar, but maybe I didn't look at resonance frequency or missed something else...

[Speedskater]

If Jay Leno still had a late night TV show, he would ask 'what could possibly go wrong?'

[lacro]

Received this yesterday from Sure (12 days to eastern US). It's a 15W version of the TPA3110D2. However, when powered with my Astron PS at 13.8V, and low distortion, it's probably only 5-8W (see specs in link below). I had it up and running 5 minutes after it hit the door. Sure uses good packaging.

 It seems to have plenty of power to drive low efficiency speakers. The sound stage is huge, there is noticeably more information/detail in the music (over YJ Blue), plenty of bass, and very good separation of instruments, but it's slightly on the bright side. I don't hear any hum. hiss, or ringing yet. I think this board probably sounds better than the stock YJ blue/black board. I paired it with my 6N3 3-Tube preamp and it sounds even better. I requested a schematic from Sure, but don't know if they will comply.

I need to do a lot more listening/comparing with the my modded YJ blue board. The board is loaded with SMD's so probably not much in modding possibilities, at least not for me. The build quality is typical of Sure, everything is straight, true and the board is super clean. Oh...no turn on/off pop! This just might be a good choice for those not wanting to do mods.
 
Here's some close-ups of the board, maybe you guys can critique the layout, good/bad?





http://store.sure-electronics.com/audio/audio-amplifier-board/low-power-100w/aa-ab32996

[dboy]

...My 2 cents...

Thanks very much for those excellent suggestions to squeeze a bit more performance from this setup. And sorry for my slow response - I have had an old friend visiting this weekend and have not been online for a couple of days. I don't think my soldering skills are up to everything you mention, but I really appreciate you taking the time to consider and write all that.

Also, after getting the feeling here last week that my cap removal scheme was seen as foolhardy I emailed Scott Frankland at Bybee Labs to enquire further. Scott got back to me superfast and was very helpful. I queried whether it really was considered best to remove all decoupling caps in the circuit powered by the MR. Scott acknowledged that this was the advice given in the app note, but continued that "we have lately found that a good quality cap across the output does improve performance so long as the stability requirement is met." The stability requirement was that the total capacitance across the output of the MR is between 33uF and 470uF. Below 33uF there may be a parasitic oscillation that is not harmful but is degrading to performance.

Now the performance of my amp with MR/no decoupling caps has settled down pretty much - and is excellent. It has come out of the very analytic stage that it went through for a while and is now warm, clear, detailed, just beautiful really. However, in view of what Scott said I will soon try implementing a different board with electrolytics reduced in value and SMD caps still in place.

[dboy]

P.S. While my friend was here I replaced the TA2020 board in an amp I made for him a few years ago with an Audiobah board. I replaced the electrolytic caps on the Audiobah board for 2xElna Cerafine 220uF 25v (which fit perfectly) and although it was not the original plan, added a 4,700uF, 40v Mlytic cap near the power input. These improvements made for a nice little amp - the Cerafines brought improvements in clarity and smoothness, while the Mlytic gave it a deeper, more rounded bass sound. I took a couple of pics since randytsuch was so kind before about my wooden boxes.