[rhing]

Just take out your scope and function generator and measure the TDA3116 output at various points, look at the signal shape.

I trust you do have a scope, bro 

Hi ClefChef, You've built amps using the Yuan Jing and Audiobah amps. can you please share photos of your scope image demonstrating the parasitic losses occurring with the Yuan Jing amp vs. the Audiobah? I'm always interested in learning more about how these amps can be improved. Thanks.

[Markvdv]

The brake is off now, but stupid as I am I did 3 things same time, I replaced the 560nF's, the jamicons for hybrid za and I lowered gain to 20dB. The pot is 10k panasonic, 30k inputimpedance for 26dB gain might be too little, 60k inputimpedance with the 10k maybe better, I am going to get the 560nF's back on pcb and listen, grrr I should have done 1 change at a time.

[Markvdv]

Filterless the output of tpa31xx is a square wave looking at it on a scope.

Size of audiobah, Brasilian ebay seller, advance controls green board is 65 x 69mm, red YJ with volumecontrol is 81 x 69mm, blueblack YJ is 103*58mm, hiampmini is 50 x 50mm

[dboy]

Size of audiobah, Brasilian ebay seller, advance controls green board is 65 x 69mm, red YJ with volumecontrol is 81 x 69mm, blueblack YJ is 103*58mm, hiampmini is 50 x 50mm

That's a relief - I was worried I was going to have to get my eyesight re-calibrated! 

[Markvdv]

Sound wasn't held back by the yellow 560nF's.

[Markvdv]

  Bad bad bad, terribly bad, if you happen to have these on your 3116 pcb too, replacing them might result in a very very big surprise, especially if you don't have another 3116 amp around

[ebag4]

Bad bad bad, terribly bad, if you happen to have these on your 3116 pcb too, replacing them might result in a very very big surprise, especially if you don't have another 3116 amp around

What did you replace them with?  What did you hear with the change?  This appears to be the stock cap on a blue/black board, at least it is what is in my stock blue/black board.

Thanks,
Ed

[Markvdv]

I replaced them with Panasonic ZA hybrids, but any other I could have put there would have made approximately the same difference. All the Jamicons, both, did if anything was suck up energy it seems. Black holes is what they are. The Jamicon TL datasheet is quite good, but I fear these are part of the plague that existed, or I don't know does 944C0 production code indicate 2009??? maybe 5 year old unused caps always behave like this??? Maybe they aren't what the sleeve says they are, but the amp didn't sound like 3110 or 3116 amps I have, far far worse.

Now I can start comparing, but too pissed I waited some days before I changed parts thinking caps would settle I will listen seriously tomorrow to blueblack compared to hiampmini, now I 'm just glad I hear normal musicproduction, blueblack is playing 24 hours a day like befor only switched to hiampmini and 3110 for 10 minutes to have a quick reference and confirmation my blueblack was awfull.

[Markvdv]

OK.
Soundstage hiampminis is wider, every sound/instruments locked better in one place then with single near stock blueblack. Bass sometimes in first moment a little stronger with blueblack but diminishing a bit, sometimes blueblack bass sound a little fuller warmer, hiampmini's bass overall has more authority. Full orchestra in violin concerto is comfortable on hiampminis, but high frequency peaks get exaggerated a bit. Edges sound smoother though than on blueblack. Little sounds from audience, people walking to their seats etc are far more audible on hiampminis, so is acoustic information about recording location, resonance. Sound sometimes seems to originate outside left of left speaker. Clarinet sounds more airy, thinner on hiamp, violin sounds more lyrical, but on solo sometimes again exaggerated in highs. Hiampmini appears to be quiter when listening to music. (Volume was set on 200hz test tone, pot on 12 for blueblack, a little under for hiampminis), when switching from violin concerto to PinkFloyd dynamic compression became a little obvious, PF sounds much louder. Here for hiampmini things did not change, everything locked in place all the time, the blueboard struggled to keep things separate, and frankly I didn't expect it to be so loud this PF recording, I would have chosen another volumelevel if I had started with PF. Roger Waters S's sharper on blueblack like his mouth noises, bass a little louder on blueblack maybe??, but blueblack has trouble, cymbals are overpowered by other instruments, less in volume, metal sound barely noticable at this level (earlier I found the cymbals quite good on blueblack) hiampminis has the cymbals fixed in place like everything else. When I was fed up with listening:) I put in Pakito Living on Video. Level here between 85-90dB, in 90 but never 91dB on android app phone 2 meters from speakers, very dynamic , not really, bass and highs both louder on hiampminis. For a disco party the hiampminis are better, but highs here from blueblack i like a little better actually.

BTL blueblack has 330uF panasonic ZA hybrid per channel, no caps on dc-input
PBTL hiampminis have 2 220uF elna silmicII per channel and 4 1000uF nichicons HZ/PW per channel on dc input
every amp board was powered by 1 indeed 12v 6A, that means 2 times 12v 6A for hiampminis. yes mascot psu sounds better, a bit calmer on hiampminis highs, a bit more powerfull on bass, but switching takes too long.

[Markvdv]

I expected the blueblack to be more detailed, and who knows with the same components it might be. I keep reading about superior decoupling of ceramics vs through hole film, which makes sense, except designer of pcb must use the strong point of the ceramic then, tiny size. I feel they don't, maybe that is reason in reality even for powersupply decoupling through hole can be better. Today I took a closer look. Key to succesfull decoupling is the 0.1uF capacitor on these boards. Ceramics have a higher esr then film, but due to the longer legs through hole film caps have higher inductance, and being bigger they usually need more signalpath lenght on pcb too.

Now on hiampmini board I measured millimeters to the decoupling cap, it is all real tiny, but hiampmini distance to capacitor solderpoint is 7 to 8 mm, blueblack 11 to 12 mm. Here pcb signalpath length adds more inductance to the ceramic on blueblack, and also more resistance, and esr already is higher for ceramic, might even be factor 10 higher then PP cap.
Now the ground side of the capacitor, for ceramic on blueboard there is a pcb track back to chip, were right on edge of chip signal path goes through a via to the other side of the board for contact to groundplane. Does this sound familiar to through hole or not??????? Pcbtrack is 4mm back to via. Now hiampmini, I think topside board has a groundplane under cap topside and bottom pcb side, so distance here could be zero or the same as the via part of blueboard, again pcbtracks add inductance to ceramic on blueblack, not on hiampmini through hole cap it seems.

Normally filmcap of 0.1uF will perform better then ceramic for esr controlled part of frequency range, here that might be even further, might even be for entire frequency spectrum?? Normally the 0.1uF film outperforms the ceramic till roughly 5Mhz I believe, after that the ceramic profits in theory from lower inductance. But here more inductance is added to the ceramic in form of pcb traces, it could very well be more then the through hole part ever had totally.

[dboy]

If I understand the above correctly then Audiobah looks to have done things reasonably well, with 2 SMD caps quite close to the chip. I have not measured track length, but I do not think it could be greatly reduced without a smaller heatsink.

[Markvdv]

there is a direct connection to groundplane, and not a 4mm trace to a via to a groundplane, so that is better then blueblackboard, the side of ceramic to ground. the other much cheaper greenboard (ebayprices) has a much shorter distance from chip to decoupling ceramics. That distance does kill ESR I think and adds inductance. Heatsink audiobah hides distance, designer thought chiptemperature was more troublesome then decoupling (for me tpa chip hardly ever gets warm without any heatsink even)

[Markvdv]

This very roughly is the difference


(just saw that powertrace again, crossing beneath all the weakest inputsignals above it on the other side of pcb, poorly chosen )

[dboy]

Interesting! So that cheaper green board looks worth a try?

I guess all design is about tradeoffs. It does seem though that the best TPA3116D2 board has yet to be made. Maybe there is a niche in the market out there for one of you knowledgeable electronics guys: best board design with well spec'ed and reliable quality components... not trying to compete with the cheapest?

I still like the sound of my Audiobah board though... for now.   

[Markvdv]

Don't think the blueblack distance is huge:) This photo is much bigger then board is!



if you look at pic you see pcb trace for top 104 decoupling ceramic, going up 7 mm and the again ~6 mm to the left to 104 ceramic after the 10nF and right below the 0 in 10nF you see the via going to ground for that side of chip and there is actually another via under the 4 in 104 for the 104 there

[dboy]

I would not have spotted all that. Thanks - the education is useful & appreciated!

That chip looks like it has seen better days though. What happened and does it still work?

[Markvdv]

some white glue residue from heatsink, it is a 3116, with original ecaps I had my doubts

glue remains flexible, is nonconductive, I think it is also used in bathrooms and kitchens

[dboy]

Ahhh... viewing on my screen I thought the top had been split apart or something similar.

[Markvdv]

  100nF ceramics went for a walk and ended up their sides



So this is what happened, the 104 ceramics are no longer in their original place, yellow dot, but now where the yellow stripe is. Far easier then expected. If I had known it would be this easy I would have placed them 1 mm closer to chip but this was easier. (current distance top ceramic between tpachip contacts and ceramic contact 2 mm, lower ceramic between 1.5mm and 2 mm) Value of the ceramics was 93nF and 94nF, so they were close to 100. 0$ mod that removes the hard/harsh edge from blueblack and adds a load of detail information, clarity.

[Markvdv]

I cut the Jamicon original ecaps open. Never did that befor so I also cut a 1 year old Nichicon 1000uF open, that is a lot messier:) There is actually much fluid in the Nichicon and cutting the aluminium also has a little air sizzle, either air going in or out, air presure inside Nichicon was different then outside. Jamicon was complete silent. The roll you lift out of the can is soaking wet with Nichicon and drips , in the empty can you see a lot more fluid. The Jamicon roll was a little wet, no dripping and no fluid is in the empty can.