I've read through a lot (but not all!) of the nearly 10K posts in this thread, and over the last few months have purchased a assortment of 3116 boards which all work, but some are better than others.
What do you all consider to be the best off the shelf dual channel / single chip board at the moment? I don't mind doing a few mods, but if I could get something that ticks most of the boxes I would be happy.
What do you all consider to be the best off the shelf dual channel / single chip board at the moment? I don't mind doing a few mods, but if I could get something that ticks most of the boxes I would be happy.
The 3118 lookalike ampboard does distort a lot, but can't compare to standard Sanwu, Sanwu with different inductors I did compare, that one listens a lot easierAgain, high current listening situation.
Haven't noticed. What signal at what PVCC? Maybe it's the inductors again.
For my application (fed via Pi & IQAudio DAC, and separate BT board, via a pre-amp for volume control), it's the Audiobah clone:
Assembled TPA3116 class D amplifier board 50W+50W (green pcb)
20.8V (~20.5V), music signal probably from 2V output
So GVDD from PVCC, not AVCC as diagram might suggest.
Is main switching residu frequency for both standard and split AVCC equal? I have seen half switching frequency as main residu, but forgot with what "3116/8" ampboard.
What is the significance of the increasing difference horizontally between positive and negative part sinewave ?
edit (at mouser tpa3128 is cheaper than tpa3118 🙂 )
Last edited:
At 24v, the nominal output for a TPA3116 @ 4 ohms is ~70Wrms. A single chip in PTBL mode can do that. Or just use one channel on a btl board as the "50W + 50W" rating is only the chip's thermal power dissipation limit so 70w on 1 channel only is acceptable.
A 36v supply exceeds TPA31xx chips' maximum Vcc.
Last edited:
Yes.
PVCC down to 6V0: 1.4Vrms Output -> 0.5Wrms into 4R
4V3: 1.9Vrms Output -> 0.9Wrms into 4R
So an increase of ~2.65dB.
Left to right, horizontal, negative wave is smaller than positive wave. Amplitude is vertical, up and down, isn't it?
For all voltages except 4.3V its 76/80us (or Ratio 19/20 or 5% mismatch) or - also for stock. So this is generator induced - my guess. For 4.3V it's 74/82us (or Ratio 37/41 or 10% mismatch. So yes, 4.3V might be a bit low for operation. (TPA3128D2 states a minimum of 4.5V.) But this wasn't the main question.
Can't confirm this on the TPA3118, IO390 tested this with the TPA3132. Maybe the LDO output was oscillating. As i mentioned earlier, incresing/decrasing voltage in the lower voltage range (4.3-5.0V) sometimes triggered some resyncing of the switching with hearable noise - like "Schhhhht" When settled to stable switching, the noise is gone.(Beside the low static noise from gain increase)
Can't confirm this on the TPA3118, IO390 tested this with the TPA3132. Maybe the LDO output was oscillating. As i mentioned earlier, incresing/decrasing voltage in the lower voltage range (4.3-5.0V) sometimes triggered some resyncing of the switching with hearable noise - like "Schhhhht" When settled to stable switching, the noise is gone.(Beside the low static noise from gain increase)
Last edited:
Well back to the board, why can it deliver over twice the voltage into a heavy load, music. Obviously cooling is a big problem for continuous sinewave, but never is relevant here playing music, maybe when in unvented small enclosure only. So why can it play music so much louder than all others I own. And can the funny looking chip clone also play music much louder than for example your minitpa3132 or any other 3116 you own when you parallel multiple speakers to have a real heavy load ?
Sanwu here compared to any other pbtl, gmarsh,dugs,ac/audiobah/sure, they all switch off/trip protection with the difficult speakers at 3-4-5Vac over speaker, playing music, not all at same level, before the Sanwu I thought those level differences were relevant, but small Sanwu doesn't trip protection putting out 11 Vac playing same sample music. Why does it keep playing when "better" boards start stuttering at less than half the outputvoltage.
I actually don'T have any other boards here for testing. I may build up a 3132 for comparison.
From my memory, the wasn't any problem running the board into clipping except thermal shutdown.
The TPA3116+BT i have also hit the rails easily with 4R speaker or dummy-load.
From my memory, the wasn't any problem running the board into clipping except thermal shutdown.
The TPA3116+BT i have also hit the rails easily with 4R speaker or dummy-load.
The speakers here drop to 1 ohm within audiorange. On 4 ohm (zobelled) speakers I can't reach levels for protection tripping comfortably with any ampboard, but with the electrostatic hybrid speakers levels for that with all ampboards, except Sanwu pbtl, are within reach.
Good question.
I'd suspect the DCR of the inductors and/or the inductance drop as f(I) then. If the inductors on the Samwu have higher DCR then the rest and do drop less on high current, it won't trip.
Doesn't the Samwu got the smallest inductors?
(Btw. The 3132 is running filterless, total different story)
I'd suspect the DCR of the inductors and/or the inductance drop as f(I) then. If the inductors on the Samwu have higher DCR then the rest and do drop less on high current, it won't trip.
Doesn't the Samwu got the smallest inductors?
(Btw. The 3132 is running filterless, total different story)
Inductors make no difference I think, replaced those chinese with two kinds of Wurth's, including same ones as on my hiampmini pbtl boards, there protection trips at ~5Vac too, with same Wurth's again no tripping at 11Vac.
(Not related to poweroutput) Btw something else: cutting the narrow analog ground trace to power ground entry bottom side had slight positive effect on hiss (gets darker), no negative effect noticed on Sanwu pbtl, but compared only with preamp source. With hiampmini's analog inputground lead needed to be at powerground entry point to have least amount of/ "cleanest" hiss somehow, but only compared from cdoutput directly to pot and hiampmini, didn't try from preamp, might differ don't know. As with other boards treble gets a tiny little "easier" sounding with preinductor snubbers, especially playing louder. All directly compared listening and switching between boards, not from memory, two speakers side by side, muting boards, on off. Still can be psychological LOL but don't think so 🙂
You can compare stereo BTL burst/shortterm outputvoltage with two parallel 4 ohm loads, or three parallel loads? Sanwu bluetooth 3116 board from your noisefix thread has same groundplane splits as Sanwu pbtl boards?
Sanwu pbtl guides main outputfilter grounds through separate groundplane to powerground entry, tpachip groundplane has different path to powerground entry, bulk and decoupling capacitor's groundplane are "tpachip" groundplane, analog inputground is seperated (mostly) and guided around the entire ampboard towards powerentry ground, narrow "inductive" ? path at that, probably unnecessarily connected to powerground entry, possibly better only through analog ground pin tpachip ? No general topside groundplane, small ground "trace" around capacitors, bottom groundplanes funny like a net, not massive. Like I think you said before, seems like they collect the copper they saved/wasted 🙂
I thought it was my imagination but mine runs cooler when blasting away. I have not put a Tc on it to document yet but is on to do list.
-
Very interesting, I was running my nobsound amp, (two 3116 chips) at new years, at party volume into 8ohm and it was stone cold, while it actually gets warm at home connected to 2ohm speakers, it actually gets warm without any input, I assumed that it was the load, but that it might become more "efficient" under pressure is kinda funny.