Believe me, I want to (if I am lucky, a friend will loan me one to audition soon). I have heard it is the BEST sounding amp at any price (although there are folks working their tweaks on that amp even). But the PCB and BOM will set you back ~$200 I think. That doesn't include the soldering which is really dedicated time. Different than time required to do other DIY things. I know as I built the first TPA3118 board on this forum from scratch using deadbug P-to-P wiring. I would not have done so if I could buy a $15 or even $50 pre-made board at the time.
Now, if the Chinese would just make a clone mod86 then we would really be in for a treat.
Until then, I tried out a Chinese LM3886 and OPA2604 and really like it, that I bought a second unit to form the basis for my new active 4 channel system. They basically sound as good or better than a 3116 with $40 worth of mods out of the box. They make some serious heat though.
one cool thing I found out that may be keeping some of the class D folks from trying out LM3886 amps is that you don't need big iron fancy toroidal trafo power supplies for dual rail PS. Just use two 19v smps laptop bricks in series with a center tap. Works like a charm and is cheap and compact. I am using $6 19v 4.6a bricks. Make sure you get the 2-prong variety to do this.
Last edited:
Change Rubycon ZL 680uf with elna cerafine 470uf because sound too bright.
Change wima mkp 10 4.7uf with this caps k73-16 2.2uf (only have this value) sound better more detail, smooth highs , but bass little weak, and now have buzz on turn off. May be caps leg too long.
Change wima mkp 10 4.7uf with this caps k73-16 2.2uf (only have this value) sound better more detail, smooth highs , but bass little weak, and now have buzz on turn off. May be caps leg too long.
Attachments
Believe me, I want to (if I am lucky, a friend will loan me one to audition soon). I have heard it is the BEST sounding amp at any price (although there are folks working their tweaks on that amp even). But the PCB and BOM will set you back ~$200 I think. That doesn't include the soldering which is really dedicated time. Different than time required to do other DIY things. I know as I built the first TPA3118 board on this forum from scratch using deadbug P-to-P wiring. I would not have done so if I could buy a $15 or even $50 pre-made board at the time.
Now, if the Chinese would just make a clone mod86 then we would really be in for a treat.
Until then, I tried out a Chinese LM3886 and OPA2604 and really like it, that I bought a second unit to form the basis for my new active 4 channel system. They basically sound as good or better than a 3116 with $40 worth of mods out of the box. They make some serious heat though.
one cool thing I found out that may be keeping some of the class D folks from trying out LM3886 amps is that you don't need big iron fancy toroidal trafo power supplies for dual rail PS. Just use two 19v smps laptop bricks in series with a center tap. Works like a charm and is cheap and compact. I am using $6 19v 4.6a bricks. Make sure you get the 2-prong variety to do this.
I don't get it, please forgive my ignorance and 'splain it to me Lucy......( the power supply thing....)
I believe a 3886 with +/-19V (double) PSU is able to put out 17 watts, so that is almost comparable with 3116
Explain to me how you get 17w because it is obviously more powerful than my 3116 when playing. The input power supply says +/-24v AC. I am powering. +/-19v DC. The amp can swing probably almost rail to rail or say +/-17v or 34v peak to peak. Are you applying class AB efficiency or something?
If you look at most amp projects that use Lm3886 a dual rail source of power is usually provided by a toroidal dual rail transformer (expensive) or a dual voltage SMPS (also expensive). I wanted to keep things cheap and simple like how I am used to playing with $6 19v SMPS laptop bricks.
Worth exploring the brick power due to low cost. Real shame toroids so expensive, the price of the toroid bi-polar supply I've built would buy over 10 of those bricks!! 😱
Yep! It's a tough thing to justify a $60 toroid trafo or even open frame dual rail SMPS because they are expensive and not plug and play. That is, you still need to get the 120vac through a jack, fuse, switch before going to amp. Since lethal voltages are now exposed you need a case to enclose it all - can't go the lazy open air class D amp route we have all been used to enjoying. But with the UL/CE/CSA approved and safe SMPS laptop brick, all I have to do is wire it straight to amp (no fuse, receptacle, switch, etc) and plug it in! Very very good for keeping it simple for someone who doesn't need a fancy $100 aluminum box to keep an amp inside just to use a toroid.
The last time I used an open frame +/-53v SMPS in an open (non boxed) setup for testing I got all sorts of grief from Irrebeo for having a dangerous amp around.
Last edited:
I've seen a few schematics to build a cheap smps. Thought about it as well.
I love my ac filter box and bi-polar (built a few years back). Next job is to build 2 new bi-polar toroid supplies for twin 3116 amps. Might encase with amp boards, might be separate supply cases...... Still deciding!
Someone has suggested to me to use an old PC tower case with a modified front panel to encase everything together!
I love my ac filter box and bi-polar (built a few years back). Next job is to build 2 new bi-polar toroid supplies for twin 3116 amps. Might encase with amp boards, might be separate supply cases...... Still deciding!
Someone has suggested to me to use an old PC tower case with a modified front panel to encase everything together!
Last edited:
Actually if you run one TPA3116 from a single 19v brick it has roughly the same voltage swing as the Lm3886 run from two 19v bricks with a virtual ground.
The TPA3116 is a bridged amplifier and as a result doubles the voltage swing that the single voltage rail can provide. So in effect it's the same thing.
Both amplifiers will have linearity issues the closer to drive them towards the limitations of the power rails, but the effective voltage swing of both amplifiers will be about the same.
+-19V gives you an effective voltage swing of 36 volts lose 2 volts per rail for losses, so you're down to 32volts. This gives you an RMS voltage of 11.35V. An 8 ohm watts figure would = 11.35^2/8, which = 16.1 watts.
This calculation is identical for the TPA3116. And you can see this on the datasheet.
Look at the graph of distortion vs power for the 8 ohm load with the amplifier being powered by a 24V rail.
This = an effective voltage swing of 48V as it's a bridged amplifier. Lose 2V per rail for internal losses and that takes this down to 44V. The RMS voltage of this = 15.6V. 15.6^2/8 = 30 watts, which is exactly what the datasheet shows. Distortion shoots up beyond 30 watts and isn't an area in which you want to be listening to music. Have you ever heard what a TPA3116/8 sounds like when you hit its voltage limits? UGLY.
There are two caveats to this.
1) If your LM3886 amplifier uses two ICs per channel and is operating bridged. If this is so then yes it will have significantly more output voltage swing and providing it can provide the current, produce far more watts into a given load.
2) The two 19V bricks with the virtual ground will be able to supply twice the amount of current vs the single brick on the TPA3116. This isn't of particular interest unless you are driving things close to their limits, but it is worth pointing out.
EDIT - If the LM3886 board has been designed to function from an AC supply voltage then you will be losing even more voltage swing by connecting a DC supply to the board. The board is supposed to be used with a mains transformer that obviously outputs AC. This requires rectification in the form of a bridge rectifier (or an arrangement of 4 diodes) and then large smoothing capacitors to remove the voltage ripple. If you power the amplifier from DC then the diodes will still conduct and pass voltage to the power rails but you will lose some potential due to the diodes inherent voltage drop. To really power the board you need to locate where the DC voltage rails are (typically with the two big smoothing caps, one of the terminals on each cap will connect to ground, the other free terminals should be a +ve and a -ve, these will be the DC power rails) and supply the DC power directly to them.
Last edited:
Suggest carefull research on buying an open frame type SMPS.
I have TWO dead Yuang Jing +26 v, 0 ,-26V subsidiary +15v, 0, -15v contraptions.
One was doa (impressive), the other failing within an hour's use.
Yuan Jing simply does not answer emails, even more impressive.
IMO buy a Brand name Laptop/Monitor brick .. thrift shops are full of them from 50 Cents to 1$.
Or if wanting an open type SMPS: a Hypex
Or adapt an old (or New) Computer SMPS If 12v is adequate to needs
Or build an Old school linear supply
Neither are inexpensive.
But as goes the adage; Buy Cheap.. Buy Twice and if trully dim; Repeatedly.
Caveat Emptor kids.
I have TWO dead Yuang Jing +26 v, 0 ,-26V subsidiary +15v, 0, -15v contraptions.
One was doa (impressive), the other failing within an hour's use.
Yuan Jing simply does not answer emails, even more impressive.
IMO buy a Brand name Laptop/Monitor brick .. thrift shops are full of them from 50 Cents to 1$.
Or if wanting an open type SMPS: a Hypex
Or adapt an old (or New) Computer SMPS If 12v is adequate to needs
Or build an Old school linear supply
Neither are inexpensive.
But as goes the adage; Buy Cheap.. Buy Twice and if trully dim; Repeatedly.
Caveat Emptor kids.
Last edited:
Thanks for the detailed explanation. What voltage SMPS brick should I be using if I want the max power out of LM3886? +/-32v DC?
What's the general recommendation for decoupling on the TPA3116? Datasheet specifies 220uf/100nf/1nf per side. How about package size, that seems to matter quite a bit with regards to inductance? Would the 100nf/1nf even add anything over a single cap if they're both the same size?
not harsh, easy and relaxing to listen, listening music for long periods not cause ear fatigue
1mm pcbtrace adds more than difference in packagesize, gmarsh Wiener uses 330pF, 330nF, 10uF, 82uF, 470uF per side, personally I think it is too little 🙂 Original first tpa3116 document states 2200uF as electrolytics needed, google it 🙂
1mm of what width, and what thickness? What do you want to say? Only capacitance value without the rest of information about ESL/ESR makes no sense at all. (Beside the missing PCB specs)
.. 😵
.. 😵