6 Channel TPA3116 + DSP

The main reason here for that is pricing. The AV receivers are too expensive.

Maybe there are some less expensive that I dont know of. Can you give a list of some reasonable prices AV receivers.




@diditaled
You take a Dolby 5.1 pre amp and put it in front of your 6-channel amp. Was that your question? But why build a 5.1 Dolby amp? There are AVR receiver out there...

@grahamgraham The diode only in the + (or, with a dual supply + and -) supply line, no diode with ground.
Yes, this diode thing is used with many D-amps. I think it is more a kind of idiot protection you need if you sell worldwide to people that do not have any basic education, like you expect them in most western country´s. I have worked with refugees and now have a different view at our (good) school system.
With D-amps it might even reduce the need to synchronize them.

@NMOS
look for PM
 
Member
Joined 2004
Paid Member
Here is a mostly completed project with that board. It's a computer monitor stand, so it's intended for very near-field listening, with small (2-1/2") subs and PR's on either side. That amp has enough power to fill up a room--it's perfect with a stereo 3-way DSP board.
 

Attachments

  • Bottom_3.jpg
    Bottom_3.jpg
    81.2 KB · Views: 552
  • Front.jpg
    Front.jpg
    56.4 KB · Views: 548
  • corner.jpg
    corner.jpg
    65.9 KB · Views: 475
Here is a mostly completed project with that board. It's a computer monitor stand, so it's intended for very near-field listening, with small (2-1/2") subs and PR's on either side. That amp has enough power to fill up a room--it's perfect with a stereo 3-way DSP board.

Nice build! I would like to make something almost exactly like that with the birch ply I will be using for the speakers.

Do you have any issues with noise? What do you do about grounding and safety with regards to the power supply?

I would like to have a go at that 3 way board but I am an SMD novice.
 
As the parts for this build start to come in from the East et al, I am starting to think about how to go about things.


In previous builds where I have used balanced inputs to the chip via a 1uf cap. I then soldered flying leads to the cap on the board. This was a pain in the neck as they aren't easily soldered/resoldered.

So, my plan is solder cables to the +&- of the board inputs and run these to a prototype board daughter board where I can use through-hole caps and connect/disconnect with ease.


Two questions regarding this:


1. I can't see any issues with doing this, can anybody else?


2. What caps do I need to use for optimal performance? From what I have read MKP film seems the best choice?

So many here, but which?
Polypropylene Film Capacitors | RS Components

Do they all pretty much perform the same?
 
Member
Joined 2004
Paid Member
Regarding noise:

1) I'm still working on grounding issues. I got a surprise by using those heat pipes to bring the heat out to the back panel and found out the heatsinks are connected to the circuit ground. I had lots of buzzing until I connected the back panel to the power cord safety ground, but now there is lots of hum whenever the amp is connected to a computer. I can make that go away by transformer-isolating the input, but I don't like that solution, so I'm still thinking about how I want to finalize the grounds.

2) The amp itself is somewhat noisy ("hiss") because the gain is set too high. I haven't looked at how the gain is configured, but I think the gain-setting resistors are under the heatsink and I don't feel like taking this thing apart right now. I bought another amp to see how the gain is set, but it won't come in for another 3 weeks or so. The gain is so high that a computer can easily overdrive the input, so I want to change the gain. Reducing the gain should lower the noise.

3) I'm not sure what you mean by "+&- of the board inputs". The board input connector only has 6 "+" inputs and one "-" connector on the original version of the board (with two bulk capacitors). The newer version of the board (with one bulk capacitor) uses a longer connector, but it still doesn't have the separate "-" inputs you need for balanced input. But you should be able to remove some of those 1uF electrolytic caps that connect the "-" input to ground, and that would give you some pads to connect to. I don't know whether using balanced inputs would make it easier to avoid shared ground problems, but it might...
 
Neil, perhaps I should have explained a little more explicitly.

Instead of using the single ended inputs provided I will remove the heatsink and fish out the differential/balanced input traces. Then, remove the coupling capacitors on both positive and negative inputs and solder in two flying leads for each channel that I will bring to a daughter board. This is where I will have the coupling caps which can be swapped out without me risking damaging the pads on the board every time I solder.

The balanced signals from the DSP board will not require a ground.

Yeah, I expected this board to have its gain set way too high so I will bring them all down to 20db with some surface mount resistors I have.

This combined with the balanced input should keep noise issues to a minimum.

I'm a little concerned about safety and wooden amp boxes. I thought perhaps a wooden frame with aluminium top and bottom.
 
Member
Joined 2004
Paid Member
I don't think you will need to remove the heatsink. Just remove all of those 1uF electrolytic capacitors and use the capacitor holes to make the connections to the daughter board.

It's usually not wise to comment on safety, but my experience is that once you convert to DC in a properly isolated and grounded power supply, the safety concerns are mostly eliminated. However, it's good to have some aluminum plate, foil or thin sheet metal close to the amp, connected to safety ground, as it there is a *lot* of radiated noise from this amp. I'm using 24V, which results in enough noise and heat that a metal mounting plate made sense.
 
Member
Joined 2004
Paid Member
There have been many, many studies and tests for distortion on different capacitor types, and as a general rule, any film type will have far lower distortion, by at least an order of magnitude, than the TPA3116 amp. Actually, even a dc-biased electrolytic, like the ones used in this amp, will have distortion levels below .001%, as shown in Figure 9 of Part 5 of the article at this link: http://www.audiodevelopers.com/misc/Capacitor_Sound.pdf

1uF is a relatively large value for a film capacitor, so for size reasons you should select a metallized stacked polypropylene. Some people swear by certain brands, but I don't think you will see any significant difference on a distortion meter or hear a difference in a properly conducted listening test. You only need to work with 3.3V peaks, but the lowest voltage you will find for these is 50V or 63V. So anything on these three pages will be fine: Film Capacitors | Capacitors | DigiKey

Of course, those electrolytics that you plan to remove will also work OK. You might want to upgrade to 10uF, to ensure a somewhat lower frequency cutoff. The capacitors are biased at 1.6V in this circuit, so a 10uF 10V cap should be fine. I was always surprised to see Cirrus Logic use electrolytic caps for audio coupling in the reference designs for their flagship products, but there aren't good reasons for not using them...
 
Member
Joined 2004
Paid Member
There are some that are smaller, but if you have the space, those are fine.

Just a note about electrolytics: they aren't recommended for passive crossovers in loudspeakers because of the relatively high ESR. With impedances in the 4 to 8 ohm range, the capacitor needs to have very low ESR, and for that it is best to use a film cap. However, for audio coupling at relatively high impedances, electrolytics aren't bad, in spite of what some may think.
 
Please, do your self a favor and change no part´s at working amps. If you use Bluetooth, no capacitor will be hear able. You only distract your self from real problems, like noise.
All you can do with such mods is make things worse.
These capacitor changes are only audible with the very best source (NO phone!) and even better loudspeakers (NO full range speakers!).

You are totally wrong about the "improvement" your mods will have in the best case. Only if you put your amp in a high end chain, you may (MAYBE) notice a very small difference. Even this small change in sound will only be there, if the original decoupling caps where absolutely the worst anyone could find.
This is what many do not realize, it is just as hard to find a cap that makes an amp sound worse, as it is to find one making it better.
Most have never been able to find any difference. Many of the cheap amps even use caps that could be labeled "Audio Special". If they where not that cheap.

Please, build a noise less combination of DSP, Amp and power supply. Then, when everything is right, build a second, identical one. If this one works just as well, compare the sound. Then modd one as much as you like. Then, compare, again. Most probably, with not seeing the amps, you will not notice any difference.
If more people would do such tests, a lot of stupid mods would be prevented and many blown amps would still work.

Do not mix up experiences some have made with Op-amps with all other audio or electronic parts. These where fake OPA´s, which took parts that had basic functionality, where tested defective from a factory because of requirements not meet and then re-printed with the name of an expensive audio part.
Only because in some country´s we will not name, such operations are not criminal, but a honorable business model, we hear these differences, some times.

PS a better PS or a good quality buffer cap, up to around 2x 4700uF very close to the Chip may be audible with dynamic music. This is something totally different to the decoupling caps that block DC and are needed, as the amps input has a few volts offset.
 
Last edited:
Hi Turbo,

I appreciate your feedback but I think you may misunderstand my intentions on the coupling capacitor swap. It isn't for reasons of fanciness it is simply because I need to access the differentials inputs to the chips and as Neil has pointed out the access point to them is via the through-holes that will be populated by the electrolytic caps.

I only wanted to know what the best possible caps I could use were as they will only cost 30p each and I doubt I will be able to use the electrolytics again (at least I need to bank on a back up if they do not work if there is a problem after desoldering them).

As for my earlier suggestion of upgrading/replacing the POWER capacitor is based on my skepticism of cheap electrolytics from the east. When I was making my last bluetooth speaker I managed to solder a 2200uf cap in the wrong orientation and it exploded. Inside was nothing but dry dry dry. This made me concerned and I thought that I would like this system to be able to handle some grunt I would replace them with Nichicons sourced by a reputable company here in the UK.

I am very much a person who does not believe in the magic of different interconnects, speaker cables and I understand the power of expectation bias.

I don't think I have made any wild claims about the improvements you say I think there will be. I am simply aiming towards a quietly operating amp for now and going the balanced/differential route seems the most sensible given that the DSP boards you recommended have these very outputs by default :)
 
There are some that are smaller, but if you have the space, those are fine.

Just a note about electrolytics: they aren't recommended for passive crossovers in loudspeakers because of the relatively high ESR. With impedances in the 4 to 8 ohm range, the capacitor needs to have very low ESR, and for that it is best to use a film cap. However, for audio coupling at relatively high impedances, electrolytics aren't bad, in spite of what some may think.

Cheers man, appreciate the help!
 
It has arrived, fantastic! I expected it to be small but it is a lot smaller than I thought.

Having a little poke around it I see many of SMD resistors are of the tinier variety but I see some (which I think are the gain resistors) under the heat sink which are a little larger.

Only problem is that the heat sink seems to be bonded to the chips (bolts removed).

I thought, maybe, heating the whole board up on a radiator and seeing if it softens a bit? Thoughts anyone?
 

Attachments

  • 20191107_151713.jpg
    20191107_151713.jpg
    1,011.7 KB · Views: 354