TPA3116D2 Amp


The output impedance on the tube pre-amp is 4K ohms - very, very high. Line Out output impedances are much less, rarely in the 100s of ohms. The input impedance on the TPA3116 @ 26dB gain is 30K ohms. And there must be a decoupling capacitor in series with the board's input through the volume pot on the way to the chip. And the volume pot's nominal resistance (25K?) to ground, in parallel with the chip's input impedance, is lowering the impedance even further. Perhaps the Pre-amp's output can't drive the ~16K equivalent input impedance at higher volumes - line level power levels.

Just seems like a mismatch ...
 
The output impedance on the tube pre-amp is 4K ohms - very, very high. Line Out output impedances are much less, rarely in the 100s of ohms. The input impedance on the TPA3116 @ 26dB gain is 30K ohms. And there must be a decoupling capacitor in series with the board's input through the volume pot on the way to the chip. And the volume pot's nominal resistance (25K?) to ground, in parallel with the chip's input impedance, is lowering the impedance even further. Perhaps the Pre-amp's output can't drive the ~16K equivalent input impedance at higher volumes - line level power levels.

Just seems like a mismatch ...
Mike, thanks for the explanation. The 3116 board has a 50k pot. I understand pot value has to be lower than input impedance? This is an 8 pin pot with switch. If I replace it with 20k or even 10k pot without built in switch and leave the two holes unused, will it cause problem? Aside from the amp being constantly on whenever PSU is plugged.
 
The output impedance on the tube pre-amp is 4K ohms - very, very high. Line Out output impedances are much less, rarely in the 100s of ohms. The input impedance on the TPA3116 @ 26dB gain is 30K ohms. And there must be a decoupling capacitor in series with the board's input through the volume pot on the way to the chip. And the volume pot's nominal resistance (25K?) to ground, in parallel with the chip's input impedance, is lowering the impedance even further. Perhaps the Pre-amp's output can't drive the ~16K equivalent input impedance at higher volumes - line level power levels.

Just seems like a mismatch ...
Mike, thanks for the explanation. The 3116 board has a 50k pot. I understand pot value has to be lower than input impedance? This is an 8 pin pot with switch. If I replace it with 20k or even 10k pot without built in switch and leave the two holes unused, will it cause problem? Aside from the amp being constantly on whenever PSU is plugged.
 
output impedance is a little worse than 4k I believe :D
buffer_6n3.gif
 

I tested the same board few weeks ago and found that:
- gain was 36 dB;
- one channel was master and the other slave;
- synchronization link was present, but the R-C values were wrong, so the slave was not synchronized and was switching at about 100 KHz (instead of 400 KHz).

After reducing gain and making synchronization correct, I discovered that the IC's and the inductors (33uH) get hot even without signal and THD is about 1.5% at 10W output power.

I hope you are more lucky than me.
 
I concur

I just found the strangest thing with my dual chip red board. I ran it with XY 6N3 tube pre amp, with the power amp at full volume, and the tube pre being the volume control. Power amp full volume, and player is at 80% volume, the tube pre at 40% is just nice in my small room, but when I turn it louder, the sound becomes bloomy and dull. I think to myself, 3116 should play louder and cleaner than this. Out of curiosity I connect the player directly to the amp, and it was so different! It plays louder, cleaner, more authority. More detail came into play, at the expense of slightly thinner vocals. It's as if the tube pre (or should I say buffer) is limiting and 'dulling' the sound. Could be impedance matching between the two, could be my system (speaker, player, etc) or maybe you guys have any thoughts on this? But I can say I'm surprised and pretty happy with this result.

Jonathan,
I find the same thing happening with some of my amps. Some sound better when I use my 2 tube Guanzo/Sain Smart preamp with them. And some sound better on their own.
Just experiment and enjoy!
Mark
 
I tested the same board few weeks ago and found that:
- gain was 36 dB;
- one channel was master and the other slave;
- synchronization link was present, but the R-C values were wrong, so the slave was not synchronized and was switching at about 100 KHz (instead of 400 KHz).

After reducing gain and making synchronization correct, I discovered that the IC's and the inductors (33uH) get hot even without signal and THD is about 1.5% at 10W output power.

I hope you are more lucky than me.
Hi fulvio, thanks for your response. I remember you posted your findings, and we exchanged a few posts then. Is the gain 26 or 36dB? I checked the datasheet and the values on my board refers to 26dB as your first post said.
I planned to reduce the gain and remove that C33 to get it synchronized, but I havent. So I guess just leave it as is? Please advise, thanks! ;);)

Post edited for easy reference:
Hi, today I’ve got my first TPA3116 board, the dual chip XH-M190 v.3; it was configured in the master/slave mode, with 26 dB gain for each channel.
Exactly like mine; one chip is set as master and the other as slave; but to get the slave synchronized with the master, a clock signal has to be transmitted to the slave and this is made through R13 and C33.
Unfortunately, the values of these components (10K and 1nF) are wrong, so the slave doesn't receive a valid synchronization signal and works completely out of specification (switching frequency of about 100KHz), with a very high ripple at the output.
As stated by TI, the correct values should be R13= 4.7K and C33= 47pF, but in my board I got synchronization simply removing C33 and leaving R13 at 10K (a very simple operation).

I really don't know if synchronization can improve the audio quality, but for sure one chip of your board is working now in a non-standard mode and the output filter is suffering for this.

Concerning the gain, I should not modify it.
 
Last edited:
Jonathan,
I find the same thing happening with some of my amps. Some sound better when I use my 2 tube Guanzo/Sain Smart preamp with them. And some sound better on their own.
Just experiment and enjoy!
Mark

Hey Mark, yep yep, I guess the fun is in experimenting and trying different combo that works in our listening space. I did however get a very nice sound from the board by doing some simple mods:
- Replacing stock 1000uF/25V caps (is it called DC decoupling cap?) with Panasonic 470uF/35V.
- Replacing stock 3.5mm socket and terminal blocks with better ones. :p
- Installing a snubber circuit, consisted of 390pF ceramic cap and 10 Ohm carbon resistor

This subtly, yet considerably, changes the sound. I'm getting more details and articulation, as described elsewhere, feels like some layers are unveiled from my music. I thought I might lose a lot of bass by going down to 470uF but not really. But is still there, and tighter too. I love it. Thanks to everyone who shared their expertise and experience.

Now on to building some cap bank. ;)

Stock board, let's just call it dual chip red board


Is there any sonic gain from this?


Trying to get them nice and straight




Finished board
 
Last edited:
my guess is that dual chip board and tube preamp are not the best choice anyways.
red dual chip has long traces where it shouldn't have, long distance between decuopling and chip input...bad overall design.
tube preamp can add nothing but distortion, which might sound preferable in some cases, but cannot be good in general.
i'd give another board try before investing in expensive parts for modding the dual chip board.
if you like 'warm' sound maybe TPA311X is not right chip for you.
 
output impedance is a little worse than 4k I believe :D
buffer_6n3.gif

When the output volume is at minimum, the output impedance is 0 ohms. The tube's output sees the 50K ohms of the pot to gnd. No signal out.

Turn up the output volume a little, say where the pot's wiper is 5K above ground. Then the tube's output will see the 50K output pot impedance in parallell with: the remaining 45K of the output pot in series with: the amp's input pot in parallel with the chip's input impedance of 30K @ 26dB. That equates to ~27K

When the output volume is at maxium, the tube's output is going to see the impedance of the 50K output pot in parallel with the 50K input pot on the amp in parallel with the chip's input impedance 30K ohms. That's about 13K ohms.

So as you raise the the output volume, the tube's high impedance output sees an ever increasing load (decreasing impedance) that it can't properly drive. And the decoupling capacitor is acting as a filter given the high output impedance of the tube amp.
 
Last edited:
Mike, thanks for the explanation. The 3116 board has a 50k pot. I understand pot value has to be lower than input impedance? This is an 8 pin pot with switch. If I replace it with 20k or even 10k pot without built in switch and leave the two holes unused, will it cause problem? Aside from the amp being constantly on whenever PSU is plugged.

Reducing the impedance of the amp's volume pot will make the situation worse. See my other post, above.
 
I am making a capacitor bank, can anybody me to a wiring diagram? How to connect the caps and how to connect to amp/PSU. I will use 8x 1000uF, thing is 4 of those are 35V and 4 others are 50V. Is there any order I sshould put them, or anyhow is fine? Thanks so much.

Wire them all in parallel, all the +leads tied together and all the - leads tied together. They don't have to all be physically arranged in single line as in the pic., just electrically. The 35V - 50V order really doesn't matter.

Cap Bank.jpg
 
Hi fulvio, thanks for your response. I remember you posted your findings, and we exchanged a few posts then. Is the gain 26 or 36dB? I checked the datasheet and the values on my board refers to 26dB as your first post said.
I planned to reduce the gain and remove that C33 to get it synchronized, but I havent. So I guess just leave it as is? Please advise, thanks! ;);)

You're right, sorry; it is because I'm testing more TPA boards at the same time.