TDA2050 amplifier board

Using an Antek transformer as a wall wart? Without a chassis? Might want an in-line fuse on the primary, in addition to the power cord. If toroids get banged around too much the secondary might develop a short. I would be inclined to mount it in a little plastic bud box, with IEC connector and coaxial power jack. It then becomes “lab equipment” and can find other uses when not powering the amplifier.
 
So I decided to take care of the DC offset. I replaced the 22k resistors with 20k resistors and put a 10k pot between both 22k resistors and the other 22k resistor so that I could adjust the offset to 0V. I wonder how much the DC offset would be if both amp chips were on one larger heatsink right beside each other.


TDA2050 bridged amplifier 2.png
 
Gonna use the other two amp boards and mount them to a single heatsink just to see if the DC offset is affected by the chips being on the same heatsink. Won't be a 100% proper test as the chips won't be right beside each other, but will be about 4" apart on an aluminum heatsink. The results of that test will determine if I do or do not do the DC offset mod as if the chips being on the same heatsink causes the DC offset to be compensated for to where it is under 20mV then there's no real need for the offset adjustment.

What I find interesting is that I can find a 2.1 channel board using the TDA2050, but I cannot find a board that runs two TDA2050 in bridged mode. If I knew how to design circuit boards I'd design one myself. Only I'd likely design it with a +/- supply so that there's no DC on the speaker + terminals which would eliminate the need for bias resistors and a DC offset adjustment.
 
Last edited:
The LM3886 will be close enough for the output you need from a bridged 2050.
Or a TDA7294.
A bridged configuration may be more complex than a simple single chip unit.

So if needed, you can think of that.

If possible, please post pictures of the original and replacement chips which were used by you.
 
Yes I'd definitely need speaker protection for a dual supply. Might even want it for a single supply.

One of the few LM-3886 boards I could find.

https://analogclassics.com/product/...NfBzQYAUeBNn2ln__wL4hE61mWcwNb8241iK_Ik&gQT=2

https://analogclassics.com/wp-content/uploads/2023/11/lm3886-and-psu-manual-chipamp.com_.pdf

The LM-3886 indeed is very tempting. Has lower distortion and higher output power, however for now I'll stick with the 2050 as I've already got what I need of those. Maybe for a future project I'll use the LM-3886.
 

Take a look here before buying lm3886 boards.
 
  • Like
Reactions: Tube Radio
The only issue with how the TDA2050 is bridged is phase shift caused by the two 22uF caps. There's some initial phase shift by the top 22uF and there's phase shift again by the bottom 22uF.


1739371576394.png



To eliminate it the 22uF cap has to be 330uF


1739371718543.png



But I don't feel that I can increase it any.


1739371940297.png
 
The settling time at switch on is affected. The designers who put together the app note probably staggered all the time constants in the “system” to minimize POP at turn on. If you would rather increase these time constants to get better bass response you either put up with a longer and possibly larger turn on transient, or put a delayed speaker relay at the output so you can “ignore” it
 
Looks like the time constant goes from 22mS with the 22uF cap to 330mS with the 330uF cap

There was a pop when I had a single amp used with a 1,000uF speaker coupling cap, but with both amps used bridged and no speaker coupling caps there's no pop.

I may parallel a cap with the existing 22uF and use my scope and DMM just to see what happens.
 
I tried 160uF caps in parallel with the 22uF caps.

When the amp powers up it starts at 4mV offset for about a second then jumps up to about a volt and drops back down quickly then jumps up to several volts and drops back down quickly.

So using larger caps there is no good.

The amp seemed to have a slight roloff below about 60Hz, however I did not do a proper test where I kept the input voltage set to a set value (the function generators I use change the output voltage slightly over the frequency range) and set the generator usually 100Hz, 80Hz, 60Hz and 40Hz then measure the output voltage and calculate the gain.

That said the amp does sound good on the speaker I am using it with at work.

Also the amp seems to be flat to well above 20kHz, but a frequency response test will show for sure.
 
The amp seemed to have a slight roloff below about 60Hz
Hanging a pot directly at the input of any power amp results in a poor frequency reponse.
1 The signal path should look something like Source - >Gain stage (optional) - >pot - >buffer stage (must) - >power amp
2 Else volume can be controlled directly from source settings - >power amp
I prefer the 2nd option as it provides best ch tracking with 0 viper noise
 
I've only experienced that issue in two instances.

1. Pot is over 100k and the first amplifier audio stage is a triode which causes high frequency roll off.
2. Pot is after an audio coupling capacitor that is too low of a value which causes low frequency roll off.

The pot is 50k and the input connects directly to it. In this instance I am using a signal generator with an output impedance of 50 ohms.

Two flaws with that test.
1. He doesn't mention the input impedance of the amp.
2. He doesn't mention the value of the pot.

However since he mentioned a buffer it can be assumed that the amp is high impedance and the pot is high resistance or that the amp has an RF filter that requires a large cap value to keep the amp stable.

Also he didn't show the output voltage of the signal generator at any point which is very important to see when doing tests like this especially when explaining why to not do something.
 
Last edited:
Here's the frequency response with no load.


1739461102098.png



So the AC gain is set by the 22k and 1k resistors.

What would happen if I swapped them to 220k and 10k which would keep the same ratio between the two resistors for gain purposes? The 22k resistor feeding the 1k/22uF junction would also have to be changed to 220k.

I keep, looking at the 22k 1k resistor junction and thinking there's a real simple way to alter the circuit to achieve the goal of better frequency response without any ill effects on the circuit.


I see how to do it. Look at the red area of the schematics.

Original


35 watt amp 1.png



Modified


35 watt amp 11.png



Now how would I modify this amp like that?


TDA2050 bridged amplifier 2.png
 
Last edited:
So here's what I'm thinking.

Replace the 22k resistors from pin 4 to pin 2 with 100k resistors which provide the DC feedback.

Install a 22k resistor from pin 4 to the 1k 22uF junction on both amps. That way the 22uF cap doesn't go to ground through any resistor and only has the DC voltage between pin 2 and the 1k 22k junction across it.

Will only require removing one resistor, replacing it with another then adding a resistor and doing the same on the second amp.

Only gonna do one first just to see how it works and so I can compare it to the unmodded amp far as gain and frequency response are concerned.

Will update the schematic after lunch.
 
Just make the 22 uF caps bigger. You’ve got 2 cascaded poles at 7 Hz that fully explains your roll off. Yeah it will cause a longer “thump” but it will be common mode, moving both outputs up/down simultaneously. The input coupling cap can be made bigger too, although that is the one you want to set the actual low frequency corner with.
 
The test I did yesterday using 160uF caps in parallel with the 22uF caps showed that the change in dc on the output is not common mode like would be expected.

I suppose part of the reason may be that the 22k resistor from the top amp to the bottom amp forms a voltage divider with the 1k resistor so there will be a DC voltage on the negative of the lower 22uF cap.

That said doing it the way I propose doesn't require the 22uF capacitor to be larger in size.

So the amp would thus be like this.


TDA2050 bridged amplifier 3.png
 
Last edited: