Trouble with bridged TDA2050 and single supply.

[BrunoARG]

Hello everyone.

I am spending some time in building a very simple medium-power audio amplifier I really need.

I found an about 60 or 80 watts, 30Vrms bored transformer so I decided to give it a useful application.

I bought a 50Wrms 8" 8ohm woofer (45~4500Hz) and a 4K~20K piezo-tweeter (about 6 ohm) to make some noise and their sound quality seems to be very good, but I have several questions:

The power source is an about 40V unregulated single supply, the filter I am using is performed by 4 2200uFx50V (total 8800uF). If the output power was 80W (in both heat sink and load), would it be fine? Line frequency here is 50Hz.

Should I use a high pass series capacitor with the tweeter? In the datasheet it recommends to use a series resistor but I don't want to (I had to go and buy it...). In a chart it said to use an about 6uF one.

I tested the circuit in a breadboard and it has a problem: It has a very high DC offset at the output which generates high -and really annoying- 100Hz ripple and power dissipation at the heat sink (it raises to about 80°C).

The circuit is the following:



What I changed are both amplifiers feedback resistors for 33K ones (higher gain) and feedback divider ones for 470 (almost the same). The lower cuttof frequency is about 30Hz. The resistor between one output and one feedback is 33K too.

What can be the problem? I tried changing some capacitors but the circuit started to oscillate at low frequencies as 10Hz or 3Hz with no input signal.

I don't know what to do, do you have any recommendation or solution for my problem? Maybe the inverting feedback needs some AC feedback?

Thabk you in .


EDIT: PD: I have even tried the original values but the DC output was still the same, or just a little bit lower, but it was there.

 

[BradtheRad]

the circuit started to oscillate at low frequencies as 10Hz or 3Hz with no input signal.

This is probably a different issue than your 100Hz ripple problem.

I made a simple audio amplifier using an easy-to-use (lower power) audio op-amp. I installed a volume control with potentiometer. I got low frequency oscillations at a certain volume range. To cure it I tried adding a capacitor in strategic positions.

I tried different configurations of resistors around the volume control. This worked. I had to find the right combination, so that oscillations stopped.

 

[D.A.(Tony)Stewart]

AC supply probably has lots of 100Hz ripple. Measure it with no load and full load signal.

Each power amp is probably oscillating above 100Khz causing overheating. If it gets hot with no load, put 2.7Ohm 0.47uF as indicated in specs or similar series snubber on each output to ground.

If DC out is not matched at Vcc/2, offset the offending one with a few uA to ground on the appropriate input with suitable R using a pot or compute with Ohms Law . Spec is Vio=20mV , your gain is 20 thus Vout =+/- 200mV maxdc offset is expected across speakers.

use a series R and C for the tweeter. You need a series R & C for the snubbers above too.

This chip is obsolete for new design.
50Hz unregulated power is expected to be noisy for audio without special care on filters, peak current limiters, LC filters etc.

 

[Audioguru]

I agree that the TDA2050 amplifiers are oscillating at a very high frequency because your circuit is missing the RC to ground at each output as shown on the datasheet and that the woofer is a high impedance at high frequencies and that the piezo tweeter is also a high impedance. The piezo tweeter has a very bad frequency response full of peak resonances and nulls between the peaks.
The 30V RMS transformer makes a +40V supply and the output of each amplifier will be about 33W at clipping. The amplifier is about 60% efficient so at an output of 66W the power transformer must deliver 110W but it is overloaded and can't.

You should adjust one of the biasing voltage dividers to eliminate the voltage offset at the outputs.

The current of the amplifier is far too high to build it on a breadboard. The resistance of all the contact strips and wires probably causes the low frequency oscillation.

 

[BrunoARG]

Thank you guys for your responses, I really appreciate!

So, to avoid high frequency oscillations I need the RC filter (I saw 1ohm-100nF ones, but doesn't it depend of the load inductance?) at both amplifier outputs.

Should I callibrate the non-inverting input of the second (not the one with the audio input) amplifier till it reachs VCC/2? I think I should do the same with the first one.

Something I did not come up with is that the DC and oscillations happens with only the woofer connected (doesn't matter, a tweeter will not solve the problem). Anyways I will take that into account and make the output filter and pot offset correction.

Thank you very much, I will be back with the (good) news tomorrow.

 

[Audioguru]

The datasheet shows ONLY 2.2 ohms in series with a 0.47uF capacitor from the output of the TDA2050 to ground so why change it and risk oscillation?
The inductance of the speaker does not matter since it is a high impedance at high frequencies and the RC at the output is a low impedance load at high frequencies (not a filter) that is needed to prevent oscillation.

The required bias voltage is somewhere around half the supply voltage, not exactly half. It can be set so that clipping is symmetrical if you want. You want to adjust one side to eliminate the DC offset voltage at the outputs.

 

[chuckey]

To minimise the DC problem, bias both the amps from the same DC point. So select one amp, the 1/2 Vcc point is the junction of the three 10K resistors, connect the 10K feeding the other + input to this point. Remove the now unused 2 X 10 K resistors and 100MF cap. The DC gain should be 1 for unless the 100 MF capacitor is leaky - connected to the - input.
Frank

 

[Audioguru]

I agree that if only a single voltage divider is used then the datasheet says the maximum DC offset voltage at the outputs will be only 30mV which is nothing.
If the voltage dividers use 5% tolerance resistors that are at their max deviation then one voltage divider will produce 6.65V instead of 7.0V and the other voltage divider will produce 7.35V instead of 7.0V and the resulting offset voltage will be 700mV.

 

[BrunoARG]

Well, I first put both output filters and it was the same, they kept overheating without input signal. When I measured that DC offset it was of about 4V.

I don't know if putting both VCC/2 references at the same resistor network would work since the input is connected directly to it, and I don't want a signal in the second amplifier.

Then, I changed the feedback resistors to 33K (higher gain) and thew DC disappeared, but the circuit didn't work, it did nothing. I changed the values back to 10K and it was still the same, nothing, nor overheating.

So frustrated, I decided to build the split supply bridged amplifier I found in several pages. It didn't have almost any DC offset between outputs, but the sound was horrible, so much clipping and distortion even at low output voltage. I put some DC at the input of the first amplifier and it appeared at the outputs too, but the sound quality was good in spite of the output DC and overheating.

Then I lost all my temper and build the split supply non bridged amplifier and it sounded really good, didn't even heat at all, the sound quality was very good and there was no DC.

This all make me think, what if I build two non inverting power amplifiers, put the original signal in one of them, invert the original one with a inverting op-amp at unity gain, and put that inverted signal at the other power amplifier? It should work, I'm gonna try that out and tell you.

 

[Audioguru]

A bridged audio amplifier that is not overloaded should sound good and sound the same as two separate amplifiers but with almost 4 times the power of one of them.
A bridged audio amplifier should have almost no DC offset voltage especially when only one voltage divider is used. The input signal is at only one amplifier because the other amplifier series input resistor is shorted to ground by the filter capacitor at the divider.
Amplifiers with a single positive supply and with split supplies should sound the same. Your bridged amplifier with a split supply had an error for it to produce clipping and distortion.

If you make two non-inverting amplifiers and use an external inverting opamp to bridge them then you should not hear any difference from the circuits you made.

 

[BrunoARG]

Finally, I don't know how, I got all my 4 TDA2050 blown. The last one I connected was working very well, so clearly even at higher power, but then I connected a bandwidth limiter capacitor in parallel with the 10K feedback resistor. It stoped working and started to output a high (about 6V) DC offset. I still don't know what happened, but time just happens and I have to move on to get the amplifier working for this week.

I don't know if starting a new thread with the amplifier I want to build so I will post its circuit here:

**broken link removed**

The question is, will it work with different supply voltages? On its page (in spanish) it says it works with +-45V. I can reach that voltage connecting transformers in series, but I had to buy higher voltage filter capacitors. I can reach +-15 with the transformers and filter capacitors I currently own. And yes, I know I won't need paralleling the output transistors since the output power won't be that high. It seems tthat the emitter resistors are wrong, they must be .47 ohms.

Should I modify the differential amplifier to work with the same current despite the lower supply voltage?
Which values are straightly important? which ones could I approximate?
Are there similar transistors I could replace for the original ones? For example, TIP122 and TIP127 for the complementary output transistors, and low power ones to their drivers, such as BC547/BC557?

And the last and least important question: How about the homemade thermal paste? I saw some made with toothpaste as thermal conductor (due to its titanium dioxide used as abrasive) and oil or vaseline as solvent to keep its doughy texture. I bought thermal paste but I gave it to a friend and I don't know if it still exists...

Thank you in advance.

 

[Audioguru]

Your attachment does not work.
You MUST NOT add too much negative feedback with a capacitor parallel to the negative feedback resistor because it will cause the power amp to oscillate at a very high frequency and get much too hot. Use a series resistor and a capacitor to ground at the input instead.

I think your datasheet for the TDA2050 is COMPLETEY WRONG! The English language one from ST Micro shows an absolute maximum supply of +-25V, NOT +-45V.
The English datasheet also does not show external output transistors and emitter resistors.

 

[BradtheRad]

Finally, I don't know how, I got all my 4 TDA2050 blown. The last one I connected was working very well...

Don't feel too badly. It comes with the territory.

I had a similar unexpected experience years ago with my TDA2002/LM383, an audio IC carried by Radio Shack. ($2.99 in 1980.) Just the easiest way I could ask, to let me make a homebrew 8W PA system. I wired it up and it worked. It was producing a bit of hum, noticeable but tolerable.

But I wasn't willing to let well enough alone. I poked around the IC with a metal prod, seeing if I could make the hum come or go. I accidentally touched both the output pin and another pin, instantly turning the IC to toast. After that it never produced a sound again.

I purchased a second LM383, and it worked, but I believe my first version produced less hum.

 

[Audioguru]

Like most audio amplifier ICs and a few opamps, the TDA2050 chip is "decompensated" so that it produces the high audio frequency 20kHz properly. Since it is decompensated then it has a minimum allowed amount of voltage gain, for the TDA2050 the typical gain as shown on the English datasheet is 33.5 times. The datasheet also says, "The gain MUST BE HIGHER THAN 24dB" which is a gain of about 15 times. A capacitor parallel with the negative feedback resistor causes the gain to be 1 so of course the IC will oscillate like crazy at a very high frequency where the phase shift changes the negative to positive. The datasheet for the little LM386 audio power amplifier says its minimum allowed amount of gain must be greater than 9 times.

It is also wrong to power the IC with a voltage that is almost double its maximum allowed voltage.

 

[BrunoARG]

Thank you for your fast replies.

It seems that I cannot edit the last post I did so I will post the circuit here.

As you have said, the TDA2050 gain couldn't be lower than 9 or 15, when I put the capacitor it got to unity gain due to its high frequency oscillation caused by the "decompensation" Audioguru came up with. I should have limited the minimum gain with a series resistor, not so useful indeed, actually I should have made a rc high pass filter at the input.

The supply voltage and output transistors I was talking about are from the circuit I "attached" and isn't available. It is the following:


Sorry about the attachment error.

 

[Audioguru]

Your new amplifier is completely different to a TDA2050. Its 10 ohms series output resistor throws away more than half of its output power and causes a high output impedance so speaker damping is poor causing boomy sounds. Its 47 ohms emitter resistors throw away most of the remaining output power. The old darlingtons might be too slow for good audio.

 

[BrunoARG]

Every output resistor might be in miliohms, otherwise as you say, the major power would be on the resistor and not on the load.

I think I will make a TDA based amplifier with split supply as I am running out of time and if I have time later, maybe in a couple of weeks, make the single supply one... I don't own any oscilloscope to "see" what is going on in a circuit but i'm planning to make an investment and buy one in the following weeks.

Thank you all for your replies, I learned a lot of things.

 

[Orson Cart]

A lot of care is needed in the layout to make these work well, check everything 3 times before powering up - READ the original data sheets carefully...! then you will have no problems in bridge or otherwise as thousands of other people have...
Also there is a small chance you have counterfeit parts which will always give you problems...

 

[Orson Cart]

Also output wires any where input circuitry gives problems, good and close de-coupling needed - there are any amount of good circuits on Google for bridging these...

 

[Audioguru]

Wasn't the circuit built on a Mickey Mouse solderless breadboard with messy long wires all over the place and many strips of intermittent contacts?