By connecting my scope across the leads of a dummy load at output of the TDA2050 power amplifier with the required external components, I see oscillation and then smoke. If the TDA2050 oscillates for a few seconds, does this destroy the IC? From what I have seen, I would think so.

The component values are those given in the application circuit of the ST data sheets.

The P.C.B. for the IC I constructed from scratch, and I think the problem might be that the power supply 100 nF decoupling capacitors are too far away from the IC at about 7 cm (2.75 inch). Could this conceivably be a flaw in my design of the P.C.B.? The 220 uF decoupling capacitors are about 2 cm to 3 cm (about 1 inch) distant from the I.C., and I guess that the electrolytic capacitors should be closer also.

What is possibly also leading to oscillation is that the power supply for the amp connects to the amp with hook-up wire, each of the three wires about 20 cm (8 inch) long. The power supply and amp are built on separate P.C.B.s. Do you think so?

Two amplifiers that I previously built do not oscillate, though I did see some oscillation with one of them that went away. The third board smoked two of the TDA2050.

If I were to make the connection from the power supply to the amp (point to point wiring with hook-up wire) as short as possible, would that help to prevent oscillation?

Thanks in advance for your response,
Pete

 

It would help if you post your schmatic, which one you used from datasheet.

Yes, caps have to be close to minimize stray L which effectively reduces the
C actually presented to its connection point.

Also oscillation into a high load for sure can destroy a part.

Regards, Dana.

 

What is the "dummy load"?

 

Maybe the TDA2050 became obsolete and is not made anymore because many of them did not work properly or were bad copies from China?

 

@danadak, The circuit that I built is the Test and Application circuit on the first page of the ST data sheets that I'm attaching (bipolar supply). V+ and V- power supply with respect to the center tap of the power supply transformer are + / - 22V. The question is whether or not the decoupling capacitors on my board are too far from the IC to be effective. Or in other words whether or not if I redesign the board with the decoupling capacitors brought closer to the IC this will solve the problem. The dummy load is 10 Ohms resistance.

@crutschow, The dummy load is five of 50 Ohm 10W cement resistors in parallel.

@Audioguru, I was careful to locate genuine ST microelectronics TDA2050. There are many counterfeits being offered, it is true. Not having worked with this power amp before, I don't know if there is often a problem of oscillation with this part.

http://www.ampslab.com/Freewares/tda2050.pdf
Thanks,
Pete

 

Your load resistors are an inductive coil wrapped around some kind of core. but a speaker also has some inductance so it should not matter.
The 100nF supply decoupling capacitors should be ceramic (not electrolytic) and be mounted very close to the supply pins and the "star" ground on the IC's pcb. C7 also must be ceramic type. The gain is fairly high at 33 times so the input and output wires should not be close together.

It will oscillate if its gain is reduced by having more negative feedback as explained in the datasheet of the little LM386.
At Elliot Sound Products, Rod Elliot says, "Note that the speaker must return to the central 'star' earth (ground) point at the junction of the power supply filter caps. If connected to the amplifier's earth bus, you will get oscillation and/or poor distortion performance."

 

@Audioguru

Thanks especially for the advice from Rod Elliot. Since the amplifier's earth bus is connected to the junction of the power supply filter caps, his advice would seem to entail providing a separate path to reference potential (ground) of the power supply. Is that what makes the difference?

This morning I shortened the length of the power supply wires to the amplifier to 7 inches (18 cm), made the 100 nF bypass caps C3 and C4 disc ceramic (previously film), and was able to modify the amplifier P.C.B. so that C3 and C4 connected to within 2 cm of the power supply pins of the TDA2050. Connecting this arrangement to +/ - 10V from my bench power supply and not loading the amp, it did not oscillate.

Well, I AM fearful of seeing more smoke, but I will be testing some more.

 

The star ground on the pcb has the speaker and everything else at 0V connected to one small spot. Then a fairly thick ground wire from the power supply 0V.

It is good that you fixed the oscillation and smoke.

 

Circuit board layout and ground management are two very common causes of audio power amplifier chip oscillation. But, the most common cause is inadequate power supply decoupling. ALL audio amplifiers chips are whiney cranky little children that will scream at the slightest provocation. Some datasheets address this; most do not. At a minimum, use the values on the datasheet you posted. Also, follow their layouts as closely as possible.

ak

 

It would help if you post your schmatic, which one you used from datasheet.

Yes, caps have to be close to minimize stray L which effectively reduces the
C actually presented to its connection point.

Also oscillation into a high load for sure can destroy a part.

Regards, Dana.

pretty much - caps close to pins and low ESR ceramic types as well as bulk capacitance electrolytics. supply tracks need to be substantial too.

A heavy capacitive load will make those burst into song, but a series CR damping network is usually a MUST HAVE.

Once I built a TV sound intercarrier sender with a combined SIF/audio PA chip - the 6MHz got into the audio amplifier and let the magic smoke out in no time flat...…...

 

To test my amplifier I have a RCA inline jack connected to the input terminals of the amp board with two short lengths of hook-up wire. A length of shielded cable terminated at both ends with RCA plugs is connected at one end to my sine wave generator. With the power connected to the amp, the amp oscillates when initially inserting the free RCA plug into the jack such that only the center post of the plug makes contact with the center conductor of the jack.

Fantastic, I've discovered one more way to make the TDA2050 oscillate and be destroyed.

 

Thanks to all for your advice. What I have assembled is a three channel power amplifier. Using my scope, where the input signal was a 1 kHz sine wave and each channel of the amplifier had a 10 Ohm dummy load, each channel can simultaneously output 14 W max. before clipping. Given the relative simplicity of the amp, I think this is pretty good and should be okay for me as long as I'm connecting eight Ohm speakers of at least average sensitivity to it.

Making the power supply bypass capacitors C3 and C4 (100 nF) disc ceramic rather than film and connecting a few centimeters closer to the power pins of the TDA2050 amazingly seems to be the primary change that stopped the oscillation. Given that capacitance is capacitance, why is disc ceramic more effective?

C7, 470 nF, (in the Zobel network, if that is giving it the correct name) I did not change to disc ceramic from film. As Rod Elliott suggests and as passed along by Audioguru, amp output reference potential is through a separate wire to the power supply transformer center tap.

If interested, my unregulated bipolar power supply includes a Hammond 187F28 transformer, 28V 100VA. Smoothing of the each + / - voltages of rectification is done with four of 3300 uF / 35V radial capacitors (paralleled).

 

A ceramic capacitor works well at very high radio frequencies where a power amplifier can oscillate. A film or electrolytic capacitor is made wound around and around causing inductance that cancels its capacitance at high frequencies.

 

With the power connected to the amp, the amp oscillates when initially inserting the free RCA plug into the jack such that only the center post of the plug makes contact with the center conductor of the jack.

The RCA plug is designed back-assward since it connects the center conductor before the outer ground.
That's why you never want to plug into an amp when it's powered since that will usually give a large 50-60Hz signal, the same as if you had a long unshielded, open wire connected to the amp.
That may be what happened to you, not an oscillation.
Although an oscillation is certainly possible from that.

 

A ceramic capacitor works well at very high radio frequencies where a power amplifier can oscillate. A film or electrolytic capacitor is made wound around and around causing inductance that cancels its capacitance at high frequencies.

Thanks, it has been quite a few years ago that I read the advice to use ceramic disc capacitors to bypass the power supply of op amps, and I did not understand why one type rather than another until now.

 

The RCA plug is designed back-assward since it connects the center conductor before the outer ground.
That's why you never want to plug into an amp when it's powered since that will usually give a large 50-60Hz signal, the same as if you had a long unshielded, open wire connected to the amp.
That may be what happened to you, not an oscillation.
Although an oscillation is certainly possible from that.

Thanks. Not wanting to experiment and possibly losing another IC, I'm not certain what I saw on my scope. In Rod Elliott's design of an amp optionally using TDA2050, Figure 1 shows a RF filter at input to the amp. This leads me to think that the TDA2050 might amplify RF at input, given an unshielded wire connected to input to the amp?

http://sound.whsites.net/project72.htm

Regards,
Pete

 

An audio amplifier should never have an unshielded input wire.
The TDA2050 has a fairly high voltage gain and produces 80kHz at -3dB so it would have interference and maybe excessive heating if a high power ham transmitter was next door or if an AM radio station was down the block.