Why is my amplifier motorboating?

Question

I have been trying to make a simple guitar amplifier and I've been having a lot of trouble trying to get it to work properly. It keeps on motorboating at a very low frequency which is driving me nuts. I'm still kind of a beginner so I really don't know what I've done wrong but I'm sure there's something.

I've tried it with and without the two bypass capacitors (the 1000uF & 100nF), played around with IC1's gain and tried a few different coupling capacitor values (the other 100nF & 220nF.)

The 741 Op Amp is supposed to be more of a preamp rather than a buffer. Its gain is very low because of a long, unsuccessful debugging session. I will replace the 50 ohm with something much higher.

IC1: UA741CP

IC2: Phillips TDA1519A

Power Supply: 12V, 3A

To me it sounds like IC2 is the culprit as changing the volume control does not simply change the loudness of the motorboating but changes its behaviour.

To describe the noise the amp is making better: signal is getting through, just not how it's supposed to. I can hear the guitar, though it sounds more like it's oscillating at the same frequency as the guitar signal rather than actually going through and being amplified. It is unusable as the amp only makes noise if the signal is loud enough (I have to hit the guitar really hard and single notes just choke out). Sounds neat though.

I have also tried with the volume control bypassed to no avail.

The other thing is that the speakers are behaving as if DC is going through them.

When I turn on the power, they suddenly push to one side and stay there (they still vibrate and all but not in the middle) which is really confusing because I'm not sure how that could even be happening.

I've also gone through an LM833 kit preamp with the TDA1519A set up in an identical way to this which also didn't work.

^{simulate this circuit – Schematic created using CircuitLab}

I thought I'd ask here before I consider IC2 a dud and try a replacement.

Answer 1

Here's what causes the motorboating: The resistive divider that biases your input amplifier is coupling power supply fluctuations into the amplifier. The output amplifier draws heavy currents from the power supply, creating voltage fluctuations on the power supply, and so you have a positive feedback loop at some low frequency.

The cure is to power that voltage divider with something that will isolate the input. One way is to use an RC filter with a very low corner frequency:

1. Disconnect the top of the upper 100k.
2. Connect a 10k resistor to the 12V supply.
3. Connect a big capacitor, say 100 uF from the other end of that 10k to ground, then connect the top of that upper 100k to the top of that new capacitor. If it still motorboats increase the RC time constant of your new filter.

Answer 2

The other thing, and usually the only thing, that causes this is grounding. Audio power amplifier chips, from the lowly LM386 to Sanken 50 W blasters, are notoriously picky about power supply decoupling and grounding.

The capacitor on pin 3 is 10x smaller that in the datasheet. Change it.

It must be as close as possible to the IC pins.

Use the exact decoupling values shown in the datasheet. Again, as close as possible to the IC pins.

Separate the power connections to the 741 buffer and the 1519 into two wires so the power amp currents do not modulate power to the buffer.

Keep the buffer and ignore all of the 741 blabber. It is not the problem. Other parts have better signal fidelity, but that was not your question.

Disconnect R1 and R2 from the buffer input, reduce thwir values to 10 K, decouple the pants off of them, and run a 100K resistor from that node to the buffer input.

The image is pretty small, but it looks like you are driving both amplifiers from one signal, and expecting the outputs to be out of phase for a bridge-tied load. This will not work, and there is no way to use this chip that way because neither amp has an inverting input brought out to a pin. You can work around this by using another opamp as a unity-gain inverter, but I recommend against this.

BTL circuits are even more finicky that regular power amps, and some parts just will not behave. The big clue is that the circuit is not in the 1519 datasheet. If there were any way for the part to be used that way, you can bet the designers would have it in the datasheet. There are other TDA parts that have the app circuit, a clear indication from the designers that it will work in that configuration. If possible, change to one of those other parts.

UPDATE: This is interesting . . .

The 1519 datasheet I pulled down yesterday had an app circuit as in the OP's first post, with both inputs shown as non-inverting and no mention of BTL operation. But the one I pulled down today (1519A) is a different part, with one input called inverting, and BTL right there on page one. However, it does not have the app schematic in the OP's post. SO - which part does the OP have, a 1519 or 1519A?

https://pdf1.alldatasheet.com/datasheet-pdf/view/19203/PHILIPS/TDA1519A.html

Answer 3

Could it be the volume potentiometer? How is it wired? Is one of its pins connected to ground? If so that would that would be something to fix, as the 220nf capacitor is there to remove DCoffset. One fix could be to connect the wiper to the capacitor, one end to pin 3, and the other to pin 1. Now it is referenced to its own internal voltage.

Answer 4

I would ditch the UA741P and switch to a more modern op-amp. The OPA1671 would work but be careful it is fast and decoupling etc. is important. You could use an old school one such as 1/2 of a LM358. Also in all the reference drawings I have it shows capacitors in each of the output legs. That decouples the speaker from the FV.

Answer 5

Why do you need the 54 years old lousy 741 opamp? The obsolete TDA1519A BTL amplifier datasheet does not have it. With a high current 12V power supply the amplifier output into your 3 ohms speaker will be 15W RMS or 30W peak and its heating will be another 30W so it draws peak currents of 60W/12V= 5A. Can your power supply produce 12V at 5A without its voltage dropping much?

Answer 6

As shown in the question, the input circuit that feeds IC2:1, IC2:9 is equivalent to:

^{simulate this circuit – Schematic created using CircuitLab}

V1 is the R1||R2||C6 buffered by IC1.

RA4/RB4 is the potentiometer.

VIN is the effective input voltage. As you can easily see, there'll always be a DC offset, since a current flows from the internal IC2V2 through input impedance and via RB5. The lower the volume setting - the smaller RB4 - the larger the offset, since there's always DC current flowing from PIN1||PIN9 to ground via RB4.

Recall that the amplifier is a DC-coupled amplifier, so the input voltages are subtracted from 6V, amplified by 100x, added to voltage on pin 3, and then output.

So the input circuit cannot look like that: the only thing you can connect to PIN1 and PIN9 is a capacitor. There cannot be a DC path.

Furthermore, R4 is much too large: 1MOhm vs 30kOhm effective input impedance. For most of its range, the output will be inaudible. Furthermore, R4 in series with 30kOhm input impedance acts as a low-pass filter. So you don't want to make it any larger than it needs to be. A value of 10kOhm would be much better: it's 1/3rd of the input impedance, and is well within the capability of LM741's output driver.

Thus, the circuit should be modified as follows:

^{simulate this circuit}

Changes:

1. Feed the supply directly to C1||C2, and from there to pins 7/5 of IC2. The positive and common rails to the rest of the circuit must be connected directly to pins 7/5, as shown.
2. R7+C7 is a low-pass supply filter for IC1.
3. R4 is down from 1M to 10k.
4. C4 is moved after R4, and couples R4 to IC2's input. The value shown in the datasheet is wrong, it should be 440nF ideally, since it drives 30kOhm instead of 60kOhm. The nearest standard value of 470nF is fine.

At mid-volume C4 is connected to 5kOhm voltage source on the left, and 30kOhm source on the right. Thus, its time constant is \$470\cdot 10^{-9}\text{F} \times 35\cdot 10^3\Omega = 16.5\text{ms}\$, maximum, or less at other volume settings. That's acceptable.

Now, generally speaking IC1 is somewhat pointless, since the input impedance as seen into IC1 is 50kOhm, vs. 30kOhm into IC2 directly. You could remove R1, R2, R3, R6, C6, IC1, and feed the input directly into top side of R4, for an input impedance of 10kOhm. That's quite acceptable for a line input impedance.