TDA 2030 amp, strange fault

[ValvoStef]

Morning all,

I am experimenting with chip amps and I built a stereo amp (on a strip board) with two TDA 2030 amps and a few components. The schematics I found on the net. It works, sort of. At low volume level it sounds reasonably good. When turning the volume pot above a quarter, a high pitch noise appears on both speakers. Turning the volume to approx the quarters the noise disappears. The odd thing is, with the volume pot at zero (connected to a MP3), the amp pulls 50 mA which increases with the volume fully up to 400 mA. However, when the high pitch noise starts, the current drops to 200 mA. My first thought was the heat sinks may be ‘floating’ but connecting them to ground made no difference. When I disconnect the input signal, at zero volume the amp pulls 50 mA, at full volume the amp pulls 5 mA as well, turning the volume slowly up, the amp then pulls 200mA and the high pitch noise appears.


Initially, I built one channel on a bread board. When testing, there was no high pitch noise at all. I did not measure the current then as it was working fine.

What could possibly causes this?

See photos attached.

[ValvoStef]

I just connected a scope, volts per division 0.1V and time base 0.5 ms, scope screen shot attached.

[G6Tanuki]

Looks like your amplifier has achieved the underlying ambition of all amplifiers and has evolved into an oscillator.

Have you fitted the Zobel network [1Ohm /220nF series R/C pair] from speaker to ground? Without this you can get some definitely-odd effects. The Zobel needs to go as close to the output terminal of the IC as possible. A few ferrite-beads [on speaker-leads and power-leads] along with making sure the power-supply-decoupling electrolytics on the board are decent low-ESR types [preferably also paralleled with something like a 0.1uF polyester or two] should help tame it.

I recall seeing some amps using similar chips where the designer had deliberately used very wide earth-traces to keep things tame (as compared to the usual mantra - supported by commercial PCB CAD programs - to etch away all 'superfluous' copper from the PCB so it can be reclaimed from the etchant).

Long 'common' signal and output earth-returns are your enemy!

[SiriusHardware]

I love the valve in a socket on the breadboard!

The most common mistake I make myself when building projects on veroboard / stripboard is forgetting to make track cuts where I meant to, leaving two random parts of the circuit connected together by a longways track. Or, I make the track cut as I meant to but then don't entirely manage to cut the track and leave an invisible sliver of track intact.

If this proves not to be the case then try separating the common connections between the two amplifiers and running only one at a time to see if the symptom is still present when each channel is run independently.

It may just be a matter of needing thicker supply and 0V return tracks, or routing all the 0V points (amp IC supply 0V, speaker 0V, input 0V) separately back to a common 0V point rather than through a shared track or wire.

[ValvoStef]

Hi,

My Zobeluses 10nF only, is this too low? As you mentioned low-ESR caps, I doubt they are as they came out of me stash. However, I used the same components on the bread board and it was working fine. I am really puzzled.

[ValvoStef]

Originally Posted by SiriusHardware

I love the valve in a socket on the breadboard!

The most common mistake I make myself when building projects on veroboard / stripboard is forgetting to make track cuts where I meant to, leaving two random parts of the circuit connected together by a longways track. Or, I make the track cut as I meant to but then don't entirely manage to cut the track and leave an invisible sliver of track intact.

If this proves not to be the case then try separating the common connections between the two amplifiers and running only one at a time to see if the symptom is still present when each channel is run independently.

It may just be a matter of needing thicker supply and 0V return tracks, or routing all the 0V points (amp IC supply 0V, speaker 0V, input 0V) separately back to a common 0V point rather than through a shared track or wire.

I’l try this next. Thanks, good tip.

[Ambientnoise]

Cannot see the scope but it sounds like instability. It could be simple input to output coupling so check your physical arrangements carefully. Also, Strip board is notorious for stray inter track capacitance causing unwanted feedback and instability. I recommend rebuilding on plain matrix board but you can try removing all unused sections of track (a hot iron will help them to lift) and ensure good I/O seperation. I presume that if you are using the same schematic then you have adequate decoupling ?
There are cheap 2030 PCBs available on the net which work fine.

[G8HQP Dave]

If you used an electrolytic cap from the volume control to the chip input check that this is the right way round.

Ensure that the input connection, coupling cap etc. are well away from the output to reduce stray capacitive coupling.

Try to ensure that loops containing current flows are as small as possible by keeping 'go' and 'return' conductors near each other. Also ensure that as far as possible the input loop and output loop are well apart except at the single point where they are joined together.

[Restoration73]

The decoupling caps 100-220uF need to be wired directly across pins 3 & 5 at the chip.
I think you're using the single rail configuration.
I strongly recommend fitting a 1nF disc across pin 1 and ground to prevent pickup
of local rf noise or transmitters. It won't have much effect on the audio.

[ValvoStef]

Originally Posted by Ambientnoise

Cannot see the scope but it sounds like instability. It could be simple input to output coupling so check your physical arrangements carefully. Also, Strip board is notorious for stray inter track capacitance causing unwanted feedback and instability. I recommend rebuilding on plain matrix board but you can try removing all unused sections of track (a hot iron will help them to lift) and ensure good I/O seperation. I presume that if you are using the same schematic then you have adequate decoupling ?
There are cheap 2030 PCBs available on the net which work fine.

Yep, the board is as per attached circuit Of course I could buy a board but this project basically came out of nowhere last Sunday and I used what I had in the stash.

[ValvoStef]

I I’d what SiriusHardware suggested, separated the supplies and tested each channel on its own. The effect disappeared. as suggested the problem may be the veto board tracks. I try what Restauration73 suggested.

[Refugee]

The board looks rather long and if there are unused tracks that run all the way to the end of the board they will pick up noise and cause your problem.
I would disconnect and earth any long unused lengths of track.
The big capacitors also look like they are a long way from the chips.

[ValvoStef]

Success: I added 100uF to pin 3 and 5 of the TDA, also added 1n ceramic cap from pin 1 to ground.The high pitch noise disappeared, without input signal the amp pulls 50 mA across the whole volume range. WOW, what a learning curve, thank you so much, guys, for all your input. Really appreciated.

[ValvoStef]

Schematics are now updated and attached. The devil is always in the detail...

[paulsherwin]

I like the heatsinks!

[ValvoStef]

Paul,

The heat sinks were salvaged from two industrial solid state relays which I fished out of the skip at work.

[trh01uk]

Great to hear you had success with this. The art of electronics is in understanding the role of all the components that are not on the schematic......

Richard

[Andrew2]

Good to hear it's working at last. These chip amplifiers are known for their 'friskiness' if they aren't decoupled very close to the relevant pins, and (in my experience with the smaller LM380 and 386) the zobel network is crucial and can even be beefed up a bit. Also the cap from input to deck is omitted at your peril!
I've seen constructional articles in the hobby mags which have taken terrible liberties and on the air (I'm a radio amateur) I've heard some oldsters telling young 'uns that they can leave out 'those little caps' as 'the amplifier will work just fine without them'! I cringe when I hear this.

[TonyDuell]

I've done a lot of work with 'traditional' digital ICs (TTL, etc) and I learnt early on that decoupling capacitors are cheap, my time in finding obscure faults is not!. More seriously, it's probably a different matter if you are a manufacturer making thousands of units and need to save every penny, but for enthusiasts making 1-offs it makes a lot of sense to decouple just about every IC power pin, etc.

[Refugee]

I built a signal generator in the 1970s and it never worked with faster chips and it had to stay with slow chips until I visited it again to find two dead fets and then I put some decoupling caps in under the bank of opamps.
The original chips I bought back in the day worked fine.
Those SMD caps saved all the drilling and mess.