# Audio/Video Issue with DIY music system

#### PhysicoRaj

Gold Member
Hello there..
I had previously worked with the TDA2030A amplifier and made an 8 W sound-box which was nice. So I thought I should upgrade my plan and decided to make a 2 channel music system on my own.
I have it all rigged up and mounted inside an old cassette player (with my own speakers) but it seems to have a problem. As soon as i turn it on, it starts to produce a near 3 Hz square wave kind of pounding in both the speakers.

I have my overall design block diagram, The amplifier circuit schematic, and the datasheet attached for reference.

The system is mono, but two channels for bass and mid-range. The spectrum is split by two second order op-amp butterworth filters, HPF and LPF both of cutoff around 150 Hz. Input is common to both filters and they are each followed by an amplifier then by respective speakers as in the image.

The whole thing (both filters and both amplifiers) are parallel on a split supply of +12/-12 volts DC from two 12v lead acid batteries.

So frequencies above 150 Hz goes to 10 W 8 ohm speaker (mid-range and high) , while below 150 Hz goes to 25 W 4 ohm speaker (bass and sub).

Coming to the amplifiers, I have used the bipolar configuration as in the datasheet, except for the components marked in red below, Which I have not used.

C8 is around 330 pF to give an upper bandwidth limit at 22 kHz. [Did I go wrong anywhere here, skipping the R5 which is supposed to be approx. = 3*R2]
R1 for both the amplifiers is a pot to vary the closed loop gain.

Now when I push the toggle switches of the supply, an intermittent 3-6 Hz square wave pounds both the speakers. Irrespective of the volume, input/no input. Plays nothing except this noise. The frequency varies with varying pot R1 value. I checked the amplifier IC's and heatsinks, they're fine and do not seem to heat up.

I have four major questions..
• Am I wrong in omitting any of those components marked? Are they crucial? What are their functions in fact? [R4, C7]
• Why are there two pairs of supply bypass capacitors? Isn't any one pair enough? [I felt C5,C6 larger caps would be enough]
• Anything else I have missed out in my overall setup?
• What could be the reason for the completely irrelevant output I am getting?
I would be grateful for any help/insight/advice.
Thank you.

#### Attachments

• 464 KB Views: 311
Last edited:

#### PhysicoRaj

Gold Member
I observed that the speakers oscillate when both +Vs and -Vs are connected.
The common ground is untouched. I have toggle switches for +Vs and -Vs. (Each followed by a 5 A fuse).
I tried connecting only +Vs and then the speakers (diaphragm) are pushed out, indicating there's DC at the output.
When only -Vs was connected (no +Vs) the speakers were pulled in, again proves negative DC at the output.
Though there's nothing at the input.

The split supply means absence of output DC decoupling capacitor. But I'm not sure how the DC level is being introduced. Is there a way that the protection diodes might be the cause? (They might be operating in break down, letting out the supply to the output).
One more theory that either the amplifiers, or the Op-Amps are saturating, and thus the oscillating square wave..but mysteriously this happens without any input and I'm unable to detect the source of this. Wrong capacitors holding charges and circulating through resistors?
I'm lost.

#### davenn

Gold Member
I have four major questions..
• Am I wrong in omitting any of those components marked? Are they crucial? What are their functions in fact? [R4, C7]
• Why are there two pairs of supply bypass capacitors? Isn't any one pair enough? [I felt C5,C6 larger caps would be enough]
• Anything else I have missed out in my overall setup?
C3 and C4 are power rail decoupling for higher freq's ... they should be left there .... will improve stability
C5 and C6 do the decoupling for the lower freq's on the power rails

R4 and C7 are part of what is called a Zobel Network is in its simplest form, a series resistor-capacitor (R-C) network that is connected in parallel with an audio speaker in order to minimize the effects of the voice coil's inductance

Dave

#### PhysicoRaj

Gold Member
• Can I leave out the R5? The datasheet says R5 and C8 form the high frequency attenuators. But it seems the cutoff is defined as $$f_H=\frac{1}{2 \times \pi \times R1 \times C8}$$ which is independent of R5.
• Can I omit the diodes? Are voice coils so inductive to produce spikes taller than supply?
• And..seems either the Zobel network or the diodes any one is enough.
Because I want to narrow down on the causes of my major problem.

Last edited:

#### davenn

Gold Member
Can I omit the diodes?
they are part of the design layout and shouldn't be causing any problems

And..seems either the Zobel network or the diodes any one is enough.
no, the resistor and cap are to stop high freq oscillations ... don't see how the diodes will do that

and again they are part of the design layout and shouldn't be causing any problems

I really don't like the idea of using 741 op-amps in an audio cct ... they are pretty crappy and old devices
you could try op-amps that are 30 yrs newer
TL081 or TL082 for example

firstly, disconnect your front end filters and make sure the 2 amps are working normally with a good audio signal in. Then add one of the filters and see what effect it has. It could well be the filters based around the 741 that is causing your probs

Dave

Last edited:

#### Tom.G

Your description of the 3-6Hz oscillation is known as 'motorboating'. It is usually due to insufficient power supply filtering, supply bypassing, or a circuit-board layout problem.

Here is the scenario:
1) A turn-on transient or random circuit noise is fed to the output stage. This causes a large Power Supply load. (As high as several amps in your case.)
2) This high load current causes the supply voltage to drop.
3) This supply voltage drop is coupled back into an amplifier stage; can be thru the power supply or into an input via a board layout problem (or both).

The frequency varies with the gain because at higher gain you have a more rapid change in supply voltage.

If the op-amps can be unplugged, do so. Three different results will help localize the problem.
1) If the motorboating stops, part of the problem is in their implementation.
2) If the motorboating shows little or no change, the problem(s) are in the output stage and/or power supply.
3) If there is a significant change in the motorboating, the problems are distributed over output stage, op-amp stage, power supply, and/or board layout.

The specifics of the board layout (at least photographs of both sides) and the power supply design would be needed to help with detailed troubleshooting.

Have Fun!

#### jim hardy

Gold Member
2018 Award
Dearly Missed
Run down batteries will cause motorboating. Measure supply voltage while it's misbehaving.

#### PhysicoRaj

Gold Member
Thank you for all those inputs guys.. really helpful.

I removed the filters(completely decoupled them) but the motorboating is still in action. So it might be problem with the main amplifiers, supply.
The 741 Op Amps are erroneous sometimes yeah.. Thanks for those numbers Dave, I never heard of those opamps.
Now it's sure the probs with output stage.

I read about the motorboating in wiki and understood that either ineffective supply regulation or, improper feedback could cause this. Thanks for pointing out this Tom. G!

As presumed, I measured some stray voltage at the input (though nothing is connected) while it's motorboating.. So indicates a faulty feedback of some sort. (A hefty 4 v (avg/rms) during motorboating!)

I checked my board closely and I don't have any shorts, or other track anomalies.

Might be due to supply regulation capacitors? 100uF was suggested by the datasheet. Should I use any other value, or remove them and check?

Or anything wrong in the feedback path I chose(see amplifier ckt in #1) I have missed R5.

And Jim, the batteries are new, I measured them and it's 12.8v open, 11.8v during turn on, 12.4v(avg? ) during motorboating. I have two 5A fuses at either ends of the bipolar terminals so current drawn (rms/continuous) is less than that.

Thanks all.

Last edited:

#### PhysicoRaj

Gold Member
Update :
I removed the 330pF C8 capacitor (and R5 was never there). Now the feedback is clean with only R1, R2 and C2, now the motorboating has reduced by 90%(by volume)

They now oscillate at a specific range of R1, but very little, though I am determined to eradicate 'em all.

Out of curiosity I plugged those filters back and it seems to have no effect on the distortion, 10% of the motorboating persists that too at typical values of the gain setting pot R1 , with or without filters. [still I'll replace the 741's with good ones.]

Yet the remaining small problem, I think, is with the main amplifier circuit itself, which has to be addressed.
Any suggestions?

Thanks.

#### Tom.G

Some observations:
1) A 1V supply voltage drop is quite high for lead-acid cells. Either they are undersized, partially discharged or damaged, or you measured at the op-amp pins and have a long run of undersized wiring to the batteries.
2) C5, C6 should be in the range of 1500uF to 3300uF and placed physically and electrically close to the TDA3020 with heavy traces/wiring between the caps, the TDA3020, the speaker return line, and the battery common. This is especially true for the speaker return line. It should connect only to the speaker and the common connection of C5, C6.
3) R5, C8, as well as R4, C7 may be needed to avoid ultrasonic oscillations, You will need a 'scope to check for sure. If ultrasonics are present they will increase power dissipation, decrease audio power, probably introduce distortion, and may damage the speaker. If a 'scope is absolutely not available, you can try connecting a bulb from a single cell flashlight in place of the speaker. If it lights at all without any input signal, you have problem.

Details on point 2) above:
A "single point Ground" is needed for Grounding. This consists of:
A) A large pad (if it is a pc board, or shortest practical, heavy wire if wired) that connects the common point of C3 thru C7, the battery Common, the speaker return line.

B) Other circuit power grounds are traces that connect to that large pad.

C) Power to other op-amps on the board should come the appropriate + and - pads C5 and C6.

D) For the LPF, the input signal ground should connect to the gnd end of the 10k, the gnd end of the 4.7uF, the gnd end of R3 at the TDA3020 input, and then to the large pad of detail A), above; in that order.

Have Fun!

#### PhysicoRaj

Gold Member
Some observations:
1) A 1V supply voltage drop is quite high for lead-acid cells. Either they are undersized, partially discharged or damaged, or you measured at the op-amp pins and have a long run of undersized wiring to the batteries.
The drop was only during turn on transient, now that the motorboating reduced, new value is 12.69v continuous.
But for the transient drop, I would say they are not fully charged. Because they are new, undamaged and I measured the voltage at the battery terminals and there's thick enough wiring.
I'll try charging them.
2) C5, C6 should be in the range of 1500uF to 3300uF
but the supply is a battery right? Would such huge values be required?
and placed physically and electrically close to the TDA3020 with heavy traces/wiring between the caps, the TDA3020, the speaker return line, and the battery common. This is especially true for the speaker return line. It should connect only to the speaker and the common connection of C5, C6.
Except for the TDA2030, that's exactly how the design is. Thick plate-like tracks, all components not far from each other, but yeah.. the TDA2030 is a bit far.. I telll you, the reason is I wanted it to be well heatsinked and easily accessible for inspection, so i mounted it on a sink and placed it near a vent in the box, this required some wiring from the PCB to the IC, about 15 cms length of 0.75mm thick single strand wires [for each lead]...
3) R5, C8, as well as R4, C7 may be needed to avoid ultrasonic oscillations, You will need a 'scope to check for sure. If ultrasonics are present they will increase power dissipation, decrease audio power, probably introduce distortion, and may damage the speaker. If a 'scope is absolutely not available, you can try connecting a bulb from a single cell flashlight in place of the speaker. If it lights at all without any input signal, you have problem.
I'll soon put the R4 C7 pair.

R5 C8? Not sure..removing this stopped the motorboating actually..
If ultrasonics are present, they will be present in amp A1 right..(High pass driver).
Or since filter is at input, the ultrasonics would emerge from inside the amp?

Single cell flashlight incandescent bulbs are obsolete here. Would a white LED [array?] work?

Details on point 2) above:
A "single point Ground" is needed for Grounding. This consists of:
A) A large pad (if it is a pc board, or shortest practical, heavy wire if wired) that connects the common point of C3 thru C7, the battery Common, the speaker return line.
B) Other circuit power grounds are traces that connect to that large pad.
Yes all these are satisfied.
C) Power to other op-amps on the board should come the appropriate + and - pads C5 and C6.
Yeah..all components are fed from the caps.
D) For the LPF, the input signal ground should connect to the gnd end of the 10k, the gnd end of the 4.7uF, the gnd end of R3 at the TDA3020 input, and then to the large pad of detail A), above; in that order
Yeah. The large pad is actually the TDA2030 common ground, which is large enough to collect back all return paths- speakers return, cap commons, and supply common. This common is connected to the common grounds of filters, and the input ground goes to the grounds of the filter components as you specified. [the filters are actually one board while the Amps make up another, a total of two boards whose grounds are connected by a thick wire.]

So the trap is the wirings from the board to the IC amps?
Because, as I mentioned in #1 I have a pocket 10w soundbox made of one TDA , single supply configuration, and it's not this problematic at all. It works on the same batteries, 47uF supply bypass/regulation capacitors, no R5, C8, R4 or C7. It's crystal clear and loud as hell .
Here it seems only difference is the IC's not being home.

Thanks for the help and advice..

#### jim hardy

Gold Member
2018 Award
Dearly Missed
A "single point Ground" is needed for Grounding. This consists of:
A) A large pad (if it is a pc board, or shortest practical, heavy wire if wired) that connects the common point of C3 thru C7, the battery Common, the speaker return line.
how the small tings of the earth confound the mighty....

#### PhysicoRaj

Gold Member
Woah Jim..That sounds familiar..!!
Wait, it was mentioned in the datasheet yeah , oh i've ignored that too!
Thanks for pointing that!

So with all commons in place, I just need to isolate the ground of the 22k??

How do i do it?

#### PhysicoRaj

Gold Member
How do I isolate input ground and speaker return?

Here's the layout for reference:

Thanks again.

#### PhysicoRaj

Gold Member
I assume I have to connect the supply common nearer to speaker return than the input resistor, so that the speaker return current goes into main ground as soon as possible and doesn't have a chance to enter the 22k??
I drew this conclusion:

correct me if I am wrong.

#### jim hardy

Gold Member
2018 Award
Dearly Missed

here's how they did it on their datasheet

looks like they used same terminal for both speaker return and 0V gnd . Center of their "star gnd " ?

#### PhysicoRaj

Gold Member
here's how they did it on their datasheet

View attachment 102412
looks like they used same terminal for both speaker return and 0V gnd . Center of their "star gnd " ?
Yeah. So my supply common goes directly to the speaker return point (star ground). I have this one correct right?
So this possibility ruled out.

I'm beginning to suspect wires that connect the TDA to the PC board [#11].
I could eliminate them, but I fear I'll lose my precious amps..because I don't trust my heatsinks.
What heatsink would suit my specs?

Thanks a lot!

#### jim hardy

Gold Member
2018 Award
Dearly Missed
I'm beginning to suspect wires that connect the TDA to the PC board [#11].
i missed that statement

Tom.G spotted it quite a while back.....

I hope you twisted those two power supply wires to minimize their inductance.

And, i see. you have not bypassed your big electrolytics C5 and C6 with little 0.1's C3 and C4 .
That would be important even without the long wires.
It seems silly at first but it's one of those basics they forget to explain.

Electrolytics are made by rolling up a long strip of aluminum foil into a coil and placing it inside a can.
The coil gives a small amount of inductance in series wit the capacitance.
At higher frequencies that small inductance becomes significant and the capacitor looks less and less capacitive and more inductive.
That's why you see a 100 uf paralleled by a 0,1 uf. At high frequency the 100uf disappears.
So let's try something:
place 0.1's as close as possible to the amp, tack them right across its pins
since there's no ground pin on the IC , first try just one from V+ to V-,
if no luck use two 0.1's and run a ground wire to their junction, and twist that ground wire with the V+ and V- wires.

Then save and read this capacitor application guide
http://www.cde.com/resources/catalogs/AEappGUIDE.pdf

#### jim hardy

Gold Member
2018 Award
Dearly Missed
Omitting R5 -C8 extends frequency response making power supply bypass even more important

#### davenn

Gold Member
2) C5, C6 should be in the range of 1500uF to 3300uF
this is incorrect .... they are not the main smoothing caps ... those values you give would be the main smoothing ones at the output of the bridge rectifier IF the amplifier was supplied by a Mains powered supply
These 100 - 200uF odd caps are for higher freq filtering of the power rails closer to the chip

but the supply is a battery right? Would such huge values be required?
that's correct .... no they wouldn't and for my above reason, not even if it was a mains fed supply

Dave

Last edited:

#### Tom.G

If ultrasonics are present, they will be present in amp A1 right..(High pass driver).
No, the 741 op-amp is only good to about 20KHz GBW (Gain BandWidth product). The TDA3020 is good to 100KHz at full output power.

Single cell flashlight incandescent bulbs are obsolete here. Would a white LED [array?] work?
Probably not. You can try a pair in inverse parallel. If they light, there is a problem. If they don't light, they don't supply any more information; their turn-on voltage is too high..

Here it seems only difference is the IC's not being home.
Very likely too far away from C2 thru C7. That's a lot of centimeters!

wiring from the PCB to the IC, about 15 cms length of 0.75mm thick single strand wires [for each lead]...
For this usage, high gain & high power, 1 to 1.3mm (18ga to 16ga for those of us stuck on Imperial or American measure) would be considered adequate.

R1 for both the amplifiers is a pot to vary the closed loop gain.
On the datasheet TDA3020 , see the footnote to Table 7 (page 19). "The value of closed loop gain must be higher than 24 dB." This indicates that the amplifier is unstable at lower gains and will oscillate.

Mentor

#### berkeman

Mentor
Thread re-opened after a bit of cleanup. Let's focus on being helpful to the OP please. Thanks.

#### PhysicoRaj

Gold Member
Thanks berkeman! And to all other replies also..

I'll add the 100nF and check. Thanks for those info and links Jim!

And yes, I have read the datasheet and the gain is sufficiently high.. Actually my oscillations are at higher gains.

I'll soon post the results.

Thanks everyone :)

#### PhysicoRaj

Gold Member
Hello again folks!
I've completely fixed the device going through all your valuable pointers.. and it works very well !

I reduced the lengths of all wires, added the smaller caps, placed the amp IC home, twisted the opposite wires, cleaned the trace gaps with a poker, refitted the filter and made the common grounds more clear.

My boom box is booming hard now, thanks to all your time and advice .

The only thing is the heating of the amp. Not very much, but pretty hot at enjoyable volumes. I just have to keep monitoring them and be careful at the volume.
The amp incorporates a thermal shutdown, but still I feel uncomfortable to enjoy the music blast while your heatsinks can cook your lunch.

Anyway thanks a ton all of you, it wouldn't be possible without such useful insights and suggestions, hail PF!

Regards.

"Issue with DIY music system"

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving