I've built a few electronics projects recently (mainly LED based stuff) and had good results but this latest one has me a bit stumped. I'm still very much learning and this is my first audio project.

I have built a CMOY type amplifier based very much on Tangent's tutorial, using at the moment an NE5532 Op-amp (mainly so I don't damage the more expensive OPA2132 I have. I have some very large 0.47uF 250v Metallised Polypropylene Capacitors as Input caps in the circuit.

I first built the circuit on breadboard with the standard 10x Gain 1K resistor on the Op-amp and after some silly issues I got it working and it sounded awesome, going very loud and without any distortion.

Moving on I then built the same circuit on Matrix board, and while I made a bit of a mess of the track joining (never used Matrix board before) the circuit itself looks fine but is presenting me with strange issues. With the input capacitor in position the sound is totally messed up at anything but minimal volume, and even then could just be so quiet I can't detect the distortion.

However if I stick my fingers on the bare LED wires and touch the Op-amp tracks the sound magically clears, and returns to distorted shortly after I remove my hand.

At first I'd thought I'd destroyed the polyprop during soldering but after trying several different capacitors and they all seem to distort the sound.

I've also found that removing the input caps from the circuit and shorting the connection clears the sound up.

I've tried swapping the Op-amp for another with no change.

I've also tried adding in two more electrolytic caps in parallel with the existing power supply capacitors and that has had no effect. I can see the voltage dropping on these once I switch the power off.

I've added bypass 104 ceramic caps to the +v and the -v, both going to Virtual ground, with no effect.

I've cleaned all traces of flux residue away with good IPA and a toothbrush.

I've also checked resistance from the 8 pin IC socket and they are equal to each other and correct as per schematic.

I also noticed that when the input capacitor was in place the voltages on the 'rails' were massively unbalanced. On a 12.4v source (regulated) it was 10.7v on the one side with 1.7v on the other, yet when the circuit it no fitted with the Op-amp it's more or less dead centre at 6.1v/6.3v.

Ideally I'd like the Input cap to close down any DC-offset as I'd hate to destroy my headphones.

Source at the moment is an iPod Touch with a Fiio E1 at low volume.

Any insights as to why when I have the input cap fitted and touch the LED and the rear of the Op-amp contacts that the sound clears?

I can take pictures if that would help but be warned the soldering is pretty bad in places and I've pulled a few tracks off and slaved resistors and caps in place trying to fix the issue.

Thanks.

 

This may or may not be related to your problem, but I question the way you are generating your ground. The use of a voltage divider with 4.7 k resistors is ok if you are only tieing ground to the opamp input, but you are drawing significant current through R+ and R-.

As a quick test short the ground to the common point between your two batteries. If this clears the sound up, then you should change the grounding scheme. Actually, I recommend you change it anyway.

 

If the soldering is "pretty bad in places", then you really should start over.

Apparently, you got it working fine on a breadboard, so all of your components were OK.

Then you tried to build it on a matrix board, and it didn't work properly anymore.

Poor soldering technique has ruined many projects. Soldering isn't intuitive, it takes a good bit of practice and the right equipment/materials to do a good job.

Cold solder joints are physically weak and have high resistance.

Cleanliness is very important. Use isopropyl alcohol (90% or better) and a small nylon brush (acid brushes from the auto parts store works well) to clean your components before soldering. If you touch a parts' leads, your finger oils will prevent solder from flowing properly. The isopropyl will remove the finger oils.

 

steveb brings up a very good point.

If you remove R1+ and R1-, you will need a DPST switch to replace your power switch SW, as you will need to break/make the circuit for both batteries.

Since this is a battery operated circuit, creating a "virtual ground" using another opamp isn't a good option.

 

The symptoms you describe would seem to indicate that there is no DC path for the non-inverting input bias current. Which says that R2 is not connected in some way (mis-wired, bad connection, etc.)

 

Thanks guys for your input, it is much appreciated.

I am using a 12.4v regulated bench power supply for the purposes of building and will probably go with a 20-24v power source with regulator in the final design. I'm looking for this amp to be desk based and don't need it portable.

I think I will give it another bash, this time on stripboard with which I am more familiar. The matrix soldering was fine until I tried to join pads together.

Tried out the OPA2132 Op-amp in the circuit and that works fine, good sound depth and volume goes pretty high. That's only on iPod earbuds though. Measured output dc offset as 8mv/20mv even with the input caps in place. I'm no audiophiles but it sounds pretty good with the 2132. A bit like the 5532 sounds when I grab stuff with my fingers or short the input caps.

Will test using my Ns-1000 headphones later as they are higher impedance and should give me some indication of performance.

What I did notice with the 2132 was a little hum during the quiet bits which I didn't notice with the 5532, although I will have to check that again later.

 

Wired it up to two batteries in series and still the 5532 misbehaves but the 2132 sound good, despite the soldering and caps hanging off loose wires along with the R5. The bass is a lot more pronounced than I expected but isn't obtrusive.

I'm still going to make it again on stripboard and perhaps use this board to do a few experiments with different resistors and things, or just do that on breadboard.

With the 2132 the voltage division is much more normal and within half a volt with my NS-1000 headphones being driven, putting the 5532 back in still throws the rails very unbalanced and distorts the sound.

Still think it's strange that it works pretty well with the 2132 and so appallingly with the cheaper 5532 amp without shoving my fingers on it.

 

Still think it's strange that it works pretty well with the 2132 and so appallingly with the cheaper 5532 amp without shoving my fingers on it.

Have you used an O-scope to view the waveforms? What you are describing sounds like an unstable oscillation with the 5532 amp. Your fingers can act as capacitance that stabilizes the circuit.

 

Why are you using huge high voltage wrong value input capacitors? I use Epcos metalized plastic film 63V capacitors for coupling audio and they have a 0.2" lead spacing and are not thick and not high.

 

I haven't got an oscilloscope (only just started with electronics as a hobby) so can't check that but I took your advice from earlier, SteveB and shorted my virtual ground to the middle of the batteries I'd just fitted instead of the 12w power supply and it immediately cleared up.

So what causes oscillation in circuits and how can I avoid this in future?

Thanks.

 

As for the input caps they were the closest I could find to Tangent's parts list, being 0.1, 0.47 or 1uF. Places like Farnell and RS components here in the UK attract large postage charges and large minimum order quantities, making a little project like this prohibitive in cost. The place I got them from only had these large polyprops(and they are truly huge taking up more space than everything else combined. I figured that capacitance and dielectric material was more important than voltage rating. Was I wrong to assume this?

Anyway I do have 2x 474k film capacitors out of an ADSL splitter which I can use.

 

Hi AudioGuru,

I'm afraid I was unable to source lower voltage polypropylene capacitors easily. RS Components and Farnell were about the only options here in the UK and expect people to buy large multiples, have large postage charges (for small components) and minimum order values which didn't sit well with me as a hobbyist.

Tangent's guide suggests that polyprops are the preferred film cap to use so I went with them. I was quite surprised by the size of the capacitors and they do take up more space than the rest of the circuit combined but I figured that capacitance was the primary value to get right and the voltage didn't really matter. Was I not right to do this with regard to audio?

 

So what causes oscillation in circuits and how can I avoid this in future?

Thanks.

That's a big question that we all have to deal with, and it challenges us all. The simple answer, in terms of theory, is that positive feedback causes the oscillations. You may use negative feedback on the opamp, but at high frequency, phase shifts can mount up, and if they get to 180 degrees within the circuit bandwidth, you are likely to have oscillations. There is an old joke about how our amplifier designs often turn out to be oscillators, and our oscillator designs often turn out to be amplfiers.

In practice, some common things that apply to your type of circuit are as follows:

1. Poor power supply decoupling (solution is more capcitance or other filtering on the power supplies)

2. Poor grounding (as you saw in your case)

3. Too much load capacitance

4. Stray inductance due to long leads and wires

There is a nice little trick you can use with opamp circuits like yours when any of the above things happen. You need to be careful applying it because you may inadvertantly limit the bandwidth too much. Place a very small capacitor (10-10000 pF) from the negative opamp terminal to the output pin of the opamp (not the signal output, but the opamp output pin itself), and use a very small resistor (1-100 Ohms) for the "optional" resistor in your schematic. This is a form of Opamp compensation that reduces the bandwidth and helps to squash any high-frequency oscillations.

Note that it's better to make a very good layout with good design and good construction practices, but the real world often messes up even the best intentions. The above trick can be helpful, if not used to excess. It also provides a way to deliberately make a lower bandwidth opamp out of a higher bandwidth opamp, in case you have a fast one, but deliberately want to limit the bandwidth for some reason.

 

Tangent's guide suggests that polyprops are the preferred film cap to use so I went with them. I was quite surprised by the size of the capacitors and they do take up more space than the rest of the circuit combined but I figured that capacitance was the primary value to get right and the voltage didn't really matter. Was I not right to do this with regard to audio?

I worked with PRO-audio equipment manufacturers who used 63V poly film capacitors for coupling audio and I also use them. They were small (like I showed) not the huge high voltage ones you found. You will not hear and will not measure the difference between a polypropelene or polyester film capacitor.

As for the input caps they were the closest I could find to Tangent's parts list, being 0.1, 0.47 or 1uF.

The value for the project is 0.1uF but yours is 4.7 times too high at 0.47uF which also make them bigger.

 

Right thanks Audioguru I will look and see if I can source some of those smaller 63v ones for a reasonable price. I need some DIL sockets anyway as the ones I have are terrible.

I was never really looking to make a pocket amplifier. This amplifier is and will always be to sit on my desk and work with my PC and headphones. I have a Fiio E5 for portability, although I may build a pocket amp just for fun if I can find some of those small caps.

@Steveb - Many thanks for that explanation and for that trick, which I think I've seen on a few Cmoy esque circuits around the internet but decided to ignore at the time.

I will have a poke around with my circuit tomorrow and see if I can find that ground issue. I will build a new circuit this weekend too if I get the time.

 

This excellent and small capacitor is available at the American branch of Farnell called Newark (but now it also says Element 14):

 

Ah brilliant I checked and the place I got the last lot of stuff from does poly films too.

http://www.bitsbox.co.uk/polyfilm.html

Wish I'd checked them out now as I could have saved postage. No matter. I will grab a few items to go with them.

Thanks.

 

I started another board, this time with stripboard and it had the same grounding issue when used with the 0.47uF and the NE5532. This one was soldered meticulously with good solder and separate flux and I used IPA to clean everything including the iron.

So I created a more stable virtual ground power circuit with the LM741 just to be sure and now just use the OPA2132 and it's pretty good at keeping it evenly balanced. The sound is better than the Fiio E5 so I am happy.

I am still using the 0.47uF mainly because I wanted to use them as they weren't cheap and they are large and yellow so look better.

I've still got the initial (not done so well) circuit so will probably do some further investigation with the 100nF poly box caps and the NE5532 just to see if that would have worked. I only need one headphone amplifier but have enough parts here to build about five of them.