Beginner's blues, weird active filter behaviour

[The Lord Flashheart]

Hi all,

I have built up my first 2 way L-R filter based on the schematic on Rod Elliot's website, project 81, with component's selected to give a crossover point of around 300Hz.

12dB / Octave Linkwitz Riley Crossover

It has a few issues with phase and distortion.

The first issue is that is distortion rises as the frequency rises, please see the photos below.

At ~70 Hz the waveforms are clean though out of phase by something like 3ms.

At ~1kHz the output from the high pass section start to show distortion, looking a little like crossover distortion but offset to the negative rail slightly.

At ~20Khz the output from the high pass section looks terrible and the low pass section shows weird spike things in the waveform.

The Op amps I used were TL071s, resistors were hand matched 1% metal film and the capacitors hand matched polyester film types.

What gives, is it the breadboard or have I missed something?

 

[richie00boy]

What are your component values? Looks like you are overloading the opamps. Also maybe lacking decoupling capacitors at supply pins.

 

[The Lord Flashheart]

My component values are all as per Rod's original design, i.e. 470nF for the filter capacitors and 11.2K for the filter resistors.

I have no load on it at all, just connecting the probes directly to the outputs via jumper wires.

I asked the question about decoupling capacitors and the consensus seemed that they weren't necessarily needed for TL071s.

I've decoupled evey IC's supply pins with 103 (10nF I think) ceramics, I only had 2kV high voltage ones if that makes a difference but the problem remains.

I've attached a photo of the circuit, it seems a bit messy I know...

Red and Black are +V and -V, green is earth, white is signal between stages, blue is filter to output buffer, yellow is feedback and orange is output.

decoupling caps are the little blue ones.

 

[The Lord Flashheart]

I'm not sure what I did exactly but aas I was checking connections I briefly got a clean sinewave on the 'scope but with a DC offset of -around minus 12V?

I can't seem to replicate it...

 

[Osvaldo de Banfield]

Try spreading the components away in the proto, perhaps too much inter-coupling through the proto capacitances.

 

[The Lord Flashheart]

Hi Osvaldo, thanks for chiming in.

I shall try that tomorrow and report back.

Would it be worth building each op amp "block" on a separate small bread board and then linking those together or are you saying I have over done it with the coupling capacitors?

 

[Osvaldo de Banfield]

As it is audio and no high currents, and no high voltages, taking some components some centimeters away as to prevent capacitive couplings between them and trough the proto inter-capacitances, perhaps it will perform OK.

What happen with the other tread about imbalance in pos and neg cycles of the amp?

 

[The Lord Flashheart]

That amp is on the shelf for the moment, I need to regroup and have another go at it. I got distracted by DSPs and active crossovers...

 

[cbdb]

Check the signal on the output of the first opamp, and scope the rails at each opamp to make sure your power supply is working properly.

 

[Osvaldo de Banfield]

OK. Enjoy prototyping. Its interesting. Me the weekend will try an interesting tube circuit: I will try to make an 6AL11 double pentode compactron behave like a 40106 or 74HC14. Let's see what happen.

 

[Mark Tillotson]

Something's definitely wrong, hard to figure out with that cramped layout, but you have a lot of loops of wire talking to each other above the breadboard which isn't great.

The rails need bulk decoupling to ground somewhere, and your ground wiring needs to run linearly along the signal chain.

You didn't mention what your power source was nor what the driving signal was from or what source impedance.

Check every IC is getting both power rails at the chip itself (you'd be surprised how well a circuit can behave with a power pin floating!).

 

[Douglas Blake]

The first thing I would suggest is that you spread your parts out a bit and tidy up those wires. Haybale wring on protoboards is just as bad as in real products.

Try to organize your layout for easy troubleshooting and measurement.

 

[The Lord Flashheart]

Check the signal on the output of the first opamp, and scope the rails at each opamp to make sure your power supply is working properly.

I checked the power supply and it works as it should, perhaps a more bit noisily that it should thinking about it.

Probing the signal between the stages seems to show it gets progressively worse rather than it going haywire at a particular stage.

 

[The Lord Flashheart]

OK. Enjoy prototyping. Its interesting. Me the weekend will try an interesting tube circuit: I will try to make an 6AL11 double pentode compactron behave like a 40106 or 74HC14. Let's see what happen.

Very cool, I don’t what that is and googling has left me more confused but it sounds cool!

I’ve been trying to avoid getting into valves, I have only one valve amplifer (a set of ancient Quads) buy so far have worshiped at the altar of linearity and hence have avoided getting any more.

I do like the way they look though...

 

[The Lord Flashheart]

Something's definitely wrong, hard to figure out with that cramped layout, but you have a lot of loops of wire talking to each other above the breadboard which isn't great.

The rails need bulk decoupling to ground somewhere, and your ground wiring needs to run linearly along the signal chain.

You didn't mention what your power source was nor what the driving signal was from or what source impedance.

Check every IC is getting both power rails at the chip itself (you'd be surprised how well a circuit can behave with a power pin floating!).

Thanks for your thoughts Mark.

The layout is cramped as I rather childishly hoped to build two channels in the breadboard, cut up a RCA cable, plug it all in and test it in the living room that afternoon...

I shall rebuild it spaced out and test it again to see.

The wires are the only ones I have so far, I found this breadboard/project lab thing on eBay and ordered a generic set of wires as I had no clue otherwise.

It’s a Knight portable mini lab with built in power supplies, signal generators and some digital stuff I haven’t a clue about. I thought it would be a perfect compact thing to do electronic stuff on that can be closed up and packed away to avoid incurring the wrath of the Better Half at home, is it possible that all the stuff under the breadboard is interfering with the circuit?

The power supply I used is the variable LM317/338 supply built into the box. It seems ok on the oscilloscope, better than the other power supply built into the thing, being TO-3 based LM7812/LM7912 +- 12V. I Set the variable power supply to +- 15V for the test above.


When you say bulk decoupling somewhere, does that mean an electrolytic capacitor of some description and if so how would I go about calculating it’s value?

When you say the ground should run linearly along the signal chain, again please forgive my ignorance but what does that translate to in terms of wire routing?

The source is a XR2206 based signal generator kit with some improvements based on stuff I found on YouTube about removing the DC offset from the output with a 1uF electrolytic cap on the output and adding trim pots to adjust the sine wave. I can’t hazard a guess at its input impedance I’m afraid.

I’ll stop trying to fit two channels on one board and rebuild it with short lengths of hook up wire instead of the long loop wires and see what happens.

Thank you for your help.

 

[The Lord Flashheart]

The first thing I would suggest is that you spread your parts out a bit and tidy up those wires. Haybale wring on protoboards is just as bad as in real products.

Try to organize your layout for easy troubleshooting and measurement.

Fair comment, will do.

 

[Mark Tillotson]

Bulk decoupling is two electrolytics, say around 100µF, one from +ve rail to ground, one from -ve rail to ground, correct polarity of course. This will stiffen up the supply rails at audio frequencies.

 

[Osvaldo de Banfield]

And? Any improvement(s)?