Linkwitz-Riley low pass, strange distortion?

[DionG]

Hi all, so im trying to build the low pass filter part of the following design of Linkwitz-Riley Electronic Crossover .



so far I have built the bit outlined in red so i decided to test it on an oscilloscope with a wave generator and i seam to get generally the right shape but its got lots and lots of high frequency stuff in there. again the input is a clean 85hz. the output looks like below


edit: shocking drawing i know

so yea, any ideas what the hell is going wrong? or is it something simple i just don't know?

its using 1% metal film resistors and 1% metalized film polypropylene film capacitors and NE5532.
also, its using 33nf and 40Kohm on the low pass section to cross at 85Hz (which it does respond to when i change the frequency)

any help is great
Dion

 

[infinia]

and the input looks clean? can you trigger on the noise freq? could be a lot of things, op-amp unity gain stability, oscillating not enough PS bypassing, grounding on the probes, or SMPS common mode noise. is it an analog scope with a high BW? measure yer ground and supplies.

 

[DionG]

...yes haha
umm input is crystal clean.
as for the rest i have no idea what any of it means.
if it helps, i took an output right after the input buffer and its the same.
could if have anything to do with ground loops or anything. also its power by a lap power supply (unfiltered at them moment). noisy? and the ground of it is connected to the ground of the input and output. are they supposed to be connected? when they weren't there seamed to be some other problem. the occiliscope was has a wave passing through it doing other weird stuff

im just lost haha

 

[jan.didden]

...yes haha
umm input is crystal clean.
as for the rest i have no idea what any of it means.
if it helps, i took an output right after the input buffer and its the same.
could if have anything to do with ground loops or anything. also its power by a lap power supply (unfiltered at them moment). noisy? and the ground of it is connected to the ground of the input and output. are they supposed to be connected? when they weren't there seamed to be some other problem. the occiliscope was has a wave passing through it doing other weird stuff

im just lost haha

Probably switching noise from the laptop supply through the grounding.
This is not a circuit problem but a measurement/grounding issue.
You really need a good linear supply.
Try a couple of 9V batteries, I bet it's clean then.
Also try to move the ground point from the ground lead from the supply to different points in the circuit. And ohh, yes: put the scope probe ground clip to the same point as the supply ground lead!

jan didden

 

[DionG]

so to get this right, use some batteries or regulated voltage supply and connect the ground on the supply to the ground on the scope but don't or do connect it to the ground on the input wave??

thanks, then i'll test it and give results

Dion

 

[DionG]

WIN!!!
so i filtered the lab power supply with some 33nf caps that i was using for the crossover and it turned the output signal clean.
thanks for the idea Jan

there is however a noticeable distortion at the peaks of high frequency waves ( maybe the power just isnt regulated enough. As in that's what im assuming until i test it with some actual power regulator chips).

Also, i had all the grounds connected together (including the input, power supply and output ground). is that the best thing to do?

thanks as always
Dion

 

[jan.didden]

WIN!!!
so i filtered the lab power supply with some 33nf caps that i was using for the crossover and it turned the output signal clean.
thanks for the idea Jan

there is however a noticeable distortion at the peaks of high frequency waves ( maybe the power just isnt regulated enough. As in that's what im assuming until i test it with some actual power regulator chips).

Also, i had all the grounds connected together (including the input, power supply and output ground). is that the best thing to do?

thanks as always
Dion

So it seems the opams were oscillating due to poor supply decoupling. It is good practise to have a local decoupling cap on each supply line physically close to each opamp, like a pair of 0.1uF. Don't use high-Q foil caps here; you want some losses to quench any oscillation tendency. Ceramics are ideal here, or some 1-10uF electrolytics.

Edit: that distortion on hf waves, what freq and level are you talking about?

jan didden

 

[DionG]

Jan,
yea i'll put some electrolytics in there from now on to help that. and it will be from an actual regulated supply and not just lab power supply lol.

umm not sure of an frequency, i can find out if you like although personally i thinks is just the crap supply as i was only using 33nf polypropylene ones, and therefore couldn't stop the lower frequency stuff. it was putting out about +- a volt on the test.

cheers
Dion

 

[DionG]

Hi, to anyone who would like to help, i need to get the +- 15v for the preamp but i only have a +-34v DC at the moment. as the LM7815/7915 can take +-35v i was just going to put it straight into the regulators (with smoothing caps on both sides of course).
I was just wondering if you think that will be ohk or whether i should do something else? it wont giving much power as it only goes into the opamps and into the LM3886 amp board.

cheers
Dion

 

[infinia]

You could use series R in front of each Vreg to take off some voltage and heat.

Maybe in your case, R could drop about 15V. R~ 15V/n*Iq , where n = # op amps , Iq = quies. current of each op amp. (note add a couple of mA to the total for the regulator part as well). Calculate the power dissipated for the R and the regulator, making sure each part can stand the heat with their power ratings.

add a shunt cap (10-100 uF) after the new series R for a noise filter and to keep the input of the Vreg stable.

 

[DionG]

Infinia, thanks for the idea, I will look into that more also i have a new problem. keep in mind that this is built on veroboard and is using a laboratory power supply as i haven't got my regulator chips yet. In the preamp i have installed 2.2uf aluminum electrolytic caps on the power supplies ring next to each op amp. up to about 500hz, the output signal 'looks' clean but up at 800hz (when the oscilloscope zooms in due to it being a low pass filter), you can see about a 20mv distortion. I think it would be about a few KHZ. my question is, do you think this is still the power supply and will be fixed when i give it a proper regulated supply or you think it might go deeper than that? thanks as always Dion edit: could crossing signal wires underneath the board make this much of a difference?

 

[infinia]

The way I build these things, 1) arrange the op amps on the board so signal flow is logical, keeping all input wires short as possible and away from their outputs. 2) wire the op amp power 1st, using twisted pairs of wires to each DIP package with small ceramic caps 0.1 uF located close to each of the IC +/-pins. I guess you could try small elecro's here as long as their physically really small, but would use the ceramics. Each of the twisted power wires tie back to a common pair of small caps 47 ~470 uF. Also this point will be at the output of the local V regs. kind of like, star power routing. This should make for a stable build keeping oscillation and feedback in check.

 

[DionG]

thanks infinia. if i cant get it to work better (ie actual regulators etc) i'll have to have a go at wiring it with less crap underneath the veroboard. Dion

 

[Soundminded]

Hi all, so im trying to build the low pass filter part of the following design of Linkwitz-Riley Electronic Crossover .

View attachment 218601

so far I have built the bit outlined in red so i decided to test it on an oscilloscope with a wave generator and i seam to get generally the right shape but its got lots and lots of high frequency stuff in there. again the input is a clean 85hz. the output looks like below

View attachment 218602
edit: shocking drawing i know

so yea, any ideas what the hell is going wrong? or is it something simple i just don't know?

its using 1% metal film resistors and 1% metalized film polypropylene film capacitors and NE5532.
also, its using 33nf and 40Kohm on the low pass section to cross at 85Hz (which it does respond to when i change the frequency)

any help is great
Dion

It looks like you've built an excellent RC oscillator. The high and low pass sections may be interacting with each other. Disconnect the high pass section and see if it goes away. If it does, you need to connect it to a separate op amp in parallel with the input where it does not have a common point at the output of the input buffer op amp. If it doesn't go away, then the circuit for the LP section is an RC oscillator itself. Just a guess.

 

[DionG]

hmm, my first attachment shows Ive only built the low pass section of the circuit, there is no high pass. ...