upgrading active xover with LM4562

[bzhei]

hi guys,

newbie here. been reading through the forums about Op-amp ugrades and i had my first try a couple of days ago. i used LM4562 as the upgrade.

i have an old 2-way active crossover lying around, when i opened it up i found 10 op-amps:

3 op-amps (2 ba4560, 1 ba4558) in the input section
2 op-amps (2 ba4558) in the filter section
2 op-amps (2 ba4558) in the highpass output section
2 op-amps (2 ba4558) in the lowpass output section
1 op-amp (1 ba4560) in between the powersupply and audio section

been aware of oscillation and dc offset issues, i decided to swap op-amps per section so i can concentrate on a small area should a problem arise.

i swapped out the lowpass section and found 1mv dc offset. i've read anything less than 20mv is acceptable??? then i played music low to high volume, no problem.

then i did the same thing to the highpass section, no problem.

then the one between the power supply and audio section, no problem again.

then the filter section, still no problem.

it is in the input section where the problem starts. since there are two kinds of op-amps, i decided to swap the ba4558 first. no problem there. but when i swap the ba4560, it still plays fine until a particular portion of a particular musical track (michael ruff's I will find you there) causes the entire crossover to 'lock up' and just oscillate.

the rest of the tracks on that cd won't cause the oscillation. i soldered the ba4560 back and the problem goes away.

the two 4560 op-amps are the first two op-amps just after the RCA inputs and they were both configured to have a corner frequency of 21kHz.

all the caps are still stock but i placed 0.1uf ceramics on the +/- pins and + to gnd and - to gnd, nothing helps the 4562.

i would really appreciate any help/suggestions offered

thanks!

[bzhei]

btw, it doesn't oscillate 3 levels down the max unclipped volume of my Alpine 4v headunit. 32/35.

i did my own troubleshooting by myself but i don't know if i'm in the right path.

i threw in a sine-wave from my function generator and found out that a frequency with 3.2volts starting 5,996kHz to beyond causes the crossover to 'lock up' and oscillate.

once it locks up, it oscillates @ 2.5kHz peak

funny thing is i was probing the output leg of the 4562 opamp and accidentally shorted the output with the inverting input. the lock up went away momentarily. so i tried to purposely short the output with the inverting input and it plays normally again until i 'unshort' it.

[bzhei]

i mean 5,996Hz and beyond

[Randy62]

Maybe a defective 4562.Have you swapped it out with another?I have never seen this with a 4562 and Ive modded a few car crossovers with them.

[bzhei]

hmmm.... i may have to give it a try. haven't really thought about swapping it since it plays well with the rest of the cd's tracks. i'll give it a shot tomorrow.

btw, mine's an Audiocontrol 2xs, i figured it's a great unit to start with. what xovers have you modded before?

thanks.

[Mooly]

You say it's oscillating... I assume you are looking on a scope to see what is happening ?

2.5Khz is low frequency for opamp instability.

10 opamps ? the 4562 draws far more current than the 4558/60. Look at the supplies ! If any opamp is fed resistively or if they are all fed of just a 78L12 type supply then if the opamps drive a low impedance filter network "hard" it may just draw enough to pull the rails down or modulate them.

[bzhei]

Originally Posted by Mooly

You say it's oscillating... I assume you are looking on a scope to see what is happening ?

2.5Khz is low frequency for opamp instability.

10 opamps ? the 4562 draws far more current than the 4558/60. Look at the supplies ! If any opamp is fed resistively or if they are all fed of just a 78L12 type supply then if the opamps drive a low impedance filter network "hard" it may just draw enough to pull the rails down or modulate them.

hi mooly,

here are the scope readings i took a while ago. maybe you can spend a little time helping me analyze it? that's the best wave resolution i can get out of my scope (kikusui 559a)

i'm using 8kHz tone (test tone cd) loaded from my alpine headunit. i'm getting my readings from the crossovers highpass RCA output gradually increasing volume level.

@ 0.281v


@0.671v


@1.724v


@2.307v


@ "locking volume", not headunit's max volume.


notice the frequency is now 2.5kHz even if i'm playing a 8kHz tone and the voltage is nearly 6volts when my headunit's max volume is around 4volts give or take.

here's the reading @ 1kHz. waveform is not very clean but i'm able to push max volume without "locking" the crossover.


and here's a photo of the power supply section. i'm not really good at this but i've been probing around it when i read your message.

readings with ba4560 installed at input stage


readings with lm4562 installed at input stage (photo still has 4558 and 4560)


the output at the diodes are strangely high with the 4562 (+/-18v) compared to +/-15.53v with the 4560.

thanks.

[Mooly]

Quite difficult to analyse without actually seeing it all for real.

In your first post you say it all seems OK until one particular opamp is replaced.

So we need to be methodical and measure (and scope) the power supply to that opamp and all the others as a first check. Lets see what that shows. The LT1072 is a low power switching PSU chip. It might be going into some form of limiting with the extra loading.

Connect the scope and measure from ground to Pin 8 of the opamps and then from ground to pin 4. The scope should show a straight line (DC) and if you AC couple the scope there should be minimal ripple on the rails. Anything over say 100 or 200 mv pk/pk may be enough to start causing problems. Ideally it should be way below that at only a few millivolts.

That has to be your first check.

Edit, you say the rails seem high with the LM4562 fitted. If that is being read on a DVM then it's possible that noise on the rails is giving a false reading. This is why we have to scope the rails to see what is going on.

[bzhei]

Originally Posted by Mooly

Quite difficult to analyse without actually seeing it all for real.

In your first post you say it all seems OK until one particular opamp is replaced.

So we need to be methodical and measure (and scope) the power supply to that opamp and all the others as a first check. Lets see what that shows. The LT1072 is a low power switching PSU chip. It might be going into some form of limiting with the extra loading.

Connect the scope and measure from ground to Pin 8 of the opamps and then from ground to pin 4. The scope should show a straight line (DC) and if you AC couple the scope there should be minimal ripple on the rails. Anything over say 100 or 200 mv pk/pk may be enough to start causing problems. Ideally it should be way below that at only a few millivolts.

That has to be your first check.

Edit, you say the rails seem high with the LM4562 fitted. If that is being read on a DVM then it's possible that noise on the rails is giving a false reading. This is why we have to scope the rails to see what is going on.

mooly,

when i read your message i already reinstalled the ba4560 @ the input section. i was checking the output and there was oscillation but not as severe as with the 4562. i'll check the rails as soon as i reinstall the 4562 but here's the data i got with the ba4560 at the input.

DC - straight line
AC ripple @ diode output (+/-) - 85.5mVp-p
AC ripple @ op-amp (+) - 37.05mVp-p
AC ripple @ op-amp (-) - 42.75mVp-p

here's a photo of the waveform magnified.

negative rail


positive rail



i'll try to reinstall the 4562 today and check ac ripple with them installed

thanks.

[Randy62]

I also did this to an AC 2xs.However I did not swap out the low pass IC's.They were only being used from 333hz and down so I did see much sonic benifit from doing so.
After my first post I started thinking the same thing Mooly brought up.
The other ones were an AC Four.1 and an LA sound.
I had to rework the power supply on the LA sound because it was being pulled down to +/-5volts.However there were 17 LM4562 op amp's in the circuit.
I had no problems with the Audio Control units.