LM4562 disappointing in x-over circuit

Rolox

Member
2019-10-31 8:05 am
I have this cheap 2 way x-over I recently bought; there's just five dual op-amps in there; it originally came with socketed NJR NE5532 wich sound quite good

I swapped first for OPA2132 (adding 0,1uF MLCC caps on power pins to ground) wich added more details and punch; but today as I still had some used LM4562 around I decided to put them in.
I really liked LM4562 in some other devices I upgraded, but in this circuit, it sounds awful: recessed midrange with vague and unfocused voices, bloated bass.

I'm surprised because LML4562 should, on paper, fare better in a NE5532 circuit than OPA2132.

In the other devices were I put LM4562 I did, however (after a lot of reading and experimenting), opt for a different local bypass solution: one 0,1uF MLCC between power pins, under the chip, and one from V- to ground. Maybe that's the explanation.

I'm not planning to work more on that x-over so I just put the OPA2132 back in and it sounds great again. but I'm a little bit puzzled at how bad LM4562 sounded in there.
 

Mark Johnson

Member
Paid Member
2011-05-27 3:27 pm
Silicon Valley
Obtain some other dual opamps in the same packages. I suggest you buy five units of each of these duals:
  • TI NE5532
  • TI LM833N
  • AD AD746
  • TI OPA2134
  • AD LT1057
  • AD OP275
  • TI RC4580

Put the 5532's in envelope#1, put the 833's in envelope#2, etc.

Have your roommate brush all five chips in envelope#1 with nail polish color#1. The nail polish should obscure the chip logos so they are not readable. (It might be wise for him/her to snap a smartphone photo of the chip logo and the nail polish color BEFORE painting).

Have your roommate brush all five chips in envelope#2 with nail polish color#2.

Do this for all of the envelopes. Each unique chip type gets its own unique nail polish color. Maybe the little jars of paint sold for plastic model airplane builders, might be cheaper than nail polish.

Now you can perform a single blind listening test. You, the listener, don't know what chip you're listening to. You are set free from external factors like price, or company reputation, or "what other people say THEY prefer", and so forth.

Take all the time you need. Maybe cycle through all of the candidates, more than once. Maybe several times.

When you've decided which color opamp sounds best TO YOU, have your roommate tell you the manufacturer and part number. Congratulations!

If your roommate moves away, or refuses to be helpful, you can remove the nail polish using polish remover / paint thinner, + rags + toothpicks + slow methodical effort.

edit- if you're feeling all science-y, you could also include a dual opamp which is known to be terrible for audio: the LM358, with its 1.5V crossover distortion deadband. Maybe if you can't tell any difference among the good opamps, you'll at least be able to identify the terrible one!

_
 

Attachments

  • lm358_inv_5kc_4v.png
    lm358_inv_5kc_4v.png
    45.3 KB · Views: 169
Last edited:

Rolox

Member
2019-10-31 8:05 am
Shock horror, just slamming an opamp into a circuit not designed for it doesn't cause cherubs to start singing.

There's no need to be cynical, I'm well aware of possible incompatibilities; however I've done this operation a good number of times and in common circuits it can be effective without having to be an engineer.
OPA2132 works great in that circuit yet is a less logical candidate to replace nE5532, hence my post.
 
I swapped first for OPA2132 (adding 0,1uF MLCC caps on power pins to ground) wich added more details and punch; but today as I still had some used LM4562 around I decided to put them in.


When you say used, does that mean they might be faulty? Or counterfeit even?
The LM4562 should perform as good or better than the NE5532, except for higher current noise.


Its worth also checking for high frequency oscillation - different opamps can have different sensitivity to oscillation if the circuit is borderline to start with (stray capacitive coupling can be a cause of this).
it sounds awful: recessed midrange with vague and unfocused voices, bloated bass.
Could easily be due to oscillations or a burnt out chip. Gross faults are usually due to very definite problems.
 
First off I know bugger all about circuit design, not ven enough to be dangerous! :)

However I have two identical analogue mono crossovers I was unhappy with because no matter what I did I couldn't get a reliable stable stereo image with them. So I opened them up to find one populated with TL072 while the other was full of LF353s so I decided to see if I could upgrade them with better ones.
The data sheet of the LM4562 looked good and I decided to ask someone who knew a LOT more about these things than myself.
I was told to try OPA2132 instead since in that circuit the LM4562 would be prone to oscillating without some extra bypass caps.
With this in mind I suspect oscillation to be the problem here.