Which opamp where?

[gnd]

I'm working on upgrading my Behringer cx2300 crossover, which I use to split signal at 90Hz between my studio monitors and subwoofer.

Now there are BE037 all the way for audio signal, and it sounds too harsh on hi freq range for my taste.
I read that BE037 are actually TL074, and that those sound more harsh compared to 5532. So I thought I'd swap some ICs.

My plan is to make small boards with two NE5532 on each, with pins which would fit into BE037 (TL084) footprint. (Or maybe with four NE5534, lets see...) I plan to bias NE553* outputs into class A, and I may play arround with some other opamps, to see the effect. Therefore I want to stick with dual/single ICs, and avoid quad ICs. Probably I'll have to upgrade power supply, but no problem, I can make external one.

I'm just not sure if replacement of all BE037 is proper. They are FET, I guess, like TL074(84), and NE5532 are bipolar.

Crossover basically has three sections:
1. inputs and hi-pass filter
2. crossover filters
3. outputs with delay, gain, limiting and balanced output

Here is link to schematics, if it helps.
http://84.255.203.119/Berhringer%2520CX2300C%2520Service%2520Schematics.pdf

Are maybe TL074 FET better in some positions?

Outputs may be better driven by NE5532, is it? And filter (at least hi freq range) I'd also prefer to change to NE5532, or possibly opa2134. How about line inputs?

Any suggestions?

Is FET vs bipolar of any significance in such line level circuits? Besides turning arround some polarized capacitors because of changed polarity of offset voltage, is there anything else that I should be aware of?

thnx

 

[JohnRoberts]

[quote author="gnd"]I'm working on upgrading my Behringer cx2300 crossover, which I use to split signal at 90Hz between my studio monitors and subwoofer.

Now there are BE037 all the way for audio signal, and it sounds too harsh on hi freq range for my taste.
I read that BE037 are actually TL074, and that those sound more harsh compared to 5532. So I thought I'd swap some ICs.

My plan is to make small boards with two NE5532 on each, with pins which would fit into BE037 (TL084) footprint. (Or maybe with four NE5534, lets see...) I plan to bias NE553* outputs into class A, and I may play arround with some other opamps, to see the effect. Therefore I want to stick with dual/single ICs, and avoid quad ICs. Probably I'll have to upgrade power supply, but no problem, I can make external one.

I'm just not sure if replacement of all BE037 is proper. They are FET, I guess, like TL074(84), and NE5532 are bipolar.

Crossover basically has three sections:
1. inputs and hi-pass filter
2. crossover filters
3. outputs with delay, gain, limiting and balanced output

Here is link to schematics, if it helps.
http://84.255.203.119/Berhringer%2520CX2300C%2520Service%2520Schematics.pdf

Are maybe TL074 FET better in some positions?

Outputs may be better driven by NE5532, is it? And filter (at least hi freq range) I'd also prefer to change to NE5532, or possibly opa2134. How about line inputs?

Any suggestions?

Is FET vs bipolar of any significance in such line level circuits? Besides turning arround some polarized capacitors because of changed polarity of offset voltage, is there anything else that I should be aware of?

thnx[/quote]

5532 and 074 are both fair performers when applied correctly. The 074 is slightly faster, the 5532 will drive lower impedance loads and has slightly less input noise voltage (but higher noise current) so may be quieter in some low impedance applications. Neither is clearly better or worse than the other, and if hanging resistors off the output stage made a useful difference the IC manufacturers would have done that decades ago..

Perhaps try modifying a single circuit block and listen to the results. Then judge for yourself what sounds better.

JR

 

[tomcat]

http://www.groupdiy.com/index.php?topic=13594

 

[gnd]

Here is an interesting comparison too:

http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm#mult

 

[JohnRoberts]

[quote author="gnd"]Here is an interesting comparison too:

http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm#mult[/quote]

072 are rated to drive 2k loads, elevated distortion plots at lower resistance are not very surprising.

JR

 

[gnd]

[quote author="JohnRoberts"]
072 are rated to drive 2k loads, elevated distortion plots at lower resistance are not very surprising.

JR[/quote]

Still, comparing to 5532, in 072 distortion rises with frequency, from 1k HZ above. In any case there is much higher distortion on highs with 072. And slew rate of 074 is higher, and that combined with higher distortion, maybe thats why some say, that it sounds more harsh that 5532.

I checked the RANE site, they have similar crossovers with schematics. Only they use mc33078, but filter section they do with TL072.

Schematics is here: http://www.rane.com/pdf/ac22bsch.pdf

Why do they use TL072 in filter section, and 33078 elsewhere? Any ideas?

thnx

 

[Samuel Groner]

Why do they use TL072 in filter section, and 33078 elsewhere? Any ideas?

FET inputs have low input bias and noise currents which can be important in filter design. IMO a TL072 is cool for a simple LED driver but has nothing lost in a pro audio signal path nowadays as there are much better (but pricier) FET opamps out there.

Samuel

 

[gnd]

Samuel:

FET inputs have low input bias and noise currents which can be important in filter design. IMO a TL072 is cool for a simple LED driver but has nothing lost in a pro audio signal path nowadays as there are much better (but pricier) FET opamps out there.

What would you recommend as good (but not too esoteric and expensive) replacement for TL072 FET opamps in filter section?

MT:

And expanding on the flip side of what Sam said the MC33078 is not FET input but PNP bipolar, lower noise and has higher current drive capability than the 072.

What I dont like about 33078 is their high distortion, increasing with frequency. http://www.dself.dsl.pipex.com/ampins/webbop/33078.htm It just grows from 300Hz on. NE5532 is much cleaner in this regard. Otherwise they are both bipolar. I ordered some MC33079 quads, just for comparison. But that increasing distortion with frequency is really not encouraging to me. NE5532 seems much more appealing. What do you think?

 

[Samuel Groner]

What would you recommend as good (but not too esoteric and expensive) replacement for TL072 FET opamps in filter section?

I'd go with OPA4132 or OPA4134 (I think you need the quad version, right?). Note the PSU requirement notes by RogerFoote, though I'd guess that for an xover a few mA more won't hurt--consoles are different obviously.

Samuel

 

[gnd]

Samuel:
Regarding quad opamps, I'm seriously thinking of making adapter boards to use duals. In my crossover quads are all mixed by usage. Like filter section is one half on one IC, and other half on another. Duals will give me more flexibility.

Roger:
Regarding PSU, I will first mount those LM7*15 on aluminium profile, and see if that is enough. NE5532 take some four time more power, is it. Anf if I bias them into class A, it will be more. But still, LM7*15 can give out 1A current, so that should be enough for 24 pieces of NE5532, I guess. When short circuited, NE5532 uses 60mA max. At normal use it takes what, some 15mA? So I'm well below 1A. I just need to cool those LM7*15 well, i guess.

MT:
As you suggest, I think I may just stick with NE5532 instead of 33078. LM4562 will not be so easy to get, I guess. Even Farnell does not have it. Anyway, 5532 will do just fine, I guess.

I'll try to get some opa2604 or opa2134. Opa4134 is gone, but I may still get 2134, Farnell has them.

...

 

[gnd]

[quote author="RogerFoote"]Be aware that 2134 and 2604 have different input current specs if that matters.. 2134s are a lot like LF353 etc.

Made quite a difference in my EGs that I built for my synth. They needed LF351/3s and the ONLY chip that would replace it in that app was 2134...

Otherwise, huge offset at output.[/quote]

I don't know... Will current specs matter in my crossover filter? There are four gain stages in filter.... Can I expect problems with offset? And what offset voltage is considered as huge? Milivolts, volts?

thnx

 

[gnd]

[quote author="RogerFoote"]I think it is just a quirk with CES3310 EG chips output stage. Just interesting is all.[/quote]

I did some quick testing in Multisim. I don't know how accurate that is....

I made filter part of my crossover circuit in 3 versionsn of opamps, and I got following output DC offset with 100mV/50Hz input:
NE5532 --- 17mV
TL074 --- 83microV
OPA2604 ---47microV

So obviously NE5532 gives much larger offset, but since it is line level circuit, no high gains, it this may not be problem with some DC decoupling capacitors after filter.

With FET opams offset is low.

Is this anywhere near expected numbers? Is Multisim usable for this?

Could I go just all the way through with NE5532, even in filter section? They are easy to get, and cheap. OPA i need to order and they are several times more expensive. What do you think?

 

[gnd]

[quote author="mediatechnology"]I think on the Bifets you're looking Ibias errors only in your analysis.

The Input Vos on the Bifets are the more significant error.[/quote]

I only measured output voltage offset.

What I did is draw filter circuit in Multisim, connected it to +-15V supply and measured DC output at the last IC with multimeter. I also connected bode plotter, to see freq response. Filter acted as expected on bode plot, and multimeter showed some DC offset on output, and that offset was different for different ICs.

How is input voltage offset showing in voltage flow? And how to measure it? Isn't any voltage offset on input reflected on the output, multiplied by gain of that stage?

 

[usekgb]

[quote author="RogerFoote"]OPA4134 is obsolete, I sure wish I had bought some when they were available... Maybe one of the industrial surplus places?[/quote]

The OPA4134 is not obsolete. You just can't get the DIP version anymore. You have to get surface mount. What a pain in the butt! I really like this chip in several applications and I really wish that I could still get a quad in DIP form. Oh well.

Cheers,
Zach

 

[clintrubber]

[quote author="gnd"]Crossover basically has three sections:
1. inputs and hi-pass filter
2. crossover filters
3. outputs with delay, gain, limiting and balanced output

Here is link to schematics, if it helps.
http://84.255.203.119/Berhringer%2520CX2300C%2520Service%2520Schematics.pdf
[/quote]

OT, but hmm, interesting delay-topology (see bottom of page 4).
I don't see the advantage of this topology vs the usual allpass-structure though.

Look for instance at this crossover from Rane (page 3):
http://www.rane.com/pdf/ac22bsch.pdf

About the same component count (ah, and oops, both using 5n6, dual 100k-pot & 511 stopper-resistors... coincidence ? )

More equal amplitude response isn't the reason says the simulator about the Beh-circuit....

 

[Guest]

[quote author="gnd"]

Why do they use TL072 in filter section, and 33078 elsewhere? Any ideas?
[/quote]

To avoid input bias current through 100K potentiometers

 

[gnd]

[quote author="Wavebourn"][quote author="gnd"]

Why do they use TL072 in filter section, and 33078 elsewhere? Any ideas?
[/quote]

To avoid input bias current through 100K potentiometers[/quote]

I see. Why is that? It is bad for potentiometer? Or some other reason?

gnd

 

[Guest]

[quote author="gnd"][quote author="Wavebourn"][quote author="gnd"]

Why do they use TL072 in filter section, and 33078 elsewhere? Any ideas?
[/quote]

To avoid input bias current through 100K potentiometers[/quote]

I see. Why is that? It is bad for potentiometer? Or some other reason?

[/quote]

Noise when tuning.