Can a 5532 replace a 4558 in single-supply 9V circuits?

[fryingpan]

I just say this because 5532s can be found aplenty while I'm having a bit of difficulty sourcing 4558s (out of stock at the shop I'm planning to buy from). The NE5532 is a superior opamp at +/-15V and has similar cost, but 4558s are ubiquitous at low voltage supplies, which makes me suspect that the 5532 doesn't like being powered by low voltages.

[ElectricDruid]

You may be right. TI's datasheet for their version recommends +/-5V (e.g. 10V) as the minimum supply. It *might* work lower than that, but it'd be outside the spec.

https://www.ti.com/lit/ds/symlink/ne5532.pdf

However, OnSemi say +/-3V, so either their version is slightly different, or they're just not as cautious as TI:

https://www.onsemi.com/pdf/datasheet/ne5532-d.pdf

[davent]

 I believe the 5532 might be relatively current hungry as well,  if i recall correctly.

[rustypinto]

You can power both 5532 and 4558 at 9V just fine. Both will have pitiful voltage swing at 9V, but thats usually not a problem for guitar signals.

5532 (and it's counterpart NJM4580) really shines with it's very low output impedance compared to 4558, even being able to drive headphones. And for that kind of performance, the 5532 supply current is ~2x (6mA-typical/supply pin) of 4558's supply current.

4558 is a nice balance of relatively low current draw (eg. for battery operation), high input impedance (eg. its a decent input buffer), and its cheap, so it finds it's way into a lot of general purpose designs.

[fryingpan]

Looking at the specs, it would appear the biggest problem with the 5532 is that it is limited to V++ - 2V (and viceversa). So not much headroom at 9V.

[fryingpan]

You can power both 5532 and 4558 at 9V just fine. Both will have pitiful voltage swing, but thats usually not a problem for guitar signals.

5532 (and it's counterpart NJM4580) really shines with it's very low output impedance compared to 4558, even being able to drive headphones. And for that kind of performance, the 5532 supply current is ~2x (6mA-typical/supply pin) of 4558's supply current.

4558 is a nice balance of relatively low current draw (eg. for battery operation), high input impedance (eg. its a decent input buffer), and its cheap, so it finds it's way into a lot of general purpose designs.

It would appear that guitar or bass signal peaks are +/-1V or so, right? Taking into account particularly hot pickups, because otherwise it should be half that.

Anyway, the 4558 is bipolar input right? For high input impedances, I'd imagine the TL0xy line should be better.

[rustypinto]

Sure, TL0xy has very high input impedance with slightly higher noise. I use TL071 regularly for input buffers. TL0xy also has high slew rate, making it an excellent choice for peak/envelope detectors.

If you are having a difficult time sourcing 4558, I recommend looking for MC33078 instead.

[fryingpan]

What about MC33178?

[rustypinto]

What about MC33178?

As a replacement of 4558 doing general purpose mixing, filtering, and output stages, and at lower power? Sure, looks good.

I never tried MC33178 myself, but it looks like it was designed specifically for battery operation. A quick glance at octopart.com reveals that you will pay a lot more for that kind of performance compared to 4558 (but not much more than NE5532). But if you can get it in the first place, and cost is not a factor, then it's not a bad choice.

[PRR]

'5532 used to be VERY much more expensive than the jellybean chips. Like $7 or $3 instead of $0.59. Our tone was not worth that much.

The '5532 has some really excellent specs at the high voltage. Performance may fall-off at lower voltage. But how much spec do you really need, or even want?

I say just try it.

[Rob Strand]

You may be right. TI's datasheet for their version recommends +/-5V (e.g. 10V) as the minimum supply. It *might* work lower than that, but it'd be outside the spec.

https://www.ti.com/lit/ds/symlink/ne5532.pdf

However, OnSemi say +/-3V, so either their version is slightly different, or they're just not as cautious as TI:

https://www.onsemi.com/pdf/datasheet/ne5532-d.pdf

The lowest supply voltage can be a fuzzy area for opamps.   You might even find LM741 datasheets which imply a minimum of only +/- 5V.

Technically the specs in the datasheet only apply at the specified voltage (eg.  +/- 15V).   Some datasheets give graphs for other parameters so you can see how things vary or poop-out.   Operating below the minimum supply on the graphs is uncertain but like I said even LM741 datasheets are like this.    Different manufacturers might have different minimum voltages on the plots!  Many opamp will "work" with lower supply voltages.

On some opamps the PSRR can be poor if you run at voltages less than say +/-4V  because the bias circuits stop working properly.  Some opamps have a "back up" bias or naturally decreasing bias so the opamp still "works" at lower voltages but others could misbehave altogether.

Where I'd say you need to watch out for the NE5532 (some people have already covered some of these):
- high input bias current.  This can cause high offset voltages with 1M input resistors and that will eat into the headroom unless Vref is tweaked to compensate.   The low noise voltage of the NE5532 is a result of the higher input bias current (and the input configuration.)!
- NE5532 has protection diodes across the input.  Some circuits can push too much current through the input diodes and fry or damage them.    The input diodes can also affect the way the circuit works when the opamp clips or in non-linear circuits like oscillators.   (That appears in the specs as a low "Differential Input Voltage".)
- higher supply current.

I suspect you can get the NE5532 to work in many circuits, or can make it work with some minor tweaks.

The NE5532 was a Philips device, here's a Philips datasheet,
https://www.semiee.com/file/backup/PHILIPS-SA5532.pdf

Quotes +/- 3V but the graphs show a little lower maybe +/- 2V at a stretch.

[fryingpan]

Well, as a general rule, bipolar input opamps are not to be used with high impedance signals anyway and they really want to see 5-10kohm before them, ideally. (At least for optimal performance). Also, high currents are unlikely with guitar signals, and even when dealing with "line level" voltages, you can only fry the diodes with voltages over the recommended input voltage swing.

I may as well use a NE5532 anyway, or solder in a socket.

[GGBB]

high currents are unlikely with guitar signals

What is a "high current" in this context? Guitar "signals" are not just generated by guitar pickups, but also by effects, amp sends, etc.

[PRR]

......you can only fry the diodes with voltages over the recommended input voltage swing.

When feedback is working, the one input follows the others, differential input voltage is near zero.

When feedback fails, such as an overdriven pedal, differential input voltage can be as much as full signal voltage. A naked guitar may be OK, but kids today are humping 12V and 18V and more into pedals "for headroom".

For 95% of pedal sockets, TL072 makes more sense than '5532, Have the '72 become scarce?

[fryingpan]

......you can only fry the diodes with voltages over the recommended input voltage swing.

When feedback is working, the one input follows the others, differential input voltage is near zero.

When feedback fails, such as an overdriven pedal, differential input voltage can be as much as full signal voltage. A naked guitar may be OK, but kids today are humping 12V and 18V and more into pedals "for headroom".

For 95% of pedal sockets, TL072 makes more sense than '5532, Have the '72 become scarce?

TL072 is a good choice too, of course. Probably even better, although I suspect that a TL062 would be a better choice for lower voltages? Still, if I wanted to replace a 4558 with a similarly bipolar input opamp, the 5532 is one and it has even lower noise. And I would imagine that feedback only really fails if you are clipping the opamp, but if you limit voltages in the feedback path feedback shouldn't fail at all. It would be working as intended, and I believe that a pedal that intentionally clips an opamp is either doing so for creative purposes (and chances are that the clipping characteristics of a specific opamp contribute enough to the tone that you are better off just using the same) or asking for trouble.

[ElectricDruid]

It used to be the case that TL082 was the "standard" version, TL072 was the "low noise" version and TL062 was the "low current" version. I used to use TL062 for battery-powered pedals back in the day, before I discovered the LF444, which was an even lower current quad package.
The TL062 is noisier than the TL072, so these days I'd use the TL072 pretty much always, and the TL082 I haven't seen for ages, although I haven't looked and I guess they're still out there.
NE5532 was a more expensive specialist thing that you could get some excellent noise figures out of if you were willing to run it with more juice. It could cope with much lower resistor values and loads and gave excellent performance like that. Not suitable for battery gear, but it used to turn up in mains-powered mixers quite a bit. These days there are probably better options, although it may still be good for the price.

[Rob Strand]

Well, as a general rule, bipolar input opamps are not to be used with high impedance signals anyway and they really want to see 5-10kohm before them, ideally. (At least for optimal performance).

For audio the 5k to 10k would be an AC impedance seen by the input but bias shifts are due to the DC resistance feeding the input pin - that would be the Vref resistor for most non-inverting configurations used in pedals.

[fryingpan]

Well, as a general rule, bipolar input opamps are not to be used with high impedance signals anyway and they really want to see 5-10kohm before them, ideally. (At least for optimal performance).

For audio the 5k to 10k would be an AC impedance seen by the input but bias shifts are due to the DC resistance feeding the input pin - that would be the Vref resistor for most non-inverting configurations used in pedals.

The Vref resistor doesn't have to be that high though. Maybe 100k or even lower is fine. The voltage divider for Vref could be a couple of 22k resistors. I mean, 9V/44k, it's 0.2mA.

[antonis]

You can always bootstrap the Vref (bias) resistor so practically no DC drift while high input impedance..

See 6 & 6a below:
(apologize for modding Ray Marston's original schematics..)



P.S.
R3 can be as low as you like but bear in mind its bootstrapped value for input HPF calculation..