rc4558 vs Tlo72

[Nolind]

Enlightenment required.

I'm forever seeing forum posts ( not necessarilly on this forum) saying " I built such  and such a project but didn't use the noisy RC4558 using a TL072 instead". Having studied the data sheet for both chips as fas as I can see the 4558 has a lower noise rating! I've certainly never had any noise issues with the projects I've buit so far.

The rc4558 datasheet quotes 8nv (followed by a complex equation I assume to mean rms or similar) whilst the TL072 quotes 18nv. I presumed lower was best. Am I right?

[StephenGiles]

Enlightenment required.

I'm forever seeing forum posts ( not necessarilly on this forum) saying " I built such  and such a project but didn't use the noisy RC4558 using a TL072 instead". Having studied the data sheet for both chips as fas as I can see the 4558 has a lower noise rating! I've certainly never had any noise issues with the projects I've buit so far.

The rc4558 datasheet quotes 8nv (followed by a complex equation I assume to mean rms or similar) whilst the TL072 quotes 18nv. I presumed lower was best. Am I right?

At 9v there is no difference. However, some guys like to think there is so that that they can impress their girlfriends

[Nolind]

  That'll be their imaginary girlfiends then!

So it's just like a Gibson / Fender   or Nikon / Canon type thing then.

[DougH]

Dude...

[MoltenVoltage]

I replaced the TL072 in the rebote with a 4558 and found it sounded significantly better - more rock and roll anyway!

[brett]

Hi
the 4558 *was* simply 2 741s in a single package.
A recipe for bulk hissssss
Apparently, modern 4558s aren't anything like 2 x 741s.  Much quieter.  But in circuits like a tubescreamer, the 4558 is significantly noisier than a TL072 (by several dB - about what most people would say is "double"). 
The NE5532 is quieter, but has low input impedance (approx 100k), so isn't suited to all circuits.
cheers

[PRR]

> Having studied the data sheet

WHAT? Blasphemy!!

> a complex equation I assume to mean rms or similar

Uh, no.

Anyway 18nV audio-band hiss is absurdly small.

Tune an FM radio between stations. Hissss. Now cut the highs, Less hiss. There's hiss at all frequencies. The narrower the frequency band, the less hiss.

One designer needs a 500Hz-3KHz telephone band, another wants 20Hz-20KHz, and the guy in the bar will only hear 82Hz-6KHz. The makers can't publish noise for every possible bandwidth.

The standard measure is for a band only 1Hz wide. This is semi-fictional: not easy to make such a narrow filter.

We want to find the RMS over our frequency band. Square the 1Hz, it's still 1Hz, easy. Sum all the 1Hz bands in our bandwidth. Take the square root of that.

20Hz-20KHz is 20,020Hz. SqRt is 141.5. If the spec is 10nV per Hz, we expect 141.5*10nV or 1,415nV or 1.4uV.

> The rc4558 datasheet....

...dates from the 1970s. In those days it was hard to make clean Silicon. Some wafers were quiet, others had excess noise due to surface contamination. Since most users are not very noise adverse, the specs were written so even the dirty wafers would meet spec and could be sold.

Silicon is very much better now. But there is NO call to update the old specs. If someone wants a low-noise chip, sell them another (higher price) part.

I would expect any TL072/082 to be a bit under 2uV in 20KHz. That's 14nV/1Hz. The 18uV spec leaves little leeway; indeed by the time the TL072 series came out Silicon was pretty clean.

The 4558 (AND 741) is a simple bipolar affair and the theoretical noise can be calculated. Input devices run about 20uA, so about 1,400 ohms each emitter. There's four emitter junctions across the input, 5,600 ohms. I would expect roughly 1.2uV across 20KHz. That computes to 8uV/1Hz, which shows the old 4558 datasheet allowed a very generous margin for dirty wafers.

Only looking at _Voltage_ noise, a clean-wafer 4558 is quieter than a bogie TL072.

BUT look at the other number: Current noise! Multiply this by the effective audio impedance at your input. Do you even know the effective audio impedance at your input? Usually it is low; occasionally it is startlingly high. Note that a TL072 has much-much lower current noise than 741/4558.

[chi_boy]

Hi
the 4558 *was* simply 2 741s in a single package.

Isn't it the 1458 that was 2 741's in a single package?  Or am I mistaken?

[brett]

Hi
that's close enough to true.  After a bit of research, I find that neither was exactly 2 x 741s. 

The 1458 contains 2 op-amps that are updated/improved versions of the 741 (source: The IC hobbyists handbook by Power).  The first 741 came out in 1968, and the 4558 (the first multiple op-amp) came out in 1974, so there wasn't too much time to improve the design.  I think that quieter, NPN transistors with higher hFE at the input was one change (?).  The 1458 (improved further?) and JFET-input op-amps came a year or two later.

Summary: the 1458 and 4558 have slight improvements on the 741.
cheers

[PRR]

> there wasn't too much time to improve the design.

Six years is a LOOONG time in this racket.

By Moore's Law, in 6 years we should expect 16 times as many bits per wafer.

Analog chips don't move as fast as simple RAM; more-cheaper transistors does not ensure a better design.

And yes a Dual has almost twice the transistors, eats-up 18 months of Moore.

However the 1.5X-3X improvement from original 101/741 to the Duals is not impressive.

Not near as impressive as the advance from '709 to '101/'741 types.

Nor as impressive as the advance, in Audio terms, from 101 types to the 5534 (do you have a date for that?). WOW. All the stuff we were doing to ensure low input noise, wide bandwidth, and strong output drive (oh the ugly things we did) were already in there. (And were not majorly improved for 25 years.)

[Perfboard Patcher]

Or maybe the way it deals with low impedances at the output?

Like in a tube screamer-type overdrive when the diodes conduct and the resistor between  inverting input and  (virtual) earth functions as load.

cheers
PP

[km-r]

^ the lower impedances are less susceptible to noise due to RF interference as this practically *shunts* high impedance RF that gets in. this is why RF circuits in radios are almost always below 50ohms impedance.

[ddpawel]

Don't look to datasheet. In my Metal Zone Clone I tried CA3240E and MC33078P. MC33078P has better noise characteristics in datasheet than CA3240E, but in Metal Zone CA3240E is really quiet and MC33078P hums like hell.
So test some op-amps and find the best one.

[Nolind]

Wow. Some pretty concise answers there!   Most of it went whooosh over my head.   I fix central heating sytems for a living . If it has an electronic problem we just ooof the old pcb out and charge the customer £250 for a new one. Easy.

I will heed advise and trust my own ears not the data sheet.  Some TL072's  just arrived in the post from EvilBay.   Of course being a guitarist and 50 years old I have some mid to high range hearing loss so I'm not sure hiss is a problem.  Not for me at least.

[Dave W]

"Noise" is a very broad term that encompasses a variety of variables. V noise, I noise, Thermal noise, etc. Each contributes to a varying degree depending on the circuit in question. To just say "oh, this one has "better" noise specs" is fools gold. Do a Google search. National Semiconductor (and other manufacturers) have nice well written docs about op amp noise on their sites.
Personally, I am of the StephenGiles school of "at 9V there is little to no difference". However, if one wants to convince themself there is; more power to you.
Besides, the noise from your guitar easily dwarfs the noise specs of the op amp.
BTW, yes, the 4558 does have better V noise specs than the 072. However, the I noise of the 072 is almost non existant. Can't say that for the 4558. Both play a role when calculating total noise. Do the research. Do the math. Try different op amps and use what you've convinced yourself "sounds better/quieter/whatever..."

[brett]

Hi

Besides, the noise from your guitar easily dwarfs the noise specs of the op amp.

Guitar noise?  Maybe my guitars are all quiet (except for hum from my noisey Gibson Lucille).
Most of my high-gain pedals hiss between tunes.  And my (lowish gain) tubescreamer, for some reason (poor design?).
Despite my crook old ears.
cheers

[petemoore]

  welpp, if big noise is the deal, and the socketted opamp swaps helped, that's one thing.
  The other thing is that either opamp should be able to duck easily under most anything else noise related, often to do with large potential power supplies and sensative inputs.
  Boost/compression circuits such as TS bring signal level up when not playing [anything 'noise' at the input is multiplied], then the max output is chopped down by the diodes, so a portion of what was amplified is de-amplified, changing 072/4558 matters tiny to the equation, can be socketted.
  Unplugging the input to the amp test for noise, if noise, it is 'east' of the amplifier input.
  For guitar, grounding the source/pickup using the guitar volume control [if it works right], or unplugging the guitar [the cable tip is now an open antenna] may further locate noise sources.
  The noise isn't getting in there is generally the solution which doesn't involve the actual opamp as a noise-source input, and peeling back the other onion layers one at a time, crying for each layer.
  Low noise solutions involve figuring out exactly what you have there, and trying fixes at places deemed as likely culprits.
 
 

[PRR]

> Maybe my guitars are all quiet (except for hum....

Hum is avoidable. Play far out in a field.

Hiss is not avoidable. There is always some hiss.

The raw guitar pickup has 5K of resistance, but with capacitance a huge impedance peak at 2KHz-5KHz, typically damped by a 250K pot in the guitar. We may say the impedance is often a nearly pure 50K resistance over the whole audio band. With infinite load, and assuming anywhere near room temperature (not absolute zero, not red-hot), the 20KHz hiss is about 3 microvolts.

That's an interesting number I had not looked at before. The hiss of a 12AX7 runs 1uV at best but often 2, 3, even 5uV. Pickup turns, magnet strength, and wire gauge has evolved so that tube noise runs from negligible to no big deal. Switching to a much lower voltage-noise device would not improve input hiss. A TL072 has higher hiss than the best 12AX7, but lower than guitar, so is not really hissier than a 12AX7.

> I noise of ... ... 4558

For a 50K source, works out to significantly more than its V noise, source noise, TL072 or 12AX7 noise.

The fact that 101-derived inputs like 4558 are popular in guitar circuits shows that such hiss is not a big deal.

And that if you want to KISS, a TL072 is usually your best bet for low hiss. You don't have to think about impedances.

[puretube]

Hi

Besides, the noise from your guitar easily dwarfs the noise specs of the op amp.

...Guitar noise?...

It`s all in the fingers... (is what I learned from the web...)