Hi Folks,

When I run this circuit at 9 V (V+/GND) it behaves very well with good noise and distortion levels. As I bring the supply voltage down, around 7 V it starts to distort audibly, as shown in the second sweep below. The harmonic distortion goes way up across the board. This is well before clipping starts (around 4 V). Any idea what is going on? Is this normal behavior for op amps? I have tried:

A .1 uF bypass cap at V+ on the op amp
Three different op amps (OPA1692, OPA180). They are different, but still distort.
A 22uF blocking cap between R3 and Vb.
Removing R3 and shorting RV1 (straight unity gain)
Removing C4,
Bench supply vs battery/dead battery.

Does anyone know a trick to fix this? I have tested a few other circuits without the rail-splitter IC and they have similar behavior.

Thanks in advance for having a look!

 

As I bring the supply voltage down, around 7 V it starts to distort audibly, as shown in the second sweep below.

I don't see how 0.061% THD would be audible(?).
Generally distortion below 1% is not heard.

According to the data sheet graph (below) for the OPA1692 with a ±18V supply, the THD sharply goes up at about 11Vrms or 16V peak, (2V below the supply voltage).
Of course that is for a gain of 1, so operating at a higher gain may affect the result.

 

I don't see how 0.061% THD would be audible(?).
Generally distortion below 1% is not heard.

According to the data sheet graph (below) for the OPA1692 with a ±18V supply, the THD sharply goes up at about 11Vrms or 16V peak, (2V below the supply voltage).
Of course that is for a gain of 1, so operating at a higher gain may affect the result.

View attachment 276293

The Golden Ears are unreachable with logic, reason and facts. Responding to them is a colossal waste of time and energy IMHO.

 

I don't see how 0.061% THD would be audible(?).
Generally distortion below 1% is not heard.

According to the data sheet graph (below) for the OPA1692 with a ±18V supply, the THD sharply goes up at about 11Vrms or 16V peak, (2V below the supply voltage).
Of course that is for a gain of 1, so operating at a higher gain may affect the result.

View attachment 276293

That THD level is the level where the cursor is at (in the software)... At 100 Hz it's 0.215%. You can hear it as a ringing tone that appears as the voltage goes down, when feeding a steady tone to it. This is probably a better example of why it's a problem, with a filter downstream:

 

I think @crutschow has pointed out that you will have less distortion when you operate the circuit with the recommended power supply voltage.

It is the nature of the circuits to need “extra” power supply voltage in order to minimize distortion.

 

The Golden Ears are unreachable with logic, reason and facts. Responding to them is a colossal waste of time and energy IMHO.

You'd have to be deaf to not hear this, but I guess what you guys are saying is that it's normal behavior for an op amp, so you have answered my question. I'm just looking for anything I can do to improve performance and extend battery time. If the op amp is the limiting factor then so be it, good to know. I'm mainly asking because some circuits seem a lot better, and I wondered if there is a simple way to improve this with the clever placement of a cap or something. I'm completely open to logic, reason, and facts.

 

Can you see the distortion in the output waveform?

 

1% is only -40dB which is clearly audible if it is loud enough to hear it.
0.1% distortion (-60dB) is not heard by most people.

 

I think @crutschow has pointed out that you will have less distortion when you operate the circuit with the recommended power supply voltage.

It is the nature of the circuits to need “extra” power supply voltage in order to minimize distortion.

Thanks! I am well within the recommended minimum supply (4.5V) so I thought I would be fine. According to the datasheet, the THD should be around .002% at my signal voltage, but the graph is for +/-18V supply. So it clearly goes up rapidly toward the bottom end of supply voltage (which I didn't know). Sounds like I need a different op amp to improve it.

 

Can you see the distortion in the output waveform?

Not really. It looks pretty much like a sine wave on the scope!

 

Not really. It looks pretty much like a sine wave on the scope!

I haven't been paying too much attention to the scope though, except to verify clipping. I'll take a look, I'm curious to see what it looks like when it's really bad!

 

All of the Texas Instruments "SoundPlus" opamps have extremely low distortion and very low noise at almost any supply voltage within its spec'd wide range up to clipping. Of course at very low levels then the noise dominates.

The graph of distortion plus noise shows very low distortion and noise of -130dB (0.0000003%) at 11V RMS which is 31.1Vp-p.

 

All of the Texas Instruments "SoundPlus" opamps have extremely low distortion and very low noise at almost any supply voltage within its spec'd wide range up to clipping. Of course at very low levels then the noise dominates.

The graph of distortion plus noise shows very low distortion and noise of -130dB (0.0000003%) at 11V RMS which is 31.1Vp-p.

Yes I really like the 1692, which I started with. For this device I needed one with lower input bias current, less power consumption, and most especially a single version. Thanks for the recommendation!

 

It sounds like you want a preamp. How about a MEMS microphone with a built-in low noise and low power preamp?
it is mono but most audio opamps are stereo.

 

I think this is your problem:

OPA145 datasheet

The input common voltage max is V+ - 3.5V.

With a seven volt supply, anything above the midpoint is out of range.

 

I think this is your problem:

OPA145 datasheetView attachment 276357

The input common voltage max is V+ - 3.5V.

With a seven volt supply, anything above the midpoint is out of range.

That's interesting. So if I biased it to say 2V (at 7V) that should prove it out. I might be able to get away with it in the rest of the circuit. I would have to ditch the 2426 and make my own driver though. Any downside to having the bias voltage ~30% of supply? I'd much rather have it just start clipping (6V would be fine) than make the harmonics. I should have posted a sweep at 6V instead of 7V (it STARTS at 7V and gets way worse) to quell the trolls; at 6V the harmonics are way up near the signal, and it just sounds like something is wrong. Clipping just sounds like you need a new battery!

 

It is worth a try. Why the odd supply voltage?

 

It is worth a try. Why the odd supply voltage?

It's a consumer product that runs on a 9V battery. The typical user tends to let the battery run down until they hear clipping. Sometimes they don't change it for a year or two. I'm happy with the way it runs > 7V or so. Below that it starts sounding bad, but it doesn't clip so the user won't understand what is wrong and change the battery.

 

It is worth a try. Why the odd supply voltage?

Woohoo! BobTPH you are the man! That fixed it. It's clean down to 4V even with a 357mV signal, then starts clipping. MUCH better. I'll have to go through and check my headroom throughout the system, but it should be adequate. If you were close by I would buy you a beer! Thank you so much for increasing my understanding of the specs and identifying the problem. Here's 9V, 7V, and 5V:

 

Great. Glad to be of help.

I think it is criminal to say that op amp works at 4.5V when the inputs are restricted to 1V or less.