It is often suggested that square wave is important indication of performance quality. The reason I ask this question is that my experience with TL072 contradicts this presumption. DS of TL07x series indicates that opamp is not stable without series output resistor but I like the sound of it much more without this resistor. As soon as the output resistor is in circuit the quality of sound is reduced, at least for my taste. Only 50R at the output is enough to make the sound dull, boring, unexciting.
Why we need to make the opamp stable with signal that will never be present in any music or speech?
Why we need to make the opamp stable with signal that will never be present in any music or speech?
You could try a series inductor with RC 'zobel' network. You can improve stability with many opamps by running at a higher noise gain. Are your TL072 running as unity gain buffers? The weakness of TL07x seems to me to be that it already contains a 128R series resistor on its output (from the internal schematic).
TL07x are notorious for requiring up to 470 ohms at the output in some situations (which I presume means unity gain and capacitive loading), and are not very fond of heavy loading in general.
Here's what the good folks at AD have to say on the matter:
Ask The Applications Engineer-25: Op Amps Driving Capacitive Loads
Due to the internal output series resistance, TL07x can be easily stacked on top of each other. Note that this means that input capacitance multiplies as well, and the minimum feedback resistor that'll require a capacitor in parallel will decrease (and the cap value increase) correspondingly. So if you have 22k || 10p for one, use 10k || 22p for two, 4k7 || 47p for 4 etc.
It be mentioned that the notion of an opamp having a "sound" per se is faulty - if anything, it's the sound of this part in a particular circuit. I would recommend always giving an example circuit.
Here's what the good folks at AD have to say on the matter:
Ask The Applications Engineer-25: Op Amps Driving Capacitive Loads
Due to the internal output series resistance, TL07x can be easily stacked on top of each other. Note that this means that input capacitance multiplies as well, and the minimum feedback resistor that'll require a capacitor in parallel will decrease (and the cap value increase) correspondingly. So if you have 22k || 10p for one, use 10k || 22p for two, 4k7 || 47p for 4 etc.
It be mentioned that the notion of an opamp having a "sound" per se is faulty - if anything, it's the sound of this part in a particular circuit. I would recommend always giving an example circuit.
Square wave response is overrated as a performance criterion. Yes, it can show the verge of instability - yet some apparent 'ringing' is simply the consequence of a low pass filter removing the higher harmonics of the square wave.
An amplifier which is near to instability at X kHz is likely to be showing phase problems at much less than X kHz.
If you can hear the effect of inserting an 50R resistor, something odd seems to be taking place. Have you tried using lower capacitance or shorter cables? Even so, I doubt that you are hearing the effect of treble toll-off from inserting only 50R of resistance.
Another possibility is that you, unknowingly, prefer the sound with some signal ringing on the interface. Which may subjectively add a brighter, more exciting, treble character. This possibility would be supported if the sound remains dull and boring after implementing one of the alternative amplifier stability solutions suggested above.
Last edited:
I guess that's the reason why I prefer not to insert output series resistor.
So long as it's not causing undue amplifier heating (at the driving op-amp or at the power amplifier), indicating outright oscillation (check cautiously with a finger tip, however, a wideband oscilloscope is the right way to check for oscillation), or significant D.C. offset (measure that), or other obvious system problems, I suggest that you simply leave things as they are for now.
Last edited:
I must say that several years ago I had oscillation problem with ESP P88 preamp configured as buffer (both input and output opamps used as unity gain buffers and the pot between them). I liked the sound of this buffer but after several hours I noticed that power amp (biased at only 15mA) was hot! ESP P88 used as buffer oscillated and caused power amp (ESP P03) to overheat. But both input and output opamps had series 100R resistor at the output, and the pot used very short coax cables to connect with the PCB. As soon as I turned the buffer preamp off, power amp stopped overheating. Therefore, series output resistors did not help stability at all.
Perhaps, your P88 implementation was self-oscillating - meaning, oscillating not due to the output cable capacitance.
I used original ESP boards then. But the same circuit on the boards that I designed does not oscillate. My PCB has pot onboard, it does not use cables to connect pot with PCB, and inverting input side of feedback resistor is just couple of mm away from inverting input. Also, I used 4558 as unity gain buffers then. May be that 4558 was the culprit, but it would be surprising because if used as unity gain buffers 4558 should not be problematic.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.