[diredesire]

I've got a few basic questions about op-amps that were not answered by searching/googling.

They're questions for the EE majors/pros out there. I'm wondering what "paper" characteristics mean in terms of audio, and what sort of ranges are good.

I've been desoldering a few components and i've come across various op-amps that i'm curious about. I've got a ton of old/crappy computer hardware lying around, and i'm wondering if i can salvage any chips to use in my amps.

My questions can be answered in detail or very short, whichever you prefer.

I'm just wondering what these characteristics will result to (theoretically, i understand that specs aren't everything).


Questions to answer per bullet: What is this? What does it mean in terms of sound. Is a higher/lower number better? What am i looking for in terms of numbers? (a range please)


Gain Bandwidth (MHz)
Slew Rate
Supply Current Per Channel (i assume this is just how much current the chip dissipates?)
Offset Voltage (mV)
Max Input Bias Current (nA)
Output Current (mA) [i understand what this is, but what numbers am i looking for?)
Voltage Noise

I am curious because i have a few LM392's on hand that i salvaged from some dead hard drives. A link to the specs is here

Thanks for your time, guys! Maybe we can sticky this, as i imagine the hacker/'salvagers' that are not EE majors or the humble builders are curious.

 

[Porksoda]

I could get started on this, but I feel someone like Tangent will give a much more complete and detailed response than I ever could.

If none of the local experts pipe up, I will chime in later when I have time to elaborate fully.

EDIT: Off hand, those look like they are designed for single-ended setups, and they are bipolar input, so they are probably not friendly with the typical high-impedence front ends that you see in headphone amps. Also, the two op-amps on the chip appear to be different, unless I read it wrong. One is a comparator and the other is an amplifier. I don't think they would work out very well as audio amplifiers.

 

[amb]

Gain Bandwidth (MHz)
This is the maximum bandwidth of the opamp when operating at unity gain (above which the gain will drop below unity). As it turns out, the bandwidth of the opamp is reduced in such a way that as you increase the gain, the product of gain * bandwidth is a relative constant (up to the maximum open-loop gain of the device). This is is therefore called a gain-bandwidth-product. An opamp with a 100MHz GBP will have a bandwidth of 100MHz at unity gain, but it goes down to 10MHz when the gain is 10, and 1MHz when the gain is 100, etc.

Slew Rate
A measure of the speed of the opamp. Given a square wave, how fast can the opamp swing its output on the edge of the waveform. This is expressed in volts per microsecond.

Supply Current Per Channel (i assume this is just how much current the chip dissipates?)
At quiescent, how much current will each opamp draw through its power supply pins. If this is a dual or quad opamp, then the spec is usually per amp within the package. You use the term "dissipates" incorrectly here because that usually refer to power. The power dissipation at quiescent of the opamp will be the supply current times the supply voltage, and this is given off as heat.

Offset Voltage (mV)
This refers to the DC offset from 0 at the input, and is an indication of the opamp's DC accuracy.

Max Input Bias Current (nA)
How much current will flow (in or out of each of the opamp's two inputs) to keep the internal transistors biased. FET opamps have miniscule input bias currents whereas BJT opamps usually have significant input bias current.

Output Current (mA) [i understand what this is, but what numbers am i looking for?)
The maximum current that the opamp's output is rated to sink or source. Of course the more, the merrier, but due to current and power dissipation limits of the output transistors within the opamp, and the heat dissipation capability of the packaging this is usually well short of 100mA per device. For headphone amps (if the opamp is to drive the load directly), low impedance headphones need greater output current capability.

Voltage Noise
An indicator of how noisy the opamp is.

Edit: My post #1000!

yay.

 

[Porksoda]

Solid work, amb, and congrats on 1k posts!

One issue I would point out is that there is input and output voltage offset. Input offset meaning that the input transistors are a tad mismatched, so where you would expect the voltage at each pin to be basically equal, they are in fact off by a few mV. Output offset is how much the output diverges from zero with no signal. Now the effects of these two are linked, but I would point out that output offset will vary with output impedence and closed loop gain, whereas input offset is relatively constant. This is because output offset is actually CAUSED by input offset, and is actually the input offset multiplied by the closed loop gain (ideally).

 

[tangent]

Quote:

Originally Posted by diredesire I'm wondering what "paper" characteristics mean in terms of audio

They're indicators at best, not predictors of sound quality.

One cannot guess from bandwidth, slew rate, offset, etc. numbers that the OPA627 sounds dark and detailed, but the AD843 sounds bright but smooth. Take every spec where the AD843 differs significantly from the OPA627, then find a chip that goes even further: I'm sure you can find such a chip among the thousands of types that are available, but I wouldn't lay money that it's brigther and smoother, just from knowing these specs!

That said, if you have a chip with 1/4 the bandwidth, three times the noise, half the output current, and a fifth the supply current as the OPA627, I wouldn't hesitate to bet that it sounds worse.

Use the specs to help you reject obviously bad choices. The rest, you'll have to listen to.

 

[diredesire]

Quote:

Originally Posted by tangent They're indicators at best, not predictors of sound quality. One cannot guess from bandwidth, slew rate, offset, etc. numbers that the OPA627 sounds dark and detailed, but the AD843 sounds bright but smooth. Take every spec where the AD843 differs significantly from the OPA627, then find a chip that goes even further: I'm sure you can find such a chip among the thousands of types that are available, but I wouldn't lay money that it's brigther and smoother, just from knowing these specs! That said, if you have a chip with 1/4 the bandwidth, three times the noise, half the output current, and a fifth the supply current as the OPA627, I wouldn't hesitate to bet that it sounds worse. Use the specs to help you reject obviously bad choices. The rest, you'll have to listen to.

Thanks Tangent, and amb, for taking the time to reply. Your posts (both of you) have been very helpful to me in the past. The op-amp i found seems to be very fast, i read another post on the board that said 5-200(i THINK) was the boundaries of a normal audio-use opamp, so it looks like this thing won't work well. I may try to figure the pinout and try it anyways, i've got nothing to lose..