Jump to content
Electronics-Lab.com Community
Sign in to follow this  
tadgesualdo

Op amp theory Q's

Recommended Posts

I'm no expert when it comes to audio, but I do know a bit about op-amps. Audio is a very personal thing.... somethings sounds good to one person, while the next person thinks it sounds bad/flawed. Whenever I buy audio gear, I go into the sound room, close my eyes, and tell the salesman to "show-me-what-you-got".  I let my ears tell me what is good and bad, TO ME, not my wallet. Sure, I'd love a Crown pre-amp and a couple of Macintosh tube amp's but THAT JUST AIN'T GONNA HAPPEN. In the 70's, music was played LOUD, and I'm sure that to some degree that has affected my hearing, so my Yamaha reciever and Klipsch Heresy II's do just fine..... spending more would just be a waste as I wouldn't be able to hear the "better performance"!!!

Back to op-amps......... the transfer function doesn't do anything more than defign the gain. When reduced to the standard form, the numerator contains, zeros, or points where there is an increase in gain (phase lead), and the denominator has poles which are points where there is a decrease in gain (phase lag). Crossover frequency and phase magin determine stability. A higher cross over frquency means better transient response. Remember also, there are two parts to transient response... one being recovery time and the other being deviation. Gain will control deviation and cross over will determine recovery time. How these parameters relate to how we hear things, I don't have a clue. Now, why would you even consider using a AD843 for audio? It has a full power  bandwidth of 3.9MHz and a slew rate of 250V/uS...... seem's like OVERKILL for a audio app's, even if you can hear beyond 20KHz!!! Yes, the "numbers" would be impressive, but how many could actually hear the difference. In the end, how many zeros beyond the decimal place really matter!!

Share this post


Link to post
Share on other sites

Indulis,

I wish you could come over and I could have you listen to this little preamp that I made for the purpose of checking out opamps.  I don't know why but the AD843 sounds much cleaner than say a TL072.  I don't know enough about these specification to understand what is over kill, this is one thing I am trying to understand.  I do know 250v slew rate is better than say 60v, but I don't know what is over kill and if you can have such.  The AD843 is inexpensive enough that even if it were overkill it wouldn't matter all that much.

So lets talk about bandwidth and slew rates.  Most opamp have bandwidth way above human hearing, does that matter as long as it is ABOVE human hearing.  What I mean is who cares as long as it is not and issue?  With slew rates, at what point is this specification addiquate for audio?  What part of the transient response equation is slew rate, recovery time, or diviation?

What i am trying to figure out is how are these specifications related to what I am hearing, in other words how can I make these specifications more practical, by understanding them?

BTW, there are a few papers out that mention that in a sense humans can hear above 20K Hz.  We localise sound by a combination of two things, amplitude and phase.  So we precieve phase difference of frequencies above our own human hearing range, even though we can't detect those frequencies.  So saying NOTHING above 20K Hz has no use, is like throwing out the baby with the bath water.    Perhaps these frequencies add very little to our listening experience, but if your listening to stereo images, then they must add some spacial properties, otherwise why don't we just listen to mono.

Can we start on the most basic level like saying that an amp with 8v slew rate will sound dull, because it can't respond even to a 15K hz sine tone?  I just want to understand how things relate to others.  What I would like to do is be able to listen to two different opamp and then after picking which one I like, look at teh specification and understand the measured differences that could be contributing to what i am hearing.  I have no clue.  Then again perhaps I am better off in ignorance as it is more honest this way.  It sounds good but I don't know why, not because I thought it would because of a spec, but because it just does.

TG :)

Share this post


Link to post
Share on other sites

A good amplifier has a bandwidth much higher than we can hear because ultrasonic frequencies intermodulate if the ultrasonic distortion is high, causing beat frequencies that we can hear.

Some people "claim" that some opamps sound better. I know that LM324 (LM358 dual) and 741 opamps sound awful, and a TL071 sounds great. I can't hear any difference of opamps "better" than a TL071, except for their noise levels at very high gain. The AD843 is just as noisy as a TL071, other audio opamps are quieter.

Noise is hiss and rumble, distortion is a hashness to reproduced sound.

I looked at the spec's for the AD843. It has a full-power bandwidth of more than double the frequencies of the AM broadcast band! The TL071 reaches 100kHz. The LM324 reaches 5kHz.
The AD843 has a high supply current so it gets warm. The others don't. The AD843 has distortion of about 1000 times lower than I can hear. The TL071's distortion is only 100 times lower than I can hear. Overkill. ;D

Share this post


Link to post
Share on other sites

I don't know the reasons why, but inverting inputs sound better to me than non-inverting.

I haven't heard any difference, but an inverting opamp doesn't have any (well, hardly any) common-mode voltage swing at its inputs, while a non-inverting opamp has the entire input voltage swing at both inputs. The inputs are slightly non-linear so slightly more distortion is produced in a non-inverting opamp circuit.

Also each opamp seems to have its own sweet spot for feedback settings.

I haven't heard any difference. If the feedback or if the load capacitance are not too high for the opamp then the distortion is 100 to 1000 times lower than I can hear. In the article you posted they show a standard method of reducing the effect of too much load capacitance, which is caused by a shielded audio cable at the output. Maybe you are reducing the opamp's feedback instead to reduce the effect of too much load capacitance.

Currently I have an idea of attempting to use positive local feedback on two opamp stages, and global negative on both of them together.

Share this post


Link to post
Share on other sites

Audioguru,

Currently the circuit I am tweaking and experimenting with is a two stage inverting opamp configuration.  I really like the idea of stages and components not effecting each other, so i try to issolate and buffer them as much as possible.  For instance on my first opamp stage I use the ratios of resistors that present a higher circuit impedance to the source circuits, and I use up some circuit gain by issolating the two stages with a law faked volume pot.  The second stage then I have complete flexibility to do the most radical things, as this section will not effect other circuits nearlly as much.  I also use a 100 ohm resistor on its output so that the decoupling cap I have on the output will not effect the second stage as much, we all know opamps don't like to drive capacitive loads. I don't mind using inverting inputs as long as I have an even number of opamp stages so that my output would be as if I were using non-inverting  inputs.

My trick besides the above-mentioned output resistor to isolate load capacitance, is adding supply bypass capacitors which are are important for opamps in audio circuits.


Could you expand on the reasons for this?  What capacitor values do you use?  What benefits does this have with the opamps performance?

I guess since opamps leave you with mostly the buffering and feedback to worry about, it becomes much easier to focus on these parts of the circuit.  So far my experiments in this area have had satisfying benefits.  I am very impressed with how good you can make opamp circuits, they perform much better than I ever would have imagined

TG ;D.

Share this post


Link to post
Share on other sites

Hi TG,
Attach your schematic so we can see what you are doing.
Modern opamps are "perfect" when they are used within their ratings and with their recommended supply bypassing. The series inductance of supply pcb tracks or wiring to the supply pins of an opamp from the supply creates a high impedance at high frequencies. Therefore fast opamps need bypass caps mounted directly at their supply pins. I recommend a 0.1uF ceramic disc in parallel with a 10uF or 100uF electrolytic. The pcb should be planned so that the noise currents from the supply through the bypass caps doesn't travel through input ground tracks.

Share this post


Link to post
Share on other sites

Audioguru,

While I figure out how to post a schematic perhaps you could answer a few more questions.  Are these supply caps your talking about in series with the supply, or between the positive and ground of the supply?  What starts to happen the further from the supply pins of the op amp you start getting with these caps?  I make my own PCBs (toner transfer method) so I make my traces as wide as possible, I imagine this helps.  Is there any other thing I can do to reduce inductance in working with the PCB?

Also if this helps I use a +/-15v supply on the opamps.  It doesn't appear at this point that I am losing anything in the audible range, but it wouldn't hurt to play it safe.  What effect would this inductance make regarding the likelihood of my opamps oscillating?

If might be helpful for me to post a pic of the PCB layout as I don't have the power traces on the schematic, they are so simple I just put them in on the second side of the PCB.

TG :)

Share this post


Link to post
Share on other sites

Hi TG,
1) The value of your input caps C3 and C4 cause a low frequency cutoff (-3dB) at 80 Hz which is bad for music having deep bass, but is OK for speech.
2) R3, R4. R9 and R10 don't do anything except add noise and should be removed.
3) What is P3 for? All it does is to overload IC1.
4) The values of R7 and R8 are too low for a TL072 to drive with low distortion. The minimum load for it to meet its spec's is 2K. See the graph of distortion vs load I found. Its gain is only 3 so your circuit with higher gain will have more distortion.
Why is your circuit's gain 100?

post-1706-14279142629355_thumb.png

Share this post


Link to post
Share on other sites

The TL072 is only 13V/µS, that's not exactly fast when compared to the 250V/µS of the AD843 mentioned earlier. A 100µF bypass cap on a opamp that's only capable of a few mA of output current is overkill. For small signal stuff, even more than an order of magnitude above audio, doesn't need anything that big. If the opamp were to draw some current, that would be a different story!! Parasitic inductance is about 2nH/inch, so you would need a LONG run before it became a factor. Yes, when it comes to power, short and fat is GOOD, but it depends on how much current is flowing!!

Share this post


Link to post
Share on other sites

Hi TG,
A sim doesn't show the 0.00008% distortion that you hear (?) with a non-inverting opamp. It also doesn't show the 0.000008% distortion that you can't hear with an inverting opamp.
Maybe the difference that you hear is caused by the inverting opamp circuit cutting-off the bass? Maybe you like to hear the extra hiss caused by the noise-gain resistor?

An inverting opamp circuit has a low input impedance. Therefore having an input balance control is difficult and the input coupling cap must be huge for good bass.

Share this post


Link to post
Share on other sites

I can say I know what I prefer, and the difference seems to be in the high frequency extention.

Nobody can hear frequencies as high as radio waves.

The first run through with the bleed resistors did sound noisier

Like I and the article said.

I will adjust the input coupling cap to be larger in the final design when I start using them, for now the 1uf will just be there to remind me.

Share this post


Link to post
Share on other sites

I thought this quote from a converstation I had with another designer might help others understand the reason  for pursuing a design with resistors that bleed open loop gain in an op amp circuit.  The idea is to allow the op amp to function in an area that it operates in a more stable condition.  For some op amps noise may be a consideration that will not make this practical, for others like the OPA637 this is a viable solution that would otherwise make the op amp unable to work in a certain application.  Some have found that most op amps, even those that are suposedly unity gain stable function better further from their oscillating point.  I think most of my problems have been with not taking into consieration the differences between non-inverting op amp implimentation and inverting.  The op amps become like different components.  We will see as I try the schematic in my last post, with the new PCB's.

"The implementation you show is correct enough, but keep in mind that
using the resistors from (-) input to ground do not actually affect the
output gain - they affect the feedback experienced by the op amp
circuit, the op amp operates with less apparent feedback and is thus
more stable that otherwise."

Audioguru, I agree with all that you said with the phase shift of various components in most audio applications, but in my situation I am not so sure it applies.  I have no crossovers in my power amp in the signal path, I have only a single capacitor in my speakers crossover and my tweeters are not reversed with the other driver in the two way speakers I have.  The caps I put in the design were there in case the the preamp circuit has an unexceptable offset that I need to decouple it from other circuits.  There are many reasons why things may sound different other than a difference in radio frequency extention.  I feel that various circuits and components increase in distortion as frequency increases in the upper ranges of the audio band because of their gain bandwidth not being great enough.  So the extention well abover human hearing is to make this distortion curve well out of audiable range.

Would you not agree that the TL0x series sounds better than the NE5532 op amp in the audio band because of distortion?  If this is so would that not be because of the gain bandwidth differences?  I am curious if others have experimented with op amp circuit design and found other interesting effects of circuit design  on them.  Like I said I like inverting input better than non-inverting, but I hope it wasn't because the circuit I used in each case was not what made the precieved differences.  Keeping an open mind is the beginging of understanding.


TG :)

Share this post


Link to post
Share on other sites

Hi TG,
There is your problem. You were using an OPA637 opamp that has full output at the top of the AM radio band! In order for it to go so high, its design causes it to oscillate with a gain of less than 5. Its high frequency response is so good (?) that in a non-inverting situation the pcb's wiring capacitance causes positive feedback making it unstable at gains much higher than 5.

The NE5532 and the TL072 have the same full output bandwidth of 100kHz.
Since the 5532 uses bipolar input transistors, it has a high input current but has low noise with a very low input impedance source like a phono cartridge (remember them?). It has a fairly high idle current and can drive a 600 ohm load well.
The TL072 has FET imputs with extremely low input current, so has low noise only with a high impedance source. It has a low idle current and doesn't drive 600 ohms. 

post-1706-14279142638114_thumb.png

post-1706-14279142638297_thumb.png

Share this post


Link to post
Share on other sites

I'm confused... Right off the bat I'll say audio is not my thing, but I am very familiar with things in the analog world. In a "first order" DC analysis of the circuit, R3, R4, R11 and R12 do nothing, as one input to all the op-amps is hard tied to ground, and the other is at a "virtual ground" of sorts. Aside from the input offset voltage, there is no voltage potential across any of those resistors, and the additional current the op-amp would have to source or sink due to these resistors is a few orders of magnitude smaller than the normal current flowing. So, my question is this, what exactly do they do from an audio perspective??? What is the math behind the phenomena, as it doesn't figure into the DC gain equation?? As a general observation, in the analog instrumentation world, none of those op-amp inputs would be tied directly to ground. In that particular configuration, they would be connected to ground thru a resistor that is equal in value to the parallel equivalent of all the resistors connected to the op-amp minus input node.

Share this post


Link to post
Share on other sites

Hi Indulis,
Those resistors from the inverting input to ground are used to make unstable opamps more stable, by adding "noise gain". I use stable opamps so have never used them.

Opamps with ridiculously high operating frequencies are made with inadequate compensation capacitors, so they oscillate if their closed loop gain is less than 5, and/or have a capacitive load. The addition of noise gain allows the opamp to operate with low closed loop gain but adds noise. Usually a capacitor is added in series with the noise gain resistor so that the input offset voltage isn't amplified.
Here is an applications note from National Semi that explains it pretty well:
http://www.national.com/an/LB/LB-42.pdf

Share this post


Link to post
Share on other sites

some op amps are on the verge of oscillation and will exibit some problems

That's why I use ordinary audio opamps, then I don't worry about instability and don't need to use RF construction methods. Without adding noise gain, then my circuits have the low distortion provided by full negative feedback and are quiet.

I think using an opamp for audio that has full output up to 1.7MHz is ridiculous. 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

×
×
  • Create New...