Op amp circuit help plz

[J. Burnett]

FvM and bcarso,

Just to prove to myself that I'm not a complete idiot I submit the following (it's hard to believe I used to do this for a living).

https://obrazki.elektroda.pl/95_1305658172.jpg

The first figure is for my interpretation of the spec (i.e., +20 db at 20 Hz and -20db at 20 kHz. I no longer have access to SPICE or any other analysis software but I'm pretty sure this is right (maybe one of you guys could run it for me?).
The second circuit is for the flat passband with corners at 20 Hz and 20 kHz. My only concern here is along the lines bcarso noted about the 324 itself. The open loop gain is only (barely) 20 db at 20 kHz which makes this a questionable design. There's not much stuff around 20 kHz so it probably wouldn't matter for a low budget audio application? Maybe you could cut it back to 5 kHz for AM broadcast quality?

Best of luck to you guys

J. Burnett

 

[BradtheRad]

I'm looking at the most recent schematic. I see one part of the feedback loop that provides gain of 820k / 82k = 10.

The other part is a feedback loop that provides gain of 8.2k / 82k. This makes 1/10. Attenuation. It has a capacitor inline. Capacitors pass high frequencies and impede lows. Therefore doesn't that part of the feedback loop attenuate highs?

I don't know what the net effect will be.

 

[bcarso]

No offense taken here J. Actually your suggestion that this might be something for driving an earbud is plausible. In this case we don't want d.c. going through such! I fear we'll have to await the poster's guidance about what this is indeed for.

And yes, if it's to define an audio passband there are better ways.

Cheers,

Brad

 

[J. Burnett]

Hi Brad,
That's right. My original interpretation of the spec. was that it was an equalization circuit such as used for RIAA compensation. Those circuits provided a 20 db boost to highs and 20 db cut to lows (when recording) and therefore required a 20 db boost to lows and 20 db cut to highs (when playing back). The intention of this circuit was to do just that. The spec called for +40 and -40 so two stages were required of course. FvM later convinced me that that might not be the correct interpretation of the spec and what was desired was a flat spectrum of 40 db gain from 20 to 20KHz with rolloff at both ends of 40 db/decade. There was a second circuit that I now see didn't get inserted and so I'll try again here:

<a href="http://obrazki.elektroda.pl/60_1305663301.jpg"><img src="http://obrazki.elektroda.pl/60_1305663301_thumb.jpg" alt="60_1305663301.jpg" /></a>
URL:
https://obrazki.elektroda.pl/60_1305663301.jpg



Again, there will be 20 db gain for each stage so two stages are required to get the 40 db. My concern is if this is supposed to be a "high fidelity" circuit the low open loop gain at 20 kHz is marginal, but probably okay. In no case, however, would a single stage be acceptable.

Hope this helps.
J. Burnett

 

[FvM]

I see only one circuit rather than two. It has a zero at 20Hz, pole at 60 hz and a zero at 6 kHz. Thus the overall frequency response is 20 db/decade cut-off below 20 Hz, flat between 20 an 60 Hz (gain +20 dB), 20 dB/decade drop between 60 Hz and 6 kHz and flat gain (-20dB) above 6 kHz.

As I altready mentioned, the original specification is meant different in my understanding.

P.S.: Add Image is basically simple in edaboard, I think.

 

[bcarso]

Again, a 324 may be all that is available for this lab (?). My condolences, as it is noisy and slow. Once a guy who worked for Compaq told me about the trick of drawing current to ground to reduce crossover distortion, as if it were a big secret, and enabled him to get barely adequate performance out of active filters.

I confess I jumped into this thread because edaboard notified me that I was about to be further demoted due to lack of activity . FvM did not mean to be derisive I realize, but properly pointed out that I may have been supposing that this was to be "hifi" rather than some basic class assignment.

But if the poster is designing hearing aids...

 

[J. Burnett]

Right on all counts FMV.

As you know there will be a 3 db difference between the designed pole and the actual frequency response. As bradtherad pointed out end gains are correct and should provide a smooth transition to the end point specified frequencies. As we've already discussed my interpretation of the spec may well be incorrect and so I've provided a second schematic (which unfortunately got dropped out on the first try but added later in my response to bradtherad). In that much simpler circuit each stage provides a simple +20 db flat spectrum with 20 db/decade rolloff (per stage) past the specified pass-band.

bcarso, thanks for the comment, it's appreciated.:grin:

J. Burnett

 

[FvM]

As you know there will be a 3 db difference between the designed pole and the actual frequency response.

Yes, I've been sketching a straight line approximation of the magnitude characteristic.

the trick of drawing current to ground to reduce crossover distortion

It works of course, as said. No secret, simply class A operation.

 

[J. Burnett]

Hey guys,

I've been thinking about FvM's interpretation of the spec as a sort of bandpass amplifier. The only way I can make sense of this is if it was an exercise to illustrate the idiosyncracies of the non-inverting configuration (the ckt given was a hopeless non-inverting circuit). The problem I see is the +1 follower phenomenon-it's difficult to get the gain below unity. Perhaps that's what the exercise was supposed to illustrate. I've taken the time tonight to put together this not very clever circuit (but then it's a not very clever problem)
<a href="http://obrazki.elektroda.pl/18_1305769086.jpg"><img src="http://obrazki.elektroda.pl/18_1305769086_thumb.jpg" alt="18_1305769086.jpg" /></a>
URL:
https://obrazki.elektroda.pl/18_1305769086.jpg

Anyway, this ckt provides the 20 db/decade slopes as requested (I still find it hard to believe any college professor would specif a slope as "db" and not "db/decade"). As I was saying: anyway, the bandpass is controlled by the rc network preceeding the actual amp and the op-amp only acts as a simple amplifier. While the 20 db/decade is theoretically achieved above 20 kHz the openloop gain characteristic will cut in about 2 or 3 octaves above 20 kHz and spoil that. You might put a resistor in series with the 100pf cap but that's a little over the top if this is just an exercise to show difficulty of getting 40 db/decade for two stages. I didn't concern myself over the details of "offset" etc.

It's been fun. Great hob-nobbing with the real brains (apparently the few still left) of the worlds. I've got to go back to the real world next week but I'll try to stop in when I can.

bcarso, Fvm, bradthe rad...take care...there's damn few of us left.

 

[BradtheRad]

Mr. J. Burnett -

Few left, indeed. Remember when there were Heathkit and Lafayette Electronics retail stores? We're lucky there's still Radio Shack stores around.

About op amps. For my own part I confess that I've found it hard to get my mind around them, in the past 32 years that I've been working with IC's.

The basics of operating an op amp is portrayed as simple and straightforward. Nevertheless the more useful op amp configurations are non-obvious.

Whenever I've considered using an op amp in a project, I've had to search through all the circuits on every page of my Forrest Mims Notebooks, or my Linear Databook of IC's (3'' thick), or the Op Amp Cookbook by Walter Jung, or the Active Filter Cookbook by Don Lancaster. Or the book of circuits in my Radio Shack electronics lab with breadboard.

 

[J. Burnett]

Brad,

Good memories. I once built a Heathkit 25 watt stereo amplifier with vacuum tubes and huge transformers sitting on top of an aluminum chassis. When something is new there's a lot of good stuff published in the trade magazines and I think I still have articles I tore out and saved (probably in boxes in the attic). Engineers would do something clever and then publish a short article explaining their circuit much better than any textbook. The catalogues/data books always had interesting "application" circuits but more often than not they were just a starting point. The same thing happened when the Commodore 64 and Apple II came on the market. Magazines like Byte and Compute! published really good articles for a few years but the technology moved on and the market for that sort of article died. Later it's difficult to get in on that bottom floor and learn all the vocabulary, clever tricks and "tribal knowledge" except via the school of hard knocks.