Op amp circuit help plz

[masab_ahmad]

how to get an op amp circuit working with the following considerations in PSPICE:
1. Gain = 40dB
2. Bandwidth = 20Hz to 20KHz
3. The frequency response should have +40dB and -40dB slope for the corners of the bandwidth.
4. Assume a load of 50Ω.
5. Assume a sinusoidal input of 100uV amplitude.

i do get the first 4 options done at 1V ac with a 1V dc offset, but when i adjust to the 5th option the circuit doesnt give any gain..plz help

pics:

 

[vimalkhanna]

You will need a video wideband opamp and a capacitor at input to cater for DC shifts in V- terminal.The 100uV sine wave must be shifting with the bias current .
Try changes in the DC offset trim.Try to improve the gain-bandwidth product of the opamp.rather than increase the nmber of stages to enhance gain.

 

[masab_ahmad]

thanks for the reply...
wouldn't the input capacitor manipulate the poles?
i also need to use the same lm324 op amp for the job..

 

[FvM]

- a DC offset (0.5 - 1 V) is also needed with the sine voltage source
- a transient analysis has to be performed in this case
- for the AC analysis, you should select a reasonable frequency range (e.g. 1 Hz - 1Meg), or at least use a suitable magnitude range in the plot. Or can you detect any useful details with a -500 - +100 dB range?
- the 50 ohm load is nearly shorting the OP output, it only works at low DC output levels. Why did you select this unsuitable load?

 

[masab_ahmad]

@fvM
thanks
i gave a dc offset of 1 V and got the following with AC sweep, with the 50ohm load,
at 1V AC the output is just what i need but when i apply a signal of 0.1mV or 100uV the same db wave form shifts down by 80db, graph is attached, now i just need to shift it up by 80db at the same input, 0.1mV....
and the 50ohm load is ok i think...

 

[FvM]

5. Assume a sinusoidal input of 100uV amplitude

I understand, that a transient analysis is meant here.

 

[masab_ahmad]

using freq of 1kHz, 1v dc offset, and 100uV amp, i get this with the same circuit...
i'm not fully understanding you...the output shows a Vpp of 0.02...which i think shows a gain of 100?

 

[J. Burnett]

Hi masab_ahmad,

Your spec is apparently for an audio amp that we used to call an equalizer. When recording analog signals we always boosted the high frequencies and cut the low frequencies to provide much improved noise reduction and therefore fidelity (the noise is proportional to bandwidth and so boosting the highs provides great improvement). Your circuit appears to be the playback equalizer which cuts the highs and boosts the lows to get a flat spectrum back. In the old days things were boosted and cut by 20 db but as technology improved it was found the 40db could be used successfully with even better performance.
We can do a quick and simple analysis of your circuit by considering three extreme cases: At the very low end of the spectrum the circuit may be simnplified to only R3 and C3 and C3 will essentially be an open circuit providing unity gain through both amps (0 db). At mid-range the circuit will contain (hopefully) only R3 and R5 which will yield a gain of 5. The two amplifiers will give a total gain of 25 which will be sufficient to provide a gain of only +27.9 db. To get your required 40 db you will need a gain of 10 in each of the amplifiers, requiring the R3 be raised to (on the order of) 10K. (Actually the complex impedance of the circuits will have to be considered here). Finally, at the extreme high frequency, only C5 and R5 will be in the circuit which is where we want a cut in gain to 0.1 (for both stages in order to yield a total of -40db). That would put the capacitive reactance of C5 at about 0.1 kohms. at a frequency of 20 kHz. The frequency response of the circuit should be a downward slop of -20db/decade throughout the passband of 20 Hz to 20 kHz, as that is all you can obtain from capacitors in single pole networks.
As noted by the other responses, when your input signal goes positive the output will go positive in the non-inverting configuration, but when the input goes negative the output will also go negative. Since you're using a single +power supply you will have to somehow bias the output upward to accomodate this fact. You need only provide this offset bias on the first stage of course. It could easity and reliably be provided by a voltage divider network to which your input transducer is returned.
Good luck...I hope some day our countries can be on better terms.

 

[masab_ahmad]

wow thanks!

 

[J. Burnett]

Approximate values might be: R3 = 820kohms, C5 = .001Ufd, R5 = 22kohm, C3 = 1Ufd
These are arbitrary values that can be scaled up or down to suit your purposes. Above 20kHz the circuit continues to roll off at 40 db/decade. Below 20 hz gain should roll off (eventually) at about 40 db/decade until it eventually comes to unity gain. It's been my pleasure. Good luck

 

[J. Burnett]

Okay, if it's an IQ test you got me. I assumed your circuit would handle the sxpecs. We need a minor modification (see figure)

 

[FvM]

@J.Burnett: Seriously, I don't understand what you want to achieve. The circuit in the original post basically keeps the specified frequency characteristic, you can see it in the posted simulation results. There are some minor problems, but not related to this point. Your suggestions are completely missing the specification.

P.S.: The low frequency cutoff is not complete, because the circuit has a +1 DC gain. To achieve an exact high pass characteristic, the circuit topology has to be changed. But I don't know if the problem specification is asking for other topologies.

 

[J. Burnett]

Hi FvM,
Where is Bochum? My family and I lived in the Frankfurt Am Main area for eight years back in the 1980s. My children remember that period very favorably. We go back from time to time on vacations-spend a week in Munchen last May.
This is an audio circuit, the spec being for the standard audio spectrum of 20 Hz to 20 kHz. There is no specification below 20 hz so the unity gain (at zero Hz) provided by the original design is not in default. The spec is for a gain of +40 db at 20 hz to -40db at 20 kHz, which implies a -2 slope over most of the frequency band specified. The stability of the circuit above the 20 kHz frequency is not specified but for anyone familiar with op-amp technology it is apparent this circuit will almost certainly not be unstable (unless some ungodly abortion of design is hatched up).
So, the problem here is to provide a feedback loop that will provide +40 db at 20 Hz and also provide -40 db at 20 kHz. The original design wisely elected to use two op-amp stages each providing +20 db at the low end and -20 db at the high end. Since a capacitor across the output to input feedback resistor will provide -20 db/decade rolloff it's apparent it will take 2 of the 3 decades specified (i.e., 20 Hz to 20 kHz) to obtain the 40 db range required of each stage. This implies (keeping the desire to make both stages identical) that we make the first pole about half a decade above the 20 Hz lower specified limit ( i.e., about 63 Hz) and the upper break point about half a decade below the 20 kHz specification. It is also recognized that, with the practical circuit obtained with this sort of circuitry, the actual frequency fesponse will be 3 db below/above the break points, and the circuit response (if properly designed) will thereafter smoothly approach the gains of +10 and -10. I included a circuit with component values to obtain such a response but it has somehow disappeared (I'll include it again with this posting). at the spec frequencies
I hope this helps but if you have further questions don't hesitate ask. As I noted in the original posting this circuit is typical of what was called an equilization in the latter days of analog recording/playback. In the earlier days of 331/3 LP vinyl the equalization was standardized to what was called RIAA which provided only +20 to -20 db at 20 hz to 20,000 hz. In the latter days of analog people like Dolby and DBX realized they could improve signal to noise further by using +40 to -40 db and were not restrcted to using RIAA in their internal designs. DBX advertised they could obtain a dynamic recording range of 100 db and I believe they really could (with an acceptable overhead). This circuit was not explicitly stated to be one of those equalizers (for the playback side) but it is otherwise identical to such a circuit.

J. Burnett[

 

[bcarso]

The remarks about input d.c. bias are very important, and it sounds like with a volt the circuit basically works. Note however that the 50 ohm resistor is now pulling about 20mA all day long. The 324 can supply this, but each section doing that will dissipate (12V - 1V)*20mA = 220mW (minimum) which will warm things up considerably. If this is for stereo audio then we'll have at least 440mW to handle. That's within ratings for a 25C ambient, but it will be getting warm.

The other drawback is that with that stiff loading the loop gain of the second section will be reduced appreciably --- exactly how much depends on 324 internal circuit details. Since the devices that provide a positive output current are compound emitter followers ("darlington"), this may not be horrible. But this loading will most likely entail a signficant increase in distortion, particularly at high frequencies where one is running out of loop gain.

However, there is one advantage. The 324/358 family is notorious for having high crossover distortion. It's been a long-standing technique to pull enough current from the output, typically by loading to "ground", to force the output stage into class A operation, thus dodging the crossover distortion. Although I've never seen anything as aggressive as 50 ohms, it may work adequately.

By coincidence, recently a client had me look at a microphone filter circuit also using such opamps. The circuit had several problems besides, but the designer(s) asked for even more gain --- a whopping 34dB per filter section, and had no loading to reduce crossover distortion. Overall, a very poor audio design.

 

[FvM]

@J. Burnett: I think, you simply misunderstood the specification. In my view it's asking for a flat 40 dB gain with 20 Hz (L.P.) and 20 kHz (H.P.) corner frequencies. But it's not very clearly stated and I may be wrong as well.

@bcarso: I wonder, if this simple exercise should be analyzed that serious. Your possibly thinking more thorough about it, than the author ever did. Personally, I'm motivated to think about real technical problems rather than artificial ones.

 

[bcarso]

FvM, although the author has not disclosed the application, and I have merely inferred that it is audio, are you saying that crossover and other distortions, and excessive power dissipation, are not "real technical problems", but artificial ones? Vielleicht etwas ist nicht in Ordnung hier.

 

[FvM]

I mean, that the problem in the original post sounds like an exercise rather than a circuit that serves a serious purpose. I simply refuse to improve this absurd design. I completely agree, that it's pretty useless to load a LM324 with 50 ohm, which implies, that the usable output voltage range is restricted to about 1V. I also agree about the performance of the amplifier for audio applications, although you don't get crossover distortions in this circuit - it's pure class A.:smile:

 

[bcarso]

Ganz gut! Ich verstehe. Perhaps we should criticize the instructor for a bad exercise. Vielen Dank.

 

[J. Burnett]

Hey Guys,
Let me make the following observation: The spec is very informal! It requires 1) 40 db gain and it says 3) the corner frequency slopes are +40 db and -40 db. Would that be 40 db/octave or 40 db/decade or, as I read it, as an informally stated gain of +40 db at 20 Hz and -40 db at 20kHz (in which the units of db would correctly be given). The input is 100Uv (is that peak or peak to peak or rms?). Assuming it's peak volts the offset would need to be somewhat more than 10 mv since the 40db gain would yield a 10 mv output. This being so, and recognizing the DC gain is unity, the DC output would only be 10 mv yielding a power dissipation in the load of (E^2/R ) 1e-4/50 which I think we can safely ignore.

Speaking of reality the requirement of 40 db/decade outside the bandwidth of interest (20 Hz to 20 kHz) sounds like a filter circuit...maybe something to prevent aliasing in the a/d converter following...but there are much better configurations, and a 10 mv signal is still pretty small...something you would mix in with other inputs in a pre-amp.
The 50 ohm load sound like an earbud??? In that case an equalizer might be desirable.
If this is just a filter there are much better configurations. Two inverting amps would meet the spec very simply, but that would remove the subtlety of question. If this is some sort of textbook exercise or test question the "super RIAA " equalization view is a much better problem. I think we should strike for a better statement of the specification.

Hang in there.
J. Burnett

 

[J. Burnett]

FvM and bcarso,

My apologies. I read what I expected to see into the spec. Either this circuit is put together by knowledgable amateur or .... I also slipped up on the non-inverting stages. The cirrcuit I put together was for an inverting stage...sorry about that...my brain stops working occasionally....two inverting stages would make this so much easier regardless of how you interpret the spec. The exercise was fun nonetheless-hope I didn't offend anyone?

J. Burnett