Op Amp (uA741CN) Help

[Mechatrommer]

well its a shame but i have to put it here. i never actually played with an op amp. last night i tried, but a simple thing is much of a discouraging. i tried closely by the book, but its not what i'm getting. Here's an example of a simple non inverting amp that i tried...

R1=100, R2=100, V+=3.3V, V-=0V, the signal that goes into +input is from my FG.
the gain should be (1+R1/R2) = 2, should be output = 2 * input right?

this is what i'm getting (yellow=signal on non inverting input, blue is the op amp output)...
top: 0 (no) signal from input, note the output is 0.9V offsetted
bottom: input of sine wave, the output? WTF?!
pls help noob here

ps: i tried
R1=10M, R2=1K, the gain should be 10,001, even a slight voltage difference in the input should show up in output right?
but none, all i get is some random distorted noise.

[Afrotechmods]

You might want to power it up with more than 3.3V, and use a dual rail power supply if you want to use that exact circuit. Keep in mind that the 741 is one of the most outdated POS op amps in existence. Also R1 and R2 of 100 ohms is a bit low - try 1k and 1k.
And with R1=10M, R2=1K you would be attenuating the signal, not amplifying it. You've got the resistors in the gain equation backwards. Even if you did reverse them a gain of 10k is too much - try to keep it around 10 or 20 to start with.

[Mechatrommer]

i reversed the 10M and 1K resistor, afaik it will turn into something like comparator? saturated at V+ when the input difference is positive, and saturated at V- when the input difference is negative. thats what i learnt, at least.

[mkissin]

If you check the datasheet (http://focus.ti.com/lit/ds/symlink/ua741.pdf) that particular opamp is only spec'd to swing to within 3V of the rails (page 5, Vom). That's worst case, but you should always design with worst case in mind. So, with you only having 3.3V rails, your opamp might not have any output at all, by design! You also have input voltage concerns at such a low rail voltage, as the input common mode voltage is also only spec'd to within 3V of the rails.

On another note, as Afrotechmods said, the 741 is pretty much the most useless opamp money can buy. Try an LM358, which is much easier to use, is spec'd for 3V rails, and allows input and output voltages that swing to the nagative rail (though not the positive rail).

[Mechatrommer]

i know 741 is the most useless op-amp today. i'll consider getting the latest op-amp such as LM358 during my next order.

If you check the datasheet (http://focus.ti.com/lit/ds/symlink/ua741.pdf) that particular opamp is only spec'd to swing to within 3V of the rails (page 5, Vom). That's worst case, but you should always design with worst case in mind. So, with you only having 3.3V rails, your opamp might not have any output at all, by design! You also have input voltage concerns at such a low rail voltage, as the input common mode voltage is also only spec'd to within 3V of the rails.

so what i suppose to do?
dont blame me man, i'm mechanical!

[mkissin]

so what i suppose to do?
dont blame me man, i'm mechanical!

Realistically, you're going to need higher voltage rails. 3.3V simply isn't enough for this opamp. If you can, use +/- 5V (or 0-10V, depending on how you want to look at it) or even higher.

You also want to avoid either your input or your output going to within 3V of any of your rails. That means, with a 0V and 10V rail, keep your input and expected output within a 3V to 7V range. Depending on your input signal source, and how you connect it, you might have to AC couple it using a capacitor.

[saturation]

The 741 is as historical a linear chip as the 555 timer is; and they are both obsolete by today's standards but they can be used for frequencies under 1 MHz.

http://en.wikipedia.org/wiki/Operational_amplifier

A problem you have is your power source, it needs positive and negative supplies.  You cannot use a single supply such as 0-10V to give +/-5V unless you use a voltage divider to split the supply, which will create quite a design headache for a newbie.

It wasn't until the late 1970s that single supply op amps changed the face of using op amps, and that didn't become popular until a decade later, as most folks designed with dual supplies; that's why tracking dual supply bench PSU became popular and are not as popular today.

Second, how far the output swings to the input is a function of gain, and if you exceed the specified rail limits, it just clips, so reduce the gain.  Since there is no output on mecha's scope, thats probably not the problem.  Rail limit was later overcome in later designs, but it will still clip if gain > V+, as expected.

Why play with the 741 or 555?  Because they are very forgiving for errors and good for experimenting, like what is in mecha's design; chances are it will still work if you put the right supplies in after an design error.

If you don't have a negative supply, take 2 9V batteries, connect them together to make 18V from the combined + and - end and use the joint were the 2 batteries meet as your ground.

  

[tjw]

Hi,
    I would follow mkissin's advice.

Specifically, power the opamp from +10v and bias the non-inverting input at ~ 1/2 the supply voltage. This is easily achieved by connecting a 10k resistor from the +Vsupply to the non-inverting input and another 10K resistor from the non-inverting input to gnd. Theory tells us that since no current flows into the non-inverting input (not strictly true in real life) the quiescent voltage at the non-inverting input will be :

Vsupply*(10k)/(10k + 10k) = Vsupply/2 and the output will also be at approximately Vsupply/2.

If you then apply a 1v p-p sinusoidal signal to the input, the output will be a 1v p-p sinusoidal but the Vmax and Vmin will be :

(Vsupply/2) +0.5 v and (Vsupply/2) -0.5 v  = approximately 4.5 to 5.5 volts if Vsupply=10v.

I think you are doing yourself a disservice by listening to advice like "....Keep in mind that the 741 is one of the most outdated POS op amps in existence. (Afrotechmods)" This is clearly incorrect. Read chapter 4 of "The Art of Electronics", and page 211 in particular.

The web is awash with information about the 741, so go to it.

The LM358 is quite similar to the 741 - yes, it is designed to be used in single supply circuits and can swing it's output closer to the supply rails but other than that works much like a 741.

I would stick with the 741 and learn everything you can about it. For example :

- why won't the output swing to the supply rails ?
- what is the importance of the 741's slew rate ?
- how do the input offset specifications affect the performance ?
- what is output offset ?
- what is latchup ?
- what is the bandwidth ?

All these questions apply equally to the latest opamp with gold-plated specs so you may as well start at the beginning with an easy to use opamp like the 741. One day you may be called on to repair a piece of equipment that uses 741's.

Regards, tjw.

[Mechatrommer]

thanx guys. i'll keep experimenting esp on the V+ V- thing. i also played with 324 that worked quite differently.
why i played with 741 (and 555?) because a have a few in my stock. so i think i'll miss the game if i dont play with em.

[scrat]

You can make it work from a single supply (or, better, the IC doesn't know wether supply you are using, since there is no 0 V pin, just two +,- ones). The problem is just it doesn't work if the inputs are too close to the supply voltages.

So if you need to amplify just the AC part of the signal, then adding an offset will make it possible even from a single supply. tjw forgot to mention that, if the amplification is >1 (= R1<Inf) and you don't add a second capacitor in series with R1, the opamp will likely go into saturation, or at least will work around an offset that is 1+R2/R1 times the input offset (Vsupply/2). See below.

You know the impedance of C is 1/(j*2*pi*f*C). Then the amplification factor for the DC component (offset on n.inv pin) is: 1 + Z2/Z1 = 1 + R2/(R1+1/(j*0*C)) = 1. For frequencies above 1/(2*pi*(R1+R2)*C1) and above 1/(2*pi*(R3||R4)*Cin) the AC input to output gain will be about 1+Z2/Z1 = 1+R2/R1 (since the caps are approximatively a short), which is the expected value.  

[saturation]

I agree with tjw fix but while that works for pulling the 741 input to 5V using a 10V supply you'd need to be aware that you are putting 5+V into the output of your signal source, which is referenced to 0V.  Most function generators will tolerate this, some will not, so rather than risk overloading the signal source's input consider truly splitting the supply.  One safety feature is adding a blocking capacitor to reduce currents flowing into this source's output; connect it between the signal source output and the 741 input. 

[tjw]

Excellent advice - shows how much can be learnt from experimenting with such a simple circuit !

Again, I'd like to mention Chapter 4 of "The Art of Electronics". This explains the effects of adding capacitors to opamp circuits and what it does to the frequency response. Essential information from authors that really know what they are talking about.

A reason for my suggestion of biasing the input in the single supply mode is that over the years I have built hobbiest-grade test equipment from published magazine articles that use this technique and it performs very well. These designs most often use an external transformer (plug pack) and a simple single supply regular regulator circuit. Advantage - low cost, simplicity and safety since no mains wiring required.

Regards, tjw.

[Mechatrommer]

You can make it work from a single supply...

i tried the circuit in the picture... not working. this thing is crazy. at idle (no signal, or 0 volt at non.inv input) the output stays at 4.6V level. gave it a 1Vpp sine, nothing, the output stays at 4.6V, nuts! its like some short circuit happened inside, but i tried different chip earlier, they all behave the same.

R3=R4=100Ohm, confirmed 2.66V at node connecting them.
Cin=1nF, so impedance at 1MHz not far from 50Ohm.
i dont put Ci and Cout, and i think it doesnt matter and wont change anything.
R1=R2=10M Ohm.

[scrat]

If you don't put C1, the cap in series with R1, of course it will not work as you want: the DC component will be multiplied by 2, hence ~5V output (it's a typical error).

R3 and R4 should be higher (let's say at least one order of magnitude): otherwise the output impedance (50 Ohm usually) of the FG + Cin impedance in series will divide the input with R3||R4.

It's not a good rule to use high (~10MOhm) resistors on the feedback, unless you want the power consumption to be very low: every small parasitic capacitance will limit the bandwidth, and every noise will easily couple, not to mention the resistance itself generates voltage noise  proportional to sqrt(R) (http://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise).

[tjw]

Hi,
    try R3=R4=10k, R1=R2=10k to 100K, Cin=100nF to 1uF and C1=1uF to 10uF - polarised electrolytic for C1 ok.

This circuit is explained on p178/179 in The Art Of Electronics.

Keep at it, it will work !

[Zero999]

The 741 is as historical a linear chip as the 555 timer is; and they are both obsolete by today's standards but they can be used for frequencies under 1 MHz.

Even that's optimistic.

The old 741 isn't much good at frequencies much over 8kHz because the open loop gain and slew rate are too low so the 741 is useless for audio.

The LM358 also suffers from the same problems but is more useful because the inputs work down to 0V and will work down to 3V.

The TL072 is a better choice for audio because it's low noise and has a high enough slew rate but it has similar power supply requirements to the uA741.

[saturation]

Its not that terrible but its certainly obsolete given newer op amps available with better responses.

At least 3 major items limit its performance substantially: gain bandwidth product, output impedance and its slew rate.  

I found this link which summarizes my experience with 741, it was written over 10 years ago:

http://sound.westhost.com/highspeed.htm

You can use it on audio applications, but how it rolls off high frequencies depending on what gain you set it at and what output voltage you require.  Folks who used the 741 were able to work around those limits 30+ years ago as that opamp was widely known and dirt cheap.

Checking the chips you mention I am shocked that the 741 goes for over $1 each from Digikey, who sells the TL072 for 60c, and the LM358 for as low 12c.  There are many better spec'd op amps far cheaper than any of those three today.

Nevertheless, if you have any old op amps around, you should play with either it to see what they do, moreso to know op amps rather than trying to design something operational in the 21st Century using 30-40 year old technology, its also not cost effective given those prices.

I remember when the LM741 was 20c each!  I guess these $1 is the "antique roadshow" prices [ this thread triggered my interest in this old chip and dust out the cobwebs!]

The 741 is as historical a linear chip as the 555 timer is; and they are both obsolete by today's standards but they can be used for frequencies under 1 MHz.

Even that's optimistic.

The old 741 isn't much good at frequencies much over 8kHz because the open loop gain and slew rate are too low so the 741 is useless for audio.

The LM358 also suffers from the same problems but is more useful because the inputs work down to 0V and will work down to 3V.

The TL072 is a better choice for audio because it's low noise and has a high enough slew rate but it has similar power supply requirements to the uA741.

[Mechatrommer]

@tjw and scrat: thanx for the circuit advice, i somehow manage to get some signal on the 741's output, the annoying thing is the dc offset (~2V) from the 741 output. Cout is needed to block it, yet the clipped signal remained (as in ua741.png ch2, except graph moved down to origin with Cout presence). i also played with another op-amp/amplifier with minimal effort/parts (shown below, ch1 yellow=input signal, ch2 blue=output), esp the tda1517 stereo amplifier i soldered out of an old soundcard, i managed to get 12Vpp out of it last night and produced the sound on the speaker, but now its gone i think, connect the wrong pin earlier, whats left is the distorted output at 1517.png. and i stumbled/interrupted with transmission line issue that i posted on another thread, luckily i didnt toss my FG to smoke. i can see all of them are not able to catch up with greater than 1MHz signal, so now i know the limitation. thanx all guys for the help.

pictures setup:
signal: 2Vpp 5KHz sine.
V+: 5V V-: gnd = 0V

[Mechatrommer]

anyway. also am reading AoE (chapter 4) esp the golden rule. There is however on page 183 that i dont quite understand...

There must always be feedback at dc in an op-amp circuit. Otherwise the op-amp is guaranteed to go into saturation.

[oPossum]

no feedback == comparator == output always at lowest or highest voltage

dc feedback (resistor) == amplifier == output clips with too much input

ac feedback (capacitor) == integrator == output always moving toward lowest or highest voltage (at rate determined by input current) - will get there eventually

[allanw]

There must always be a path for DC current to flow.

http://www.opamp-electronics.com/tutorials/practical_considerations_bias_current_3_08_15.htm

[Mechatrommer]

o ok, misunderstanding, i thought "feedback at dc" means either inv or non.inv input must be tied to V+ i know the basic/ideal model of an op-amp (hi Z inp, low Z out, inf gain) and what it does. but as i said, what i learnt in theory is not what i get in practice. just try a simple voltage follower with 741 (-ve feedback shorted to ouput), does it follow the voltage on +ve feedback? not gonna happen i think with 741 (a voltage "not" follower), tried that.

[scrat]

uA741 is singing (cfr [youtube][/youtube]):

"I will follow in(put)
ever since you put me on I knew
there is an input too deep
an input so high it can keep
keep me away
away from work"

It works if the input stays between the two output limits (gnd+3V, +Vcc-3V). So give it a DC offset input, and a properly high supply. It must work!

[Mechatrommer]

It works if the input stays between the two output limits (gnd+3V, +Vcc-3V). So give it a DC offset input, and a properly high supply. It must work!

this is the golden rule i'm looking for! simply put, i was violating this rule (due to my FG limitation ±3.5V max, and err... knowledge). to be specific from my test (GND-Vcc supply), the limit is (GND+2V)-(Vcc-0.3V) for the particular voltage follower circuit, distortion will be prominent working near those limit. so now, yes, its working! thanx scrat.

i also was told earlier by mkissin, but i did not quite get it.

You also want to avoid either your input or your output going to within 3V of any of your rails. That means, with a 0V and 10V rail, keep your input and expected output within a 3V to 7V range. Depending on your input signal source, and how you connect it, you might have to AC couple it using a capacitor.

[scrat]

Attention to details (which is something I really lack of) is what usually drives not understanding simple errors.
I always blame there's something unexplainable happening, just to find myself wrong after a while (most times, not always..).