Do i need to buffer op amp outputs?

Thread Starter

Fluxor1964

Joined Jun 11, 2015
182
I just built a function generator using an LM324, I have the waveforms I want available to me on the relevent pins of the chip, do I need to buffer these before I stuff them into the base of a transistor to test some transistor theory, or can I go direct from the chip?

Neil
 

crutschow

Joined Mar 14, 2008
28,542
You don't need to buffer as long as the output current (including through any feedback resistors) is below the maximum op amp rating, usually about 10-20mA.
 

Thread Starter

Fluxor1964

Joined Jun 11, 2015
182
I've been reading about impedance, something about going from a high to low impedance may cripple the waveform...... That won't be an issue here?
 

crutschow

Joined Mar 14, 2008
28,542
I've been reading about impedance, something about going from a high to low impedance may cripple the waveform...... That won't be an issue here?
The op amp has a very low output impedance as long as the maximum output current is not exceeded.
So, for example, with a 15V supply it should be able to deliver 10Vpk into a 1kΩ load without significant distortion.
 

AnalogKid

Joined Aug 1, 2013
9,630
All of your outputs are directly out of an opamp output pin (a low impedance point), so it all depends on the load current and the frequency.

Load current - the LM324 is an old, reliable workhorse, but its output stage is not as "stiff" as newer devices. I think any output current over 10 mA will introduce amplitude and crossover distortion.

Frequency - again, the LM324 is over 40 years old, and opamp design and fabrication techniques have improved a lot since then. Neither its gain-bandwidth product nor its slew rate are very high. Even at unity gain or an inverting gain of 2, the part does not have enough forward gain and slew rate to keep the frequency response flat much above 5 kHz or 10 kHz. When the sine wave starts to look like a triangle wave, that's slew-rate limiting.

ak
 

Bordodynov

Joined May 20, 2015
2,949
This operational amplifier operates without any hiccups with a load resistor of the order of 100 kOhm. This is due to the fact that under high-resistance loads, negative levels are formed by the built-in current generator of 50 μA. (100 kΩ * 50 μA = 5 V). For a low-ohmic load (greater than 50 μA, a PNP transistor should be turned on.) In effect, the output transistors operate in "Class B." This leads to distortion.
To reduce distortion, an additional resistor is required, which will transfer the output stage to the "class A" mode.
 

OBW0549

Joined Mar 2, 2015
3,566
In post #10, @Bordodynov has given a good demo of one of the main drawbacks of the LM324: crossover distortion, and what to do about it-- pre-loading the output with a resistor to the negative supply to force the chip to constantly operate in Class A mode.

Overall, the LM324 is not a very good op amp by today's standards: it's noisy, has high input bias current and a lot of input offset voltage drift; these make it useless for precision applications and anything with high source impedances. Neither the inputs nor the output work rail-to-rail, making it awkward to use in some circuits. Its output sink current is rather weak, at only 5 mA. And it's not very fast, making it not very useful for a lot of audio applications.

Back when the LM324 was first introduced (when dinosaurs still ruled the Earth) it was pretty nifty; but these days it doesn't have much going for it except being dirt-cheap and readily available almost anywhere. Otherwise, it's just a mediocre op amp.

EDIT: Regarding the need to buffer an op amp output, there is another circumstance besides the need for more output current than the op amp can deliver by itself: driving capacitive loads (such as a load connected to the op amp through a long coaxial cable). Op amps generally do not like to drive capacitive loads, as these create phase shift in the output which can lead to overshoot and ringing in the op amp's transient response, or even continuous oscillation. Buffering the op amp output is one way of dealing with the problem.
 
Last edited:

OBW0549

Joined Mar 2, 2015
3,566
What would a better choice of op amp be, preferably with the same pin out?
Just about all quad op amps use the same pinout; it's pretty standard anymore. I reckon there are a few exceptions, though I don't know what they are offhand.

Provided you're not trying to go too high in frequency with your function generator (like >10 kHz or so), the LM324 may actually be adequate provided you add the pre-load resistor suggested by @Bordodynov.

I've settled on two quad op amps as general-purpose "jelly bean" parts: the TL074 for designs using traditional +/-10V, +/-12V or +/-15V supplies, and the LMC6484 CMOS op amp for low-voltage, single-supply applications (15V total supply voltage, max). The LMC6484 has rail-to-rail inputs and outputs, which makes it handy for designs that operate off a single +5V supply.

Neither one of these is as cheap as the LM324, with the TL074 selling on Digi-key for $0.57 each in quantities of 10, and the LMC6484 for $2.61 apiece. But for me, the extra benefits they give make the higher cost worthwhile.
 

Thread Starter

Fluxor1964

Joined Jun 11, 2015
182
I need a little more frequency than that, like I said I play with audio so I need 20khz, I'll play with that resistor idea when I get a chance.

Thanks.
 

RichardO

Joined May 4, 2013
2,271
Unless you use a very, very fast op-amp for the comparator function you will have distortion in the tips of the triangle wave.

Here is a circuit based on a clever CMOS 555 timer based triangle wave generator posted here on ACC some time back.

Eventually you will want a sine wave output so I added a sine shaper to the design. I also added a level shifter to get a square wave for driving logic circuits.

AAC_555_Fg.png AAC_555_FG_ckt.png
 

Attachments

RichardO

Joined May 4, 2013
2,271
I couldn't help myself. All I intended to do was clean up the resistor values around the sine shaper circuit. I did do that.

I could see the non-linearity in the original circuit. So, I added a real integrator to get better triangle wave linearity. This required new timing cap values. Fixed those.

Then I decided that the resistor values in the level shifter were drawing more current than needed and fixed that.

In the process I noticed that C5 was drawn with the wrong polarity. Fixed that.

Added and fixed some notes.

I think I am done now. Maybe i need to be fixed. :D

upload_2017-7-18_16-10-53.png
 

Attachments

Top