LM 393 discrete components circuit

Thread Starter


Joined Sep 28, 2021
Hi all, I saw this article at here . I tried to simulate it in LT Spice, however, not an expected output. What mistake have I made? . I have attached my LT Spice below for amendment. Thank you for reading and have a nice day :)



Joined Oct 2, 2009
Q8 and R3 do nothing.
Q8 is called an open collector output, i.e. the collector is uncommitted. It needs to sink current from a positive voltage source.

R3 should be connected to a positive voltage source, not ground.


Joined Aug 21, 2008
While you are editing your simulation file, you probably should make R1 larger. As it is at the moment when the inputs are near ground you would be running well over 100 ma through your differential pair. At the very least that would cause very high input bias current. Something like 10k for R1, or even higher, would be more appropriate. R2 and R3 also should be made much higher in resistance. As it is now those two resistors would dissipate several watts if the inputs were close to ground. Notice that Eric, in his post above, used current sources.


Joined Aug 7, 2020
Where did you find that article?
The information about the speed of the output stage is complete nonsense.
It is the presence of C1 which slows down the output stage of the 741, nothing at all to do with its having a push-pull output stage.
In fact, open collector outputs are slower, because they rely on a resistor to pull the output high, not a transistor to drive it high.
The open-collector output is used so that it can interface to logic at a different power supply voltage to the comparator.
Nor does the push-pull output stage "enable linear amplification" - you can make a linear amplifier with an open collector output/


Joined Jan 29, 2010
hi M,
You can trim R1, R3, & R5, to give closer uAmp currents if required [ R1 and R3 should be the same value]

Post your project when done.


Joined Jan 23, 2018
There are some very good application notes by the folks at National Semiconductor, written many years ago and they can be very educational. The interesting thing about fast comparators is that they are fast, and they can oscillate very well unless prevented. So a discrete component comparator may produce a of of high frequency signal

Audioguru again

Joined Oct 21, 2019
Why are you blasting away the input with +10V and -10V when the absolute maximum allowed negative input is -0.3V to avoid destroying the base-collector junction of the input transistor?
Simulation software does not read the datasheet and does not realize that.

Thread Starter


Joined Sep 28, 2021
Hi Mrss
This is one method using resistors, very basic.
View attachment 263263
Hi sir :), I have actually built the circuit and it works well :). However, I want to modify your circuit so that it can self-generate a square wave PWM as shown in diagram above. I have tried to follow the diagram to implement it in your circuit but output is weird. I have attached my LT Spice file below , hopefully to have some guidance on this :)


Thread Starter


Joined Sep 28, 2021
Hi sir
Hi Mrsss, @MrsssSu
This is your circuit corrected.
View attachment 263721
Hi sir, actually I disassembled the circuit a few days ago, so I have to assemble it back again. I might have forgotten about the correct polarity of transistors. I try to test the new updated circuit but output is 0. I have labelled the npn transisors as white dot and pnp with no label on the breadboard. Are my transistors correctly placed as in your schematic? I have checked that all components and resistor values are correct. I am using a UF 4007 (fast-diode)