Need help with comparators

Hi Az,

it cannot be lower than -0.3V, not 0.3V.

Kai

Right sorry.

I'm testing it and i noticed with scope that the output fluctuates before switching to ground. It's possible to avoid this behaviour?

Thanks

Hi Az-eddine,

On a 5V supply, the valid input range is -0.3V to 3V (over temp). Both of your inputs can be out of that range supplies less than 5.5V.

330 ohm is a heavy output load at 5-12V supply...That is 15mA at 5V, and 36mA at 12V. Current limit is ~16mA. That may be the output  "fluctuation" you are seeing. A 3.3K for pulling-up a transistor base is more like it.

Can you please provide a schematic of what you are doing, and possibly a scope photo of the "fluctuation".

Please also review the LM393 family application note, particularly sections 2.2 and 3.1

 Application Design Guidelines for LM339, LM393, TL331 Family Comparators

Yes, a schematic would be helpful.

Kai

Az-eddine

we will await your response

Chuck

Hi,

I read that the valid input range is -0.3V to 3V so the test was done with Vref 2.5 Volt (i'll use something to drop input signal in the final stage) and power supply 5V.

With fluctuation I mean an oscillation of the LM393 out when Vin is arround Vref.

Please find attached the schematic.

Hi Az-eddine,

when your input voltage is 0...+5V or by other words 0...+2.5V behind the voltage divider, then your Vref should not be +2.5V as well. It's better to choose a smaller Vref then, as shown here:

Also, I would add a hysteresis by the help of R7. (Curve "VP_4" is the voltage directly at the +input of LM393.)

az_lm393.TSC

Kai

Great support as usual, thanks Kai

Az-eddine,

I believe you have the information you need at this point but please let us know if anything additional is needed.

Chuck

Sorry How can I change this applying VG1 on positive input of comparator and fixing the voltage on negative input?

Sorry but I'm new with comparators and it is not so clear how it works and how to choose positive feedback resistors

Hi Az-eddine,

you could add a second transistor stage as an inverter. By this the hysteresis setting scheme at the +input of LM393 could be kept unchanged.

Kai

anyway please find attached the oscillation during switching.

drive.google.com/.../view

Sorry but can upload your pictures directly. TI has a policy against accessing external shared drives.

Chuck

sorry, I can't see that either.  A simple image would work like a JPG.

And when you are getting the ringing at the output when the input crosses the threshold, is this on an actual board?  How are you observing this?  Comparators have fast fall times at the output, so if careful consideration to layout and supply bypassing is not applied, you can get coupling to the input.

Now you should be able to see it.

same behaviour passing from low to high or viceversa.

I'm using a bread board with bench power supply.

Thanks

I can now see your output response but I would need to see your input at the same time.  Also, is this from a circuit board?  Do you have a good board layout and decoupling capacitor scheme for your supplies?

In regards to your question of how to make the circuit non-inverting, this app note shows what that would look like.  You would need to do some additional math to design for a specific amount of hysteresis or you can use the simulation tool to play with the circuit.  Kai's suggestion of adding an inverting stage so the hysteresis part of the circuit is unchanged is another option but adds circuitry.  Either way is fine.

file:///C:/Users/a0412175/Downloads/sboa313.pdf

Chuck

do want to see what happen on IN+ /IN- pins?

this is my new circuit that is working fine on simulator.

IO checked all inputs and no oscillation occurred.

Good work

hysteresis is our friend against noise coupling to the inputs

Chuck

The diagram above does not contain hysteresis?

I designed all looking to a TI app note.

the picture you sent has hysteresis included with the positive feedback resistor.  It may be minimal but it is there.

Chuck

Hi,

this is my new circuit. Any suggestion to improve circuit with capacitors somewhere?

Thanks

I cannot see any hysteresis

And you should not take a chopper OPAmp for a comparator.

Kai

Why?

Can you please tell me wich one I have to use(I need VCM Common-mode voltage range approx from 0.20V to 4V max with 5Volt power supply)

Az-eddine

Chopper amplifiers require negative feedback to operate properly and should not be used as comparators.  This circuit seems to be digressing.  My apologies because I thought you had resolved your issue.  Comparators do not have any problem with the voltage range you described, so I do not understand why you are moving away from using one.  In all the previous examples, you were trying to use an open drain output device which doesn't seem to add any value to your circuit.  All you seem to need is a simple comparator with a push pull output such as the TLV7011.  And if you desire faster response time and a bit more accuracy, the TLV3201 is pin compatible.  Either of these will serve your need.

Chuck

Hi,

I had input voltage issue because input over 3V is gray region for lm393 for example  (VCC-2V) in worst case.

Got it.  Well that is not a problem with the devices I suggested.  They are both rail to rail on the input side.

Chuck

Hi,

thanks in advance here is the new schematic please let me know what do you think about the whole circuit.

The mos used does not need fly wheel diode, is designed to drive inductive load, relay etc...

I'm a litte bit worried about how i'm using the PC817.

Is there an alternative to TLV7031DCKR? (difficult to find).

Thanks

Hi Az-eddine,

I would do it this way:

az_lm393_1.TSC

And please no cap in parallel to the load (R6), but only a free-wheeling diode, if not already mounted within the relay. A cap in parallel to the load would cause a heavy inrush current potentially destroying the switching transistor when turning-on.

Kai

Good suggestion Kai, thanks

Alternatives to TLV7031 are the TLV7011DCKR  and TLV3201AIDCKR.

You just won't need the pullup resistor R5 that Kai included in his schematic since they are push pull output devices.

Chuck

Hello Az-eddine,

If you must use the optocoupler, I would place a pull-down resistor on the gate to of the MOSFET to ground to make sure it turns off.

Otherwise, any leakage through the optocoupler photo transistor, or gate, while "off" could cause the gate to float up, possibly into the linear region where it could possibly turn the MOSFET "half-on" and cause it to possibly overheat or keep the relay on (the drop-out voltage of a relay is much lower then the pull-in). A 10k to 100k  resistor should suffice.

And as Kai said...the cap is not needed across the coil. That cap can also lead to nasty resonances or ringing.

But the flyback diode across the coil is a MUST! The MOSFET clamp diode alone will not help - and can even make things worse.

Remember, when de-energized, the relay coil will generate a back EMF voltage of the opposite polarity (and usually much higher) than the previously applied coil voltage, due to the magnetic field collapsing back through the coil. For that +12V coil, the reverse transient could be -70 to -120V - or more!

Without the diode across the coil, the negative coil transient will make the MOSFET diode conduct  and you will dump the full negative transient across the power supply. NOT GOOD!

Put a 1N4002 type or similar power diode *directly* across the coil terminals (assuming the relay does not have one built-in). The diode will clamp the reverse voltage to -1V or so and life will be good.

The mosfet in qestion is described as below:

• Provides Robust Interface between D.C. Relay Coils and Sensitive Logic

• Capable of Driving Relay Coils Rated up to 150 mA at 12 Volts

• Replaces 3 or 4 Discrete Components for Lower Cost

• Internal Zener Eliminates Need for Free−Wheeling Diode

• Meets Load Dump and other Automotive Specs

• SZ Prefix for Automotive and Other Applications Requiring Unique Site and Control Change Requirements

Hi,

I had to move to rail to rail comparator due the input range voltage limit at 5V that is max 3V.

Anyway thank you very much

Hi Az-eddine,

you don't have to move to a rail to rail comparator because of the voltage divider R1 and R2 (in my schematic). Your 4V input signal decreases to 2V directly at the -input of LM393 which is well within the common mode input voltage range of LM393 

Kai

Ah great is clear now but I don't understand why did you use two transistors?

Fair point by Kai but I will need to leave the transistor question to him.  I would only be guessing the reason.

But if you only need the single channel version of the LM393B, that would be the TL331B.

Chuck

Hi,

I tested the circuit above (with R5 and R8 4.7K)  and the T1 is always ON.

What is the issue?

Sorry Az-eddine

We are on holiday break today and will review on Tuesday.

Chuck

Hi Az-eddine,

please show your current schematic and some scope plots.

Kai

I am seeing the output of the comparator switching at these voltages on the circuit you posted a while back called "az_lm393_1.TSC"

Chuck

Hi Az-eddine,

Az-eddine Haida said:

the circuit proposed does not work, the transistors was always on.

I guess you have killed the transistor by omitting the free-wheeling diode.

Kai

Hi Kai,

it was present since the beginning.

Thanks

Hello Az-eddine,

I still do not see a diode across the relay.

As I previously stated, the flyback diode inside the NUD3112 DOES NOT protect against the back EMF of the coil that is in SERIES with the diode.

Assuming the scope photo is 5V/div, there is about -3 to -4V of undershoot. Does not seem that the clamp is working too well

I would also scope the 12V supply and watch for transients when the relay de-energizes.

Have you confirmed the relay has a built-in diode? What is the *exact*, full part number of the relay?  I would add an external diode, anyways.

In my experience, circuits that involve inductive loads (relays, solenoids, motors, etc) almost always fail due to missing or improper back-EMF suppression or poor layout causing ground or supply transients.

And what is the value of R4? Blow-back through the MOSFET gate capacitance can take out the comparator output, so R4 needs to be able to current limit these transients. I would start with 4.7k.