ULN2803 some output pins got damage

[dhakeparag81]

Hello Guys,

I am using ULN2803 to drive some segment but it is seen that most of the time some outputs of that pins got damage.
These Seg-A1 to Seg-H line are going parallel for internal display and for connector on which we can connect multiple segment PCB's.
I have attached the ckt diagram please give some advice.

 

[KlausST]

Hi.

The shown circuit is not problematic.
There is no reason for getting damaged.

You need to show us the whole story.

However:
The HCT device is not specified to work with 3.3V
And the COM connection should be hard connected to 12V ... low impedance, best with a fast capacitor to GND.

Klaus

 

[dhakeparag81]

Hi.

The shown circuit is not problematic.
There is no reason for getting damaged.

You need to show us the whole story.

However:
The HCT device is not specified to work with 3.3V
And the COM connection should be hard connected to 12V ... low impedance, best with a fast capacitor to GND.

Klaus

There is 12v or 24v selection in locally power generation ckt, so someone may use 24v supply or 12v, because of that im not using pin 10 to 12v.
at the output side there is generally opto (pc817) ckt for which uln2803 provide gnd to turn on specific segment.

Which ic should be used for 3.3v?

 

[FvM]

There's to room for PC817 in posted schematic, please show complete circuit.
Freewheeling diodes must be connected to a voltage greater or equal than highest load supply. It can be unconnected for purely resistive load. In case of doubt I'll connect a zener diode clamping output to a legal value.

 

[dhakeparag81]

This is just a sample ckt,
other pcb's on which this type of ckt connected through ULN2803 output.
ULN2803 oututs are fetched on connector there is no other ckt on pcb for uln2803.
from below ckt controller will sense the input and then turn on the specific segment. This ckt i used for dot matrix to know which segment is ON and accordingly pattern will be generated on dot matrix, but the concept should be the same.

Also we are using this for 24v so all zener diode are of 24v

--- Updated Jul 27, 2024 ---

In ULN2803 some of the output pins damage (short- output is ON)

 

[betwixt]

I think the full intention of this device needs to be known. Even using a 50V/500mA driver for a load with 1K in series with it seems like overkill and why all those driver transistors which can only serve to starve the ULN2803 from input current, it is designed to be driven by 5V TTL levels and may not function well with 3.3V.
If the transistors are needed as signal inverters, why not invert the drives to the HCT373 either in software or with an inverter IC.

The picture of the bright segments doesn't seem right either, they seem to be passing WAY too much current and that is impossible in the given schematic.

Brian.

 

[dhakeparag81]

I think the full intention of this device needs to be known. Even using a 50V/500mA driver for a load with 1K in series with it seems like overkill and why all those driver transistors which can only serve to starve the ULN2803 from input current, it is designed to be driven by 5V TTL levels and may not function well with 3.3V.
If the transistors are needed as signal inverters, why not invert the drives to the HCT373 either in software or with an inverter IC.

The picture of the bright segments doesn't seem right either, they seem to be passing WAY too much current and that is impossible in the given schematic.

Brian.

those bright segment are damage (uln2803 ouput damage),
i have tried to drive the uln2803 from hct373 but hct373 are not able to drive the uln so i have added 2n3906 transistors.
also previously i have used TIP122 transistor with opto pc817 ckt, to reduce the pcb size i change it to uln2803

 

[FvM]

0.1 uF capacitors generate high current peak, probably not so good for ULN2803.

 

[betwixt]

I think you are trying to fix the wrong problem. The ULN2803 is a darlington driver, its output stage needs about 1.4V to turn it on but it has a potential divider built in so it expects 5V TLL levels for normal operation. You have added a transistor driver stage that itself is powered from 3.3V so there is no way it can drive the ULN properly, there is still insufficient voltage. Why not simply use the 2N3906 to drive the display and omit the ULN completely, that satisfies all the voltage/current requirements.

What still puzzles me is those bright segments in the photograph. Unless you are trying to dim the display by driving with PWM, the ULN output stage should be fully conducting to turn the segments on so it should pass almost identical current to it being shorted out. In any case a 500mA rated output stage should not be damaged with an LED and 1K resistor in series with it. You would have to power the LED from over 500V to do that! The current sinking through the ULN outputs in your schematic can never exceed about 10.4mA and the zener diodes can never conduct. All they would do is let current directly through the display, bypassing the ULN completely if the supply was more than about 24V.

Brian.

 

[dhakeparag81]

I think you are trying to fix the wrong problem. The ULN2803 is a darlington driver, its output stage needs about 1.4V to turn it on but it has a potential divider built in so it expects 5V TLL levels for normal operation. You have added a transistor driver stage that itself is powered from 3.3V so there is no way it can drive the ULN properly, there is still insufficient voltage. Why not simply use the 2N3906 to drive the display and omit the ULN completely, that satisfies all the voltage/current requirements.

What still puzzles me is those bright segments in the photograph. Unless you are trying to dim the display by driving with PWM, the ULN output stage should be fully conducting to turn the segments on so it should pass almost identical current to it being shorted out. In any case a 500mA rated output stage should not be damaged with an LED and 1K resistor in series with it. You would have to power the LED from over 500V to do that! The current sinking through the ULN outputs in your schematic can never exceed about 10.4mA and the zener diodes can never conduct. All they would do is let current directly through the display, bypassing the ULN completely if the supply was more than about 24V.

Brian.

That ULN2803 is driving on board display as well as its pins are provided at connector, so that more display pcb can be driven through it, and it damage when more displays are connected through that connector, its not like how many, bcoz some times only 3 display pcb attached to it also damage the uln ic.
Some of the board of uln2803 are working fine.
i know there is issue with the ckt as u mentioned at u r first point.

I am using only 2N3906 for other display that is not glowing in picture.
also im not using pwm its direct bus RC0-RC7 for all input and outputs that are on board.

I dont understand here if 3.3v is not sufficient then ULN ic should not work properly at first, but that is not a case when i have tested nearly 100 board. all the time expected segment was ON.


I am looking for solution to existing issue with minimum ckt changes.

--- Updated Jul 27, 2024 ---

0.1 uF capacitors generate high current peak, probably not so good for ULN2803.

Are u talking about PC817 ckt?
i added that capacitor for noise filtration.

 

[KlausST]

Which ic should be used for 3.3v?

I don't know.

But your schematic shows a 74HCT373 supplied with 3.3V. This is not according specifications.

Klaus

 

[dhakeparag81]

I don't know.

But your schematic shows a 74HCT373 supplied with 3.3V. This is not according specifications.

Klaus

Sorry my mistake its 74HC373 not HCT

 

[aliarifat794]

Have you tested the 7 segment display separately?

 

[Easy peasy]

Some 12V zeners will draw significant currents at 12V applied thru the ULN2803, due to spread of parameters - either remove these or use 13V zeners

 

[dhakeparag81]

Have you tested the 7 segment display separately?

there is no issue with display. Becoz after replacing ULN2803 everything works fine.

 

[KlausST]

Hi,

there is a lot of misinformation in this threaed.
* the shown schematic is not complete / not according real circuit
* the "HCT" is indeed a HC
* the information that "a 3.3V powered HC373 can NOT drive the ULN" is wrong (there is no need for discrete transistors)
* the power supply information is unclear: 3.3V, 5V, 12V, 24V .. when and how they are used, where they come from..
* then there is information about a "dot matrix" display
* there is no need for optocouplers

*****
With this bunch of confusion I´m not willing to investigate what´s rigth or wrong.

--> focus on one circuit only, show all it´s use cases, show the whole schematic, show the PCB layout, give clear informations you are 100% sure of, give measurement values....

****
IF (only IF) the ULN is destroyed, then there are only 3 main solutions to get it destroyed:
* overcurrent (absolutely impossible with the shown circuit ... unless you are hiding other fails like the series resistors)
* overvoltage - even very short spikes. Here we simply don´t have all informations (like requested above).... or eventual ESD problems
* overtemperature of the die. Caused by external or internal heat sources. We don´t know ambient temperture. But even with high ambient temperature the internal power dissipation is expected to be very low, so overheat can be excluded with the given informations.

Klaus

--- Updated Jul 29, 2024 ---

it is designed to be driven by 5V TTL levels and may not function well with 3.3V.

TTL output levels are 0.4V/2.4V ... independent of power supply.
I don´t think there is something like 3.3V TTL levels.
3.3V CMOS output levels are expected to be lower then 0.8V (@ LOW) and higher than 2.0V (@ HIGH)

So according HC373 datasheet and ULN datasheet ... the use of the discrete BJTs makes no sense at all.
The HC373 HIGH level ... when driving direlctly the ULN - is expected to be higher than VCC - 40mV
and the ULN should be able to drive a 300mA (!!!) load with an input level of 3.0V

So the drop of 40mV ... does not explain malfunction at all.
It seems the OP prefers trial-and-error method to using simple math according OHM´s law or using a voltmeter to verify levels.

Klaus

 

[FvM]

i added that capacitor for noise filtration.

Didn't ask why you put it there but if it might cause driver damage. There's no ULN2803 SOA spec that would mark the load capacitor as acceptable. A safe place for filter capacitor is PC817 LED. Anyhow there are probably more likely ways to damage ULN2803, I agree with KlausST.

 

[dhakeparag81]

Hi,

there is a lot of misinformation in this threaed.
* the shown schematic is not complete / not according real circuit
* the "HCT" is indeed a HC
* the information that "a 3.3V powered HC373 can NOT drive the ULN" is wrong (there is no need for discrete transistors)
* the power supply information is unclear: 3.3V, 5V, 12V, 24V .. when and how they are used, where they come from..
* then there is information about a "dot matrix" display
* there is no need for optocouplers

*****
With this bunch of confusion I´m not willing to investigate what´s rigth or wrong.

--> focus on one circuit only, show all it´s use cases, show the whole schematic, show the PCB layout, give clear informations you are 100% sure of, give measurement values....

****
IF (only IF) the ULN is destroyed, then there are only 3 main solutions to get it destroyed:
* overcurrent (absolutely impossible with the shown circuit ... unless you are hiding other fails like the series resistors)
* overvoltage - even very short spikes. Here we simply don´t have all informations (like requested above).... or eventual ESD problems
* overtemperature of the die. Caused by external or internal heat sources. We don´t know ambient temperture. But even with high ambient temperature the internal power dissipation is expected to be very low, so overheat can be excluded with the given informations.

Klaus

--- Updated Jul 29, 2024 ---

TTL output levels are 0.4V/2.4V ... independent of power supply.
I don´t think there is something like 3.3V TTL levels.
3.3V CMOS output levels are expected to be lower then 0.8V (@ LOW) and higher than 2.0V (@ HIGH)

So according HC373 datasheet and ULN datasheet ... the use of the discrete BJTs makes no sense at all.
The HC373 HIGH level ... when driving direlctly the ULN - is expected to be higher than VCC - 40mV
and the ULN should be able to drive a 300mA (!!!) load with an input level of 3.0V

So the drop of 40mV ... does not explain malfunction at all.
It seems the OP prefers trial-and-error method to using simple math according OHM´s law or using a voltmeter to verify levels.

Klaus

Sorry to create confusion its not intentional.
when connecting Display PCb externally uses 24v supply and gnd is common for both the external display as well as on board.
by switching gnd and opto ckt we try to operate extrenal display.
By measuring voltages on 74hc373 @ high its 3.31v and @ low 0.08-0.09v

I dont have the info on the external display side, i will get that first.

 

[betwixt]

The more information you give, the better our answers can be.

Firstly, the ULN cannot be damaged by the display, it is simply too highly rated to suffer a breakdown from excess current or voltage. In fact it could handle more than 20 parallel displays.

Where the almost certainly will be a problem is those zener diodes. Zeners are not "open circuit" until their rated voltage, 24V one plateau at 24V when passing current but there is some conduction at slightly lower voltage. If you are supplying the display from a 24V source it is very possible that current is flowing through the display to ground through the zener diodes, not the ULN2803.

A better strategy to protect the ULN if it needs it, it to connect the COMM pin to the same supply that feeds the display. Note this is not 12V, it must be the voltage feeding the display or slightly higher or the internal diodes on the COMM pin will be forward biased and conduct current from 24V, through the display then down to the 12V rail.

Also, this is a long shot, check those SMD resistors are "102" and not "201", the markings are very similar. One is 1K but the other is only 200 Ohms.

Brian.

 

[dhakeparag81]

The more information you give, the better our answers can be.

Firstly, the ULN cannot be damaged by the display, it is simply too highly rated to suffer a breakdown from excess current or voltage. In fact it could handle more than 20 parallel displays.

Where the almost certainly will be a problem is those zener diodes. Zeners are not "open circuit" until their rated voltage, 24V one plateau at 24V when passing current but there is some conduction at slightly lower voltage. If you are supplying the display from a 24V source it is very possible that current is flowing through the display to ground through the zener diodes, not the ULN2803.

A better strategy to protect the ULN if it needs it, it to connect the COMM pin to the same supply that feeds the display. Note this is not 12V, it must be the voltage feeding the display or slightly higher or the internal diodes on the COMM pin will be forward biased and conduct current from 24V, through the display then down to the 12V rail.

Also, this is a long shot, check those SMD resistors are "102" and not "201", the markings are very similar. One is 1K but the other is only 200 Ohms.

Brian.

Tnx
I will try some boards without zener.
for supply to COMM pin currently not possible as there is no pad to attach to it.
Those are not 200 ohm, i checked it.
ill post the result.