LM317 voltage drop when load is connected

[Prince Charming]

Hello, I've designed the circuit in the picture and aside from its main purpose, when the Q1 MOSFET is being triggered or is triggered recently the yellow led must be kept open. When the door is switch is closed while the yellow led is on, the pink power led must blink also the buzzer must go off after a while. The relays, pink power led and yellow led are all connected to LM317 (U2) as loads. My problem is that yellow led is dimming and lightening while the pink led is blinking (basically when a new load is connected), or both pink and yellow LEDs are dimming even further when RL2 is on. I checked the voltage across them and I saw that voltage on each load drops when another is connected. There is something I really messed up while using LM317 voltage regulator, as it doesn't do its sole purpose of existence

Note: U1 LM317 works just fine as the load of its doesn't change. The MQ2 is connected properly in real life forget the schematic, I used sole sensor not the module anyway. Also the cables coming from below and left are just 24 volt supply which is generated by a boost converter.

Expert suggestions and guidance is required.

Thank you

 

[KlausST]

Hi,

electronics is with voltage and current.

When there are problems with voltage, then give (very important)
* the expected voltage
* and the voltage you measure
also ... voltage often/usually depends on current.
* thus always give the current.

*****

All in all the design des not make much sense to me.
Parts are overcomplicated, like
* U1 ... just to power a LED ..that still includes a series resistor
* RL1 simply could be replaced by transistor

Parts are not well designed, like
* the threshold levels without noise filter and without hysteresis call for trouble.
* Reverse current may flow through the coil of RL3...

Klaus

 

[Prince Charming]

Hi,

electronics is with voltage and current.

When there are problems with voltage, then give (very important)
* the expected voltage
* and the voltage you measure
also ... voltage often/usually depends on current.
* thus always give the current.

*****

All in all the design des not make much sense to me.
Parts are overcomplicated, like
* U1 ... just to power a LED ..that still includes a series resistor
* RL1 simply could be replaced by transistor

Parts are not well designed, like
* the threshold levels without noise filter and without hysteresis call for trouble.
* Reverse current may flow through the coil of RL3...

Klaus

Thanks for the comments.

The expected output voltage of U2 is 3.1 but when the power led is blinking it drops to 2.8, that's no problem because it is close enough but when the RL2 is on, the voltage drops to 1.8 and that's disaster.

I can't measure current as the parts are soldered but expected LED current is 0.02 amps and the relays are 3V, 0.120 amps and 25 ohms of resistance.

I am aware that 2 relays will have 240 miliamps of current in total and that would exceed the maximum current rating of 2N7000 (Q2) which is 200 miliamps. But the coil of RL3 was originally directly connected to ground until I realized U2 somehow continues to work even though Q2 transistor is off. I figured out that there is a path from U2 to ground from RL3 coils. That's why I had to replace the ground connection with drain connection. I will replace 2N7000 if it blows up, I take the responsibility.

U1 was originally designed to power both LED's but I realized a logic mistake and replaced the anode of yellow led to U2. U2 was supplying 3.1V instead of 2V that's why I had to add a resistor.

I could replace RL1 with a transistor I guess, I used the think C1 wouldn't be able to discharge if I did but looking again I think you are right. It is not a big issue for now but thanks.

I don't see which reverse current can flow through RL3.
I thought I did enough filtering by using recommended capacitors for LM317s, what other filtering do you suggest?

 

[KlausST]

Hi,

I can't measure current

for debugging it´s urgent to know about the currents. I see no way around this.
In worst case you need to desolder a wire and put a amperemeter in.

The power LED has no current limitng resistor. This is quite unusual for a LED. What exact vendor and part number is it? Link to datasheet (directly at the manufacturer site) please.

U1 was originally designed to power both LED's

LEDs usually are not voltage driven, thus a constant voltage for a LED makes not much sense. In either case a LED needs a current limiting resistor .. and for the LED it makes no difference whether the source is 3.1V or 5V.
The U1 supply is just added effort.

I don't see which reverse current can flow through RL3.

Maybe "reverse" was not the correct expression. Do a current measurement when ON and when OFF.

I thought I did enough filtering

You talk about power supply.
But I talked about the threshold levels in the signal path. Like input of Q1 for example. I expect either slow edges or oscillation/chattering.

But you talk about power supply bypassing (capacitor). U3 is missing a power supply capacitor.
And the whole circuit should have a bulk capacitor.

All in all: If I had to do this project, I´d most probably use a microcontroller. Less parts, maybe even lower cost, and I guess when you take the soldering time and debugging time into account it´s faster to write a little piece of software. ... if you had some experience with microcntrollers.

Klaus

 

[Prince Charming]

What exact vendor and part number is it?

I also don't know the manufacturer but in the website I bought it says 3 Watts, 3.0 -3.2 volt, 0.7 amps at the description.

thus a constant voltage for a LED makes not much sense

Well, I totally forgot about it, I didn't take diode's resistance depends on its bias into account. But I guess that would give me a dimmer or brighter led with a constant brightness. I will supply constant current to LEDs from now on, thank you. This still doesn't explain why additional load decreases the output voltage of LM317, why doesn't it compensate the error signal?

Oh wait a second, I added an 68 ohm resistor between output of U2 and ground to see if more load will cause anything and yes all lights went dim. Then the big question came up to my mind, is this LM317 working as a constant current source? But how? I connected in constant voltage configuration.. I wish I could test this by measuring the output current of U2 but the bottom of the perfboard is the ugliest shit you can ever imagine, there are tons of cables running on top of each other If I try to desolder the connection i will burn many cables. But considering the test results I think we can say the LM317 is compensating voltage for current change. Can you tell me why?

Like input of Q1 for example. I expect either slow edges or oscillation/chattering.

Well I don't have an oscilloscope but RL1 turns on just fine and only purpose of Q1 is doing that. But there is a fact that Q4 turns on very slowly due to slow increase in gate voltage and it makes the buzzer sound like like it is reborning every time it turns on. It starts with low and painful voice and over time it buzzes normally. I must have somehow made the buzzer voltage like 0 or 24 and nothing in between. The first thing came into my mind is using another relay but come on they are taking a lot of space and I like the slow start characteristic of the buzzer as it warns you for the last time before it buzzes impulsively

U3 is missing a power supply capacitor.

Well true, I ignored that, the biggest reason is not my lack of B12 but it is lack of knowledge. Why is it so important to put filters everywhere? As long as the timer does it's blinking job, and yes it does make the led blink, it is fine. Do I put that filter to lengthen the lifespan of the 555 timer (U3)? There is an output capacitor at the 5V power supply so I think that would act as input voltage filter for the 555timer (U3) anyway.

And the whole circuit should have a bulk capacitor.

Believe me it does because of the power supply module which is made by someone else. The circuit is turning of with a delay when I cut the power.

If I had to do this project, I´d most probably use a microcontroller.

Yes it would be easier but it wouldn't teach me anything and isn't a good idea for mass production (Although I am not planning to make a mass production of this circuit I still want to learn the practices of it.)

 

[KlausST]

I also don't know the manufacturer but in the website I bought it says 3 Watts, 3.0 -3.2 volt, 0.7 amps at the description.

Yes, always good for a surprise and extra work.
BTW: 3.2V and 0.7A isn´t even close to 3W.

the bottom of the perfboard is the ugliest shit you can ever imagine, there are tons of cables running on top of each other

Another chance for a bad surprise.

I like the slow start characteristic of the buzzer as it warns you for the last time before it buzzes impulsively

Oh, a new feature ....

Why is it so important to put filters everywhere?

There are many documents, descriptions, tutorials and videos about this topic. There are many reasons.
You don´t have a scope, thus you don´t see what happens. EMI/EMC is one reason.
Maybe you "hear" it when your radio goes off when the LED starts to blink. Or your buzzer makes a short beep when your refrigerator turns off.

Just when a relay turns ON or OFF does not mean this is "clean" switching.
It´s more like you have a car without speedmeter .. but you tell by the sound of the motor you are not speeding.
I mean: it´s possible to calculate the speed by the audible frequency of the motor. But the ears and the human will not give very accurate results .. and is highly subjective.

There is an output capacitor at the 5V power supply so I think that would act as input voltage filter for the 555timer (U3) anyway.

Surely not. Good thing is that the NE555 is old and slow. With a modern chip this is a "no-go".

Klaus

 

[Prince Charming]

BTW: 3.2V and 0.7A isn´t even close to 3W.

Very true, bad marketing. The brightness is enough so no problem.

EMI/EMC is one reason

I didn't think this circuit is at a level to consider EMI/EMC.

Just when a relay turns ON or OFF does not mean this is "clean" switching.

Well I know that but as long as it doesn't create any visible drawbacks i don't think about it. As you said maybe it doesn't bother me because I work with low currents, slow components and low frequency switching.

Well the unexpected behaviour of LM317T which works as a current source even though it is connected in voltage source configuration is still unsolved and I'll use others ways to regulate voltage in the future.

Thank you for attention.

 

[KlausST]

I didn't think this circuit is at a level to consider EMI/EMC.
..
Well I know that but as long as it doesn't create any visible drawbacks i don't think about it.

We professional designer have to guarantee the compliance with EMI/EMC rules for every circuit we design. And we must take responsibility for it.

BTW: you talk about "low frequency switching". Indeed it´s not the actal frequency that causes high frequency problems. but it´s the rise rate at the signal edges. The rise rate is the same, even at low frequency, But for sure higher switching frequency make things worse.

Klaus

 

[danadakk]

If LEDs low current ( I think you mentioned 20 mA) then a current diode
easy fix for that part of the [problem.

The Case for a Constant Current

LED technology is used far more extensively than any other forms of lighting technology that have come before it.

www.diodes.com

https://www.diodes.com/assets/Datasheets/BCR430UW6.pdf

Linear LED Drivers

Our linear LED drivers are constant current LED drivers for a broad range of applications, with a low BOM cost, and no inductor required.

www.diodes.com

They could get rid of your LM317's and all the associated R and C needed for their
functioning.

The 317 can be configed for constant current, but has high delta V requirements to keep
it regulating -

Bypassing, always look at datasheet recommendations for bypassing. Usually in the
application section of datasheet.



Regards, Dana.

--- Updated Jul 10, 2022 ---

I see a clamp diode for relay turnoff on one relay but not the other. Relay
transients can smoke an IC in a few uS with the HV transient they produce
if not handled by protection diode. Note use a fast diode as well.

Take a pic of top and bottom of your perf board and post for recommendations.

The 2N7000 driving a buzzer, you need a scope to look at its drain to see if buzzer
producing any transients.

You can get a scope, PC based, by

PC SOUND-CARD SCOPE INTERFACE FACILITATES DC RESTORATION

PC SOUND-CARD SCOPE INTERFACE FACILITATES DC RESTORATION: To start with I must appreciate acknowledge and thank ‘Christian Zeitnitz’ for the fantastic Sound-card Oscilloscope & signal generator software available at his website on which this Instructable is based. While working with the PC sound…

www.instructables.com

OR

https://www.electro-tech-online.com/articles/diy-oscilloscope.964/

Whatever you do make sure you protect the sound card inputs, or you could hose
your PC. Gives you a function generator, spectrum analyzer, and scope, ~ good for 20 Khz.


Regards, Dana.

 

[Prince Charming]

https://www.diodes.com/assets/Datasheets/BCR430UW6.pdf

Wow a brand new IC for me, thank you.

I see a clamp diode for relay turnoff on one relay but not the other.

Well that diode was actually to discharge C11 capacitor. The C1 capacitor on the other hand was purposefully discharged slowly through the variable resistor so it didn't need a shortcut to discharge. I will search for clamp diodes next time i use relays.

You can get a scope, PC based

I would like to learn the regular way of doing things and would prefer to stay away from extra confusion but thanks for this method, I didn't know it existed and it is impressive. I may buy scopes or beg my uncle to use his when I really need it

The 2N7000 driving a buzzer, you need a scope to look at its drain to see if buzzer
producing any transients.

That's very respectful of professionals, they even care about transient of buzzer drivers. By the lack of knowledge I ignored but I started to learn about using parallel capacitors for MOSFETS.

Sorry for the late response I was working on another project, I lost interest to this one and already mounted it to its place. It is working as expected except when the optional output relay is on, the lights go very dim. I learned my lesson about led current regulation thx to you both.

Take a pic of top and bottom of your perf board

That made me laugh as the board is disgusting. I will lastly leave pictures here but I am expecting no suggestions to this junk

 

[Audioguru]

Each one of the tangle of wires is an antenna that picks up all kinds of interference.

The new (2010) datasheet of the LM317 shows that it has excellent voltage regulation when its input=output voltage is 5V (yours is 2V and 3V) and its current is 10mA or more. Your 220 ohms resistors use a current of only 5.68mA which is fine for a more expensive LM117.

The IRFZ44 Mosfet and most IRF Mosfets need a 10V gate-source voltage to fully turn on. Some of them barely conduct a few mA with your 5V or less. You need IRLZ44 (see the L for Low?) Mosfets.

 

[Easy peasy]

L = logic level drive, i.e. fairly fully on for 5V G-S

 

[Prince Charming]

Yep that's a good point, I will consider my gate voltage and choose my MOSFET according to that. Going only above the threshold is not enough I know. Thanks.