The LM317 is quite the hot head ...

[EasyGoing1]

Hello ...

So I've been playing with this circuit that uses a digital pot as the adjustment resistor for an LM317. I feed the voltage from the output of the 317 through a hall effect current sensor then to the leads on a discharged LiIon battery... using an Arduino, I monitor the current flowing to the battery and adjust the 317 to keep the current stable for the main portion of the charge cycle...

But I noticed right away, that even at 400 milliamps, the 317 gets REALLY HOT and I was under the impression that the 317 is rated for 1.5 amps... I couldn't imagine the thing would survive 1.5 amps for any real length of time...

Is this normal for the 317 or do you think I'm doing something wrong?

Thank you,

[budm]

Re: The LM317 is quite the hot head ...

The key point is what is the Voltage drops measured between the input and output pin? You do use filter cap at the input pin and output pin, right?
It's the same power dissipation due to the Vdrops that you have to watch out for just like that solenoid driver transistor that you were working on.
It is not the amp rating that you go by, its the Voltage and current.
What kind of Voltage are you feeding into the circuit?
And what is the Output Voltage is set to?

[stj]

Re: The LM317 is quite the hot head ...

how big is your heatsink?

[mariushm]

Re: The LM317 is quite the hot head ...

LM317 like most linear regulators can dissipate up to 15 watts in the form of heat, if it's properly cooled.

See datasheet : https://cdn.badcaps-static.com/pdfs/...4077910403.pdf - all values are reported for power dissipation below 20 watts.

The amount of power dissipated is calculated with the formula Power = (Vin - Vout ) x Current

On page 4 of the datasheet you have 6.4 thermal information where you can see there for a generic to-220 the resistance junction - ambient is 23c / w ... so with one watt dissipated, the regulator will be ambient temperature + 23c or around 50-60c

The regulator can work at up to 125c, so without any heatsink it can only do around 2-3 watts of dissipated power continuously if you don't use a heatsink.

[EasyGoing1]

Re: The LM317 is quite the hot head ...

The key point is what is the Voltage drops measured between the input and output pin? You do use filter cap at the input pin and output pin, right?
It's the same power dissipation due to the Vdrops that you have to watch out for just like that solenoid driver transistor that you were working on.
It is not the amp rating that you go by, its the Voltage and current.
What kind of Voltage are you feeding into the circuit?
And what is the Output Voltage is set to?

Im giving Vin 12 volts and it never exceeds roughly 4.2 to 4.5 on the output.

And YES I have caps on both in and out pins

how big is your heatsink?

I just put a pretty big one on there cause I already fried an LM317 ... this heat sync has a thermal resistor stuck on it that actually works ... I'm taking readings right now to find out what the values are at known temps so that I can use it in my project to monitor things and make adjustments as needed.

The amount of power dissipated is calculated with the formula Power = (Vin - Vout ) x Current
...
The regulator can work at up to 125c, so without any heatsink it can only do around 2-3 watts of dissipated power continuously if you don't use a heatsink.

OK then it checks out ... now would 4 watts be very hot on your finger?

But Vin at 12 volts Out at 4.5 and current of half an amp is right around 4 watts.

I adjusted Vin down to 7 volts ... we'll see how that goes...

[budm]

Re: The LM317 is quite the hot head ...

OK then it checks out ... now would 4 watts be very hot on your finger?

But Vin at 12 volts Out at 4.5 and current of half an amp is right around 4 watts.

Yes, It would be. You should check the temperature so you can see what the temperature rise is so you not exceed the junction temperature spec..

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Yes, It would be. You should check the temperature so you can see what the temperature rise is so you not exceed the junction temperature spec..

After lowering the source voltage to 7, I ran it at 800 ma with almost zero increase in heat ... that seems to have been the problem (Vin too high).

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Incidentally, I'm using my Arduino Due to make my own 18650 LiIon battery charger. Why? Because I wanna see if I can do it... Ive got all the core components working within acceptable tolerance ... the 712 or whatever hall effect current sensor is less than ideal because its resolution doesn't seem all that great to me. In fact, the Arduino itself isn't that ideal in terms of measuring voltage etc. the internal reference voltage has to be known at the time your taking a reading or your calculations will be off and in this scenario, a tenth of a volt matters!

Anyways, if any of you have a link to a decent article that explains the three charge states of a LiIon but from a perspective of the power supply doing the charging ... everyone talks about constant current mode and constant voltage mode, but no one explains HOW a constant current is maintained from the power supplies point of view ... I assume that voltage on the power supply has to be constantly changed in order to maintain the desired current ... and that begs the question ... say your charging a battery at half an amp ... but you have to kick the voltage up to 5 volts in order to get that amperage ... want exceeding the voltage rating of the battery damage it? Or are they fairly forgiving?? And if so, how else could you keep a constant current flowing into the battery if not by adjusting the voltage on the power source?

Also, when determining if the battery is ready for the next stage, shouldn't voltage on the battery be read WITHOUT the charging power supply attached? And if thats the case, can I charge the battery through a transistor so that I can switch off charging then take a read of the battery??? Cause I tried this and it was no bueno...

[mariushm]

Re: The LM317 is quite the hot head ...

Uggh ... google "charging lithium ion" .. ADD "Application note" in front of your text to get loads of documentation.

Li-Ion Battery Charging Basics, Featuring the FAN5400 / FAN5420 Family of PWM Battery Chargers


Application Note: Safe Use and Charging of Lithium Ion Batteries
https://cdn.badcaps-static.com/pdfs/...da8774d7db.pdf

AN1149 Designing A Li-Ion Battery Charger and Load Sharing System With
Microchip’s Stand-Alone Li-Ion Battery Charge Management Controller
https://cdn.badcaps-static.com/pdfs/...5127eb7b61.pdf

Designing Applications with Lithium-Ion Batteries
https://cdn.badcaps-static.com/pdfs/...924440ad9e.pdf

[sam_sam_sam]

Re: The LM317 is quite the hot head ...

You might even want to try 5.5 to 6.0 volts this might even work better

Test this a lot before you trust it
Test to it ----> try to make it fail <-----> do this with nicad batteries first

[EasyGoing1]

Re: The LM317 is quite the hot head ...

A thought occurred to me this morning (it happens once in a while) ... since the LM317 uses a simple voltage divider to determine what its output is, couldn't you directly control the adjustment pin with a micro controller like an Arduino ... using an analog out pin run through an RC filter to clean up the pulse width etc.??? (yes even caps to ground on the input and output ... quit bustin my chops man! lol )

Seriously though why would this not work? a lm317 all by itself with a micro controller controlling the voltage on the adjustment pin changing the output voltage as we see fit???

[eccerr0r]

Re: The LM317 is quite the hot head ...

No matter what you still need some sort of feedback control loop to make sure your PWM controlled voltage is accurate to what voltage you have set on the output. You could also use a DAC to use as the reference voltage for the LM317, and the feedback control loop is a lot easier for this method as no filtering is needed.

Anyway, Lots of possibilities.

[stj]

Re: The LM317 is quite the hot head ...

the duo is ARM cpu based,
your i/o pins have a 3.3v limit.

[EasyGoing1]

Re: The LM317 is quite the hot head ...

the duo is ARM cpu based,
your i/o pins have a 3.3v limit.

I'm aware of this ... and I'm also aware (after running some calcs anyways) that the range in output that I could achieve with this is rather small ... but my point in bringing it up, is that it had literally just occurred to me that the LM317 uses a reference VOLTAGE to determine the output level ... I thought there was some other magic happening between adjustment pin, output pin and the brain of the device which was somehow also dependent on resistance values being detected between pins 2 and 3 etc. ... until I thought about it and realized that a voltage reading is really all you need to design a device like the 317...

I just thought it was interesting and wondered if it could actually work (applying a voltage directly to the adjustment pin with respect to ground instead of using a voltage divider...

[stj]

Re: The LM317 is quite the hot head ...

maybe an opto-isolator could be used

[budm]

Re: The LM317 is quite the hot head ...

http://www.edn.com/design/analog/436...h-a-PWM-signal

[eccerr0r]

Re: The LM317 is quite the hot head ...

http://www.edn.com/design/analog/436...h-a-PWM-signal

Woah wtf. Is that really going to work with a LM741? Having the null is not that big an issue, I'd think either a negative supply is necessary (~ -1.25V) or a LM6142 or other RRIO op amp is needed to get a minimum voltage of 1.25V. Or a negative supply with the LM6142 to get the output to 0V...

[budm]

Re: The LM317 is quite the hot head ...

LM317 is not made to go down to 0V output, 1.25V is the min output, look at the spec sheet, and if you want it to be 0 then look at the Apps notes.
https://cdn.badcaps-static.com/pdfs/...4077910403.pdf

[eccerr0r]

Re: The LM317 is quite the hot head ...

Yes, as said, with a RRIO LM6142 and a negative supply will give a negative output, which will get the LM317 down to 0V.

Or a LM741 with a negative ~3V supply which will both deal with the 741 not reaching negative rail plus get negative enough for the LM317. Either way, a negative supply is necessary.

But that edn article seems to have a design error.

[mariushm]

Re: The LM317 is quite the hot head ...

You can find voltage regulators that have voltage references inside with much lower voltage, like 0.6v or 0.25v even , but generally these are designed for max 6v in , max 5v out or something like that.

For ex. LT3021 (max 10v in , max 9v out, 0.5a max, 200mV reference, 300mV max dropout voltage) : http://www.linear.com/product/LT3021

or LT3086 (max 40v in, 32v out, 400mV reference, 2.1a max out, programmable current limit ) : http://www.linear.com/product/LT3086

These are both kinda expensive (6-10$ each) and in general low voltage references are much harder to make, so chips are more expensive.

Another trick I've seen often used is to use a diode in front of the linear regulator, or the classic use of a voltage reference/zener diode and a negative power supply :

TI: AN-181 3-Terminal Regulator is Adjustable



Here you have a 6.9v voltage reference - you can probably find some cheaper or more common zener diode or a lower voltage one ... you basically need anything above 1.25v to make it possible to output 0v

You can make negative power supplies easily with voltage doubler ics (they can be configured to be voltage inverters as well or voltage half-ers) but the majority of the cheap ones only work with up to 5v (still you only need a few ma of current on the negative psu so you could use a 7805 or something cheap to get 5v out of a big tranformer winding, then use a voltage inverter to get -5v and use a zener with voltage lower than 5v ... ex 2.5v AN431 or a lot of TL431 (lots of companies make them) with default reference below 5v