LM317 circuit help? Output voltage different then expected

[Clarkdale44]

Hello

I have an CFL emergency light which works on 6v, and i only have a 12v 7.2ah battery around. So i found this circuit to regulate the voltage from 12 to 7.
I thought i can use it for other devices which runs on 3v and 5v.

I made it exactly as the schematic below but the output voltages are not like they should be.

When i connect it with my 12v battery , max output voltage is 4.4v.
When connected to 12v ac adapter(voltage is non regulated so on mm shows 18v) max output is 4.4v sometimes 4.5v.
When connected with 12v 2.5A power supply it max output is 4.5v.

I want it to regulate voltage from zero to the max voltage provided by power source. For example if power source is 24v then it should regulate from 0 to 24v using pot.
Currently using 10k pot, do i need to change that?



Regards!!!

 

[Clarkdale44]

My mistake, found out LM317 has different pinouts, i thought it would be same as 780x IC's. Will fix this tomorrow and update as i am out of solder wire for now.

 

[picbits]

To get down to 0 volts output, you'll need to find a -1.2v supply from somewhere to connect to the pot as the 317 will only regulate down to +1.2v minimum otherwise.

 

[MikeMl]

The 10K pot makes it such that you will be using only a small part of the rotation. Use a 2K pot if you have one. When you get it rewired and running, feel the tab on the LM317 (you are using a TO220 package?). If it burns your finger, then it is too hot and is going into thermal shutdown, reducing the output voltage. If that happens, you will need a heatsink.

 

[Clarkdale44]

The 10K pot makes it such that you will be using only a small part of the rotation. Use a 2K pot if you have one. When you get it rewired and running, feel the tab on the LM317 (you are using a TO220 package?). If it burns your finger, then it is too hot and is going into thermal shutdown, reducing the output voltage. If that happens, you will need a heatsink.

The issue was i mistook its pinouts, i thought it would be same as like 780x. Rewired and now its working like it should be. And yeah it was damn hot when i tried running my cfl circuit using 7.5v, i have a decent heatsink installed, i hope its enough, else gonna have to put even bigger one.
For now 10k pot is alright, if needed i will change in future, or when i build another one.

 

[crutschow]

The LM317 has a minimum output of about 1.2V and a maximum output of about 3V less than the supply voltage.

You might consider the LT3083 which can go down to 0V with a single pot and has a maximum output of slightly more than a volt below the supply voltage.

 

[audioguru]

The resistor from the output to the ADJ pin on an LM317 is supposed to be 120 ohms so that the supply voltage does not rise when its load current is low (and destroy the low current Cmos circuit you might try powering). Then the value of the pot needs to be recalculated. Many circuits wrongly show 240 ohms for the resistor because the datasheet has many examples of circuits for the more expensive LM117 that can use 240 ohms.

 

[eTech]

The resistor from the output to the ADJ pin on an LM317 is supposed to be 120 ohms so that the supply voltage does not rise when its load current is low (and destroy the low current Cmos circuit you might try powering). Then the value of the pot needs to be recalculated. Many circuits wrongly show 240 ohms for the resistor because the datasheet has many examples of circuits for the more expensive LM117 that can use 240 ohms.

You can also connect a resistor across the output to maintain a minimum current load of 3.5mA so regulation is maintained...

eT

 

[audioguru]

You can also connect a resistor across the output to maintain a minimum current load of 3.5mA so regulation is maintained...

No.
If the output voltage is variable then how can you set a certain output current?
The minimum load current for an LM317 is a maximum of 10mA for some of them. With only 3.5mA then some will work and some will not. A 120 ohm resistor from the output to the ADJ pin provides a load current of (1.2V/120 ohms=) 10.0mA to (1.3V/120 ohms=) 10.8mA.

 

[MikeMl]

I have had the discussion about the 120Ohm resistor between Vout and Vadj with Audioguru before. What he says is correct if you operate the LM317 with no load. If the minimum external load current is more than ~10mA, then your 1K resistor from Vout to Vadj will work just fine...

If the load provides the minimum required load current, then the voltage setting resistors can be selected based on the current that flows out of the Vadj pin. This calculation lets you use the 10K pot to set the voltage and it works just fine.

 

[eTech]

No.
If the output voltage is variable then how can you set a certain output current?
The minimum load current for an LM317 is a maximum of 10mA for some of them. With only 3.5mA then some will work and some will not. A 120 ohm resistor from the output to the ADJ pin provides a load current of (1.2V/120 ohms=) 10.0mA to (1.3V/120 ohms=) 10.8mA.

Yes...its an option.

The TI spec sheet states the minimum load current for regulation is 3.5 ma, so for that device it will work.

Obviously, the resistor size needs to be adjusted for the appropriate minimum load current for the device being used.

But I agree that the program resistor should be chosen correctly in the first place.

eT

 

[audioguru]

The TI spec sheet states the minimum load current for regulation is 3.5 ma, so for that device it will work.

No it might not work. Some LM317 ICs need a few mA of load current, "typical ones" need 3.5mA and maximum ones need 10mA. But all of them work perfectly if you calculate the resistor to draw at least 10mA.

 

[tvtech]

Now I am doing it again...bored so I comment....

The LM317T is a venerable little beast. My little project uses one. For my purposes a 180 Ohm resistor is used for the job. Sure, things change from design to design...

All I know is that in my case...no issues ever. Like a perfect match???

Going on for more than a Year of testing..the little E3's seem to be getting brighter, the LM317T is not moaning....nothing is abnormal...and as people that have followed or read my threads would know....this little light was put through the mill...

Rated at 220VAC it successfully ran for 3 runs @ 400VAC. For 5 Minutes each way. No blow ups. No repairs. A Year ago. And still running...

Funny stuff. People want to buy them......I say no. Testing not complete...They say how "much longer"?

Truth is the casing is a problem. And yes it's an X2 Cap design. I have to mill parts and keep prying fingers out of "the insides"...

I am probably around 90% there with casings. ABS and all.

Forgive me if I ramble a little bit. I hope you Guys don't mind.

I absolutely hate SMPS. Everyday of my life on CRT I have to deal with the same crappy Chinese SMPS design we are seeing here lately for CRT sets.
It's like one bright spark there designed a "maybe good" SMPS for TV's....and it has been copied flat out by all and sundry.

All the same. Blow up the same. And are reliably the same.

And yet...a Company maybe like Delta Electronics in China makes the very best SMPS for Lappies Etc.
My SMPS for my Acer lappie has worked solidly since 2006...

And is still running as I type here......and that is a sealed enclosure that gets "really" hot. It has weathered games, a lot of Mains surges, like stuff that should make it blow up....

I have said this before and I will say it again....and I will shout as much as I can so people get the message:

ALL SWITCH MODE POWER SUPPLIES ARE NOT CREATED EQUALLY. THERE ARE PROFESSIONAL DESIGNS THAT ARE UNBURSTABLE.

I once said that PROPER SMPS design is a "dark art". People disagreed with me.

Even I dabbled with SMPS...anyone remember the Viper series from Thompson.....I purchased 100 Viper 50's from America. Brand new.

I had someone in New Zealand winding the little Chopper Transformers for me....switching at 100 KHZ....full design and guidelines given by Thompson...

I received five little Chopper Transformers from the Guys in New Zealand...only ONE worked. Kind company there that wanted to help me. They did not charge me postage or anything.....FREE from there to here.

I laid out the board and traces according TO THE BOOK. Vipers should have been singing...

Switch on...Bang. 4.7 Ohm safety resistor popped. Fail. Viper to heaven...One gone...99 left.

And so I soldiered on. Next sample. Same board...different little SMPS transformer..

Switch on....hey it's working


I am so happy I take the little thing and show my friends. In it's little enclosure it is capable of charging a 7.5 aH Sealed Lead Acid Battery much quicker than any commercially available product......like other duties too...maybe trickle charging a car battery in Canada where it is cold.

I see Dollar signs.

Next little Chopper Transformer arrives. Very confident now. Good stuff here...and ******* bang.
And the next one...bang.

So I think I culled it at around 6 wasted.

I believe I have around 94 of the original Viper 50's left. What an experience.....

You know, on a seriously funny note.....for me...and this goes back to around 2002...

I built a SMPS that actually worked. As a rookie. Well, only one...but it worked.

Regards,
tvtech

 

[MrAl]

Hi,

The load current is not the only issue, but lets start with that.

If the load is 10ma then there is less to worry about. But you do have to make sure that your load is always 10ma. That means if you can adjust down to 5v for example then the load must be a max of 5/0.010 or 500 ohms. That actually is something we might typically see too. But that's not always the case because the voltage might be adjustable down to 2v, and then we need a much smaller resistor.

So it is usually recommended that 120 ohms is used between output and adjust terminal and that takes care of that as well as another issue.

The second issue is the temperature sensitivity of the bias current.
The bias current is part of what sets the output voltage even though it is usually ignored. It flows through the lower resistor and therefore it raises the output voltage somewhat over the normal voltage ratio calculation set point.
The problem is, the sensitivity of the output voltage is dependent on the ratio of the bias current to the normal set current through the lower resistor. If the bias current is small relative to the current through the 'feedback' resistor (like 120 ohms) then the temperature sensitivity is at a minimum. But if the bias current is larger then the temperature sensitivity goes up meaning the output will change more with temperature than is normal for a proper design.

With 120 ohm resistor the feedback current is going to be 10ma. That means the bias current is 200 times lower than the normal current, or about 0.5 percent. With a resistor of 1200 ohms, the feedback current is about 1ma, so the bias current is only 20 times the normal current, or about 5 percent.
So the difference between using a 120 ohm and 1200 ohm resistor (with appropriate load current) is the temperature sensitivity is increased by about 10 times, which basically bites. If that does not bother your application then you're ok, but if may bother it if it changes temperature by any reasonable amount.

So the 120 ohm resistor solves two potential problems.

Should you use 120 ohms? Well, if you have a constant load current then you may want to consider a higher value, as long as the bias current temperature issues does not affect your application. If you dont have a constant load current where it can decrease below 10ma then you really need that 120 ohms.

On the more practical side, i've used 200 ohms for years with the LM317L and never had a problem. Temperature range about -10 F to about 100 F. Application was voltage reference.

Also, the lower resistor should be physically solid and stable because if it becomes partly disconnected the output voltage will rise and blow out the circuit being driven. This means using a 1/2 watt resistor sometimes rather than a 1/4 watt or 1/8 watt resistor (leaded components).

 

[audioguru]

I've used 200 ohms for years with the LM317L and never had a problem. Temperature range about -10 F to about 100 F. Application was voltage reference.

The LM317LC is different from the larger higher current one,. The "minimum output current to maintain regulation" is only 2.5mA instead of 10mA for the larger higher current one. Then the resistor from the output to the ADJ pin can be 500 ohms.
But the bias current is the same for both of them.

 

[MrAl]

The LM317LC is different from the larger higher current one,. The "minimum output current to maintain regulation" is only 2.5mA instead of 10mA for the larger higher current one. Then the resistor from the output to the ADJ pin can be 500 ohms.
But the bias current is the same for both of them.

Hi,

Well, not exactly. 2.5ma is only valid for voltage differentials of 15v or less. Higher differentials call for a minimum of 5ma worst case. 1.2v/200ohms=6ma which more than saves the day. 240 ohms would just make it.

On the brighter side, the adjustment pin bias current change is only about 10 percent of nominal, so temperature tracking should be quite good.

 

[eTech]

I like this thread...

eT

 

[audioguru]

2.5ma is only valid for voltage differentials of 15v or less. Higher differentials call for a minimum of 5ma worst case.

Whose datasheet did you see? I have the datasheets from Texas Instruments for the LM317L for July 2004 and revised October 2011. They say the minimum output current required to maintain regulation is a maximum of 2.5mA when Vi - Vo= 35V which is its maximum voltage rating.

 

[MrAl]

Whose datasheet did you see? I have the datasheets from Texas Instruments for the LM317L for July 2004 and revised October 2011. They say the minimum output current required to maintain regulation is a maximum of 2.5mA when Vi - Vo= 35V which is its maximum voltage rating.

Hi,

That was straight from National around the year 2006. I tend to trust their data a little more actually.
You might also notice that TI uses a 240 ohm or 120 ohm resistor in many of their reference circuits. 240 ohms implies 5ma min too.

 

[audioguru]

Hi MrAl,
I guess you know that National Semiconductor was purchased by Texas Instruments in 2011. Then most of the National datasheets were revised showing Texas Instruments.
I found a reference of an English 2006 datasheet and its brief shows a max Vi - Vo of 40V and a Japanese datasheet of 2006 showing a minimum output current required to maintain regulation of 5mA max when Vi - Vo= 40V.