LM338T strange behavior

[dimircea]

Hi,

I searched the forum, but did not found a similar issues as mine, so I post it as a new topic.
I work on a project where a Raspberry PI and an Arduino will 'speak' together. Since the project will be powered from 5 Akkus, each having 3800mAh, and also because the total consumption of the project can go up to 1A, I use a LM338T with an attached heatsink. Yes, I am aware that the voltage can go as low as 5.xV and the output of the LM338T can go under 5V when the accus are low, but very likely at the end I will use 6 accus and this will  not be a problem anymore.
So I created the circuitry with LM338T, 220Ohm fixed value for R1 and 0-2.2k pot for R2 (I want to be able to fine adjust the voltage). I also added a 10uF Cap on the output (but not yet a 0.1uF ceramic on the input, I wait for my package with 0.1uF caps).

Now comes the problem:

1. If I don't have a load on the circuitry, then the voltage of the LM338T is almost the same with the input voltage ( 0.4V-0.6V less) - that may be ok, since a minimal 3.5mA load is specified in the datasheet

2. I add a 10 Ohm 20W resistor as a load. Now my power supply (that provides the input for the LM338T shows around 400mA, and the output voltage is 7.5V). The output voltage from LM338T is 5V (with the R2 set to a value of aprox. 660 Ohm). Actually, I was expected 500mA since  5V are dissipated in the resistor and according to the Ohm Law we have 5V = 10 Ohm * I, so I = 0.5A (on, this is an approximation since the resistor is actually 9.4Ohm, but still...the difference is big).

3. This is my real problem: I remove now the resistor and the output voltage of the LM338T drops to 3.9V and stays there until I connect back the 10 Ohm resistor. If I connect a small load (like an LED set to 10mA current) in place of the big resistor, then the voltage stays at 3.9V, but if I change the output of the power supply from 7.5V to 6V then the voltage output of the LM338T goes to 5V (which is the normal value). Now I can set back the voltage on the power supply to 7.5V and the voltage on the output stays at the normal value of 5V. Also if I stop the power supply when the voltage was 3.9V and the 10mA led was connected, then when I turn it back the voltage goes to the value that it should be, that is 5V.

Please note that all the tests were made using a linear power supply set to 7.5V.

Do I miss something or this is behavior is just strange... I have 2 pieces of LM338T and both acts the same, so can't be a defect in both...or ?
The diagram to wire the LM338T I used is exactly the one from the datasheet (the simple one with two resistors and 10uF cap on the output).

Thanks in advance for your time and answers.
Mircea.

[wraper]

What you wrote is some kind of brainfuck. Are you sure you have one ground? By that I mean single ground wire coming from power supply where you connect R2, load and multimeter at single point. and use at least 8,5 V voltage, even if 7,5 v is enough to provide 5V output voltage (what I doubt) then just barely.

[Jon86]

Well I kind of made sense of that...
Is this circuit soldered or breadboarded? I've had this kind of behavior happen with components soldered over the specified temp.

[dimircea]

Well, I tried with 8V, 9V and also 10V and no big change (maybe 0.1V differencecomparing with the strange 3.9V result).
I used 7.5V at the end since this will be mostly my voltage (at the end I will  use 6 x 1.2V (approx 7.2V under load)  NiMH Accus with 3800mAh)
I first tried on a breadboard and had this issue (was even stranger since the voltage does not drop to 3.9V but to 3.5V sometimes, and to 4.2V some other times), but then I used a solder board since I was thinking on some kind of strange capacitance or for any reason the LM338T start oscillating and do strange things...but no change.
The circuitry is new soldered (last 3-4 days) so can't be that the solder points are already dead, and also tried for hours to find if there is any strange connection, but...nothing.
And yes, I use just one GND and one VCC from a linear power supply that is capable of 1A up to 15V.

One more question: it is normal that if I measure the resistance of the circuitry between the GND and the LM338T output when no power is supplied (no connection with my linear power supply) to get 220 Ohm ?...which is actually the value of the fixed R1 resistor.

Thanks again!
Mircea.

[Jon86]

Have you tried with another meter?
The only thing I can think of is that somehow you've killed both ICs?
If I were you I'd build up the circuit again from scratch, as carefully as possible, maybe solder the test leads onto the board, and then check and double check your supply. You've probably got one tiny problem somewhere that's being repeated.

[dimircea]

I hope I did not fried both of them  .
I have two digital multi-meters: a cheap one (less than 20€) and a UNI-T61E which costed me over 100€ like 2 years ago.
Both reports very similar readings (with some tenths of volt difference, but this is not important now for me).
On another side, I used 300 degree Celsius when soldered the chip on the solder board. while the second one I never solder it, I just tried it on a bread board...

What about the 220 Ohm resistance between Output and GND when the LM338T and the entire circuitry is not powered (the R1 value I used is exactly 220 Ohm). It is normal to read the R1 value as resistance ?

[Jon86]

Well if your pot is down at 0 when you measure it, then yes, you're just measuring R1.

[dimircea]

Yes, my pot is down at 0 when I measure this (otherwise it adds the value of the resistance on the pot at the readings). Thanks for clarifying this!
 I was thinking that I also should have some resistance from the chip and not only the R1 value.

[Jon86]

Nah, using your meter on resistance youre just trying to feed voltage into the back of a transistor, you shouldn't get any current going through the regulator.

[dimircea]

Okay, thank you.

When I directly ground Adj Pin and don't use other resistance I read around 1.33V (nominal should be 1.25V as per datasheet) with the 10 Ohm resistance as load. Without resistance I read the value of the input voltage minus around 0.4-0.6V (depending on the input value, when I use 7.5V I get 7.02V at output)
Doesn't this mean that the chip is ok?...Do you know a method to check if the LM338T has some damages (dead are not for sure, but not sure if some damage occurred...mistake maybe) ?

[Jon86]

That sounds fine, but the feedback could be messed up... I think there's a minimum current that you need for the chip to be stable, but you should be meeting that with the 220...

[dimircea]

I will order another chip and see if this is doing the same (actually I will order a few of them, to be sure that I still have some  )
I will come back with news next week when I will have the new chips.

Thanks to everyone for using the time to read and answer to my thread!
Mircea.

[wraper]

Okay, thank you.

When I directly ground Adj Pin and don't use other resistance I read around 1.33V (nominal should be 1.25V as per datasheet) with the 10 Ohm resistance as load. Without resistance I read the value of the input voltage minus around 0.4-0.6V (depending on the input value, when I use 7.5V I get 7.02V at output)
Doesn't this mean that the chip is ok?...Do you know a method to check if the LM338T has some damages (dead are not for sure, but not sure if some damage occurred...mistake maybe) ?

Try to measure output with ~200-500 ohm load, if it is more than 1.3V then IC is dead
EDIT. I found one LM338 from ST, with load less than couple of kiloohms it keeps 1,23V otput voltage, without load it rises and is ~4V at 10V input, 7V with 30V input.

[sync]

Maybe it oscillates. If you have an oscilloscope check the output of the regulator when it's in the 3.9V state. Or you can try to use a DMM in AC mode. It should only measure a few mV max or something is wrong. But even then it's not a proof that the regulator is not oscillating.

Where did you buy the LM338?
And for 1A a LM317T is enough.

[dimircea]

I used my Oscilloscope (Tektronix 2246A) and checked to output (one can ask why you didn't done it before, but was so that I destroyed my last probe last week and needed to buy new ones...the packet didn't arrived yet, but I borrowed one from a friend).
Seems that the LM338T oscillates at 0.5KHz when the output is 3.9V and does not oscillates when is in the 5V state (the expected one).
So what can be the reason for this oscillation ? I will add the 0.1uF as far as I get them, but will that help ?
I bought the LM338T and most of my other parts usually from eBay or Pollin.de....don't really remember now which one was.

Also tried the test with 330 Ohm and with 550 Ohm. and it shows 1.27V...so maybe I don't have a dead LM338T...but just a stupid one 

I used the LM338T in place of LM317 since I had two of them in stock and no LM317 for now...and maybe later I can have a higher load, maybe up to 2A since I plan to add a small motor to rotate a camera module at the project - it is a 'smart' RC Truck which will have an Arduino used as sensor shield and will also control the RC receiver (is a standard RX2 based one) and will use the Raspberry PI as brain (getting data from Arduino and sending commands to Arduino to controll the motors).

[Jon86]

That sounds strange, I've operated an LM338 without any caps before and got perfect clean performance over all the specified ranges, including loading and unloading it. Ah well, try out those caps, maybe change the resistor values a bit, add some more capacitance here and there and see how it goes 

[madires]

I bought the LM338T and most of my other parts usually from eBay or Pollin.de....don't really remember now which one was.

I wouldn't be suprised if ICs bought via eBay behave strangely.

[codeboy2k]

For battery operation, you should consider a switcher anyways, not a linear regulator like the LM338T

You might be able to find a buck-boost controller that can handle your load up to 1A internally with moderate heat sinking; or if not, its easy to add an external MOSFET to any run of the mill controller IC.

You'll get more efficiency and thus more battery life from a switcher vs a linear regulator, plus you can get more energy out of your battery pack when it goes below 5V.  The dropout voltage of the LM338T can be 1.8V at low amps, and up to 3V at higher amps and over the temperature range.  Using batteries with only 7.5V max or so is just too close in my opinion to try to get a reasonable regulated 5V from it for any length of time. This is why I suggest you should seriously consider a buck-boost switch mode regulator.

You can use the TI Web-Bench application to help you chose components and build it yourself, or spend more money and chose a ready made module, if you're not comfortable with making a switcher... but it's not necessary... building a DC-DC switcher is not that hard and only requires a few components.

[Strada916]

schematic or photo of circuit would be a great help. Also check the pin outs. Have you tried another reg ic?

[dimircea]

I think I will order this module:
http://www.ebay.de/itm/3A-DC-DC-Converter-Adjustable-Step-down-Power-Supply-Modul-4-5-28V-1-3-20V-NEU-/400581191070?pt=Netzger%C3%A4te&hash=item5d447fe59e
It is not expensive, it is switch based (as you say, better efficiency and less heat to dissipate). I will use the LM338T only until I get the module then I will replace it with the one from the link.
However I am still confused about the behavior of the LM338T...used LM317 before without such problems, and same wiring as I done for LM338T now. Today I will try to replace the 10uF cap with a bigger one and also add the 0.1uF on the input and see if this change something or not.

Thanks for your help!

[sync]

Seems that the LM338T oscillates at 0.5KHz when the output is 3.9V and does not oscillates when is in the 5V state (the expected one).

500Hz? That is strange. Usually regulators oscillates from a few 10kHz to MHz. I think I saw a regulator in (thermal?) protection mode oscillating at low frequencies.
I would not rule out dodgy parts. Especially from Ebay.

[dimircea]

And now the cherry on top of the cake!
I changed the 10uF to 47uF  (I did not had another value in between)....but no change (maybe a bit...in place of 3.9V it was oscillating between 3.9 and 4.0V).
However, in the main time the post arrived and I have my order with 0.1uF caps.
As a last resort trial, I added one between LM338T input pin and GND ...and that was all! The LM338T acts as it should!
Now I get a quite clean 5V with ripples of less than 0.01V between GND and Output for a load of about 420mA ( 1 x 10 Ohm 20W resistor).
The input voltage is still 7.5V and the output is stable...and it is even stable with 6.7V (under this value the voltage on the output starts to drop).  I have to check with a load of 1A to see where it is stable, but for now it is ok so.
Anyway I order the switch DC-DC step down module because of the efficiency and also less heat to dissipate.

One in all, thanks to everyone for answers and indications!
I enjoy to be member of this forum!
Wish you a nice day!

PS: sorry if my English is not the best since is not my native language (I am from Germany)

[grumpydoc]

As a last resort trial, I added one between LM338T input pin and GND ...and that was all! The LM338T acts as it should!

I'm sort of surprised that no-one picked up on this, even when the suggestion was made that the regulator was oscillating as you'd mentioned in your first post that you had ordered some caps to use (and so, obviously, hadn't got one connected).

The datasheet does say that you need input decoupling if you have more than 6 inches of track/wire from the main decoupling caps.

[dimircea]

Yes, there are more than 6 inch since I use a power supply and the cables from the power supply to the board where is LM338T mounted have around 30-35 cm, plus a few cm more inside the power supply until the main caps.
But now I am happy that it works and did not killed  both LM338T ICs (even if I somehow know that I did nothing wrong since is not a complex circuitry and also just followed the diagram from datasheet that worked with LM317 in some of my other projects).

Just as a mention, I am not a first level beginner in electronics but also by no way a professional nor expert...for me electronics is a hobby that I enjoy and want to combine with my real job that is IT Specialist and Teacher for IT technologies at BTU Cottbus (a German University of Technologies)