Zero Volts from a LM 317

[nick.sek]

Hey everybody, I just wanted to show you a simple approach to get 0 volts from a LM 317 voltage regulator. Now of course it has a not so good feature, which is it constantly eats 1.4 volts (in theory). It obtains zero volts by using diodes. If you guys have any other approaches on this let me know.  So here the link below:

[Rerouter]

kinda large problem with your design buddy, your sense line is before the diodes, and diodes tend to vary quite a bit with current, far better approach is use 2 diodes on the  external negative, and have only the set pin pulled down by the true negative, this way you dont loose regualtion with the rest of the circuit about 1.6V less than the input,

edit: actually i'm partly wrong, my method would be similarly wrong to yours,

edit2: if you dont mind a little error introduced from input voltage fluctuations, use a resistor and a capacitor just to smooth it out, then have a second resistor bias the set pin high, and you would only need to calculate it for your 0V point, as you are setting it manually instead of digitally,

[amspire]

There are much better designs to do this using chips like the TL431 or opamps to cancel out the LM317 offset, but you still need a negative supply. In the suggested approach, you do not need a negative supply, but the regulation would be worse then a simple emitter follower supply.

It is very limited in power handling, it's short circuit protection is poor in that it is only design for infrequent short term failure shorts, rather then constant overloads that any power supply supply is expected to cope with.

To me, the common feature of almost every LM317 based power supplies is that they are pretty bad power supplies.

Richard.

[MikeK]

Would you be able to do it without a negative supply, by using an inverting op-amp to offset it?

[free_electron]

@mikek : and how do you think the opamp would pull its output negative ? it can't go beyond its rails ...

As amspire said LM317 stink as 'lab supply'  . An LM317 is a really good regulator to make a FIXED voltage that cannot be obtained using of the shelf regulators like 78xx series. its twin, the lm337 can replace the79xx series. It also saves you from having to carry all the 78xx and 79xx in stock. al you need is a big bucket of LM317 and LM337 and a pile of trimpots ... Perfect. be my guest.

There is another thread here on the forum form someone that also makes an LM317 based supply. I posted along reply there with all the common pitfals desinging robust power supplies , what diodes to use , where and when to put  caps , how to do slow start and slow start recovery , reverse polarity protection and preventing backfeeding the supply.

But using the darn thing to make a lab supply is sheer madness !
You can make a really good lab supply with a couple of opamps and a darlington transistor. It will regulate down to zero , and have an adjustable current limit as well.

In my book : it ain't a lab supply until you have an adjustable current limiter.

[Bored@Work]

I have seen that diode "solution" before. I think it is a common design in the ham community. That doesn't make it in any way a good idea or good design. If you want to do the electronics community a favor pull that video from YouTube. BTW, I am not a big fan of videos when simply showing the schematics in an image would have been done.

The only reasonable way to get an LM317 down to zero is to pull the adjust pin below zero. And that isn't without problems. The negative adjust voltage directly affects the output voltage, so it needs to be stable.

There is no silver bullet when building a lab power supply from a single IC from the '70th. Even the cheap Chinese "lab" power supply makers haven't found a way. And the Chinese are masters in leaving out "unimportant" components and stretching component specs.

[Zero999]

I like the op-amp method. It may not be the simplest but it needs minimal adjustment and is pretty reliable. Of course, you'll get better results from a low offset op-amp and tight tolerance resistors but I just used the old 741 because I've got loads of them lying around.
http://www.silicontronics.com/index.php?action=ezportal;sa=page;p=19

You do need a negative supply which is easily generated with a voltage doubler, if you're using a transformer.

[nick.sek]

I'm excited to proto-type this circuit, thanks for showing me this op-amp ) volts power supply!

[free_electron]

but, you don't NEED and opamp if you got a negative supply !
just tie the bottom of the resistor divider to the negative voltage and you are done !. the lm317 doesn't give hoot. All it's after is having 1.25 volts between ot and adjust..

Code: [Select]

                 _______
(+20)-----------|IN  OUT|---------+---------(out)
                |       |         |
                |  ADJ  |       240 Ohm
                 -------          |
                    |             |
                     -------------+
                                  |
                               potmeter 5k
                                  |
                                  |
(GND)-----------------------------)---------(GND)

                                  |
(-1.2 volt)-----------------------


no need to go mucking about with opamps.

here you go
                               --------------|<|-----------------
                              |     _______                      |
    ---0---|>|----0---0-------0----|IN  OUT|----0-----0-----0----0----0----(out)
)||(   |          |  _|_           |       |    |     |    _|_        |
)||(    ---|<|-   | |___|          |__ADJ__|   240    |   |___|       |
)||(           |  |  ---               |        |    ---   ---       ---
    ---0---|>|-|--    | 1000u           --------0    /_\*   | 100u   /_\
       |       |      |                         |     |     |         |
       0---|<|-0------0-0---0-------------------|-----|-----0---0-----0-----(0)
       |                |  _|_                  0-----0         |
      _|_               | |___|                 |    _|_        |
     |___|     --|>|----   --- 100u           5kpot |___|10u    |
  100u---     |             |                   |    ---        |
       |      |             |                   |     |         |
        ------0--|<|--------0----------------1k-0-----0-|<|-|<|-

all diodes 1n4001 except * : 1n4148


fully protected lm317 supply with slow start and adjustable down to 0 volts...

[Zero999]

but, you don't NEED and opamp if you got a negative supply !

Do you really think I didn't know that?

all diodes 1n4001 except * : 1n4148

fully protected lm317 supply with slow start and adjustable down to 0 volts...

Diodes are fine if you keep it in a temperature stabilised oven, otherwise the output voltage will drift by as much as 200mV.

The diodes also won't be matched to the LM317's 1.2V to 1.3V (1.25V typical) reference so the minimum voltage setting won't be 0V. Going from the typical vales on the datasheets, it'll probably be around 50mV but could be anywhere between -150mV and 200mV, worse when temperature drift is factored in.

The circuit you posted should also use 120R and 2k5 for the potential divider as the LM317 has a worst case minimum load current of 10mA.

An alternative solution might be a TL431 and adjustment pot which isn't much more complicated but it'll be a hell of a lot more stable than diodes.

You don't need an op-amp but it's a good way to make a stable -1.25V reference which is almost perfectly matched to the LM317.

[amspire]

but, you don't NEED and opamp if you got a negative supply !

Do you really think I didn't know that?

all diodes 1n4001 except * : 1n4148

fully protected lm317 supply with slow start and adjustable down to 0 volts...

Diodes are fine if you keep it in a temperature stabilised oven, otherwise the output voltage will drift by as much as 200mV.

As you vary the the output load from 1mA to 1A, a pair of 1N4004 diode will have a forward voltage change of 0.6V in total. That is just too much variation to be useful in a lab power supply.

What is the use of adding diodes to get 0V out, if the output can vary by 0.6V? Given the choice of decent regulation down to 1.2V or horrible regulation down to 0V, I would go for 1.2V minimum any day.

Richard.

[IanB]

The circuit in the datasheet shows a 1.2 V zener. Surely that's got to be better than ordinary diodes unless you are really penny pinching?

[amspire]

The circuit in the datasheet shows a 1.2 V zener. Surely that's got to be better than ordinary diodes unless you are really penny pinching?

You can get two terminal 1.2V reference ICs, but not 1.2V zeners. The trouble with reference ICs is they cannot take the full power supply current and that was the whole point of this thread.

I think everyone understands that if you have a negative supply, there are lots of ways to get a LM317 to go to zero volts. Not so easy if you do not have a negative supply though.

Richard.

[caroper]

Just out of interest.
I was experimenting the other day with getting the 317 down to Zero volts and actually managed to drive the output negative by a couple of volts. I immediately trimmed the offset back up but it raised a question I had not previously considered.

I was initially surprised that it went negative at all but will it damage the 317 if it is driven negative for extended periods?

I was running the test with no load and a bypass Diode, but I would imagine there is a possibility of the 317 attempting to sink rather than source if it is driven negative.

[codeboy2k]

@free_electron

Are you reading Dave's blog on a dialup terminal ? VT100? ADM-3A ?   



I respect your knowledge, it's clear you know what you are talking about, all your posts are a wealth of great information. Truly.
... but...text schematics from usenet ??

We ought to take a collection to buy you a free copy of Eagle capture

lol:) just poking fun at your ascii art..

[Mechatrommer]

but, you don't NEED and opamp if you got a negative supply !
[ the code ]...

sorry master but with respect... ascii drawing is dead dinasour. i know the feeling of hard to change or in love with what we already used to, but face it, its dead. download ltspice or tina spice you can draw schematics in no time (capture screen, that Prt Scrn button on keyboard and paste/save in paint if you dont have proper graphic tool)... and easier to read too love ya!

[free_electron]

As you vary the the output load from 1mA to 1A, a pair of 1N4004 diode will have a forward voltage change of 0.6V in total. That is just too much variation to be useful in a lab

If you look at my schematic you will see that the output current does not go throught the diodes... The two diodes are only used to make a -1.2 volts reference. So there will be no variation.

As for temperature drift : silicon diodes drift about -2 mV per degree c. So if the supply gets 50 degrees above ambient the supply will drift 200mv. Keep the diodes away from the heatsink ..
Not a problem and not a reason to go mucking about with zeners ( that also drift ) and opamps. If you really want to make it stable use a tl431 based reference.

But this whole excercise is crazy. Adding 5 dollar in parts to get a 0.5 dollar regulator, that wasn't designed to go to 0 , to 0 ...

Just spend those 5 dollar on an dual opamp some resistor and a transistor and make a real power supply with adjustable current limit.

As for ascii art : i like ascii art. Its easy to make , you dont need to muck about with saving as gif ( cant use jpeg because the compression messes it up sometimes ) or png , dont need to upload it to some shady imegahoster. You can just slap it in.

[SeanB]

ASCII art survives a run though with Exchange server and outlook, while an attached image can be mangled, resized to zero or even substituted with a same sized item from the cache.............

[codeboy2k]

But this whole excercise is crazy. Adding 5 dollar in parts to get a 0.5 dollar regulator, that wasn't designed to go to 0 , to 0 ...

Just spend those 5 dollar on an dual opamp some resistor and a transistor and make a real power supply with adjustable current limit.

This is the most important lesson to take away from this exercise ==> it's never good, nor necessary, to over-engineer anything.
free_electron is absolutely correct here. When we have learned to make the right decisions, that are appropriate for the situation, only
then are we rightfully ENGINEERS.

As for ascii art : i like ascii art. Its easy to make , you dont need to muck about with saving as gif ( cant use jpeg because the compression messes it up sometimes ) or png , dont need to upload it to some shady imegahoster. You can just slap it in.

Ascii art doesn't actually bother me, I was poking fun at you.  For me, I grew up on BBS's and USENET news, so I accept it. It is sometimes a little
more work to understand it, and I am sure its more work for you to create it, but it's just not necessary anymore given the tools that are
available today. (most of them free)

ASCII art survives a run though with Exchange server and outlook, while an attached image can be mangled, resized to zero or even substituted with a same sized item from the cache.............

I've never worked in an MS Exchange environment, so I can't comment on that . But I have also seen other mail servers mangle ASCII art just as easily as you say
exchange preserves it.  So it probably goes both ways.  I think if you're reading and posting on a web forum using an intermediary service, not browsing directly, then all bets are off anyways.  And the forum seems to preserve most images that are uploaded, and can be uploaded directly, no need to host them on another server. The forum accepts png images just fine, see attached.


Cheers!

[free_electron]

ASCII art survives a run though with Exchange server and outlook, while an attached image can be mangled, resized to zero or even substituted with a same sized item from the cache.............

or worse : be actually a fake image that causes a buffer overlow and downloads nastyware onto your computer ... Adobe just got nailed ... download a 'specially crafted tiff file' to install rootkits ... they are now going to fix it. and we all know about 'poisoned' gif files and other formats that cause trouble.
I have yet to see the first ascii art that can infect a computer ...

  And the forum seems to preserve most images that are uploaded, and can be uploaded directly, no need to host them on another server. The forum accepts png images just fine, see attached.

i didn't know this forum hosted its own images ... i am very active on a few others where you need to use an external hoster like imageshack (imagine projectile vomit smiley here ) or some other 'shady' free image hosters...

anyway : i still draw faster schematics in ascii art than in a drawing program just fire up notepad and crank away.

[george graves]

LOL at the ascii art,  It's 2012 - just use a jpg.

[baljemmett]

LOL at the ascii art,  It's 2012 - just use a jpg.

Ugh, no; JPEG is not designed for that sort of image.  Use a PNG.  Or just use whatever the hell you want to use when you're giving someone advice for free

[Uncle Vernon]

Simple device for achieving 0V output from the LM317 or other adjustable 3 terminal regulators.

[free_electron]

@ uncle vernon : bwahaaaaa. That's a good one. You are perfectly right. And you can put it in series or parallel withe the output , there is no drop ( pun intended) and the polarity doesn't matter either.

[Bored@Work]

Ugh, no; JPEG is not designed for that sort of image.

Nitpicking: It is, but nobody uses it.The JPEG standard also contains a lossless compression JPEG version which would work well for things like schematics. But similar to JPEG 2000, it never became popular and you would have to be very lucky to find an implementation.