Wierd overheating of LM780x's

[Mark Fortune]

Hi all,

A couple of months ago I built a circuit that involved connecting a few
LM780x voltage regulators to the same 12v supply. The circuit was very
simple, Gnd pin connected to ground, Input pin connected to +12v and
output was tried both open and with a small load - an LED and resistor.

On thier own, I experienced no problems whatsoever; However, when I
connected two or more LM780xCV regulators in parallel (that is, using
the same + and - rails) I noticed they all got extremely hot (too hot to
touch!). This was the case with whatever combination I used - 2x
LM7805CV 5v regulators, a LM7809cv and LM7812cv, an LM7805CV and
LM7809CV etc etc. This was the case even when there was no load on any
of the regulators.
Originall I just used 2 5v regulators, the idea was to wire two 5v
regulators in parallel so that they could handle a load of up to 2A
instead of just 1A.

I even tried wiring them in series - IE the input of the LM7805CV
connected to the output of an LM7809CV - they both still heated up to
the extreme.

In contrast, when connecting an LM7805CV to the +12v supply, and putting
on fairly large loads only warmed the device - I could still touch the
heatsink and not burn myself.

I'm not entirely sure what the problem is. For reference the power
supply I am using is an old AT computer PSU rated at 300w which i've
modded so that I can tap into various voltage outputs.

One thing I did notice today - I rigged up a LM7805CV for a regulated 5v
output, and hooked up my scope to the output. I noticed a slight voltage
variation of about 200mv on top of the 5v output. On closer inspection
this turned out to be a sine wave of about 15MHz - placing a 10nf
capacitor across the output and gnd seemed to eliminate this, so I was
just seeing a flat 5v.

I'm wondering now, this sine wave I saw... was it something to do with
the regulators internal regulation mechanism? and if this is the case...
could it be why all those months ago my regulators were getting hot
enough to cook my breakfast on? My guess is that the regulators are
competing against each other - with one pulling the supply voltage down
a little... the other compensating by adjusting its output which alters
the supply voltage again, etc etc.. creating some sort of feedback
between the devices?

any ideas? I would love to know what's going on here.

Oh and btw, the output pins of the devices are not connected in any way,
they are driving individual loads or are held open.

Thanks,
Mark

 

[Michael A. Terrell]

Hi all,

A couple of months ago I built a circuit that involved connecting a few
LM780x voltage regulators to the same 12v supply. The circuit was very
simple, Gnd pin connected to ground, Input pin connected to +12v and
output was tried both open and with a small load - an LED and resistor.

On thier own, I experienced no problems whatsoever; However, when I
connected two or more LM780xCV regulators in parallel (that is, using
the same + and - rails) I noticed they all got extremely hot (too hot to
touch!). This was the case with whatever combination I used - 2x
LM7805CV 5v regulators, a LM7809cv and LM7812cv, an LM7805CV and
LM7809CV etc etc. This was the case even when there was no load on any
of the regulators.
Originall I just used 2 5v regulators, the idea was to wire two 5v
regulators in parallel so that they could handle a load of up to 2A
instead of just 1A.

I even tried wiring them in series - IE the input of the LM7805CV
connected to the output of an LM7809CV - they both still heated up to
the extreme.

In contrast, when connecting an LM7805CV to the +12v supply, and putting
on fairly large loads only warmed the device - I could still touch the
heatsink and not burn myself.

I'm not entirely sure what the problem is. For reference the power
supply I am using is an old AT computer PSU rated at 300w which i've
modded so that I can tap into various voltage outputs.

One thing I did notice today - I rigged up a LM7805CV for a regulated 5v
output, and hooked up my scope to the output. I noticed a slight voltage
variation of about 200mv on top of the 5v output. On closer inspection
this turned out to be a sine wave of about 15MHz - placing a 10nf
capacitor across the output and gnd seemed to eliminate this, so I was
just seeing a flat 5v.

I'm wondering now, this sine wave I saw... was it something to do with
the regulators internal regulation mechanism? and if this is the case...
could it be why all those months ago my regulators were getting hot
enough to cook my breakfast on? My guess is that the regulators are
competing against each other - with one pulling the supply voltage down
a little... the other compensating by adjusting its output which alters
the supply voltage again, etc etc.. creating some sort of feedback
between the devices?

any ideas? I would love to know what's going on here.

Oh and btw, the output pins of the devices are not connected in any way,
they are driving individual loads or are held open.

Thanks,
Mark

Do you have the recommended capacitors on the input ant outputs of
the regulators?


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[mc]

Any capacitance on the input (across the supply rail)? You may have
invented a new kind of 780x push-pull oscillator!

Any oscillation observed when they're getting hot?

 

[Eeyore]

Hi all,

A couple of months ago I built a circuit that involved connecting a few
LM780x voltage regulators to the same 12v supply. The circuit was very
simple, Gnd pin connected to ground, Input pin connected to +12v and
output was tried both open and with a small load - an LED and resistor.

On thier own, I experienced no problems whatsoever; However, when I
connected two or more LM780xCV regulators in parallel (that is, using
the same + and - rails)

Not really surprising !

What made you think that was a sensible thing to do ?

Graham

 

[Michael Black]

Not really surprising !

What made you think that was a sensible thing to do ?

Graham

Wait, the original post is a bit disjointed, but it seems he's not
talking about "in parallel" as we might think, but simply feeding
two regulators from the same source, which of course is legit. After
all, they were originally referred to as "on-card regulators", the
intent being that you'd have one master supply feeding a bunch of
small regulators on each board.

And way into the post, he does mention seeing oscillation on
an oscilliscope, which went away when he bypassed the regulators.

The things oscillate, because they are high gain devices, and
everyone has known from the start that they oscillate.

Michael

 

[Walter Harley]

Hi all,

A couple of months ago I built a circuit that involved connecting a few
LM780x voltage regulators to the same 12v supply. The circuit was very
simple, Gnd pin connected to ground, Input pin connected to +12v and
output was tried both open and with a small load - an LED and resistor.

On thier own, I experienced no problems whatsoever; However, when I
connected two or more LM780xCV regulators in parallel (that is, using the
same + and - rails) I noticed they all got extremely hot [...]

As others have suggested but not quite stated outright: these regulators
need capacitors to stabilize them. Otherwise they oscillate, which is what
you're seeing.

You need to read the data sheet for the chip, and do what it says. It will
tell you what values of capacitors to use, where to place them, and so
forth. Go to the manufacturer's web site to get the data sheet.

Reading the data sheet is a pretty good idea for *any* chip you use, not
just regulators.

 

[Eeyore]

Wait, the original post is a bit disjointed, but it seems he's not
talking about "in parallel" as we might think, but simply feeding
two regulators from the same source, which of course is legit. After
all, they were originally referred to as "on-card regulators", the
intent being that you'd have one master supply feeding a bunch of
small regulators on each board.

And way into the post, he does mention seeing oscillation on
an oscilliscope, which went away when he bypassed the regulators.

The things oscillate, because they are high gain devices, and
everyone has known from the start that they oscillate.

Indeed. I guess he was just lucky to begin with.

Graham

 

[sparc]

all voltage regulators deviate from the voltage they are supposed to
provide. So when u connect two in parallel you get a one at say ...12V
the other at 12.1 V. These regulators have a very small internal
resistance as seen by the load ... so a 0.1 V diff cld be working
across a few milliohms ! hence the heat ... (mercifully these regs have
thermal runaway protection so they dont just give up the ghost right
away)

u shld nev parallel regulators without an active circuit that ensures
equal current sharing between the regs (automatically ensuring equal
voltages).

 

[mc]

all voltage regulators deviate from the voltage they are supposed to
provide. So when u connect two in parallel you get a one at say ...12V
the other at 12.1 V. These regulators have a very small internal
resistance as seen by the load ... so a 0.1 V diff cld be working
across a few milliohms ! hence the heat ... (mercifully these regs have
thermal runaway protection so they dont just give up the ghost right
away)

Yes, but the OP said that only the inputs were in parallel. The problem
seems to be no capacitor on the inputs.

 

[sparc]

mc <----> with just the inputs paralleled, how does he hope to draw
twice the current at the outputs ?

Mark <--> What is your input voltage ? Linear regulators essentially
burn off power @
(Vin - Vout) Iout, which, at 200mA and a diff of say ...10V (5W) can
be enough to roast your fingers on, not to mention the device. Do you
have adequate heatsinking ?

220u seems to be quite inadequate ....try 1000u instead ..athough that
wont make much diff to your heating problem.

 

[sparc]

sorry ...didnt read the entire thing thru ...

seems like u have a decoupling problem here ....try 0.1u ceramics or
tantalums at the input to decouple the regulators (10n arent quite good
enuff). Also, increase ur smoothing caps to 1000u min. If they still
heat up .....puke ...

 

[Dave Platt]

Input voltage is around 12v, from an old switch mode AT computer PSU. I

wouldnt say i'm heatsinking the regulators adequately at the moment, but
like I stated before individually they seem to run fairly large loads (I
think about 500mA) and only get "slightly warm" without a heatsink

You may have another problem here, if you're using a traditional
multi-voltage PC switcher supply. Many of these PSUs *need* to supply
a significant amount of current to their 5-volt output - if the load
falls below this point, the 12-volt supply may not regulate properly,
and the supply may either feed you bad voltages or may shut down
entirely.

I will try 1000uF, I just grabbed a 220uF at random and noticed the
oscillations dropped off when I connected it so I used that.

It's also a good idea to put a small bypass cap right at the input
terminals, with a minimum amount of lead length (no more than an inch
at most).

 

[Mark Fortune]

Yes, but the OP said that only the inputs were in parallel. The problem
seems to be no capacitor on the inputs.

I tried it again last night, a 5v and a 9v reg connected to teh same
supply rail, 220uf cap on the supply rail, a 10nF cap from the output of
each regulator to ground and a load on each regulator - oddly the 5v was
cool to the touch, but the 9v was getting hot with about 200mA running
through it - although not super hot as before.
I will dig out the spec sheet when I get a spare five minutes and have a
browse through it, see if I cant pick up a few clues there.

Originally I was hoping to use these regulators to build myself a
regulated power supply with various output voltages that could all be
used at once. I wasnt really expecting these kinds of thermal issues
though... I just expected it to get hot when I drew a lot of current
through them.

Thanks for all the advice people! got a few nice pointers there.
Mark

 

[Mark Fortune]

mc <----> with just the inputs paralleled, how does he hope to draw
twice the current at the outputs ?

I was perhaps being a little vague, my appologies. I had hoped to draw
twice the current with the outputs in parallel as well using similar
regulators, although I understand now that it is perhaps not as simple
as that. The overheating problem I was experiencing was observed with
just the inputs wired in parallel though.

Mark <--> What is your input voltage ? Linear regulators essentially
burn off power @
(Vin - Vout) Iout, which, at 200mA and a diff of say ...10V (5W) can
be enough to roast your fingers on, not to mention the device. Do you
have adequate heatsinking ?

Input voltage is around 12v, from an old switch mode AT computer PSU. I
wouldnt say i'm heatsinking the regulators adequately at the moment, but
like I stated before individually they seem to run fairly large loads (I
think about 500mA) and only get "slightly warm" without a heatsink

220u seems to be quite inadequate ....try 1000u instead ..athough that
wont make much diff to your heating problem.

I will try 1000uF, I just grabbed a 220uF at random and noticed the
oscillations dropped off when I connected it so I used that.

I think perhaps in light of the posts i've recieved on this subject
there's a lot I dont know about these little devices which perhaps I
should. So i'm gonna take some good advice and bury my head in the
datasheets before I continue experimenting.

thanks.

 

[mc]

I tried it again last night, a 5v and a 9v reg connected to teh same

supply rail, 220uf cap on the supply rail, a 10nF cap from the output of
each regulator to ground and a load on each regulator - oddly the 5v was
cool to the touch,

Outputting how much current?

but the 9v was getting hot with about 200mA running through it - although
not super hot as before.

3 V drop x 0.2 ampere = 0.6 watt, enough to get a chip quite warm.

 

[Lionel]

I think perhaps in light of the posts i've recieved on this subject
there's a lot I dont know about these little devices which perhaps I
should. So i'm gonna take some good advice and bury my head in the
datasheets before I continue experimenting.

I've built hundreds of similar supplies, & always put 100nF monoblock
caps directly between each input pin & ground, & each output pin &
ground. If you're running long leads between the unregulated supply &
the 78xx reg's, it may be helpful to place 10uF (approx.) tantalums
between the 78xx input & ground.
If you do all that, your PSU should work fine.
Don't forget to heatsink the reg's & remember that if you're using any
negative reg's (79xx series) you'll need to isolate the heatsink tabs.

 

[Mark Fortune]

I've built hundreds of similar supplies, & always put 100nF monoblock
caps directly between each input pin & ground, & each output pin &
ground. If you're running long leads between the unregulated supply &
the 78xx reg's, it may be helpful to place 10uF (approx.) tantalums
between the 78xx input & ground.
If you do all that, your PSU should work fine.
Don't forget to heatsink the reg's & remember that if you're using any
negative reg's (79xx series) you'll need to isolate the heatsink tabs.

I had a look through the datasheet the other day and one circuit they
showed (a dual rail supply) suggested using a 330nF cap between the
inputs and ground and a 0.1uF between the output and ground.
I tried this but im still getting similar results - the 7805 is cool to
the touch, and the 7809 is getting too hot - i've even changing the 7809
for another 7809 in case it was faulty - still no change. I did put a
big capacitor across the + and - input rails - 2200uF and this didnt
change a thing... I will try some 100nF caps across the pins and see if
that makes a difference, but i'm starting to suspect its something
deeper, as i'm still getting the over-heating even when i'm using
different supply rails from the same switch mode supply (ie the +5v for
the 7805 and +12v to power the 7809, although the 7805 only outputs 4v
from this as would be expected). to note however, the 7809 is cool to
the touch if I take the 7805 out of the circuit, with or without load.

I think i'll do some more experimentation at the weekend, try different
capacitor sizes across the input/gnd output/gnd, maybe even dig out one
of my old transformers and try building a new unregulated, unswitched
supply for them...

ooh you know what i've just had a thought, the fly leads from the power
supply are fairly long (about 10") and the 0v/ground wire is pretty
thin... that might be a potential problem... will experiment with
thicker grounding too.

Regards,
Mark

 

[Walter Harley]

[...]
I had a look through the datasheet the other day and one circuit they
showed (a dual rail supply) suggested using a 330nF cap between the inputs
and ground and a 0.1uF between the output and ground.
I tried this but im still getting similar results - the 7805 is cool to
the touch, and the 7809 is getting too hot [...]
to note however, the 7809 is cool to the touch if I take the 7805 out of
the circuit, with or without load.

If you've bypassed the input and output as described and you're still having
the overheating, then you're right, there's something more going on.

It does seem possible that part of the problem is the PC supply you're
using. Take a look at the output of the PC supply with your scope, to see
whether there's oscillation. You might briefly try powering your circuit
from a 12V lantern battery, or from two 9V batteries strung in series, to
see if the problem goes away; if so, then it's an interaction between the PC
supply and the downstream regulators. But in general, as long as there's
enough load on the PC supply you should be okay. You may need to put a load
on the 5V part of the PC supply to get the 12V to regulate properly - try
for something that draws a couple of watts, e.g. a 4 ohm 10W resistor, or
even a 10W automotive bulb (turn signal, overhead lamp, etc.).