Linear voltage reg drop out or failure?

[Al (astral highway)]

Bit of a head-scratcher, this one.

A circuit has 3 x voltage regulators, D-PAK versions of 7824,
7809, 7805

Problems began in the cct which I have traced back to the power supply and there we are…

There’s a nice voltage overhead of 31V after the bridge rectifier.

But suddenly we have 22.6V, 7.7V and 3.9V. This is off load. There’s just a bunch of flip-flops and comparators and idle gate drivers with no drive signal, no current to draw here.

The 9V regulator is the hottest at 40C, which is coolish but however normal off load. The other two are only about 28C, which is too cool.

I’m completely flummoxed here.

The devices appear to be grey market, although I only just noticed that. They are D-PAK versions so loathe to desolder from their pads until I have some hunch .

Cheers folks

[Nickthedentist]

Is it really off-load,or are you just assuming it is because it should be?

Is your meter definitely beyond reproach?

Have you tried scoping the various rails?

I always assumed series regulators shouldn't get warm at all under no-current-draw conditions, but I could be wrong there.

[Al (astral highway)]

Hi Nick,

Yep, definitely off load - that’s to say there are a few uA for the idle logic and a few mA for a couple of LED’s.

The voltage regulators always do run warm.

I’m going to take DC measurements on the scope by way of checking my off-brand DMM.

What’s odd is that there’s roughly a volt missing from the output of each regulator, regardless of its design output.

I did desolder the D-PAK 5V regulator and tacked another device in place. Same result , 4V… so it’s not a fault of the regulators. Nothing else has changed. There was a bulging cap that I spotted before it blew up but that’s the danger of using 35V rated components so close to their limit.

[sortedradio]

It's worth checking they are not oscillating, and that your ground is good.

[Al (astral highway)]

Originally Posted by sortedradio

It's worth checking they are not oscillating, and that your ground is good.

Hi Martin,

It would be bizarre if they suddenly developed this fault out of the blue. But it may well be time to check in case it’s a question of one or two of the caps going awry. The ground is solid.

Just ‘scoped output !

Sawtooth on output after bridge rectifier. It must have blown a junction or two.
Easy enough to fix.

[cmjones01]

Another vote here for them being unstable and oscillating. 78xx series are pretty stable, usually, but under some circumstances they go mad. Measuring the DC output voltage gives strange results and the giveaway is that they get warm even when they're not delivering any current. Look at the rails with a scope. It may be only one of the regulators that's actually unstable, and its output is affecting the others. It's a long time since I've used the 78xx series (they're basically obsolete in the commercial world these days), but I think the capacitor on the input is most important, in contrast to modern LDOs in which the output capacitor is critical. Make sure there's a low-ESR (ceramic is a good bet) capacitor of 0.1uF or so close to each regulator on the input side.

Chris

[dave cox]

They all have a similar quiescent current, so fed from the same rail, at no load and similar heat-sinking - they should all be at an similar temp. Probably the 9V one is doing something odd - as already mentioned. making dc measurements in the presence of ac (esp at higher frequencies) can give quite amusing results. I recall ye olde worlde AVO 8 was one of the very few instruments that didn't go totally nuts when immersed in RF :P

The ripple looks OK, 100mV p-p @ 100Hz - if I am reading the scope setting right

dc

[Al (astral highway)]

Chris: hmm yes, there’s a 0.1uF ceramic SMD cap right on the pin of each, to GND. Couldn’t be closer.

The only fault I could find was that one of the SMD electrolytics was poorly bonded on its ground pin from when it was originally factory soldered. But I fixed that yesterday.

I did find bursts of oscillation on the 5V device, which I had just replaced. So it’s not a fault intrinsic to the device and must as you say be from upstream.

It does however look as if the 9V device could be in a spot of bother. I suspect it was chronically overloaded once too often, although it never reached above 65C IIRC.

Are there (D-PAK) devices that aren’t 78xx but the new generation you’re describing there? Same pinout? If so I’ll replace all three at the same time.

Dave: haha , yes, shenanigans with RF messing things up.

You’re of course exactly right about the ripple . I was rushing out and didn’t look properly. The bridge rectifier tests fine.

Both: there seems to be no remedy except replacement of all three regulators. I’ll wait to see what you say, Chris , about the new generation of devices.

[Al (astral highway)]

On second appearances, there should really be a snubber to protect the 9V regulator from switching noise coupled back into it.

[G6fylneil]

There was a telemetry unit which I used to production test. It had a 5V logic supply derived from the 12V analog supply. Occasionally the 5V regulator, a LDO type, was unstable, the supply voltages read OK. The the 5V rail was well decoupled, and the logic worked OK but noise was put on the analogue supply, which then gave problems. Changing the 5V regulator always fixed it. The unit had a BABT approval, which limited what I could do.

[duncanlowe]

Older voltage regulators were notorious for being susceptible to external RF if they didn't have a fair capacitor on the input. So if it does have one but it's degraded that could easily be the fault.

Back in the nineties, I had one part of a complex system that kept failing when we subjected it to RF at a particular frequency. Display, LEDS and control of the outputs all went bezerk. It was on the third iteration of 'solution' where there was no improvement in performance, I looked at the circuit. Absolutely no capacitors on the 5v LM regulator. Went to the test facilities component drawers and borrowed an electrolytic, tagged across the input to the VReg. Instant cure. NEVER trust suppliers have a clue! They might but just as easily they may not. And this particular device, was developed and built a few miles from the Sutton Coldfield transmitter, and the suscebtibilty was, yes you guessed it around th 88MHz region!

[Al (astral highway)]

G6fylneil, interesting to note this history of intrinsic instability. It seems my experience has many footsteps in the snow, and in high spec. equipement.

Duncan, ahah , I do suspect the electrolytics now. The they are pennies-cheap, unbranded SMD alu electrolytics . Your post prompts me to replace them.

Not to go OT but as I look around, there are no branded devices on board at all. The IC’s have had their identities weakly concealed by transposing some of their alphanumerics.

[Cruisin Marine]

A circuit has 3 x voltage regulators, D-PAK versions of 7824,
7809, 7805

Problems began in the cct which I have traced back to the power supply and there we are…

There’s a nice voltage overhead of 31V after the bridge rectifier.

But suddenly we have 22.6V, 7.7V and 3.9V. This is off load. There’s just a bunch of flip-flops and comparators and idle gate drivers with no drive signal, no current to draw here.

7805 has IP voltage range of 7-25 volts.
7809 12-35 volts
7824 ?
Are the D-Pak versions different?
This is from an old Maplin Catalogue.

[Al (astral highway)]

Originally Posted by Cruisin Marine

7805 has IP voltage range of 7-25 volts.
7809 12-35 volts
7824 ?
.

True. They are in series, so one output voltage is the next input voltage.

The hardest life on board in this case is the 9V reg. which supplies current to fast pulse switches and experiences transient spikes from the parasitic inductance of the board traces in return. It also has to drop 15V from the one upstream.

The 5V one has an easy life. The 24V reg has an unremarkable one.

There’s no difference from the style of case, same insides.

[Cruisin Marine]

Is there a possible problem with the common/earth or 0v return path perhaps?

[Phil__G]

How much current are you expecting from the 7805? with 26v across it you'll need a serious heatsink to draw more than a few mA - 2.6 watts at 100mA

[jamesperrett]

Originally Posted by Phil__G

How much current are you expecting from the 7805? with 26v across it you'll need a serious heatsink to draw more than a few mA - 2.6 watts at 100mA

If I understand the previous posts, the 5V regulator is being fed from the 9V regulator's output which, in turn, is being fed from the 24V regulator.

[Al (astral highway)]

Originally Posted by Phil__G

How much current are you expecting from the 7805? with 26v across it…

Please see post #14.

If you’re interested in the topic , please can I ask you to at least skim through the thread, otherwise it will be derailed with popcorn- style musings with no bearing on the set-up of the circuit or origin of the problem, and perhaps skipping the experience that’s been contributed.

Much appreciated

James, that’s right.

[McMurdo]

some designs rely on downstream loading effects for the regulators to remain stable. It won't hurt to add an input and output capacitor across the pins as per the datasheets if they dont already have them nearby.

[cmjones01]

Regarding alternatives for the regulators, I find it hard to know what to recommend in this application. Generalizing somewhat: these days linear regulators are only used for small currents (<100mA) or small voltage drops (5V -> 3.3V being a popular one), because efficiency is everything. This is especially relevant when the components are all themselves tiny and packed in to a small space with no heatsinking. For larger powers, regulators are almost always of the switch-mode type. To give an example, attached is a photo of a buck regulator on a board I'm working on at the moment. It takes a nominally 12V input and regulates it down to 5V at up to 5A. It's based on the TI TPS565208 chip (which has since become hard to find, but that's another story...). No heatsink, it's never given any trouble, and you can see its size relative to my thumbnail. Just imagine what a linear regulator capable of dropping 12V to 5V at 5A would look like, with a heatsink to dissipate 35W.

The 78xx regulators might well be the right choice for your application. It's just that I've forgotten how to use them!

Chris