Caps needed across LM7805?

I just don't use LDO where at all avoidable. And they usually are avoidable, meaning no tunnel of death :-)

LDOs have other pathologies which often aren't mentioned in the datasheet or possibly the chip guys didn't even know about them. My worst one was a LM29-something. Client insisted on keeping it against my advice. Phssst ... *BANG*. Hmm, we had babied the ESR so what the hell ...? Call into the mfg. The engineer there grew concerned and suggested a phone meet with the chip guys present. One could here some shuffling there in their room. More and more people came in, papers were perused, lots of mumbling in the background. Suddenly one guy let of an "Oh drat!" Turns out this LDO didn't "like it" if the source impedance got too high and we were feeding it from a week isolated path.

No imagine scenario #2: Joe Digital sees there's a nice 3.3V rail and happily sprinkles the usual 0.1uF caps here and there plus a big fat

10uF ceramic just to make sure. This brings the combined ESR down way below the minimum ... phssst ... *BANG*.
--
Regards, Joerg

http://www.analogconsultants.com/
 Click to see the full signature
Reply to
Joerg
Loading thread data ...

As a rule with the 7085 we would always just use a normal 100uF NON- ESR electrolytic near the regulator. We found that when driving long lines the 78xx series could oscillate wildly if one used the normal datasheet 100nF. Then is just became a habit to use the 100uF. Never had trouble doing that.

Rocky

Reply to
Rocky

Thank you both for the explanations and references. I understand about half. I will have to take another stab at it later.

Alan Nishioka

Reply to
Alan Nishioka

In message , kittykgl writes

Nonononono. 78xx regulators can and will oscillate if these caps aren't present.

OK, a story...

Many years ago (20 or more) I got the job of repairing and testing customer returned linear voltage droppers (24v to 12V DC for use in trucks) which used a 7812 driving the base of one or more 2N3055 like transistors (various types, BD, 2SC, 2N, MJ etc...) depending on the 'rated' output current.

A good proportion appeared to work perfectly well measuring the output with an old AVO meter and a fairly heavy dummy load, the ones that didn't were shorted pass transistors or other silly faults.

One unit had a label attached saying FM radio goes off when connected. I thought the silly trucker had shorted his accessory feed or disconnected it fitting the dropper but no. I tested it with the meter and my resistor bank load and it appeared to perform fine but I turned on my workshop radio, sure enough I couldn't hear *anything* apart from static when this unit was turned on.

None of these devices had the recommended capacitors or any kind of noise filtering on the input/output of the 7812 and they were oscillating wildly with lots of RF being generated (and possibly amplified by the pass transistors) as my 'scope proved. Fitting the datasheet recommended caps on the input and output killed the oscillation stone dead.

Measuring with and old AVO indicated no problem, measuring with a Fluke meter (8022B IIRC) gave me a different voltage depending on where about on the output wire I probed, I can only assume I was finding sub wavelength points.

I retrofitted caps to all the fifty or so in the batch and then to the several hundred the importer had.

--
Clint Sharp
Reply to
Clint Sharp

It may not have even been the regulators. The emitter follower configuration is also prone to oscillation if the load has substantial capacitance (Colpitts), so you have several chances for trouble there. I don't think you could make that old slug 2N3055 oscillate at FM-radio frequencies with a sledgehammer (it could barely do HF audio), but some of the old 2SC power parts have pretty high ft.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

.

The RCA ones were fine. It was the Motorola ones that gave real problems. They had a much better HF response and would readily oscillate at RF frequencies. This was an issue in some electronic iginition systems we built. Rocky

Reply to
Rocky

The 2N3055 specifications are not very demanding, so there have been rumors, that many manufacturers relabeled chips as 2N3055 if it was out of specs for the intended high quality/high price part.

I have also heard some decades ago the story that some out of specs VHF power transistor was relabeled as 2N3055, but this caused a lot of stability problems in a circuit layout not suitable for VHF power transistors :-).

Of course, this might be an urban legend.

Paul

Reply to
Paul Keinanen

The Motorola ones (MC7805, not LM7805) have weasel words in the datasheet that should alert the careful (ie. paranoid) designer.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

I meant the 2n3055s Rocky

Reply to
Rocky

In message , Spehro Pefhany writes

My test loads were banks of metal clad 25 watt wire wound resistors mounted in IBM XT PSU cases, switchable for different loads and a DC fan running on the higher current settings. I suspect rather low capacitance.

ISTR reading somewhere that the 78xx parts would oscillate at around

50-55MHz so I suspect that was the source, whether or not the transistors were amplifying it I have no idea and I can't remember what transistors were fitted to that particular dropper as the manufacturer covered them to prevent the collector shorting out.

I didn't remove the covers on many unless I actually needed to replace transistors (I was testing and repairing 30+ a day..) I did see a fairly mixed bag of transistors, I guess the factory was fitting whatever came cheapest so it might have been something with fairly high ft.

Shhh, don't tell that to the people who had amplifiers with them fitted... I used to make good money repairing those too.

--
Clint Sharp
Reply to
Clint Sharp

Well, that clears it up. While 2.2mF is literally the right capacitance, in the USA I rarely see the prefix milli for capacitance

- such a value is usually written as 2,200uF, as is shown on the schematic.

What made me think it was 2.2uF is that: it wouldn't be unusual to see a capacitor of perhaps 2.2uF right at the input of a regulator, especially if the regulator had some distance of wiring from that

2,200uF power supply capacitor. This makes a lower impedance on the regulator's input, which also keeps it from oscillating.

Joerg wote of just this in another post in this thread:

LDO's certainly appear to be more sensitive to source and load impedance than regular "High DropOut" linear regulators, but all linear regulators are high-gain devices, and are likely to oscillate if connected to the "wrong" impedances.

Reply to
Ben Bradley

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.