78L05 placed in reverse?

[RoGeorge]

I've thrown away some dead batteries yesterday, and found in the electric waste dumpster bin a defective ambient light lamp with movement sensor, the ones you plug into a wall power outlet, and the lamp automatically turns on a few LEDs when somebody walk through the room, and only when it's dark in the room.  Took it home for parts, in the hope that the PIR sensor will still be good to play with in some other project.

The surprise was when I opened the lamp.  The design is faulty, and apparently the lamp shouldn't work at all, in the first place.  I don't get it.

Inside there is a 78L05 in reverse, fed with voltage at its output pin.  It is not an assembly error, the footprint layer drawn on the PCB is also wrong, and the 78L05 was soldered according to the footprint draw.

I desoldered the 78L05 and placed it correctly.  It was not dead, just that in reverse it outputs some 3.4V instead of 5V (tested this outside the board, with a power source limited at 10 mA).  Now the +5V is OK, but the lamp still didn't work.

While checking the rest of the parts, the schematic (raised from the PCB) looks like nonsense.  I'll double check this afternoon, but still, the 78L05 was in reverse for sure, no doubt about the 220Vac to 5Vdc part of the schematic.

My question is, how did this schematic worked, in the first place?

Is this some 78L05 trick I didn't know, or is it just a wrong design?  Maybe a fault clone of some other working lamp?  Didn't they noticed in production that none of the lamps were working?  Even if they didn't test it, how did they manage to sell something that never worked?

 

[T3sl4co1l]

Drops a few volts, AFAIK.

It's zener regulated in the first place and the opamp is probably good for 3-30V (LM358??), it doesn't really care.

More surprising actually that they bothered with a 78xx at all.

Tim

[MagicSmoker]

I can't imagine anything useful from connecting a 78xx regulator backwards. In fact, the datasheets recommend an inverse bypass diode from output to input if there is significant capacitance hanging off the output terminal to protect against reverse biasing the emitter-collector junction of the internal pass transistor.

But for more fun 78xx regulator stuff check out this classic web page: http://sm0vpo.altervista.org/info/regulator.htm

[magic]

10V zener across the transformer and then a series resistor and capacitor? Clearly some garage entrepreneur designed that thing

If you look at the schematic on that site above you will see at least two obvious paths from the output to the input:
1. through forward biased BC junctions of the current limiter transistor and the output pre-driver, plus that series resistor before the former.
2. through avalanched EB junction of the output driver.

For very low currents 1 is used an nothing bad happens at all, at higher currents 2 happens and damage is possible if the transistor overheats. Perhaps some loss of performance too if it doesn't melt outright, I don't remember what the effects of EB breakdown were supposed to be.

[duak]

I remember being caught up in the 70's with the same basic regulator (78xx or 79xx) pinouts notated differently between the manufacturers (Fairchild, National and Motorola).  I can't remember if it was an error or something to do with corporate naming conventions or if it was related to the packaging.  I don't think I had got to the PCB stage yet so I was able to sort it out in the prototype.

[T3sl4co1l]

10V zener across the transformer and then a series resistor and capacitor? Clearly some garage entrepreneur designed that thing

Well, it's alright, it's pulsating DC at that point.  I get what they were going for, it's maybe not how I'd do it, but it's not costing a lot, whether extra, or in general. 

If you look at the schematic on that site above you will see at least two obvious paths from the output to the input:
1. through forward biased BC junctions of the current limiter transistor and the output pre-driver, plus that series resistor before the former.
2. through avalanched EB junction of the output driver.

For very low currents 1 is used an nothing bad happens at all, at higher currents 2 happens and damage is possible if the transistor overheats. Perhaps some loss of performance too if it doesn't melt outright, I don't remember what the effects of EB breakdown were supposed to be.

My experience has been, at modest currents at least (1-100mA?) it looks like about two diode drops.  It goes through internal structures, so it's not going to handle full current, let alone surge current -- hence the back diode as traditionally recommended [when the input can be pulled down before the output has discharged].

Tim

[RoGeorge]

I am now convinced it is no smart trick there, only plain stupid, and it never worked.

As a proof, simply disregarding the whole schematic mess by shorting the CE of the last PNP transistor that controls the 3 series LEDs, the whole thing can barely light up.  Measured current through LEDs is only 0.5mA.

Also, all the components are good.   

[SL4P]

Doesn’t  everyone KNOW that a regulator installed backwards, becomes an UNREGULATOR. 

[RoGeorge]

A regulator backwards will become a rotaluger.   

[T3sl4co1l]

Pfft, everyone knows a regulator becomes a regusooner.

Tim

[VK5RC]

Doesn't the regusooner arrive before the regulator then the regureallylator?

[magic]

It could actually be a fun party trick to run some common regulators through a curve tracer

[MagicSmoker]

Doesn't the regusooner arrive before the regulator then the regureallylator?

I think it regulates its input voltage, instead of its output... 

[SL4P]

Installed backwards... that’s called REVERSE ENGINEERING

[Bud]

A regulator backwards will become a rotaluger.   

Quite a usable name though 

[Medved]

10V zener across the transformer and then a series resistor and capacitor? Clearly some garage entrepreneur designed that thing

That Zener is maybe the only thing here that is actually not wrong. It is not supplied from any transformer, but from (nearly ideal) current source formed by the input series capacitor.
Removing the Zener would mean the output voltage would rise up to virtually 320V, but in reality to a level where something blows up.

What is wrong with it?
Now assume the thing worked when it was new, regardless how crazy it is connected. With these things supplied via such capacitive dropper circuit there use to be two common problems:
1) The input ballasting capacitor (the few 100's nF/275VAC in series with the line input) loses capacitance. For safety reasons it must be of a "self recovery" type, but the "selfrecovery" actually means it by itself just clears the short circuit once its dielectric breaks down, it uses to do it by either evaporating part of the conductive layer around the damaged spot (mainly older "metallized paper" designs), or by splitting the conductive layer into many sections and each connected via a thin trace acting as a fuse, which then disconnects the damaged section. Either way, as the capacitors connected like that do suffer dielectric breakdown events (because of teh various mains spikes and generally long time exposure to rather high voltage for the dielectric material thickness), the consequence is the capacitor as whole gradually looses its capacitance. And that means the current feeding the circuit gradually decreases, till it goes below a level where the circuit is still able to work. Typical consequence is erratic operation (after it "turns ON" the first time, it does not want to turn OFF anymore) and very low LED brightness or when the thing uses relay to switch the higher power load, the relay gets insufficiently driven, starts to rattle and so usually fails very soon after.

2) Second problem uses to be dried out filter capacitor behind the rectifier. Part of the reason is the usual proximity of that Zener which dissipates quite some power so generates heat, part because of the permanent ripple current load. Consequence when the capacitor is not really good quality, it tends to fail (the rubber seal rots out, allowing the electrolyte to loose water). As a consequence, it becomes unable to maintain the supply ripple within acceptable limits and so the circuit fails. Usually into a "motion permanently detected"

[Medved]

I think the design originally asked for the 78L05 to be the normal way, just for some error it got reversed. The error could come either from the thing to be originally designed with some other similar regulator with different pinout, which then was supposed to be replaced by the 78L05, but someone did not checked the correct orientation. Or the information about the need to assemble it "against the silkscreen" got lost on the manufacturing floor over time, as they were cheap to update the silkscreen on the PCB.

[Brumby]

Doesn’t  everyone KNOW that a regulator installed backwards, becomes an UNREGULATOR. 

It was my impression that they became a DEregulator.

Apparently they have been utilised in some government applications - with very mixed results.

[S. Petrukhin]

Most likely it was a 7805 made upside down. This was known at the production.