A question for you LM324 experts on failure problems.

How long are the lines/traces/wires from power supply to the rail near those LM324? How much in local bypass capacitance? Turning on a system, can cause some ringing there, easy to check with a DSO. Whack the unit, do some wild power cycling, and so on.

Running them at 31.2V total while abs max is 32V is like Russian roulette. Did you consider replacing them with an opamp that can take more? There are opamp with abs max 44V and such. The MC33174 comes to mind, not sure if it'll fit the bill in your case:

An old Motorola design, I have used them a lot in the last century.

This is the manual which has the schematic towards the end.

This is old stuff but it works very well and alot of it in service in various parts of our multi-plants. This problem is starting to get more announced now that these units are getting older. So after closer inspection I think what is happening is that maybe the supply caps are weak and the load shift may be causing some extra higher voltages to reach the less loaded side of the rail when instability takes place.

P.S. Even though that schematic states 15V +/-, I get almost 16 volts with the DMM.

Jamie

Well I ain't the brightest bulb around here :-) but for old gear my rule #1 is replace all the electro's, seems to sort a lot of strange happenings.

Cheers ....... Rheilly

The supplies are unregulated. They could spike really high in some circumstances.

John

and

his

Not to mention that power line voltages have been creeping upwards over the last several years, in some parts of the US at least. Hard to find concrete documentation of this phenomenon, but anecdotes abound.

-- john

On a sunny day (Tue, 25 Oct 2011 09:06:17 +0800) it happened Rheilly Phoull wrote in :

Right, replace by tantalums ;-) LOL No seriously ;-) But I think Joerg is right, out of spec for the supply voltage and needs a different chip too.

That's a really dodgy circuit with unregulated supply rails right up at the abs max supply voltage. Best to just redesign it, or at least sub a higher voltage quad if you're not selling many of them.

Best regards, Spehro Pefhany

Unregulated 15V assuming stable mains ... oh-oh. That does require a high voltage amp and also some voltage clamping. Clamping might not be too easy because TVS have large tolerance. Also, the dancer remote connection at pin 10 of the LM324 is totally unprotected against EMI and ESD. So is the gate of Q1, or 1Q as they call it. Looks like this board was designed by a rookie :-)

There is a chance that a lot of these units are now out there and still work, sort of, but where the LM324 chips in there have bruises and aren't 100% anymore.

Sometimes I had the impression this was done because it increases revenue. I had to complain hard when light bulbs started blowing. It had crept up to 127V early in the night. So finally they lowered it to around 121.5V.

Trivial, actually. Barely more than trival if you actually want results over time - any facility that keeps logs would have that, but probably few have it posted for public consumption. 121.1-120.9 right here today

- whoops, down to 119.6 now, something must be drawing a bit of power around here. 120.6-120.7 yesterday. 237-242 on my well pump circuit in another state, typically. So long as you don't require all the bells and whistles (and chart output and datalogging), power line monitors can be quite affordable. If I bothered to use their software, I think I could even get logging from some of the UPSes I use, but I don't bother to since I don't feel a need for it. Likewise, most of the fancier Fluke meters will datalog...

Jamie schrieb:

Hello,

there is a contradicition in this manual. The mains voltage shall be 120 V +- 10 %, but the input of the transformer shall be 115 V. With +- 10 % variation, the input voltage is between 108 and 132 V. If the 10 V secondary are measured with 115 V, 132 V will result in 11.48 V. But transients and overvoltages can cause more than 11.5 V.

Bye

"Ecnerwal" wrote in message news: snipped-for-privacy@news.eternal-september.org...

Our APC UPS logs give a nice running line voltage measurement. Summer time usage is interesting.

Cheers

As others have pointed out, the design stinks. It can't protect itself from the "outside world". Sounds like some problems have cropped up recently. If that's true, and the circuit, as poor as it is, has been reliable over many years, you need to look at the outside world for the source of the difficulty. Recent electrical work in your plant could be the culprit. For example, a bad neutral connection could raise the voltage on one side of 240 when a large load is placed on the other side.

Meanwhile, how hard would it be to add a couple of series diodes to the +15 bus, and a couple to the -15 bus? Maybe you could do it while replacing the ~30 year old 470 uF filter caps. Not the best design upgrade, but it's cheap and may be easy to do.

Ed

Or maybe a 7812 and 7912 three terminal regulators, even the L or M variety. It should work fine at 12 volt rails as it only goes to 10 volts on the output.

Some diodes to the rails on the input would provide some transient protection.

I agree, the caps should be replaced considering their age.

tm

and

his

ith

Adding regulators is the kosher thing, but dropping in LF444A's gets you 44 volts supply tolerance =3D 12 volts (total) extra margin.

They're pretty well the LM324s of BiFET op amps, so that matches too. Pin compatible. Quirky outputs--caveat emptor.

LF147's too. No kinks IIRC. Nice parts.

LT1639 or LT1014, if you're flush.

-- Cheers, James Arthur

I have pretty much concluded how to destroy a 324 much in the same way as happen recently. Line voltage being high that day may have contributed to it how ever, this unit was part of 20 in a cabinet using the same 120 Volt supply and they are all still working. This is the one that when the GAIN pot was adjusted is when the output of the last stage shorted on one side there by causing the (-) rail to over come the (+) output when it was suppose to be in the (+) swing. So I only got 0 to

-10 output and it was working backwards on top of that.

This is what I found, the day of destruction, the DC rails were measuring +- 15.8V, the next day when I really got into it, the DC rails were 14.7 Volts. Maybe the critical DC rail voltage is making the op-amp shaky with the following of what I did.

In any case, I found that if I bias the circuit to force the final output Op-amp to sit at 0 volts there seems to be a Hi-Z on the output. I know that lack of shoot through current on the output does make for a poor op-amp to be used in audio designs and such how ever, I was able to actually destroy the 324 output op-amp merely by having a low current, a vary low current induce AC signal placed there of around 18 volts or so.

I am thinking that maybe the drive this unit was connected to may have had a small linkage that really does not amount to anything, hitting a peak when the output op-amp swung through the cross over at the time I did this. You wouldn't think adjusting the gain on this circuit while in operation would cause such an upset in the circuit but it does and it appears that if something stray, even at a very low current sitting at the output during this cross over period could cause a problem..

I found that using a TVS diode or simply placing a 5k ohm R on the output to common fixes the problem. I was not able to destroy the op-amp with this type of problem.

Jamie

nd

the

it

is

ecs

ake

p?

ng

ll.

e,

ps

..

ith

If I understood that, the pull-up output transistor is fried.

he

Yep. Fried.

I don't understand that sentence. If you mean you can burn out an LM324 output with an 18 volt overload, okay, that's plausible.

Not clear to me why that would help.

-- Cheers, James Arthur