Zener diode behavior

[Ian - G4JQT]

A couple of years ago I built some equipment that needs to operate 24/7. It worked continuously until recently when I was called in to fix it.

When I cleared the fault, I checked the voltages against those I wrote down in the manual which I measured when I installed it.

The 35-volt line was now about 40 volts. This turned out to be due to a 5-watt 36-volt zener apparently having crept up to about 41 volts.

I have never come across any data suggesting that a zener voltage may rise with use. Has anyone else experienced this?

Fortunately it isn’t a problem yet, but will it continue to rise? The zener does not run hot, more like “very warm” - warm enough to easily touch.

Thanks,

Ian

[Ed_Dinning]

Hi Ian all zeners have a temperature coefficient of voltage . This is positive above about 6 volts and negative below that voltage; string the right types togetgher and you can get a zore TC (you can also add forward biased Si or Ge diodes to the string as well.
There are also different production processes for Zeners which give different TC's. Also be aware that the zener voltage has a tolerance and will also vary with the forward current through it.
If you have one of the Mullard industrial data books from the 70's/80's these show all the possible variations.

Best regards, Ed

[XTC]

I've not seen a zener behave like that, but I can well imagine that running one beyond its spec could permanently change its properties. Maybe it was over run as as a consequence of the fault? Has the current limiting resistor changed value? As Ed says, the zener voltage is slightly current dependent.

Considering the cost of a new zener, I'd be inclined to treat this as a dodgy component and swap it for a new one of a good make.

Pete.

[Brian R Pateman]

You don't say what the original fault was. Could it have caused the zener to have been subject to excess current for a period? (A leaky capacitor for example).

I've found zeners to be pretty reliable but if they are in reliability critical circuits I tend to over-rate them by a factor of 40 - 50% in terms of forward current. My line here would be to change it as Pete suggests and to use the next rating up in terms of current.

Regards,

[Ian - G4JQT]

Thanks all for the replies.

Yes, considering the price of a new zener, a replacement would now be prudent!

The fault was a dry joint (oh, the shame of it!!) which developed between the bridge and the large smoothing cap prior to the zener. The zener's current limiting resistor only runs slightly warm. The zener feeds the base of a BD131 which feeds a 2N3055 for a stablised output. So in this case the fault just took the volts away from the zener.

I was looking for an IC regulator to do all this at the design stage, but couldn't find one that could comfortably supply a peak of 3A at 36 volts. Those that did would be running near their maximum rating, which for permanent use was pushing things a bit.

The dry joint was an absolute classic. When I'd finished the unit I checked everything over carefully with a magnifying glass, gave the hole unit plenty of "bashing tests" and it passed those every time. But as the apparently poor joint is at a heavy current point, and presumably got warm, then as it got worse, even warmer, until it eventually gave out. It appeared as a very obvious grey solder joint that wasn't that bad when I finished it originally.

This required an expensive round trip to swap the unit over.

I have been warned...

Ian

[Ed_Dinning]

Hi Ian, have a look at some of the older app notes from Nat Semi on the LM317 and similar devices. These are good to 36 volts and it is possible to couple them to series pass devices to increase the current ratings.

Ed

[XTC]

I read somewhere that a good rule is that the zener's current should be about 10% of the maximum derived from the wattage

You could string a 24V 5W and 12V 5W zener or three 12V 5W zeners together in series. Not pretty, but it works and the power dissipation per device would be far less than for a single 5W. I believe that when you go for a 36V zener over 5W they are TO220 or stud packages, and that could cause problems with mounting.

Pete.

[Ian - G4JQT]

Thanks Ed and Pete.

Max zener current would be 139mA for 5W at 36V. It's running pretty warm now with about 20mA, so 139mA would suggest to me the zener would get pretty hot, so Pete's 10% rule seems a good idea.

I've got about 20mA through the zener, so that's about 0.72 watts, still well under 5W, but a bit more than Pete's 10% suggestion of 13.9mA - if that’s what you meant Pete?

Putting some zeners in series to make up the voltage would also be a good idea. I did consider this, but when I did the calculations 0.72 watts seemed well within max rating even when run continuously. It was running all last summer in a concrete-built storage area and ambient temp was probably above 35C from time to time. This may not have helped…

Ed’s LM317 suggestion was a chip I looked at, but I couldn’t quite make out from the application sheets exactly how to add just one bypass transistor (the LM317 current rating is only slightly under what I needed), as it seemed to show a design for much more current than I needed, and the max voltage was still only 36v (I wanted a bit of extra if needed it), without adding other diodes, transistors, etc. By the time I’d done all that I could have done it “the old fashioned way” with the same or fewer components – which I did!

Thanks again.

Ian

[XTC]

Running a zener at 10% of the maximum current allowed by the wattage, is a rule of thumb, which operates the device in a sensible region of the zener voltage/current curve and keeps the power dissipation at a really conservative value. It is only a rule of thumb and I wouldn't have thought that 0.72W and 20mA was pushing the zener to anywhere near its limits.

The important thing is the junction temperature. 5W gives an acceptable junction temperature at the conditions the maker specifies, usually 25C ambient. If the device is run at higher temperatures and especially, with short leads, it has to be derated, so the 5W is also something of a rule of thumb, or at least, only applies in specific conditions.

Zeners are mass-produced, and it's quite possible that this one was a manufacturing rogue.

Then there's the question of the temperature inside the cabinet. If the temperature in the building was 35C occasionally, the temperature in the cabinet could have risen much higher. If the temperature is getting very high, it will shorten the life of all the components including electrolytic caps.

Lead mounted devices dissipate much of their heat through their leads and lead length starts to be important as the device is pushed towards its limits. It sounds as if visiting this site involves a lot of travelling, and if the equipment is running 24/7, may involve visiting out of hours. My solution would be to assume this wasn't a manufacturing fault in the failed zener and replace it with two or three 5W zeners, leaving longish leads to help the heat dissipation if possible. You could also use a higher wattage zener, but that could involve problems with mounting and heat sinking.

HTH

Pete.