Aside for techies only: diode current ratings

This will only be of interest to (some) ‘techies’.
In another thread ( click here ) ...

bernardgreen said:

Peak currents are time limited, the thermal mass of the diode determines how much peak current can flow and for how long before the diode overheats to the point of destruction. A bridge rectifier rectifying AC will be dissipating the heat from 4 diodes. When the same bridge rectifier is used with a DC supply only two diodes are dissipating the same amount of heat that 4 would be when the supply is AC. Running on DC the maximum safe continuous current may be less than the maximum safe continuous current for AC. The derating for use on DC depends on the construction and method of heat disipation inside the bridge rectifier

Click to expand...

Yes, I agree with that, but I'm not really sure what/why we are arguing about. As I said, the use of a diode bridge to achieve supply polarity independence for DC equipment is very well established, and I have used in countless times (for currents varying from trivial to very large) over the years/decades. All that matters is to make sure that the diodes can cope (at operating temperature) with the DC currents involved - not a problem with present-day components. Nor did I necessarily mean custom-made 'bridge rectifiers'; on the contrary, I've usually used discrete diodes for this purpose.

Turning to technicalities ... as for the maximum peak current being time-limited, that's true, but that current limit rapidly falls when the over-current is frequently repeated, as obviously is the case with AC applications). If you consider the ubiquitous IN400x diodes, they are rated for a maximum continuous rectified current of 1A, but a 'non-repetitive peak forward current' (one half cycle of 60Hz size wave, approx 8.3mS) of 30A. However, when that half cycle is constantly repeated, the maximum permissible peak current rapidly falls:
As you can see, after just 100 cycles of 60 Hz (less than two seconds), the peak permissible current has fallen from 30A to about 8A (corresponding to ~5.6A RMS), so goodness knows how low it gets with continuous AC.

Kind Regards, John

In almost all diodes peak current events will not alter the diode in the long term provided the events are under the line on that graph. Any number of events can occur provided the overall average current, (averaged over several minutes), is less than the maximum continuous current.

Problems arise when the diode has been carrying it maximum rated continuous currrent for some time and is also close to it maximum junction temperature. A single peak current event will then even though apparently under the line will be enough to take out the diode or alter its junction structure enough to weaken it. Hot and stressed and therefor highly active ions migrate vibrate a bit too much and get across the semi-conductor junction

Some specialist diodes, mainly RF and microwave will suffer a degree parametric change when a peak occurs, designing with these includes eliminating peaks with protective circuitry or knowing how many unavoidable peaks can occur before the diode has to be replaced..

bernardgreen said:

In almost all diodes peak current events will not alter the diode in the long term provided the events are under the line on that graph. Any number of events can occur provided the overall average current, (averaged over several minutes), is less than the maximum continuous current.

Click to expand...

Yes, that's obviously implicit in what the graph is all about. However, as I'm sure you understand, my point was that the graph covers less than 2 seconds of continuous AC, so one wonders how low the peak current would have to be to still be 'below the line' after a few hours of continuous AC. The curve is, admittedly, starting to 'flatten out' by 100 cycles, but it is still very much 'going down' at that point. ... or am I missing something?

Kind Regards, John

There is ( too much ) equipment which is so border line on design that after being switched off it must not be re-powered within so many seconds or minutes of being turned off. On investigation very often this time delay is to allow components hot from "normal" operation to cool down before being subject to a peak current during power up.

bernardgreen said:

There is ( too much ) equipment which is so border line on design that after being switched off it must not be re-powered within so many seconds or minutes of being turned off. On investigation very often this time delay is to allow components hot from "normal" operation to cool down before being subject to a peak current during power up.

Click to expand...

That's obviously unsatisfactory, but presumably is only really an issue with equipment which has an 'inrush current' at power-up. With the LED lamps, fed from a DC supply, we were discussing, I see no reason why there should be any such 'inrush' - so switching back on very quickly would be no different from leaving it on in the first place.

I suppose another related issue is that of equipment with 'crucial' cooling fans which don't have a fan 'over-run' after equipment power-down (and obviously would have no over-run were the supply to fail completely).

Kind Regards, John

most modern flat type LED lamps are `soft start` to prvent inress failure,
the older `tandard type` LED lamps (multiple leds crammed together), these we found failed sooner, and light output dimmed to under a quarter of rated light output in a matter of months.

Oasis

oasistechnical said:

most modern flat type LED lamps are `soft start` to prvent inress failure,

Click to expand...

I presume you're talking about 'inrush current' at power-up. I have to say that, if (as we were) we're talking about a lamp supplied with DC, I can't really see why there should be any such 'inrush' (in contrast with, say, the large inrush current on cold start-up of an incandescent lamp). The power supply providing the DC may well, of course, result in an inrush (AC) current (into the PSU) on power-up, but that's irrelevant to the lamp. Am I missing something?

Kind Regards, John