Flyback Diodes and the 1N4148

I keep a large inventory of 1N4148 diodes. These are small signal diodes that every ham builder is undoubtedly familiar with. I use many of them in my station, almost exclusively for antenna switching matrices. They have seen greatest use in my 80 meter 3-element vertical yagi. The 1N4148 diode is cheap in small quantities and it is easy to find in electronics stores and online.

They are not always used appropriately. To some, one diode is pretty much like another diode, other than specialty devices like Zener diodes and PIN diodes. Commercial products are full of them, whether appropriate to the application or not. The 1N4148 will do the job in most cases and most people never experience the events that demonstrate otherwise.

Other than switching, perhaps the most common application of the 1N4148 is a flyback diode (pic reference) on a relay coil. (Flyback diode seems the generally used term but there are others: surge suppression, reverse EMF protection, etc.) When a coil is turned off, the magnetic field will discharge in the reverse direction of the applied current. The flyback diode dissipates or shunts the energy to ground to protect sensitive electronics driving the relay coil. Other inductive loads, like motors, require similar measures.

Calculating the required ratings for the diode is not difficult, although I've never felt the need. For the small relays used in my many antenna switches a 1N4148 is seemingly adequate. Unfortunately the calculation can be very misleading when used for antenna switches in our stations. There are other considerations than its main function. 

I previously ranted against the 1N4148 while I was repairing a Hamplus 2×8 antenna switch a few years ago. I ranted again when a lightning strike wiped out the reed relays in the Beverage switch. The reed relays have which integrated flyback diodes, which is convenient. The datasheet for those reed relays says nothing about the diode ratings. Based on how the requirements are typically calculated, the integrated flyback diode is likely no better than a 1N4148.

The reason I am returning to the subject is due to the continuing fallout from my recent lightning strike. As time went by more problems appeared in the station. Semiconductor junctions don't always fail immediately, continuing to function while weakened until failing later. This occurred with the antenna switch when one morning I could not select the 20 meter stack. 

Cleaning contacts and checking the control wires was no help. I narrowed down the fault when I checked the control line voltages and currents while the stack was selected on the station's left radio.

The antenna switch was brought indoors. The coax and control cables were wrapped in plastic to protect them from the weather. An earwig colony that infested the weather cover over the switch was disrupted in the process. They will surely find another home!

Testing in the shack confirmed my guess that the flyback 1N4148 for the misbehaving switch position was shorted. Two other diodes had measurable reverse current that promised imminent failure. I heated up the soldering iron and removed all 15 of the 1N4148 flyback diodes. The 1N4007 that replaced the original failed 1N4148 tested good so I left it alone.

As you can see from the picture I was not gentle with their removal! I should have done this the first time, but it's a lot of fussy PCB work that I was eager to avoid. Although they are easy to remove, installing new diodes in PCB holes plugged with solder is difficult to do while being careful to avoid damaging the diodes and PCB traces.

The job is done and the switch has passed initial testing. Once I complete testing it will be re-installed and I can get back on HF. For the time being I am only able to operate 6 meters, and that's okay since the "magic band" has been pretty good lately.

So what have I actually accomplished? Presumably the diode replacement is questionable since the 1N4001 and 1N4148 are, per the flyback diode calculation, equally adequate with respect to their current and voltage ratings. Yet I here I am, again insisting that the 1N4148 is a poor choice.

From the Vishay datasheet we see that the PIV rating is about the same as for the 1N4001. It is of course 1000 volts for the 1N4007. While repairing the antenna switch I ran out of 1N4007 diodes and when ahead and used several 1N4001 in the antenna switch without a qualm. 

My concern is the survival of the diode and not only its ratings as a flyback diode. Of course, no diode will survive or protect against a severe lightning strike, but most strikes are secondary or inductive and many semiconductors will survive if they can withstand the surge current and heating. Not only can a more robust flyback diode survive the abuse, they can protect downstream devices if the surge can flow through the diode to ground. 

That is where the 1N4000 series of diodes is superior to the 1N4148. The surge current for all of them is 30 A, or 15 times that of the 1N4148. The power they can shunt and dissipate is much greater and that can make all the difference. The 1000 PIV rating of the 1N4007 makes it more likely to survive than a 100 PIV device, however that is often not the major factor. Hence my willingness to use the 1N4001 rather than wait and order more of the former.

A device such as a MOV is better for protecting control lines, but every little bit helps. A diode that survives has a chance to also protect. MOVs have other issues with respect to lightning protection which I won't delve into, especially since my knowledge of them is limited. In many applications they should prove adequate if the power supply and switching electronics can withstand having control lines shorted by a MOV reacting to the high voltage of a lightning strike. Failed flyback diodes usually burn open or short, usually the latter in my personal experience.

Power supplies need protection as well for the same reason: inductive loads. In the schematic extract we see a protection diode across the output of the small Pyramid linear power supply that was damaged in a previous lightning strike. It's a low voltage and high current 1N5402. It survived the strike.

Going through the schematic of the Astron RS35 power supply damaged in the recent strike there is no equivalent diode across the output terminals. There is an SCR used in the crowbar circuit but I doubt it provides any protection from reverse EMF. It also appears to have survived the strike.

I have heard that the Pyramid power supplies are better engineered than those by Astron and this is evidence supporting that view. After all, a 35 A power supply can be employed with a variety of inductive loads, and those loads may not have diodes.

The digression about power supplies is relevant. It brings us back to a question about how to best deploy flyback diodes in the antenna switching systems in use at my station. Do they add value? Is just one enough and can the one in the power supply do the job on its own? Why is there one on every relay? 

I have flyback diodes on all my home brew antenna switch relays. This includes stack switches, Beverage switch, 80 meter yagi mode and direction switches, and there will be more. All use high side switches. The Hamplus antenna switch has a flyback diode on each control line (low side switching) but not on every relay: there are 16 lines and 24 relays.

The schematic depicts a circuit where every relay coil has its own flyback diode. An alternative is shown on the right that we'll discuss shortly. There is only one switch (box with an 'X') per relay coil, either high side -- switches the +12 VDC, with the other side grounded -- or low side -- switches the grounding, with other side connected to +12 VDC. The Hamplus uses low side switches, with each control line running back to an operator controlled switch.

When the (high or low side) switch opens, the relay coil and diode are isolated from the circuit upstream of the switch. The magnetic field collapses and the energy, with nowhere to go, is dissipated in the flyback diode and coil. Without the diode the field would collapse more slowly, but in both cases it happens fast.

Why have a flyback diode on every coil? Could just one suitably selected diode do the job? It could even be the one in the power supply. I toyed with this approach when I got frustrated with the time and effort to de-solder and re-solder the 16 flyback diodes in the Hamplus switch.

First, with a remote diode, such as in the Pyramid power supply, there is a propagation delay measured in microseconds. That isn't a lot but entails a slightly elevated risk to local electronics due to the distance between the switch and flyback diode. The second and more important point is to understand what kind of circuit we're protecting. The power supply is (or should be) already protected, so it's the low/high side switch we should be concerned about.

If the switch is a relay or mechanical switch there is little to protect. The contact will "spark" a minuscule amount, however for many relays that can increase the lifetime of the relay by, what is sometimes called, wetting the contacts.

If the switch is solid state there is risk. With just one flyback diode (option at the right of the schematic) in the equipment or power supply the semiconductors, typically transistors, in each switching circuit will not be well protected. It is better to use a flyback diode on every control line which energizes one or more relays. That is the case for the Hamplus switch with its 16 diodes and 24 relays.

I have flyback diodes on every relay coil or commonly switched groups of relays. All the diodes are 1N4001 or 1N4007 so that they are better able to withstand lightning surges than the 1N4148. I probably should have replaced all 16 in the Hamplus when I first repaired it. 

I have altered the design for my station automation system to use Arduino GPIO controlled relays, not Darlington transistors, for antenna switches that are exposed to lightning surges. Flyback diodes on the relays coils is inexpensive and protects solid state switches for the control lines. For relay switching of the control lines, having the diodes present is no inconvenience.