0

I have a few switching dc-to-dc power supplies (36V to 3.5V) which gets destroyed when I connect the positive input line before the ground.

When I connect the positive line first, some very quick connections-disconnections of the ground occur spikes at the input which destroy the power supply. But, when the ground is connected first and positive line afterwards everything is fine. Even when I try to connect-disconnect the positive line quickly the power supply works just fine as long as its ground is permanent connected.

The load current is irrelevant to the effect.

That problem has been solved by adding a TVS diode in the input of power supply but I still can't understand what causing that effect.

How is that possible? Why don't I see spikes when the ground is connected first?

MrBit
  • 1,973
  • 4
  • 27
  • 54
  • Can you post a part number or link to these power supplies? – evildemonic Mar 19 '19 at 20:18
  • Schematic, manufacturer and part number, schematic, input and output ground relationships, schematic, external source and load, schematic, isolated/non-isolated, schematic - anything? – AnalogKid Mar 19 '19 at 20:30
  • I've seen that effect in other cases too, like in development boards. For example in an arduino due which is powered with this http://www.ti.com/lit/ug/slvu371/slvu371.pdf power supply. Arduino communicates with an RFID card reader through a serial port pin (wire about 1m). at the end of the wire I see high spikes (when the positive line is connected to power supply and ground is connected-disconnected quickly) and that destroys the card reader too. they are two different cases two different boards and I'm trying to find out if there's an explanation behind this. – MrBit Mar 19 '19 at 20:44
  • @MrBit: note that you have ignored the comments requesting part numbers and schematics and provided information about a different problem. Please read the comments carefully and try to provide all the relevant information in your question. – Transistor Mar 19 '19 at 22:36
  • I found an interesting answer in a relative question https://electronics.stackexchange.com/a/340211/51770 – MrBit Mar 20 '19 at 05:40
  • 1
    Somewhat similar to my answer, it seems. – Garrett Mace Mar 22 '19 at 03:38

1 Answers1

3

There are, by necessity, other wires and ground paths that you are not describing in your question. If everything were isolated to just the two power input wires, it would not matter what order you connect them.

However, since you are experiencing these problems, there is some other path to ground through communication lines, enable pins, or the ground connection on the output side (if this is a non-isolated DC-DC converter).

Remember that AC-powered devices are referenced to earth ground. For example, if you have a power supply plugged into a wall outlet and a computer also plugged into a wall outlet, there is a ground path through the wall or (quite common) a voltage offset relative to earth ground between the two systems. Some power supplies output voltage that is actually negative in relation to earth ground; if you connected a circuit powered by one of these to a computer's USB port, large voltages may be applied to the USB supply rails or communication pins.

By not connecting the ground first, you do not have control over the voltages applied to other pins and cannot ensure they remain within specification. In systems that require hot-swapping, your solution with the protection diode is an acceptable method to clamp voltages to within specifications during the fraction of a second the connection is being made. You may want to attach similar diodes to any other control or communication pins that are involved.

In summary, ground voltages of two systems may differ significantly, or even be oscillating quickly relative to each other. Additionally, pins other than power and ground may attempt to carry power if they are the only current path available. Either control the connection sequence, or protect individual pins if connection sequencing is not possible.

Garrett Mace
  • 146
  • 3