I have been working to redesign a faulty circuit board.
One of the problems I have identified is the power supply on the board.
The input voltage is nominally 13.8V DC from an AC switchmode powersupply but can drop sharply by about 1.5V when a number of solenoids that are connected to the same voltage activate and draw a large current. When the device is operating off the backup lead acid battery the input voltage becomes 12V.
Currently they are using a small SMD 7805 LVR to produce the 5V supply for a microprocessor and RFID chip. The LVR chip is overheating and heating up the whole circuit board (some parts are too hot to touch) which I am going to fix by using a TO-220 and proper heatsink.
However the change in the supply voltage causes a 100mV change in the 5V output before the LVR recovers. I think this may be causing the RFID's output frequency to fluctuate temporarily causing it to lose track of tags.
What is the best way to fix this? Would it be better to switch to one of those miniature DC-DC switchmode supply daughter boards combined with a LDO?
Current figures: Worst case current through LVR is ~400mA, have measures up to ~300mA. From memory each solenoid draws ~200-300mA when on, but they are latching solenoids so they are only momentarily on. Each circuit board operates one solenoid, and depending on the product there can be up to 50 circuit boards and solenoids driven from the same 13.8V power supply.
They were having worse problems voltage drop due to running the circuit boards off a shared 5V line run over several metres of ribbon cable (with two conductors), but they then put the LVR to run off the 13.8 to get around that. They then had problems with the 13.8V line also run over the ribbon cable (five conductors) so they started running 16AWG lines to junction boards along the ribbon cable. They also chucked some big electrolytic capacitors on the ribbon cable but I don't think it made much difference from my testing.
The relays are near the board due to the physical layout of the product so not much can be done with that except to carefully design the board to prevent noise getting into it. It may be possible for the RFID antennas to pick up a bit of noise as they are on a long lead up to 0.5m, but only some of the lead is on the circuit board side of a steel plate, the wires then pass through a small hole so I don't think they pick up noise after that.
Should also mention all 50 relays sometimes get activated at once (can't change that unfortunately). However I have seen the voltage drop with just one relay activating which is puzzling, and it seems to be mainly when the relay is pushing out (instead of when it pulls) unless I am imagining things.