3
\$\begingroup\$

In my current project I need a high-side PMOS to disconnect a load once a over-current or (downstream) over-voltage condition is detected. Obviously without a permanent latch, once the PMOS opens, an over-current as well as over-voltage will disappear, which would let the comparators re-close the PMOS. This is all fine since I generally assume transient faults, that should disappear after power-cycling. However, should a permanent fault occur, the PMOS would alternate between open and closed state rapidly, which in case of a short would lead to high high pulsed current that drain the batteries quickly.

Therefore I need a minimum-off timer to keep the PMOS in open state for a short (20ms - 50ms) duration after a fault has been detected. This would allow return into operational state for transient faults, while also minimizing the impact of permanent faults.

I know that there are ready-made IC like the LTC4364 and similar from MAXIM but these seem to be in short supply at least in small quantities and are also rather expensive.

My current approach, without the minimum-off timer, looks like this:

schematic

simulate this circuit – Schematic created using CircuitLab

My first bet was the venerable 555 timer or other general purpose retriggerable monoflop but my supply voltage requirement of Vin,max 36V rules out all parts that I am aware of. Please correct me if I am wrong!

This leaves me with a discrete solution, resorting to a BJT monoflop. It is triggered by the OK open-collector output of the comparator. When a error occurs, OK is pulled low and C1 discharges, all while the PMOS gate is pulled up by the PNP of the first schematic. When the PMOS opens, the comparator also stops pulling OK low but slowly charging C1 keeps it there for some time.

schematic

simulate this circuit

Please review by schematics! I am especially unsure about the working of the monoflop, since I have the feeling I might be using it in a slightly unusual way. I also have not yet breadboarded the thing but only simulated in LTSpice so far (which works fine).

Clarification: Vin, max 36V is only the maximum fault withstand voltage. Maximum working voltage is 12V.

\$\endgroup\$
1
\$\begingroup\$

I'll cut to the chase and suggest a modification to your main diagram: -

enter image description here

Use a diode as shown so that activating Q1 is done fairly quickly when an "overload" occurs. To make this work effectively Q1 should be a P channel FET - this will allow C1 to charge to nearly the incoming power rail - in fact add a zener diode across R2 to limit this charge to 10V (in case the gate gets damaged by over-voltage).

R2 can be chosen to be a high value that slowly discharges C1 when "OK" returns high.

\$\endgroup\$
5
  • \$\begingroup\$ I appreciate your suggestion since it has a much lower part count. I tried it in LTSpice but I have trouble stabilizing the system. When simulating a permanent short only the first switch-off is reliable. Repeated switches see a lot of chatter and have a much shorter hold-off period. I will investigate further. \$\endgroup\$ – Arne Apr 4 '14 at 15:54
  • \$\begingroup\$ Maybe try Q1 (MOSFET) with a lower Vgs(threshold). Is your 36V input falling when the output is shorted? \$\endgroup\$ – Andy aka Apr 4 '14 at 16:14
  • \$\begingroup\$ OK I will try that. The setup works fine for single events and also for the initial turn-off of permanent faults, but fails on successive turn-offs. \$\endgroup\$ – Arne Apr 4 '14 at 16:20
  • \$\begingroup\$ Where does OA2 receive its power rails? \$\endgroup\$ – Andy aka Apr 4 '14 at 16:34
  • \$\begingroup\$ OA2 gets power from before M1. \$\endgroup\$ – Arne Apr 4 '14 at 19:02
0
\$\begingroup\$

Your input volts make ICs more difficult BUT you could afford to waste 600mV under overcurrent conditions so its reasonable to place a low value resistor in the source that would turn on a SCR when Vbe is reached . This SCR could be just two GP transistors .The use of two transistors rather than a small SCR gives better speed and more control of parameters like holding current . I have used BC337 and BC327 or BC557 and BC547 or BC817 and BC807 . Then I used transistor arrays because they are now cost effective and save space . The anode of the role your own SCR connects to the gate of the P channel via a 10 ohm gate resistor turning it off fast at overcurrent . In my application I was worried about wiring inductance so I placed a Schottky freewheel diode to protect the P channel from voltage spikes at turnoff . I was also concerned with nuiscence tripping so I monitored Freewheel diode conduction to auto reset the electronic fuse . Finally I placed a 100 microhenry coil in series with the output .This discrete design basis is now employed for DC power system protection.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.