The PSU Crowbar is designed to protect a radio from over voltage damage. This can occur if the series pass transistor of your 13.8V high current linear power supply goes short circuit.
Often the power supplies we use don't contain any internal crowbar and so have killed radios in the past.
A crowbar is a feature in higher end bench power supplies and often the over-voltage protection level is configurable. However, it is not normally fitted to simple battery eliminator type supplies designed to replace SLA batteries.
These power supplies are generally capable of delivering tens of amps at 13.8V to our radios.
This design is suitable to a power supply up to 30 Amps at 13.8V, and is designed for standard linear supplies.
Operation on a switch-mode supply has not been verified.
IT IS CRITICAL THAT THE POWER SUPPLY HAS A CORRECTY RATED FUSE FITTED FOR SAFE OPERATION
ALWAYS HAVE A SUITABLE FUSE ON THE OUTPUT OF THE POWER SUPPLY BEFORE THE CROWBAR
I have a kit available. This contains the PCB and all parts mounted to it, including the SCR. However you will need to source your own enclosure, mounting hardware, cable and fuse.
Please contact me if you're interested. See my profile
Theory of operation
The circuit consists of a pretty hefty SCR, D1 placed across the supply rails. If this device is triggered it is capable of handling a hefty current for a short period.
We arrange a voltage sense circuit so that we can trigger the SCR if the rail voltage goes over a threshold. Once triggered, it will continue to conduct and short the rails until voltage is removed (by blowing the fuse on the power supply output, or by the operator turning off the supply).
The circuit is not designed to constantly handle the full rated current, it is designed to rapidly clear the fuse and disable the supply.
WARNING: Do not operate this device without the correctly rated fuse in the power supply output.
U1 (TL431) is a precision voltage reference and in this design is configured as a voltage switch. When the voltage at the reference input (pin 1) exceeds 2.495V, it will turn on hard pulling the base of the PNP transistor Q1 down.
When Q1 turns on, it will pull the SCR's gate high, delivering enough current to trigger it via R7.
The threshold voltage used to switch U1 on is set by the voltage divider R1, R2 and R3. The values provided allow the trigger range to be adjusted between approximately 13.8V to 25.4V. See the calibration section below. C1 provides a small time constant to prevent noise from triggering the crowbar. C2 provides some general noise filtering/decoupling.
R8 is used to pull the gate low to prevent noise from triggering the SCR, and R6 is used to strongly pull the base of Q1 high to prevent it turning on during the short startup transient current draw of U1.
The PCB is designed to be mounted into a section of 30mm x 15mm x 2mm (wall thickness) aluminium square tubing. This is available from Bunnings in Australia, and I'm sure similar hardware stores elsewhere.
Most parts are mounted directly to the PCB. Resistors and capacitors are straight forward, if you follow the board markings and BOM.
The internal height of the tube is not enough to allow TO-92 packages to be mounted vertically and so I have indicated on the overlay they need to be laid back. The flat side is up, so you can read the markings. Bend the centre lead down close to the body, then the two outside legs are bent down to match the holes on the PCB. Try the keep the legs short as you can to ensure the case doesn't fowl any other parts. I used some tweezers to make the bends.
The caps should be carefully positioned low. The LED is best mounted to so the it rests against the edge of the PCB and the lead run back along the board to the pads. This way it will shine out the tube end.
The SCR, D1 is a large package WITH AN ISOLATED TAB, the tab is used as one mounting point, with the legs bent up approximately 3mm from the body (where the legs narrow). The legs are then fed through the slots, soldered to the PCB and trimmed off.
It can be helpful to temporarily mount the SCR and PCB on the outside of the aluminium extrusion using the screws in their holes so you can align the position of the bends and hold the SCR in the right position for soldering.
If you are substituting a different SCR, you will need to deal with tab isolation and mounting but remember, there can be substantial current when it fires, so keep the leads short!
The crowbar circuit is designed to be connected by two stout wires which are connected AFTER the DC fuse and before the radio, so it can short the radio end of the fuse to clear it.
These leads are soldered directly to the large pads and if possible this solder is extended back to the SCR leads. This is the high current path so care should be taken to minimised lead resistance. I used a stainless steel tool to hold down the wires while soldering. The solder won't stick to the steel easily.
VERIFY CORRECT POLARITY There is a
+ printed on the board for the positive
Once you complete calibration (see below), the assembly can be mounted into the pre-drilled tube. Two M3 bolts are used to secure the SCR and the PCB. A small insulating spacer (perhaps some washers or PCB scraps with a 3.5mm hole drilled) will be needed to raise the PCB off the tubing. Note that the hole dimensions are approximate and may vary if you bend the SCR lead at a different place.
To calibrate the device, access to a bench power supply with current limiting is required.
- Set the power supply to the desired trigger voltage, 16V is a reasonable value (13.8V +15%) or 16.6V (13.8V + 20%). However I recommend reading the data sheet for your radio to determine a safe value.
- Set the current limit on the supply to 200mA (0.2A)
- Set R2 fully clockwise. This will be the maximum trigger voltage.
Connect the PSU Crowbar to the output of the power supply, turn it on. The power supply should remain in constant-voltage (CV) mode. If not, check your assembly of the PCB and that the power supply is set for less than 20V
Slowly turn R2 anti-clockwise until the power supply enters constant-current mode (CC).
Turn off the supply output or disconnect the crowbar. Back the supply voltage back down to 13.8V. Do not adjust the current limit!
Reconnect the crowbar to the supply and turn on. It should remain in CV mode. If not, you may have overshot the setting or have an assembly fault.
Now slowly raise the power supply voltage until the crowbar triggers. Remove the crowbar again and check the trigger voltage. You may need to tweak R2 slightly and re-test.
If you don't trust the voltage reading of the power supply, I'd suggest you put a meter across the output and use that for voltage readings.
Note: setting R2 fully anti-clockwise will likely cause it to trip at the nominal 13.8V level of most of the battery eliminator supplies.
It is possible to adjust R1 and R3 so the threshold voltage can be moved to suit different supply rails. For instance a 24V supply.
DO NOT EXCEED a 30V supply or a 30A max current
What do I do if it triggers?
- Turn off the supply as soon as possible and remove mains power.
- Fix or replace the supply, you likely have a shorted pass transistor in the regulator It is critical the the fuse is replaced with the correct value, and any new PSU's fuse is also inspected prior to service
- Disassemble and inspect the crowbar circuit and discard if there is any damage
- Re-test the crowbar using the calibration procedure
I assembled the circuit and connected it to a bench power supply set for 13.8V and a current limit of 1A.
The Crowbar in this test is configured to fire at just over 16V.
The power supply allows me to enter a new voltage setting and apply it, so the voltage step occurs quickly.
The 4 channels of my 'scope where connected to the circuit:
- Input Voltage (yellow trace)
- SCR Gate (blue trace)
- Output of U1, R4 and R5. (purple trace)
- Q1 Collector (green trace)
Note the time scale is 20us a division.
We see channel 1 (input voltage, yellow) ramp up. Channel 3 (U1 output, purple) follows the voltage up (note it is 5V a division).
We can clearly see when U1 begins to turn on in Channel 3, as it starts to slope down. This begins to turn on Q1. Channel 4 (Q1 Collector, green) starts to rise. It hits about 2.2V and at the same time Channel 2 (other side of R7, 33R) hits about 0.8V. The gate is now sinking approximately 40mA (2.2V-0.8V)/33ohms (The threshold current for this SCR is specified at 50mA.)
About 2us after the voltage at the collector of Q1 began to rise, the SCR has fired and the input voltage collapses.
The complete process takes approximately 10us from over voltage to shorted supply. The fuse should blow shortly after this point and isolate the radio and crowbar from the power supply.
Additional tests were made with 5A, 10A, and 30A blade style fuses in line. The circuit was adjusted to trip at the lowest voltage (about 13V). The leads were then placed across a fully charged 12V battery, causing the crowbar to activate and open the fuse.
In all the tests, including the 30A one, the fuse blew instantly and no discernible heat was generated in the SCR, even without the aluminium case.