0
\$\begingroup\$

I'm building a load bank for testing high power lead acid batteries, and expect to pass up to 800 amps at 12 VDC. I could do it with contactors but would like to use an SCR with a pot and allow the current to be ramped up with the pot. Any thoughts?

\$\endgroup\$
  • \$\begingroup\$ You want just a switch? Or you want to control the amount of current? \$\endgroup\$ – Chupacabras Jun 21 '19 at 12:06
  • \$\begingroup\$ You want to control 0-10kW DC via a pot? \$\endgroup\$ – JimmyB Jun 21 '19 at 12:27
  • \$\begingroup\$ How are those 10kW being disposed of? I think that (car?) lead acids can deliver but not withstand hundreds of Ampères for more than 10 seconds or so at a time without overheating. \$\endgroup\$ – JimmyB Jun 21 '19 at 12:28
  • 1
    \$\begingroup\$ An alternative idea would be to use a powerful electric motor (e.g. starter motor) as the load; attach to it a brake (e.g. bake disc/rotor and caliper/pads from a car). You can then vary the torque, and with it load, smoothly via the brake. Benefit would be that both the starter motor and the brake are built to handle those 10kW at least for a few seconds. \$\endgroup\$ – JimmyB Jun 21 '19 at 12:39
  • \$\begingroup\$ ignoring the fact that SCRs are not easily controlled when used with DC, 800A through an SCR is over 1000W of heat produced in the SCR. \$\endgroup\$ – Jasen Jun 22 '19 at 3:11
2
\$\begingroup\$

I could do it with contactors but would like to use an SCR with a pot and allow the current to be ramped up with the pot. Any thoughts?

It's a foolish idea for at least two reasons: -

  • An SCR will latch once conducting and stay latched hence to disconnect the "load bank" from the battery you will need a contactor. Of course it makes sense to just use the contactor for full functionality.
  • Current won't ramp up with either a contactor or an SCR with or without a pot. If you require a variable load impedance then use something else that can be controlled on/off in a PWM manner. I'm thinking banks of MOSFETs and an inductor here.

There will be kilowatts of heat to get rid of so design your load bank rig carefully.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Was thinking about "power" MOSFETs too. 50A+ FETs are readily available, but heat dissipation will be an issue, esp. if they stay 'on' for more than a second or so. \$\endgroup\$ – JimmyB Jun 21 '19 at 12:42
  • \$\begingroup\$ 10 milli ohm x 50 amps squared is a power of only 25 watts and relatively do-able. You can get MOSFETs with sub milli ohm on resistance too. It's do-able with care. \$\endgroup\$ – Andy aka Jun 21 '19 at 12:53
  • \$\begingroup\$ And PWM-ing... I guess, for those tests, a steady, not pulsed, current is needed/preferred both on the source and on the sink side. \$\endgroup\$ – JimmyB Jun 21 '19 at 13:03
  • \$\begingroup\$ Yes, with active cooling, 25W per FET may be possible, but challenging. \$\endgroup\$ – JimmyB Jun 21 '19 at 13:04
  • \$\begingroup\$ @JimmyB the inductor will smooth things and if designed correctly can be quite a low ripple current sink within reason but, it will be a substantial design to get right and at a decent efficiency. \$\endgroup\$ – Andy aka Jun 21 '19 at 13:37
1
\$\begingroup\$

SCRs are not appropriate switches for DC because they are difficult to turn off. There are commutation circuits that you could build, but it would be a lot easier to use an IGBT or MOSFET.

To control current, you would need to use pulse width modulation (PWM). With PWM, the current drawn from the batteries would be pulses, not steady DC. You would need to determine if that is appropriate for testing the batteries. The current could be smoothed with filtering, but that adds to the equipment required.

The use of the plural "contactors" implies that you are thinking of a bank of small resistors rather than one large resistor. You could easily use a potentiometer to sequence the connection and disconnection of resistors. You can use contactors, solid state relays, MOSFETs or IGBTs.

If you are designing something for industrial use, you should consider a system that would return power to the AC power system. That is much more complex, but it would reduce operating cost and save money in a situation where the testing is going on for an extended period of time.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ "you should consider a system that would return power to the AC power system" - Hard to imagine that would make sense. If the OP tested a battery at 10kW for 10 seconds at a time, that'd be 100kWs, or about 0.03kWh, or about €0.01 monetarily. That is of course assuming that the OP is only intending to test the current capability of batteries, not to drain them completely, and that he does not do thousands of those tests a day. A system that is able to dump bursts of 10kW into the grid is probably not cheap, and the grid operator will not be happy with that burst-load pattern. \$\endgroup\$ – JimmyB Jun 21 '19 at 13:49
  • \$\begingroup\$ @JimmyB: We can only imagine. Only the OP knows for sure how this equipment will be used. All options should be on the table for consideration. \$\endgroup\$ – Charles Cowie Jun 21 '19 at 14:14
  • \$\begingroup\$ I do know you have to reverse bias the SCR to turn it off, so I guess you would have to interrupt full load current, which is about 800 amps. Like I said, I could do it with contactors, but I have an opportunity to play around with using semiconductor, I'd like to do that \$\endgroup\$ – Pete Moss Jun 21 '19 at 17:13
  • \$\begingroup\$ PWM with SCRs is rather complex. You need a commutation circuit consisting of a second SCR and an LC circuit. You need to supply a current pulse from the commutation circuit that is equal or greater than the current carried by the main SCR. You need SCRs that are designed for a short commutation time. Those who designed such circuits 50 years ago switched to transistors as soon as they were available with sufficient voltage and current ratings. \$\endgroup\$ – Charles Cowie Jun 21 '19 at 17:35
  • \$\begingroup\$ I'm not stuck on using SCR, I don't know what is the best component for the appliaction \$\endgroup\$ – Pete Moss Jun 21 '19 at 18:44

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.