Protecting Super-Caps

Lorry batteries?

On 3/26/20 12:54 PM, Winfield Hill wrote: . . .

A colleague of mine used an LTC3780 which has output overvoltage protection.

How about a truck battery?

Cheers

Phil Hobbs

Isn't the whole point of half sine 8.33/10ms to represent mains power line so why not use the mains power line via appropriate transformation?

piglet

torsdag den 26. marts 2020 kl. 20.54.45 UTC+1 skrev Winfield Hill:

there's plenty of ICs that in combination with a fet with cut off the supply when exceeding a set voltage. that would fix the fat finger on the knob problem

but I think you would also need balancing, and at those current you probably don't want to clamp each capacitor, maybe something like this?

They setup a simple system, I suppose. A Variac and maybe step-down transformer, an SCR or Triac, with a half-cycle trigger controller, and a big old power resistor to the diode. Yawn.

We can impose a shaped, controlled, up-to-500A current pulse to a diode under test. And more important, we can interrupt the current at any point during a 8.3ms pulse, and quickly measure the diode's forward voltage at 10mA, and thereby determine its Tj. Repeated tests with varying time delays can let us map the junction temp vs time. This is better than simply determining a diode failure point. Many manufacturers claim their spec is for 175C Tmax. How do they know?

Yes

I've purchased some of these, but they never give a schematic, and I have no idea what they do, or how good (or bad) it might be.

Do you have three phase mains available ?

A simple six pulse rectifier will have only 4.2 % (RMS) ripple without any storage capacitors.

If this is still too much, use 12 or 18 pulse rectifiers, which requires additional transformers (or at least multiple secondary windings), so that various wye/delta configurations can be used to generate the required phase shifts.

Since the required DC voltage is quite low (and hence the primary/secondary turn ratio is quite large) and the pulse width is small, the mains fuses should handle the situation easily.

Yes, its in the building, good idea. Ripple is no problem, because the active op-amp-controlled MOSFET current source will simple insure that the current remains constant. And high CMMR and PSRR difference amplifiers provide the output signals.

It's only necessary to keep a minimum say 5 volts across the MOSFET. For long-pulse, low-voltage 500A operation, I'm thinking of a IXFH400N075T2 MOSFET, which can have under 2V Vds drop at 500A. A 2.5mR sense resistor takes up 1.25 volts. So it's really a matter of the maximum D.U.T. drop.

Obtaining 500A transformers might be an issue.

Does the building have 120/208 V or industrial 277/480 V ?

In a six pulse rectifier, the voltage drops to 87 % before the next phase takes over (more than 12 % pp) Even a small storage capacitor will help ride trough the dip, i.e. keep the voltage at least say 90 % of peak voltage for less than 1 ms.

A single three phase transformer with 200 A secondaries should suffice for continuous duty.

If you can't find a there phase transformer with 200 A secondaries, look for single phase transformers with 200 A secondaries.

If you use single phase transformers, each transformer will need to handle only 1/3 of the power (or 1/6 if 12 pulse rectifier with six single phase transformers) is used.

With single phase transformers, look also for transformers with 208 V primary voltage (or 277 or 480 V if available at your site).

Thanks Win, I wasn't advocating going back to an SCR, you can use your MOSFETs to switch but meant to suggest using stepped down mains as the high current supply.

piglet

a stack of toroid transformers, ignore the secondaries and add a new with some welding wire?

next issue is the six 500A diodes

Also sold are toroid power transformers with only the primary winding, the user to provide the secondary. Of welding cable.

Actually, widely available albeit expensive. Probably is hockey-pick diode range. One possibility is Infineon's "High Power Thyristors and Diodes" line. Get the "Selection Guide 2019/2020" from the Infineon website. Vishay, IXYS, et al, also have candidates.

These high-power diodes have considerable thermal capacity, and so can be pulsed pretty aggressively. Schottkey diodes may also be available, but I have not checked.

Joe Gwinn

Youch! sounds like a buzz box ac/ dc arc welder... those things scare me some.

George H.

A LiFePo4 starter battery could do it. They have internal resistance in the neighborhood of 10 mohm. This one has 500 CCA, so it should deliver at least 10 V at 500 amp at room temp.

They make welding caps that type of job.

They are noramlly much larger in termina lugs and some of them.

A low ESR and low value cap for a welder is about 10 x the size of a normal cap and mount with screws to the steel for heat sinking ...

I've seen ring type oscillators for induction heating use these caps due to the required low ESR. Btw, ringing induction heater supplies are very interesting in design, simple but works.

Here is an interesting solution that uses high-voltage caps coupled thru a transformer to produce low-voltage, high-current pulses:

Do you know where that was published, and its date?

That's an awesome battery, small and lightweight, it looks like it's just the ticket. My SuperCap scheme would only work for one or two half cycles, whereas the diode guys want to test for up to 100 cycles. I'd like to be able to do a good fraction of that, before the MOSFET overheats. At an RMS average power level of say 1.4kW, a IXFH400N075T2 MOSFET can handle several cycles, and with two FETs, the heatsink should handle continuous heat.

OK, this brings up the issue of fuse-protection, otherwise a short could lead to an explosion. I'm having trouble finding an appropriate fuse, but the battery's built-in shutoff will do. The PCB has a 30A auto fuse, but it'll blow in 10ms, so I could short across it with the BLP-20500M.