Active pre-charge/inrush control

Addy said:
I built a variation of the V3 inrush limiter design to use with my 12s 20Ah ebike pack. I'm using a normally open keyswitch to control it.

I usually estimate about 20A per FET for minimal heat sinking. You should have no problems with heat as long as the board gets some air flow or against something that gets air flow. Try running it hard for a few minutes straight, then stop and check the temperature.
 
A normally open switch won't keep the feedback capacitor charged when the output is off. The gate needs to be connected to source with a double throw switch. You are probably getting inrush sometimes.

I have parts on the way for my next version, this time for sure! I'm staying with sequential contacts. I'm convinced that's the best solution for a general purpose gadget to connect and disconnect 116vdc with load current happening. It avoids the need for a switch and allows me to use it when I don't have access to the pre-contactor voltage.

I'm giving up on hard switching, too many unknowns and blown fets to continue that approach. I don't know the real impedance of my inverter. And I can't even measure it, but I think the actual inrush is what's killing my fets, hundreds of amps, could even be 1,000. Datasheets seem to say that's ok for a few microseconds but the results say otherwise. I still believe that a 0.1uF cap is a low impedance gate drive. It has less voltage rise than a 100ohm gate resistor for any amount of current for up to 10uS and will absorb transients better as well. But it's academic, I'm going back to linear feedback.

So here's my new (sort of) circuit. I found a way to reduce the parts count of the sequential ramp circuit I posted previously using only one low power fet. And I've found ways to suppress the nasty little transients that happen on connection. This will ramp up in about 1 second with less than one amp even if the inverter is on and drawing 50 watts idle.hot ramp.PNG
I've carefully selected a new fet but will need two to get the low on resistance I want. The pair are slightly more capable than the big single as far as soa, power and current, and not as expensive. They are 150v but that should be ok, I hope. https://www.mouser.com/ds/2/205/DS100228A(IXFH160N15T2)-347934.pdf There are very few mosfets that can meet the specs I need at this 28s voltage. Or at least that's what the datasheets say, or don't say.

I'll build and test this in a few days, maybe this will be the design that really works.
 
KeithT said:
A normally open switch won't keep the feedback capacitor charged when the output is off. The gate needs to be connected to source with a double throw switch. You are probably getting inrush sometimes.

Maybe you're thinking of the switch in a different position. Here's what I did:
inrush limiter.jpg


Unless I'm missing something, I think this would work exactly the same as the double throw version. When I tested the circuit all I ever saw was a smooth ramp during turn-on.
 
Yes you're right, that should work fine with the switch there, I've seen a different version that wasn't fine. I had a weird fet failure today though and it was exactly this circuit but without the RC feedback. I just needed something to switch a dummy load so I could calibrate a current sensor. Resistive load so no ramp was needed but 30 amps so I fet switched to avoid arcs. Well, it blew up right away, shorted and smoked. I replaced the fet with the same type and added a 100 ohm gate resistor and it worked perfectly. Wow.
 
Here's another recommendation to avoid zeners, maybe that's what happened. https://www.microsemi.com/document-portal/doc_download/14693-eliminating-parasitic-oscillation-between-parallel-mosfets That engineer likes ferrite beads. Well, if it works, great. If it blows up, try something different. But it seems the only way to find out is apply power and see. The large feedback capacitor and series resistor in the ramp circuit probably make the zener safe. But a gate resistor seems to be a good idea. Nobody knows what value is correct though. Oh and paralleling is trouble too, maybe I should have bought another $22 fet...
 
So yes, the circuit with sequential contacts and a linear ramp works, it drives my inverter and connects without arcing. I checked the waveforms with a scope, no parasitics, the ramp speed is as predicted. The big inverter input capacitor is precharged without inrush, and fet on resistance is low, but they do get pretty warm at high current. This device will prevent any damage to the contactor in my Zero packs. So maybe this project is done.
ramp.jpg

But...

This circuit will not handle any significant load current at turn on, power dissipation is too high, fets will fail. So not having a switch isn't really helping much, the inverter should be switched off or at least unloaded or there will be fet failure (hint: that's not an opinion!). And the load should be switched off before switching off the inverter. And I have a switch to enable the Zero contactor too. So the holy grail of a switchless DC battery power connection device that can handle lots of current, charge lots of capacitance, and avoids any arcing on connect or disconnect remains unrealized. Would a different gate drive allow hard switching into a big capacitor with huge current but low dissipation? For now my mosfet budget is overdrawn...
 
If I remember correctly, no matter how fast you switch the FETs, the energy they (and the series circuit resistance) must dissipate is equal to the capacitor bank's fully charged energy (1/2*C*V^2). Plus the I^2*R for any energy the inverter consumes during the turn-on period. Depending on the speed of switching, the heatsink does no good unless the switching is slow enough for the heat to transfer to the heatsink. In other words the FET device mass must be sufficient to absorb this energy while remaining cool enough without aid when switching times are low. I suppose if you switch fast enough a greater proportion of the energy will be dissipated by the wires and stray resistance, but that probably helps very little.

One technique is to series a resistor with the FET that can take the brunt of the energy, and then use a second FET to short that resistor after the capacitors are nearly charged. A cascade of FET/resistors could be used to get to any desired power level if needed.

A contactor could always be used to unload the output of the inverter, if the inverter doesn't have a disable input. Inverters often have control busses that can be used to adjust parameters and send them commands, these are often simple modbus interfaces.

A low cost micro could be used to implement the necessary FET timing, and send any commands or enables to the inverter.
 
I guess another approach would be to add series inductance and switch the FETs fast. The inductance would slow down the current so the FET has time to get fully on before the current gets too high. Of course this would cause other issues, like a huge voltage spike when turning off.

A two stage circuit with a resistor as the dissipating element sounds like it would work well.
 
Alan B said:
The additional inductance might not be good for the inverter.

Agreed.

I suppose you could close a switch across the inductor once it got going, but too many other issues.

A long time ago I designed a circuit that had a resistor precharge and a voltage monitor that would detect when the precharge was complete or nearly complete, then allowed the main contactor (or FET) to close. Most EVs use something like this. The circuit is pretty complex to do with strictly analog parts, but with a microprocessor would be not bad. If the load fails to get up to voltage in some relatively short time (due to shorted or excessive load), it would time out to keep the resistor from cooking and give a fault indication. If the load is within 5v or so of the source, it should be safe to close the contactor.
 
fechter said:
Alan B said:
The additional inductance might not be good for the inverter.

Agreed.

I suppose you could close a switch across the inductor once it got going, but too many other issues.

A long time ago I designed a circuit that had a resistor precharge and a voltage monitor that would detect when the precharge was complete or nearly complete, then allowed the main contactor (or FET) to close. Most EVs use something like this. The circuit is pretty complex to do with strictly analog parts, but with a microprocessor would be not bad. If the load fails to get up to voltage in some relatively short time (due to shorted or excessive load), it would time out to keep the resistor from cooking and give a fault indication. If the load is within 5v or so of the source, it should be safe to close the contactor.

Sounds like a good approach.
 
I haven't been getting notifications for this topic, but I've been busy doing exactly what is suggested above, two fets, one with resistor to limit current, the second to give low on resistance. I'm in the final stages of that design, with a big detour into SOA analysis and FET selection. I've created a really nice tool for SOA simulation and comparison that I will publish when finished.
soa-model.png
soa-plots.png
And I've learned a lot from Linear Technology and their HotSwap app notes and models. Live backplane insertion is a very similar problem, too bad their chips can't handle my voltage. http://www.linear.com/solutions/5239 http://cds.linear.com/docs/en/datasheet/4260fc.pdf
It looks like this two stage design will handle hot connect into most any load, if the FET's are up to it. I'll post more soon.two-stage.pngtwo-stage-sim.png
 
Your precharge resistors are pretty small. Should make for really fast cap charging. Most EV systems I've seen use much higher resistor values. Seems like the inverter would have some kind of delay between getting power and actually starting up. You should be good if the precharge is faster than that.

The voltage checking feature I was describing would be a really nice safety feature in case the load is shorted.
 
I used a linear tech part for my inrush limiter. cant recall the part number right now.
it a chip controlling a low side mosfet with a current sense resistor. Used a constant current source ic to supply the LT part because it uses a internal zener. could post a schematic if there is interest.

Manny
 
Yeah the resistor size is arbitrary, but there are some tradeoffs. Finding power resistor specs in joules isn't easy but there's this http://prod.sandia.gov/techlib/access-control.cgi/1981/810569.pdf
The design uses a zener to regulate the second fet turn on voltage, needed because the pack could be 90 to 116V. Fet current is under 200 amps for both, lots of parts that can handle that, but my design is not final. The SOA analysis says the parts shown (used because LTSpice has thermal models) could handle up to 10,000uF, 0.05 ESR and 3 ohms. I have some ideas for an overcurrent shutdown at turn on, not too complicated, but I'm not sure it really makes sense. Protecting against every kind of failure isn't possible, a fuse will keep the smoke minimal.

I'm doing this design mostly for the challenge, what I have now is usable. But I think a simple, no contactor, no microprocessor, no switches, hot connect/disconnect device can be achieved reasonably, and will become important as more people start to repurpose old lithium. I thought about a processor, power supply and interface are problems, and an inductor too, also trouble. But the two stage fets with resistor circuit seems to be a solid solution. Here's the basic idea, probably needs a fuse:
hotbox.png
I've named the concept HotBox, someone should trademark that, I'm not a marketing guy!
 
I've added protection for over current at turn on, no microprocessor needed, just a handful of discrete parts. Three transistors do it, Q7 turns on with the first fet and charges a timing capacitor, Q8 turns on if the load voltage doesn't come up fast enough, Q9 turns off the fets, latching the circuit off until disconnect. It seems to work very well, handles low resistance or shorted output, and too large a load capacitor, keeping the first fet well within ratings and the second never turns on at all. I'll add a pair of LEDs to indicate power on or fault conditions. The low precharge resistance is needed to make the decision quickly, 5mS, keeping the first fet from over heating. This doesn't protect against over current after turn on, that would be very complicated, requiring a current shunt, high gain differential amplifier and processor. So I'll stick with a fuse for that.hotbox-ocp.png
 
Looks great. Analog circuits are sort of a dying breed but you'll never need a software update. Downside is the parts count. I think your shorted load protection is perfect. Once the thing gets going, you generally have something in the battery BMS and a fuse to protect against shorts. With 0.5 ohm precharge resistor, it should come up pretty darn fast.
 
Yes, quite a few parts, but a small circuit board will hold it, and only the power components have any significant cost. Cheap compared to a contactor, and should work with practically any battery pack and load, no software hacking required, just hook it up. This circuit has an initial delay of about 0.3 seconds to make sure it is fully connected, then the precharge is quick at around 3 mS for the values shown. It can't be done much faster than that because of time constant of the ESR in the load capacitor.

The fets are switching pretty hard, in microseconds. There are actually two positive feedback loops. First around the delay timer to drive the precharge fet and resistors, and another around the load voltage hand off transistor and second fet. Both loops have feedback around two inverting stages and switch quickly once they fire. So the switching power loss in the fets is very low, less than 0.3 joules total. The first fet won't even get warm, no heatsink needed. The resistors take the dissipation during precharge, but they won't get hot either because it happens fast. The load stores about 17 joules so the fet efficiency is better than 98%. Heat in the second fet will be created due to the on resistance when load current is high, with a low but not negligible temperature rise. Because the 0.5 ohms is large compared to the on resistance of the second fet, almost all of the load current will be handled by the second fet. I'll do some more thermal analysis and final fet selection before I build this, but it looks very promising. Real world testing before too long.

I should mention that a switch can be used in the make-last contact circuit. So if desired this circuit could be connected and then turned on and off with a very low current control switch. But I think being able to hot swap packs without a switch is really useful in some cases.
 
Well, lots of folks on the forum would want to use something like this for a bike controller, which won't have much load other than the caps. Some run a small dc-dc for lights and accessories in parallel. This should work well in that application. A switch version would be more universal for bikes as often the pack is left connected and we just want to disconnect the pack when the bike is off.
 
That would be nice to have a clean & closed topic on this to be able to follow the evolution ^^
28 pages is way too much to read haha
 
6 years of evolution takes some space. Probably hundreds of FETs were blown in the course of this thread.

Once KeithT finalizes his design, it would be great to make a separate topic for it.
 
Not final until I build and test, but I've firmed up the design enough to order parts. I went for the same fets I've blown up before, but this circuit will treat them with a lot more care. I'm including a switch, found it interesting that it only needs to handle about one volt dc at a milliamp. High DC voltage rated switches are expensive. I note that almost half of the high parts count is 100k resistors. I have thought about a starting new topic, and I'll do that when it's time, but for now the evolution continues...
hotbox-parts.png
hotbox-pcb.PNG
 
It's getting real. Yes, a lot of parts. On to the metal work and power wiring...hot box perf.jpg

And speaking of evolution, here are the fet sizes I've been through, starting with TO-220, then TO-247, then TO-264
fets.jpg
 
So construction is done, nice package! Sadly, destruction is done too. More casualties. The fault detector triggered even with no load. Deciding it was non-essential, I disabled it with a clip lead. Circuit then worked with a light resistive load, tried the inverter and... time to order more fets. This will work someday! :cry: hot box.jpghot box openjpg.jpg
 
Bummer... Earns you a fired guinea pig award for sure.

Did both FETs fry?

I'm looking at duplicating your operating principle only with gate driver chips.
 
Back
Top