Idea for main power disconnect..........

[recumpence]

I have been thinking about a "Perfect" kill switch. You cannot just kill the center leg of the throttle POT because it is poccile the ESC can go into a fixed throttle mode and take off. It is rare, but I have had it happen to me. So, a positive disconnect is needed. For high power applications, a contactor is really not feaseable and a remote mechanical plug puller is really ghetto. So, annoyed at this delema, I think I have come up with a workable solution.

My idea (and bear with me, I am only slightly electronically inclined ) is to use a few 4110 FETs in parallel as a main power switch. One main power lead would get its power through the FETs. So, a simple switch, even a keyed switch, would send power to the FET gates and turn on the FETs and also turn them off. This would, in effect, act as a main power contactor. It would even be good to eliminate the plug-in spark associated with plugging the pack in. You could either setup a very simple cap charging circuit that would fire the 4110s only after the caps were charged, or you could merely use enough 4110s to handle the innitial cap charge hit.

Anyway, that is my idea. What are your thoughts? Are there any inductance issues or anything else you can think of that would thwart my otherwise awesome plan?

Matt

 

[Dirt-E-Dan]

Matt,

I'm using a Curtis/Albright 60V 200A contactor, controlled by a key switch that I bought from Advance Auto and I've had no problems. Just curious, but is it not feasible because of weight? However, I did blow one of those up, but that's my own fault. I didn't cover up the extra tabs...

Daniel

 

[GGoodrum]

The new LiPo-based LVC I'm doing has an active cutoff section that uses four 4110 FETs to cutoff power whenever one of the LVC optos trip, due to a cell voltage dipping below 3.0V. These opto outputs are all ganged together, in parallel. This signal can also be tied directly to a controller ebrake input. To implement the "kill switch", all you have to do is connect a switch to the "BRK" and "GND" pads on the board below, and the FETs will get turned off.

 

[fechter]

I think that would work. Using 4ea 4110s, it will dissipate around 12W at 100A, which is not too bad. A heat sink may be needed. To drive the gates, I think you could get away with a very simple 12v zener diode and a pair of resistors.

 

[recumpence]

Cool. So, what about the 5 volt regulated output from my Fechter limit board? Would that drive the gates?

Oh, would the FETs handle the load of charging the caps when turned on, of would I need a separate slow cap charge sequence first?

Matt

 

[MitchJi]

Hi,

Cool. So, what about the 5 volt regulated output from my Fechter limit board? Would that drive the gates?

Matt

How about an RC Throttle/Limiter/slow cap charge/power disconnect combo board?

 

[12p3phPMDC]

I'd look for a SOT227 package fet...
you can assemble them without boards (4110s too...but a bit ghetto)...
They have screw terminal connections and much more robust
current handling per pin. 200A vs. 75A for T0220 or variation.
A simple 12V output into a 100K pull down resistor into the gate would
turn on the fet.

 

[recumpence]

Cool! I'll try that!

:EDIT:

Holy ****! Those things are freakin expensive!

Forget that! I will go with multiple 4110s.

Matt

 

[12p3phPMDC]

Well,

Its all relative to the amount of work it takes to get the 4110 solution going.

Future electronics has these for $24 per.

http://www.vishay.com/docs/94541/fb180sa1.pdf

It's also isolated from the heat sink side....so, no need to
worry about insulating the case to the heat sink. Less cost and
time involved.

Expensive?

Yea, alot more than a
a single 4110. Ok, so, parallel 4-4110s or so. Even two. 3 bucks per 4110
from Digikey. Then add a bunch of stuff.
Without a board, you've gotta do enough soldering and wiring and
some kind of connectors. Add a circuit board and cost goes up...
shrink wrap, wiring, etc..its hard to get solution that is nearly
as nice mechanically and thermally as a SOT227.



With To220 packs, I'd make a power stick, with all the drains, uninsulated, tied to a bar of copper.
Then heat sink the copper with insulating thermal epoxy.

 

[Tiberius]

I have been thinking about a "Perfect" kill switch......

My idea (and bear with me, I am only slightly electronically inclined ) is to use a few 4110 FETs in parallel as a main power switch. One main power lead would get its power through the FETs. So, a simple switch, even a keyed switch, would send power to the FET gates and turn on the FETs and also turn them off. This would, in effect, act as a main power contactor. It would even be good to eliminate the plug-in spark associated with plugging the pack in.

Hi Matt,

I have been working on this, and I think I can see a way to do it - a combined inrush limiter and master power switch. Its the next thing after the Ideal Diodes. So far the design is on paper, but it does look possible to ramp up the start to avoid the connection spark.

There are two downsides of any electronic master switch. 1) There is some small power loss, and the max current limit may be lower than you want. 2) Can you be sure of a 100% disconnect in an emergency. On the other hand, that last consideration applies to a mechanical switch too.

Voltage limit for a readily feasible electronic master switch would be 100 V, so say 90 V to be sure. What current are you looking for?

Nick

 

[recumpence]

For most setups, it will be seeing about 150 amps on occasion. My big trike will be pulling 300 amps on accelleration, though.

My thought on cap charging is two fold;

#1 Enough FETs in parallel could handle the current inrush when turning on the gate without any resistance added for slow cap charging.

#2 What about a resistor parralleled with the big FETs that would charge the caps slowly. A second resistor would be added from the output of the charging resistor to the FET gates. When the power is innitially activated, the charge resistor would begin charging the caps. At first, the output voltage through the charging resistor would be low because of the cap charging load. As the caps increase their voltage (state of charge), the output voltage of the charge resistor would reach the critical voltage for the FET gates and the FETs would turn on.

Could it be that easy, or am I missing something?

Don't slam me if I am off my rocker on this. Again, I am pretty green (pun intended ) on complex electronics. I understand just enough to be dangerous!

Matt

 

[Tiberius]

The important thing is the max continuous current. The timescale for continuous is the heating up of the FETs, so say 10 secs when they are on heatsinks.

The most economic solution may be to do what I decided on for the ideal diodes. Design for X Amps max to meet the majority of cases, and then people needing more can parallel them up. But it depends on how much circuitry is needed.

As for the inrush limiting, I plan on a circuit that will use the FETs as the control elements so it will also stop any fast transients whether the connection is upstream or downstream of the master switch.

Nick

 

[12p3phPMDC]

A simple RC circuit that drives the gate could be used
to turn on the fet slowly. This would also allow the controller caps to be charged slowly.

Rc and C plus the Gate capacitance would be used to control the ramp rate.
When you close switch S, The capacitor would charge and the voltage would
start to rise. When it reaches the gate threshold voltage the fet turns on, but
not hard. Current starts to flow into the controller caps. The gate cap will
continue to charge and increase in voltage. Eventually, the fet will be turned on
all the way.

When you open S, Rpd (pull down) will discharge the gate cap.
A better way may be to use a double pole switch and ground the gate through a small resistor
when the switch is off to insure turn off.

 

[recumpence]

Yes, that is what I was describing!

Awesome!

So, now we just need component values and I will order up some parts and start experimenting.

Matt

 

[12p3phPMDC]

Here's another quickie...

This one would have symmetrical start and stop speed.



Like Tiberius said, we'll have to be careful with the fet heating.

Unless you go to a more sophisticated start up controller, i.e. one
that pwm controls the fet, you'll want to limit the time in the active region where it'll
heat up fast, but have it be slow enough to limit cap charging and arcing.

A double pole double throw switch could be used to soft turn on and hard off.

Edit:
Probably want a bleed off resistor permanantly installed on the caps....this circuit would
keep them charged

 

[12p3phPMDC]

Component values...

Ok, I'm going to assume
that the circuit should be fully on in 0.5 second.

The mosfet turns on at an average of 3 volts over temperature and device variation...
If we ramp from 0 to 12 volts in about 1 second, it seems like
the devices won't heat up too much if we keep it fast enough but slow enough
to keep it "soft" start.

The fets will be largely on at ~6V.... but charge them all the way to 12V
because of the high currents involved.

So, lets set the time constant to 0.5 second.

Let C = 10 uF. tau = 10uf * R , R = 0.5 / 10 uF = 50 kOhm.

This will charge up the gates to ~7.5 Volts in 0.5 second. They will
continue to turn on more up until they settle @ 12 Volts.

If you want to speed up the circuit, just drop R or increase C.

50kOhm will keep the initial charge current small.

You'll need a regulator to isolate the switching circuits from the controller....
This is the harder part of this whole thing.
It may be better to to just find a regulator with a soft start feature that
has a 50V-75V input and a 12V output....hmmm....


This looks like a great little 5 or 6 component buck regulator.
Set it up for a 12V output and keep the RC circuit.
http://www.national.com/ds/LM/LM2576.pdf

 

[julesa]

Might want to throw a zener diode on the gate for the FET(s) with breakdown voltage around 10v, so it stays completely off until the voltage rises far enough to put the FETs solidly on...?
(I'm just another electronics noob)

I was trying to do something very similar here... haven't had a chance to really dig into it yet (I just got my batteries and haven't had time to start building yet), but the last diagram in the thread linked below is pretty close to what I was planning to do. The only design changes I made since the thread below are adding a transistor to make a Darlington pair, and I was thinking I might need to add a resistor to allow the cap to discharge after the circuit gets turned off:

http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=10650&hilit=+suggestions

But I dunno, maybe using a FET is a better way to go.

 

[recumpence]

If you get a chance to set it up, please post your results. I have about 3 hours of work left to finish one trike, then another 4 to 5 hours to build trike #2 before Wednesday evening when I go out of town. So, I will not be able to try this out for another week at least. :wink:

Matt

 

[12p3phPMDC]

Julesa,

Your circuit looks like it'll work.
But, in order to do the soft start feature,
you can't put a zener on the gate, unless, you pwm into the
zener, which wouldn't be necessary at that point.
But for a "don't come on until T delay circuit", your's will work.

This would be a simple "ramp for T" circuit
and no relay required...

 

[Grinhill]

(bump)

Has anyone got one of these circuits working yet?

If so, please post details...

 

[julesa]

Works great so far but I don't have a lot of run time on it yet -- I just installed it on the bike yesterday. My AmpedBikes.com controller had a 22A factory limit, and after putting a little solder on the shunt it's now limited at 31A. That's also a recent modification. I'm going to take the bike up some nasty hills for an abuse test in an hour or two, so we'll see if anything gets hot or if any magic smoke escapes.

The 50K pot adjusts the relay's turn-on delay from 0 to about 10 seconds. Like I posted above, I added another transistor and a 22K resistor across the 100uF cap to give it a discharge path when power gets turned off. Sometime in the next few weeks I'll draw it up as I built it. The little transistor is a 2N3904, the bigger one's a 2N3053, the relay is an automotive relay rated for 35A continuous duty.

Sorry for the fuzzy cellphone pix. The terminal block, from right to left:
12V+, 12V-, on-switch return, 48V output+, 48V input+, 48V in/out-

The 48V->12V DC-DC converter (not pictured) is mounted with thermal compound to an aluminum plate, which is mounted with thermal compound to the steel toolbox my batteries are living in. Hopefully the steel will transfer enough heat so that won't be a problem... :?

 

[317537]

What about a second brushed controller with a switch onto the brake cut offs and the device is set to slow power up via the throttle input "of the brushed controller". At least you would have PWM as the power transmitter and its almost a a ready made solution and been used before for other reasons.

Can work as a current adjustor and all sorts.

 

[julesa]

I just took my bike up some nasty hills, but unfortunately by the time I reached the first one it started raining like crazy, and I didn't want to open the battery box to feel whether anything was getting hot. I liked the power boost. The controller was barely warm.

 

[317537]

Nasty hills used to concern me.

I designed this controller box for weather proofing and target ventilation onto the crucial areas.

The red stuff is just insulation tape glue.

 

[317537]

Aluminium is best to use but that looks pretty beefy. And is that nominal voltage on the relay? Relays get stuck too, but that does look nice. I was always considering relays but borked at the nominal voltages the ones that are available.

PWM with a 555 timer, transitors or similar setup to a BMS output would be better for heat dissipation on the outputs.

My red switch is rated at 240v @ 30 amps and hasnt been stuck yet. Simple, reliable and adequate.