Simple FET Driver

ol

us

th.

ir

tch

ve

The input protection circuit that was there to meet unstated requirements i s gone. With the need for a large power source from a previous motor upgra de the connector changed from a barrel connector to a DIN with 7.5 amp rate d contacts, so we are using them in pairs. The likelihood of someone apply ing an incorrect power source becomes much less so the "protection" input c ircuit is now gone. Oddly enough the guy is opposed to adding a simple fus e and seems to want a resettable circuit breaker.

Much of the design was not ready for review and much of the time was spent discussing connectors and pinouts. I asked what the power off current cons umption from the battery was and he didn't have the info. There was no sig nal to provide an indication of presence of input power, just one to indica te that power was within a valid range, but not rated to be valid below 3V. Not much good that!

His approach to the switching circuit is to use the LTC4416 which is a $4 p art. It's also not at all clear to me that it won't have the same shoot th rough current issue. They don't show an example circuit exactly as we are using it with the reverse current flow protection.

I'd have a simple fuse in series with the battery. Fuses do sometimes suffer from the customer mentality of "the fuse nust be faulty because it keeps blowing so I'll put metal foil around it" so some kind of breaker may be better in some cases if it can't be forced on when it doesn't want to be on.

There are some things I don't care about protecting a person from. At some point if stupidity is extreme or deliberate enough, I'm just not going to worry about it.

I've worked on this further. It occurred to me to set up the driver circui t a bit differently. Rather than drive the two sets of pass transistors fr om inverted signals, I thought it might work better to use a differential p air so that as one turns on the other turns off. I managed to get that to work pretty well on the transition when external power comes up, but not as well when power fails. I also realized I was leaving out an important com ponent, the rather large caps on the motor circuit, 2000 uF. That greatly changes the picture with the shoot through but adds another dimension, the surge current in the caps.

So with this simulation the input power up is handled very well with the ca pacitor surge current less than an amp each (two capacitors, so 2 amps tota l). But on removal of input power is not so smooth with up to 3 amps each on the caps and shoot through of 8 amps into the line and 14 amps from the battery. Those may be tolerable numbers given the short duration of 1.5 ms or so.

There is also the power up and down of the unit. My simulation has the ful l motor current on during these times, which is not realistic. Turn on is the worst with 10 amps into the caps (5 each) and the same amount from the line of course. My sim doesn't test this on battery, but I expect the same since it's pretty much the same circuit. Turn off shows 4.5 amps out of t he caps, but that's the motor current which would normally be off when the machine is turned off, so no problem.

Here is the latest version.

Version 4 SHEET 1 1752 916 WIRE 288 -608 272 -608 WIRE -608 -576 -672 -576 WIRE -352 -576 -608 -576 WIRE -96 -576 -352 -576 WIRE 32 -576 0 -576 WIRE 64 -576 32 -576 WIRE 96 -576 64 -576 WIRE 240 -576 192 -576 WIRE 272 -576 272 -608 WIRE 272 -576 240 -576 WIRE 336 -576 272 -576 WIRE 416 -576 336 -576 WIRE 544 -576 512 -576 WIRE 576 -576 544 -576 WIRE 608 -576 576 -576 WIRE 816 -576 704 -576 WIRE 32 -544 32 -576 WIRE 240 -544 240 -576 WIRE 336 -544 336 -576 WIRE 544 -544 544 -576 WIRE 816 -512 800 -512 WIRE -688 -496 -704 -496 WIRE 800 -480 800 -512 WIRE -704 -464 -704 -496 WIRE 240 -464 240 -480 WIRE 336 -464 336 -480 WIRE -608 -432 -608 -576 WIRE -352 -432 -352 -576 WIRE -80 -432 -96 -432 WIRE -16 -432 -16 -528 WIRE -16 -432 -80 -432 WIRE 16 -432 -16 -432 WIRE 32 -432 32 -464 WIRE 32 -432 16 -432 WIRE 112 -432 112 -528 WIRE 112 -432 32 -432 WIRE 432 -432 416 -432 WIRE 496 -432 496 -528 WIRE 496 -432 432 -432 WIRE 544 -432 544 -464 WIRE 544 -432 496 -432 WIRE 624 -432 624 -528 WIRE 624 -432 544 -432 WIRE 416 -400 416 -432 WIRE 592 -384 576 -384 WIRE 720 -384 704 -384 WIRE 800 -384 800 -400 WIRE -704 -368 -704 -384 WIRE 16 -368 16 -432 WIRE 592 -352 592 -384 WIRE 720 -352 720 -384 WIRE 416 -288 416 -320 WIRE 464 -288 416 -288 WIRE 480 -288 464 -288 WIRE -608 -272 -608 -352 WIRE -512 -272 -608 -272 WIRE -352 -272 -352 -352 WIRE -352 -272 -432 -272 WIRE -304 -272 -352 -272 WIRE -256 -272 -304 -272 WIRE -224 -272 -256 -272 WIRE -80 -272 -144 -272 WIRE 16 -272 16 -288 WIRE 64 -272 16 -272 WIRE 80 -272 64 -272 WIRE 16 -256 16 -272 WIRE 416 -256 416 -288 WIRE 592 -256 592 -288 WIRE 720 -256 720 -288 WIRE 720 -256 592 -256 WIRE 736 -256 720 -256 WIRE -352 -240 -352 -272 WIRE -256 -240 -256 -272 WIRE 656 -192 608 -192 WIRE -80 -176 -80 -272 WIRE -32 -176 -80 -176 WIRE 224 -176 192 -176 WIRE 368 -176 224 -176 WIRE 608 -176 608 -192 WIRE -432 -144 -464 -144 WIRE -352 -144 -352 -160 WIRE -80 -144 -80 -176 WIRE -432 -128 -432 -144 WIRE -464 -112 -464 -144 WIRE 224 -112 224 -176 WIRE 256 -112 224 -112 WIRE 368 -112 336 -112 WIRE -608 -96 -608 -272 WIRE -560 -96 -608 -96 WIRE -496 -96 -560 -96 WIRE 368 -96 368 -112 WIRE 496 -96 480 -96 WIRE 560 -96 496 -96 WIRE -384 -80 -432 -80 WIRE -256 -80 -256 -160 WIRE -256 -80 -384 -80 WIRE -608 -64 -608 -96 WIRE -496 -64 -544 -64 WIRE -80 -64 -80 -80 WIRE 608 -64 608 -80 WIRE 16 -48 16 -160 WIRE 112 -48 16 -48 WIRE 368 -48 112 -48 WIRE 416 -48 416 -160 WIRE 416 -48 368 -48 WIRE -464 0 -464 -48 WIRE -400 0 -464 0 WIRE -352 0 -400 0 WIRE -240 0 -288 0 WIRE -224 0 -240 0 WIRE 368 0 368 -48 WIRE 688 16 656 16 WIRE 848 16 768 16 WIRE -608 32 -608 16 WIRE 848 32 848 16 WIRE 48 48 32 48 WIRE 32 64 32 48 WIRE 480 80 416 80 WIRE 544 80 480 80 WIRE 560 80 544 80 WIRE -544 96 -544 -64 WIRE -512 96 -544 96 WIRE -400 96 -432 96 WIRE 848 128 848 112 WIRE 224 144 208 144 WIRE 336 144 304 144 WIRE 368 144 368 96 WIRE 368 144 336 144 WIRE 432 144 416 144 WIRE 544 144 544 80 WIRE 544 144 496 144 WIRE 32 160 32 144 WIRE 208 160 208 144 WIRE 416 160 416 144 FLAG -432 -128 0 FLAG -608 32 0 FLAG -672 -576 Vin IOPIN -672 -576 In FLAG -560 -96 Vth FLAG 432 -432 VbatOn_n FLAG -80 -432 VinOn_n FLAG 288 -608 V_main IOPIN 288 -608 Out FLAG 64 -576 VinSS FLAG 576 -576 VbatSS FLAG 816 -576 Vbatt IOPIN 816 -576 In FLAG -704 -368 0 FLAG -688 -496 Vin IOPIN -688 -496 In FLAG 800 -384 0 FLAG 816 -512 Vbatt IOPIN 816 -512 Out FLAG -400 96 Vref IOPIN -400 96 In FLAG -240 0 Vin FLAG 32 160 0 FLAG 48 48 V_main IOPIN 48 48 In FLAG -304 -272 VinOn FLAG 656 -192 PwrFail_n IOPIN 656 -192 Out FLAG 464 -288 Vbatdrv FLAG 64 -272 Vindrv FLAG 208 160 0 FLAG 480 80 PwrOn_g FLAG 112 -48 SoftStart FLAG 704 -384 Vbatt IOPIN 704 -384 In FLAG 576 -384 Vin IOPIN 576 -384 In FLAG 736 -256 V12_always IOPIN 736 -256 Out FLAG 560 80 PwrOn IOPIN 560 80 In FLAG 240 -464 0 FLAG 336 -464 0 FLAG 656 16 PwrOn IOPIN 656 16 Out FLAG 848 128 0 FLAG 416 160 0 FLAG 192 -176 Vref IOPIN 192 -176 Out FLAG 336 144 Vreg FLAG 608 -64 0 FLAG -80 -64 0 FLAG 368 -96 0 FLAG -384 -80 Vcmp FLAG -400 0 U1pwr FLAG -544 -64 V4Ref FLAG 496 -96 Vcmp FLAG 848 16 PwrOnDrv FLAG -352 -144 0 FLAG -80 -272 Vinon_g SYMBOL nmos 368 -256 R0 SYMATTR InstName M7 SYMATTR Value 2N7002 SYMBOL res -624 -80 R0 WINDOW 0 -11 40 Right 2 WINDOW 3 -11 65 Right 2 SYMATTR InstName R7 SYMATTR Value 27K SYMBOL res -624 -448 R0 SYMATTR InstName R8 SYMATTR Value 47K SYMBOL res -368 -448 R0 SYMATTR InstName R10 SYMATTR Value 10K SYMBOL res 16 -560 R0 SYMATTR InstName R11 SYMATTR Value 2K SYMBOL res -416 -288 R90 WINDOW 0 2 93 VBottom 2 WINDOW 3 1 29 VBottom 2 SYMATTR InstName R12 SYMATTR Value 330K SYMBOL voltage 800 -496 R0 WINDOW 123 0 0 Left 0 WINDOW 39 10 98 Left 2 WINDOW 3 -324 -53 Invisible 2 SYMATTR SpiceLine Rser=0.1 SYMATTR Value PULSE(10 13 0.1s 0.1s 0.3s 0.45s) SYMATTR InstName Batt SYMBOL voltage -704 -480 R0 WINDOW 123 0 0 Left 0 WINDOW 39 -20 -49 Left 2 WINDOW 3 -8 153 Invisible 2 SYMATTR SpiceLine Rser=0.05 SYMATTR Value PULSE(0 12.5 0.2s 0.1s 0.1s 0.3s) SYMATTR InstName In SYMBOL res 16 48 R0 SYMATTR InstName LOAD SYMATTR Value 2 SYMBOL res 400 -416 R0 SYMATTR InstName R14 SYMATTR Value 47 SYMBOL res 528 -560 R0 SYMATTR InstName R16 SYMATTR Value 2K SYMBOL res 0 -384 R0 WINDOW 3 35 66 Left 2 SYMATTR InstName R15 SYMATTR Value 47 SYMBOL schottky 704 -352 R0 SYMATTR InstName D1 SYMATTR Value MBRS140 SYMATTR Description Diode SYMATTR Type diode SYMBOL schottky 576 -352 R0 SYMATTR InstName D2 SYMATTR Value MBRS140 SYMATTR Description Diode SYMATTR Type diode SYMBOL schottky -288 -16 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D3 SYMATTR Value MBRS140 SYMATTR Description Diode SYMATTR Type diode SYMBOL res -416 80 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 1Meg SYMBOL res 784 0 R90 WINDOW 0 -12 72 VRight 2 WINDOW 3 -12 57 VLeft 2 SYMATTR InstName R2 SYMATTR Value 1Meg SYMBOL pmos -96 -528 R270 WINDOW 0 88 16 VRight 2 WINDOW 3 65 -18 VRight 2 SYMATTR InstName M1 SYMATTR Value AOD4185 SYMBOL pmos 192 -528 M270 WINDOW 0 87 62 VLeft 2 WINDOW 3 66 75 VLeft 2 SYMATTR InstName M2 SYMATTR Value AOD4185 SYMBOL pmos 416 -528 R270 WINDOW 0 88 38 VRight 2 WINDOW 3 65 -21 VRight 2 SYMATTR InstName M3 SYMATTR Value AOD4185 SYMBOL pmos 704 -528 M270 WINDOW 0 83 66 VLeft 2 WINDOW 3 63 74 VLeft 2 SYMATTR InstName M4 SYMATTR Value AOD4185 SYMBOL nmos -32 -256 R0 SYMATTR InstName M5 SYMATTR Value 2N7002 SYMBOL nmos 416 0 M0 WINDOW 0 62 34 Left 2 WINDOW 3 61 8 Left 2 SYMATTR InstName M8 SYMATTR Value 2N7002 SYMBOL Comparators\\LTC1841 -464 -80 M180 WINDOW 0 28 -18 Left 2 WINDOW 3 44 82 Right 2 SYMATTR InstName U1 SYMBOL polcap 224 -544 R0 WINDOW 3 24 56 Left 2

SYMATTR InstName C1 SYMATTR Description Capacitor SYMATTR Type cap SYMATTR SpiceLine V=35 Irms=1.95 Rser=0.03 Lser=0 mfg="Panasonic" pn="ECA1VFQ102L" type="Al electrolytic" SYMBOL polcap 320 -544 R0 WINDOW 3 24 56 Left 2

SYMATTR InstName C2 SYMATTR Description Capacitor SYMATTR Type cap SYMATTR SpiceLine V=35 Irms=1.95 Rser=0.03 Lser=0 mfg="Panasonic" pn="ECA1VFQ102L" type="Al electrolytic" SYMBOL voltage 848 16 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 WINDOW 3 -16 101 Invisible 2 WINDOW 0 -25 84 Right 2 SYMATTR Value PULSE(3.3V 0V 0.4S 1us 1us 99.998ms 0.5s) SYMATTR InstName PwrOn SYMBOL cap 496 128 R90 WINDOW 0 -11 41 VRight 2 WINDOW 3 -12 29 VLeft 2 SYMATTR InstName C3 SYMATTR Value 22nF SYMBOL res 320 128 R90 WINDOW 0 -13 75 VRight 2 WINDOW 3 -13 56 VLeft 2 SYMATTR InstName R3 SYMATTR Value 330 SYMBOL nmos 560 -176 R0 SYMATTR InstName M9 SYMATTR Value 2N7002 SYMBOL res -272 -256 R0 SYMATTR InstName R4 SYMATTR Value 3.6K SYMBOL cap -96 -144 R0 SYMATTR InstName C5 SYMATTR Value 150nF SYMBOL voltage 240 -112 R270 WINDOW 123 0 0 Left 0 WINDOW 39 -20 -49 Left 2 WINDOW 3 37 76 VRight 2 WINDOW 0 37 38 VLeft 2 SYMATTR Value 4V SYMATTR InstName Vref SYMBOL res -368 -256 R0 SYMATTR InstName R5 SYMATTR Value 10K SYMBOL res -128 -288 R90 WINDOW 0 -12 72 VRight 2 WINDOW 3 -12 57 VLeft 2 SYMATTR InstName R6 SYMATTR Value 100K TEXT 160 -384 Left 3 !.tran 1sec TEXT -552 -544 Left 2 ;Nominal range\n11.4 < Vin < 12.6 TEXT -416 -64 Left 2 ;11.2V rising, 10.6V falling TEXT 656 -160 Left 3 ;To Control Board TEXT 560 160 Left 3 ;From Control Board

The outputs of 4000B series CMOS logic hates have an impedance of about 400 ohms, and can source a milliamp. To control 10 amps, the beta would need to be 10,000. A darlington pair maybe, but a driver stage seems simpler and more robust.

Joe Gwinn

the

two

to

e on

cuit.

is

much

OS

ing

w too

rying

it,

where

what

wer

hard

s a switch over from line to battery when input power fails. Each switch is two pFETs to prevent back feed of current. The battery is 12V nominal SLA and the input is a 12V DC for running a motor with 10 amp peaks and 4 amp n ominal. I use a comparator on the input DC to detect when it is absent whic h switches between the two power sources. Using transistors with pull ups m eans the pull up is slow and there are some 50 us where current passes thro ugh both switches between the two power sources. Maybe this isn't much of a real problem, but 15 or more amps seems like something to minimize the tim e duration of if the current itself can't be minimized.

sition time will be much shorter and the issue of shoot through current wil l be less significant. But I'm probably showing my ignorance of what is nee ded for this circuit to work well. One guy on the team has picked a LT part at $4 to control the pFETs. It's a nice part, but when was an LT part ever cheap? It just doesn't seem like a hard circuit for an experienced power d esigner to pull off.

Beta applies to a junction transistor. I'm using FETs which are typically characterized by transconductance. Since the CMOS device can provide the full voltage swing there should be no problem controlling the maximum curre nt the FET can handle. The current can be an issue with switching speed ch arging the gates of the pass FETs, but the present design is using a 5 mA c urrent limiter using a reference voltage on the gate of a FET and a source leg resistor of 330 ohm. Then there is a 47 ohm series resistor after the controlling FETs, so nearly 400 ohms.

If the circuit could be timed appropriately, a 400 ohm driving resistance w ould be very useful. The problem I'm seeing is from the difficulty of cont rolling the switching on vs. off of the two paths. Enable one too early an d there is shoot through, enable too late and there is a gap during which t he voltage droops. Add in large capacitors on the load and too fast of a s witching time and the capacitor sees large surge currents from the differen t voltages, especially when being turned on.

I'm wondering if any of this is avoided by using the "magic" power path con troller from LT. I bet it has the same problems.

There is a reason why regular FET driver ICs can source and sink some several amps.

I have seen CD4000 hex drivers or hex inverters used for FET drive with ALL sections paralleled. That can also help speed up that gate drive.

If your FET is small though and especially if you aren't going to switch it fast, then 1 mA might be plenty. Maybe.

It's perilous, though, to feed a capacitive load that way; the bond wires on CMOS logic might explode if you go over 110 mA (or whatever the package limit is nowadays). The power and ground wiring for a CMOS logic chip aren't busbars....

Are you suggesting there would be a current significantly higher than V/R? With a 400 ohm output resistance I would expect that to be sufficient to prevent an explosion. Let's see, 12V/400 = 30 mA.

Am I missing the point of your post?

This is not a switching power supply. These events will happen with an expected frequency of 1 microhertz or so.

Well, yes. CMOS, in particular, can go to 18V, and at low temperature, R isn't 400 ohms either. I've blown up wires with 4000 series CMOS, in a liquid-nitrogen level monitor, because... it did get cold, and a 2n2222 only has emitter wire for 500 mA ( it was the level-shifter that applied 12V power). Opened the (dead transistor) can, and the wire was completely gone/evaporated. The cold plastic-package chip survived, though.

It's not necessarily heat that does it in, but shock; dI/dt, not I^2.

ires

/R? With a 400 ohm output resistance I would expect that to be sufficient t o prevent an explosion. Let's see, 12V/400 = 30 mA.

isn't 400 ohms

evel monitor,

( it was

an, and

rvived, though.

expected frequency of 1 microhertz or so.

Ok, we will put in the spec that this ventilator should not be used in liqu id nitrogen.

I'm sorry, I have no freaking idea what you are talking about. Why are you talking about liquid nitrogen temperatures? This device may be used at 20

So what exactly is your concern?

te

wires

R isn't 400 ohms

level monitor,

mA ( it was

can, and

survived, though.

u talking about liquid nitrogen temperatures? This device may be used at 20

If one parallels six inverter outputs to do a drive, the power pins on the IC see a high transient current that (with all six outputs driving a near-short) can crea te mechanical stress that damages the wire. The old ULN2003 is infamous for having a pa ckage limit that means you can't use the full drive of all the outputs at once.

ive

gate

nd wires

, R isn't 400 ohms

en level monitor,

r) can, and

p survived, though.

you talking about liquid nitrogen temperatures? This device may be used at

e IC see a high

eate mechanical

ckage

I was not planning to parallel all six inverters. I was planning to use on e to invert the signal while the other two are drivers from the original si gnal and the inverted signal. So one driving up and one driving down.

I'm not trying to drive anything fast at all. In fact, I want to have a co ntrolled transition so there are not large surges through the power FETs. That's the reason for using drivers that are relatively even handed with th e pullup and pulldown. 400 ohms is great being a good value to limit the c urrent. Gives the circuit some time to turn off and on.

I don't think this circuit is going to be used though. The guy doing the b oard is in love with Linear Technology and uses their parts for everything. A 50 cent buck switcher is being done with a $2 LT switcher. He has a $3

-$4 power path control chip in the circuit. The total battery drain when off is not as low as it should be, but I don't think he's going to get rid of it because of that. His approach doesn't turn off all power to the boar d because I think the chip can't do that. I also don't see anything in the data sheet that indicates the design will prevent shoot through when switc hing.

Hmmm.... I just thought of a simplification to the circuit. A Schotky diod e of sufficient capacity would handle the line input side even if it does d rop a bit more voltage than the FETs will. It won't allow shoot through at all in that direction. There is still the issue of the capacitor surge ho wever, but only when the battery voltage is lower than the line minus the d iode drop.

But how do you know if it is an issue at all? How many amps are you switching?

There is another LT power prioritiser that modulates V_gate in order to keep the switching currents sane, forgot the number, 14-something.

Best regards, Piotr

$4 part. It's also not at all clear to me that it won't have the same shoot through current issue.

The currents depend heavily on the details of the circuit in a way that I d on't trust the simulation to be accurate.

The currents tend to be from the battery into the external 12V PSU or from the caps to the PSU or from the battery. I'm willing to bet the guy design ing the circuit has not looked at these currents at all. I don't see an LT spice model for the 4416 or I would run the simulation myself. Shoot! I j ust looked harder and rather than power they have it under "SpecialFunction s". So I'll give this circuit a test run and see what it does.

So how would they measure the currents? I guess the voltages across the FE Ts? The internal workings of the chip are a bit complex.

So I ran the simulation with the LTC4416 and the huge capacitor charging cu rrent is still there. So much for the purported soft start. 16 amps into the cap when the input power drops and the battery is fully charged at 13 V . So the battery is the irresistible force and the capacitor is the immova ble object?