Low RDSon Logic CMOS Gate

[Klaus Kragelund]

Hi

I need a CMOS gate with low voltage drop at about 30mA current source and sink. The tiny logic NL27WZ14 has 600mV drop at 32mA which equates to 18ohm RDS on.

I am searching for a device with less than 5-10 ohms and it would need to be ok to parallel to get lower resistance. Also, it must not have to much crossover shoot-through (so probably schmitt trigger type) and must have low dissipation at high operational frequency (1-10MHz)

Any one have a part in mind?

Thanks

Klaus

 

[hamilton]

Massive shoot-through.

Please describe shoot-through.

 

[Klaus Kragelund]

I don't think that will fix shoot-thru. 'HC14 only fixes slow inputs,

so maybe the simple-minded solution is to simply slow down the input

to get firm turn-off of drive-low before drive-high (and vice-versa)

is activated... say take 10-20ns to get thru the hysteresis ?



Or maybe a 1/2-H bridge driven by non-overlapping drives... as

demonstrated recently with my delay block?

Been through some posts, but did not find it. Can you point me in the rightdirection?

Regards

Klaus

 

[Fred Bartoli]

Klaus Kragelund a écrit :

Hi

I need a CMOS gate with low voltage drop at about 30mA current source and sink. The tiny logic NL27WZ14 has 600mV drop at 32mA which equates to 18ohm RDS on.

I am searching for a device with less than 5-10 ohms and it would need to be ok to parallel to get lower resistance. Also, it must not have to much crossover shoot-through (so probably schmitt trigger type) and must have low dissipation at high operational frequency (1-10MHz)

Any one have a part in mind?

Thanks

Klaus

Mucho thanks for posting this...

In an attempt to answer I had a look at a recent design and noticed that
a 5V powered LVC14 escaped all the design reviews :-(

It survived all the stress test: 27MHz switching, with PCB temp=100°C
while driving a 250pF Ciss Mosfet...

How did you know ?-)

 

[Klaus Kragelund]

Use something like an HC14, with one section buffering the input, then

driving several other sections, up to 5, in parallel.



More extreme on speed - sub ns edges - would be an NL37WZ17US, again

with one section buffering the input to drive the other two.

I found maybe a slightly better one, the NC7SZ14. The device lists CPD of 24 (Power dissipation capacitance, but no information on the unit. Icc = Cpd * V * f.

(page 5 of http://www.fairchildsemi.com/ds/NC/NC7SZ14.pdf)

I guess it's in pC (pico coloumb), which corresponds to 100uA/MHz.

Regards

Klaus

 

[whit3rd]

I need a CMOS gate with low voltage drop at about 30mA current source and sink. The tiny logic NL27WZ14 has 600mV drop at 32mA which equates to 18ohm RDS on.

I am searching for a device with less than 5-10 ohms

One can use discrete complementary MOSFETs to make a CMOS-like inverter,
of course. Ferrite beads can help with the shoot-through problem, your
frequency of interest seems low enough not to care. It may matter what
voltage you're powering from, as well. At 5V, there's more options than
at 2V.

 

[Klaus Kragelund]

This is the way I generate precise delays (on-chip)...



http://www.analog-innovations.com/SED/DelayCircuitForNarrowPulseWidths.pdf



This delayed signal is then used to create non-overlapping drives for

such things as full-H bridges, and commutating switches used in

synchronous rectification and integrate control loops.



Analysis is left as an exercise for the student ;-)



Hints:



(1) These are 10ps inverters (TSMC 0.18u process)

(2) This is internal to a monolithic chip, so no ESD to get in your

way, so left end of the cap flys above VDDD and below GNDD without

clamping or consequence.



This snap-shot is from a chip I designed last fall when I did an

extended stay -) on Long Island and met Martin Riddle.



Designed entirely on my laptop, the chip worked perfectly to

specifications first pass thru the foundry, as do ALL of my designs...

I never do "designs" without component values >



Klaus, I'll see if I can excise a clip showing how it inputs into the

H-bridge without breaking any NDA restrictions.



Is your posted E-mail address valid?

Hi Jim

Thank you for taking time to elaborate on this ;-)

The email is:

[email protected]

It is valid and hotmail spamfilter works, I hardly get spam any more...

Regards

Klaus

 

[Klaus Kragelund]

Klaus Kragelund a écrit :




Mucho thanks for posting this...



In an attempt to answer I had a look at a recent design and noticed that

a 5V powered LVC14 escaped all the design reviews :-(



It survived all the stress test: 27MHz switching, with PCB temp=100°C

while driving a 250pF Ciss Mosfet...



How did you know ?-)

That's a secret ;-)

Funny how some circuit work when they are not supposed to and other circuits that look clean and nice has "gotchas"....

Cheers

Klaus

 

[Klaus Kragelund]

One can use discrete complementary MOSFETs to make a CMOS-like inverter,

of course. Ferrite beads can help with the shoot-through problem, your

frequency of interest seems low enough not to care. It may matter what

voltage you're powering from, as well. At 5V, there's more options than

at 2V.

I have 3.3V, can make 5V if I need to. This is for a high efficiency device, so any powerloss is critical. Shoot through if a FET pair can easily go up to several amps, which won't be noticed if you are not looking.

Cheers

Klaus

 

[Fred Bartoli]

Klaus Kragelund a écrit :

I found maybe a slightly better one, the NC7SZ14. The device lists CPD of 24 (Power dissipation capacitance, but no information on the unit. Icc = Cpd * V * f.

(page 5 of http://www.fairchildsemi.com/ds/NC/NC7SZ14.pdf)

I guess it's in pC (pico coloumb), which corresponds to 100uA/MHz.

Regards

Klaus

The unit is pF.
Multiply by 1/2 Vsupply^2 and you have the energy quantum lost at each
switching.

Total power loss is PL = f x Cpd x Vsupply^2

 

[Nico Coesel]

Hi

I need a CMOS gate with low voltage drop at about 30mA current source and s=
ink. The tiny logic NL27WZ14 has 600mV drop at 32mA which equates to 18ohm =
RDS on.

Maybe look for a MOSFET driver chip.

 

[[email protected]]

Hi

I need a CMOS gate with low voltage drop at about 30mA current source andsink. The tiny logic NL27WZ14 has 600mV drop at 32mA which equates to 18ohm RDS on.

I am searching for a device with less than 5-10 ohms and it would need tobe ok to parallel to get lower resistance. Also, it must not have to much crossover shoot-through (so probably schmitt trigger type) and must have low dissipation at high operational frequency (1-10MHz)

Any one have a part in mind?

Thanks

Klaus

http://www.intersil.com/en/products...ches/switches-muxs-crosspoints/ISL43L210.html

(from John Devereux, earlier this year)

Break-before-make.

DigiKey had competing parts, too.

 

[Jamie]

But look at the capacitances! There's no free lunch.

Talking about Digikey, they getting any better at stocking more of what
they advertise? I have been leaning towards the other parts suppliers
lately..

Jamie

 

[[email protected]]

Maybe look for a MOSFET driver chip.

that was my first thought, something like fan3111, but..

datasheet doesn't say what the dropout is, it maybe a bit slow for
10MHz and who
knows how much power it will use

maybe a buffer/linedriver

-Lasse

 

[[email protected]]

If a CMOS gate's input is roughly midway between Vcc and ground, both
p-channel and n-channel fets can be partially turned on in the front
end. Current "shots through" the fets from Vcc to ground. That can
draw a lot of supply current and potentially heat up the chip. It
always happens briefly when a normal, fast input transitions, but it
can get much worse if an input transitions slowly, or parks at an
intermediate voltage. Schmitt-trigger gates are designed for slow
inputs, so generally have moderate peak shoot-through currents.

Paralleling Schmitts is interesting. At some input voltage, some of
the gates will see a "1" and some may see a "0", so their paralleled
outputs will fight one another and pull a *lot* of supply current.
Running the signal through one Schmitt section, and then driving a
bunch more sections in parallel, is safer.

We almost fried some Tiny Logic gates, powering them from +5 but
driving them from 3.3 volt logic from an FPGA. They got really hot.

HC or HCT ?

http://www.ti.com/lit/ds/symlink/sn74hct04.pdf
http://www.nxp.com/documents/data_sheet/74HC_HCT04.pdf

TI <3mA per pin at 2.4Vin
NXP <590uA at 2.9Vin

that shouldn't be a problem driven with a 3.3V output

-Lasse

 

[whit3rd]

I have 3.3V, can make 5V if I need to. This is for a high efficiency device, so any powerloss is critical. Shoot through if a FET pair can easily go up to several amps, which won't be noticed if you are not looking.

A dual MOSFET (P-channel/N-channel pair) like Rohm EM6M2 can easily handle the task.
If you add a couple of 2 ohm drain resistors, it limits the shoot-through to well under one
amp, but the output resistance only gets two ohms drop (still well below your 5 ohm
goal for 3.3V power supplies). Note the PFET has rather high capacitance, that's
required to get the low on-resistance.

For best shoot-through limit, it gets complicated (you need something more than
a logic drive, it takes turn-on delayed but turn-off prompt).

 

[Klaus Kragelund]

A dual MOSFET (P-channel/N-channel pair) like Rohm EM6M2 can easily handle the task.

If you add a couple of 2 ohm drain resistors, it limits the shoot-through to well under one

amp, but the output resistance only gets two ohms drop (still well below your 5 ohm

goal for 3.3V power supplies). Note the PFET has rather high capacitance, that's

required to get the low on-resistance.

...Snip..

The EM6M2 has 1nC of gate charge. Running that at 5V and 200kHz will consume 1mW. I was trying to reduce the gate charge losses by using a logic IC

On that subject, anyone know the value of the gate charge for a run-of-the-mill gate inside a chip?

Regards

Klaus

 

[[email protected]]

But look at the capacitances! There's no free lunch.

For sure there are higher Rds(on) parts with lower capacitances,
right? I was just throwing out an extreme example, since I'm an
extremist .

I suspect the make-before-break convenience, integration, and
capacitances in other switches will be attractive.

I didn't have time to screen parts--I mostly have to post and run
these days.

James

 

[[email protected]]

This problem is greatly overstated.
While ago I tested static cross conduction of HCT04 gate powered at +5V.
The worst case was about 4mA, at 0.9V at the input.

Vladimir Vassilevsky
DSP and Mixed Signal Consultantwww.abvolt.com

I've destroyed 74AC parts that way. Smoked 'em.

 

[[email protected]]

I've destroyed 74AC parts that way. Smoked 'em.

HCT stuff isn't tuned for speed so has little crossover current (none,
ideally). AC logic is a whole different kettle. Better decouple AC gates
well. ;-)