How to quickly turn on/off N-FET switching high side of 55V?

[Winfield]

Well, two batteries can deliver current at one time if they're at the
exact same voltage, but yes uninterrupted current flow is necesarry.
And yes, it'd be problematic if one battery started charging another,
especially at 500A (the batteries I'm using are rated for 20A
continuous, 30A for up to 10 seconds, though I'm not sure about how
they handle really large transients). It's unfortunate that I have to
throw some power away in a Schottky, but as far as I can see my
options are very limited, unless I'm missing something. Doing source
to source FETs will require incredibly accurate timing along with a
large capacitor to make uninterrupted, again unless I'm missing
something. Thus it seems like the Schottky + single FET is the best
solution?

Here's one other thought that just struck me: what if I had two FETs
source to source as discussed. However, what if I put in place a
switching scheme so that the top (connected to battery) switch was
switched separately from the bottom FET, and that the bottom FETs are
driven in such a way that only one or less of them can be on at any
given time. Additionally, there would be a Schottky in parallel with
each bottom FET. The idea is that by placing Schottkys in parallel
with the FETs hot swapping would be enabled, but the second FET could
be switched on a moment after the other bottom FETs were switched off,
decreasing power loss.

Does this make any sense at all?

Thanks,

-Michael

Yes, it makes perfect sense. Schottky diodes to insure
a continuous voltage availability (Fred's suggestion of
an inductor would do that as well, but would require you
to consider the flyback voltages from the stored energy
in the inductor during switching -- might be reasonable),
and the paralleled MOSFET switched on later, when it's
safe, to reduce heating.

The setup would be a bit different than the one in Fred's
drawing following your post.

.. F1 d___s SD1
.. -- BATT1 >--/---|___|--+---|>|---+---+-------o buss
.. | FET-A | | s___d | |
.. | | '--|___|--' | LOAD
.. | S1A | FET-B |
.. | S1B | o
.. | | |
.. | F2 ___ SD2 | GND
.. +-- BATT2 >--/---|___|--+---|>|---+---+
.. | FET | | ___ | |
.. | | '--|___|--' |
.. | S2A | |
.. | S2B |
.. | |
.. | F3 SD3 |
.. +-- AC POWER >--/-----------|>|-------'
.. |
.. GND

The primary issues are figuring out when to switch the "B"
MOSFETs (e.g., turn on a delay-time after the "A" mosfets,
but **only** if V_BattN is > than V_battM, and turn off
immediately with "A" mosfets), and devising the appropriate
gate-drive voltage (the same as for the "A" mosfet). This
extra trouble would save you a little more than 10.5 - 6.3
= 4 watts, call it 5 watts, over wiring 4 sections of two
60CTQ045 diodes in parallel. I dunno if it's worth it.

This looks so much more simple:

.. F1 d___s SD1
.. -- BATT1 >--/----|___|-----|>|---+-------o buss
.. | FET | |
.. | S1A | LOAD
.. | |
.. | F2 ___ SD2 | o
.. +-- BATT2 >--/----|___|-----|>|---+ |
.. | | | GND
.. | S2A |
.. | |
.. | F3 SD3 |
.. +-- AC POWER >--/----------|>|----'
.. |
.. GND

 

[Fred Bloggs]

Yes, it makes perfect sense. Schottky diodes to insure
a continuous voltage availability (Fred's suggestion of
an inductor would do that as well, but would require you
to consider the flyback voltages from the stored energy
in the inductor during switching -- might be reasonable),
and the paralleled MOSFET switched on later, when it's
safe, to reduce heating.

The setup would be a bit different than the one in Fred's
drawing following your post.

. F1 d___s SD1
. -- BATT1 >--/---|___|--+---|>|---+---+-------o buss
. | FET-A | | s___d | |
. | | '--|___|--' | LOAD
. | S1A | FET-B |
. | S1B | o
. | | |
. | F2 ___ SD2 | GND
. +-- BATT2 >--/---|___|--+---|>|---+---+
. | FET | | ___ | |
. | | '--|___|--' |
. | S2A | |
. | S2B |
. | |
. | F3 SD3 |
. +-- AC POWER >--/-----------|>|-------'
. |
. GND

The primary issues are figuring out when to switch the "B"
MOSFETs (e.g., turn on a delay-time after the "A" mosfets,
but **only** if V_BattN is > than V_battM, and turn off
immediately with "A" mosfets), and devising the appropriate
gate-drive voltage (the same as for the "A" mosfet). This
extra trouble would save you a little more than 10.5 - 6.3
= 4 watts, call it 5 watts, over wiring 4 sections of two
60CTQ045 diodes in parallel. I dunno if it's worth it.

This looks so much more simple:

. F1 d___s SD1
. -- BATT1 >--/----|___|-----|>|---+-------o buss
. | FET | |
. | S1A | LOAD
. | |
. | F2 ___ SD2 | o
. +-- BATT2 >--/----|___|-----|>|---+ |
. | | | GND
. | S2A |
. | |
. | F3 SD3 |
. +-- AC POWER >--/----------|>|----'
. |
. GND

Now he's saying it's 10A continuous. This is nothing for an SCR, maybe
0.1V x 10A=1W additional Pd over an SD, if that, vastly simplified
transient triggering, and they shut themselves off. A single MOSFET to
the load controls load power as well as SCR commutation when it turns off.

 

[Winfield]

Yes, but John has the waveform correct, with the addition
of the gate capacitance, the voltage will be the integral
of the charging-current pulses through the gate resistor's
relay-contact connected time.

Well, no, not if the gate resistor is large enough to make
nice slow turn-on waveforms, e.g. 15V/us as Fred suggested.

Here are some pics of the behavior of a tiny Omron surface-mount
relay, an equivalent to the Fujitsu part.

The first "on" bounce is consistantly about 5 usec wide, 50x or so
longer than needed to fully charge a gate. Turnoff is interesting...
a brief increase in resistance, then a fall, but no bounce.

Contact risetime is very clean and under 10 ns, probably limited by
my setup. ftp://66.117.156.8/Omron.zip

With different drive, pullin/dropout times could be reduced.

Well, the issue is the gate resistor. I doubt we'd like
a value small enough to get ns gate-voltage risetime, but
if 5us of contact can be guaranteed, then even 15V/us will
be fast almost enough, after accounting for all the gate
charge. Myself, I'd prefer say 5x faster yet, which means
the first relay-contact tick should fully charge the gate.

 

[Winfield]

Thanks, John. Up to ~ 65K operations in a holiday
(unattended, worst case) weekend, so that one is a
no go. Interestingly - I had already started on an SSR
replacement board, but haven't found the reliability
specs for switching, so I'll have to test it myself.
No big rush - the G2R serves well for now. But one
of those tiny ones like the Fujitsu sure would be nice.
Fortunately, it's not mandatory.

65k operations is not that many for a relay, and some
relays, such as the NAiS NR long-life series, specify
rapid operation, 500 operations/sec for the NR types.
But, whew, all the noise! All the mechanical mashing!

You might consider optical-coupled MOSFETs, in a nice
miniDIP package, for about five bucks. I especially
like the Panasonic (formerly Mitsubishi NAiS) parts,
which they call PhotoMOS. Clare has their OptoMOS,
and Toshiba, omron, CEL and IR make good stuff as
well (IR was an early leader but hasn't introduced
much in the last few decades). DigiKey shows a nice
selection, pp1979-1987 in the Sept-Dec 2007 catalog.

On of the "largest" photoMOS relays is in Panasonic's
AQV series: the AQV251 has an ON resistance of 0.25 to
1 ohms (depending on whether you need to switch ac or
not), and is rated to about an amp (it uses big internal
MOSFETs). The higher-current parts have lower voltage
ratings, only 40V in the case of the AQV251, so that
might be an issue.

You can also roll your own opto-relay, with MOSFETs of
your choice, high-current and high-voltage, using the
infra-red-LED to PD-stack parts made by IR, etc., as
we discussed earlier in this thread.

 

[Rich Grise, Plainclothes Hippie]

OK, can you cite me touting any panaceas lately?

The first time I read this, it came out, "can you cite me touting
my pancreas lately?"

;-)
Rich

 

[Rich Grise]

All true, but how do you expect John Fields to grasp the idea?

Win Hill explained it in his post - it seems John was illustrating
the risetime of the first 'tick', and included the effects of the
gate resistor, which I didn't.

So he's not wrong - I misinterpreted his diagram and flew off into
my discourse. ;-)

Now, I'm not sure about the status of the "I know more about contact
bounce than you do" part of the thread. ;-)

Thanks,
Rich

 

[Paul Hovnanian P.E.]

Hi John - I don't have a desired slew rate. I just want the switch to
turn on/off without exploding. Sorry for not being clear about this
earlier.

-Michael

Not explode? Well then, you've come to the wrong newsgroup. ;-)

Anyone have a solution that uses an optoisolator? Maybe a bit of an
overkill for 55V.

 

[Fred Bloggs]

Yes, it makes perfect sense. Schottky diodes to insure
a continuous voltage availability (Fred's suggestion of
an inductor would do that as well, but would require you
to consider the flyback voltages from the stored energy
in the inductor during switching -- might be reasonable),
and the paralleled MOSFET switched on later, when it's
safe, to reduce heating.

Or he could use a big capacitor to hold up the load like so, giving the
turning off sw a 10us headstart on the turning on sw:
View in a fixed-width font such as Courier.

 

[Winfield]

I found this self-boost charge pump topology for
high side drive in integrated circuit form:
View in a fixed-width font such as Courier.
.
. HV
. |
. D1 D2 |-
. 12.6V>----+-----|>|--+-----------+-|>|--+----- .-||
. | | | |+ | | |>
. | | [Rpu] === DRIVER--+ |
. | | | C2 | | | | |
. | | .-------------+-----+-----|----+
. | | | | | | |
. | |+ -| | | | |
. | C1=== Q2 ||---+ | | |
. | | <| | | | |
. | | | | | | |
. | '----+ | | | |
. | | | | | |>
. | - | | '-||
. | D3 v | | |-
. | - | | |
. | | | | +---> out
. +-+LMC555 +------' | |
. | | | | -
. | | 10KHz |- IN>---' ^
. | |----------|| Q1 |
. | | |> |
. | | | ---
. +---+ --- pgnd
. | sgnd
. ---
. sgnd

That's an interesting circuit, where'd you find it?
It solves the problem of making a capacitor-coupled
flying V+ gate-drive voltage, for the awkward case
of when the destination voltage's V- is floating.

. Q1,Q2 SSN1N45B

What's your fascination with the ssn1n45b mosfet?

 

[Fred Bloggs]

Fred Bloggs wrote:

I found this self-boost charge pump topology for
high side drive in integrated circuit form:
View in a fixed-width font such as Courier.
.
. HV
. |
. D1 D2 |-
. 12.6V>----+-----|>|--+-----------+-|>|--+----- .-||
. | | | |+ | | |>
. | | [Rpu] === DRIVER--+ |
. | | | C2 | | | | |
. | | .-------------+-----+-----|----+
. | | | | | | |
. | |+ -| | | | |
. | C1=== Q2 ||---+ | | |
. | | <| | | | |
. | | | | | | |
. | '----+ | | | |
. | | | | | |>
. | - | | '-||
. | D3 v | | |-
. | - | | |
. | | | | +---> out
. +-+LMC555 +------' | |
. | | | | -
. | | 10KHz |- IN>---' ^
. | |----------|| Q1 |
. | | |> |
. | | | ---
. +---+ --- pgnd
. | sgnd
. ---
. sgnd

That's an interesting circuit, where'd you find it?
It solves the problem of making a capacitor-coupled
flying V+ gate-drive voltage, for the awkward case
of when the destination voltage's V- is floating.

Exactly, the high side reference node can be at any impedance..

Park & Jahns,"A Self-Boost Charge Pump Topology for a Gate Drive
High-Side Power Supply," IEEE, 2003.
"... features high voltage and current capabilities for use in
Integrated Power Electronic Modules (IPEMs)."
There is also an adaptation for a bipolar boosted drive.

What's your fascination with the ssn1n45b mosfet?

Nothing in particular, found it on Fairchild's site, and available from
Mouser for $0.60 in unit quantity, in stock, and TO-92.

 

[ehsjr]

65k operations is not that many for a relay, and some
relays, such as the NAiS NR long-life series, specify
rapid operation, 500 operations/sec for the NR types.
But, whew, all the noise! All the mechanical mashing!

You might consider optical-coupled MOSFETs, in a nice
miniDIP package, for about five bucks. I especially
like the Panasonic (formerly Mitsubishi NAiS) parts,
which they call PhotoMOS. Clare has their OptoMOS,
and Toshiba, omron, CEL and IR make good stuff as
well (IR was an early leader but hasn't introduced
much in the last few decades). DigiKey shows a nice
selection, pp1979-1987 in the Sept-Dec 2007 catalog.

On of the "largest" photoMOS relays is in Panasonic's
AQV series: the AQV251 has an ON resistance of 0.25 to
1 ohms (depending on whether you need to switch ac or
not), and is rated to about an amp (it uses big internal
MOSFETs). The higher-current parts have lower voltage
ratings, only 40V in the case of the AQV251, so that
might be an issue.

You can also roll your own opto-relay, with MOSFETs of
your choice, high-current and high-voltage, using the
infra-red-LED to PD-stack parts made by IR, etc., as
we discussed earlier in this thread.

Thanks, Win. I gave them a quick look, and at least
one looks like a good possibility, certainly worth
close consideration. Nice call!

Ed

 

[IanM]

You can save yourself a lot of circuit design by going with a P-channel,

then the drive is a simple TTL-> HV current source-> V+ referenced MOSFET
gate. Digikey has these in stock for $5.40:
http://www.vishay.com/docs/73474/sum90p10.pdf
The mounting will be non-trivial.

hello,

why use the N-FET whan you can do as Fred suggests here?

regards

Ian

 

[Winfield Hill]

Fred Bloggs wrote...

hello,

why use the N-FET whan you can do as Fred suggests here?

I agree, using a P-channel mosfet would be great - far
easier, if a good enough one can be found. Generally
one gets better N-channel types and often doesn't even
bother looking for an acceptable P-channel part. The
parts Fred has found aren't bad, but as he said, since
they come in TO-263 (D2PAK)packages, the mounting to
obtain optimal heat removal will be non-trivial.

I have ST's stp80pf55 mosfets in my stock drawers.
These may be a better choice, because they come in
TO-220 packages, easy to clip a heatsink onto, and
cost only $2.46 at DigiKey. Its typical rRds(on) is
16*1.2 = 20 milli-ohms at 100 deg-C, so at 10 amps
it'd dissipate under 2 watts, assuming a small 3W
clip-on heatsink. But if it was meant to switch 30A
for more than a second or two, I'd use two or three
of these 80pf55 mosfets in parallel. However, even
at that rate it would be much more simple than the
other approaches we've been discussing here.

stp80pf55
fuse diodes
batt --X----+--- S D ----|>|----+--- bus
| G |
2.7k | :
| | etc
+------'
|
TTL C
------- B
E 2n5551
|
2.2k
|
gnd

Note: paralleled mosfets should have individual
gate resistors, say 100 ohms each.

 

[gearhead]

 I agree, using a P-channel mosfet would be great - far
 easier, if a good enough one can be found.  Generally
 one gets better N-channel types and often doesn't even
 bother looking for an acceptable P-channel part.  The
 parts Fred has found aren't bad, but as he said, since
 they come in TO-263 (D2PAK)packages, the mounting to
 obtain optimal heat removal will be non-trivial.

 I have ST's stp80pf55 mosfets in my stock drawers.
 These may be a better choice, because they come in
 TO-220 packages, easy to clip a heatsink onto, and
 cost only $2.46 at DigiKey.  Its typical rRds(on) is
 16*1.2 = 20 milli-ohms at 100 deg-C, so at 10 amps
 it'd dissipate under 2 watts, assuming a small 3W
 clip-on heatsink.  But if it was meant to switch 30A
 for more than a second or two, I'd use two or three
 of these 80pf55 mosfets in parallel.  However, even
 at that rate it would be much more simple than the
 other approaches we've been discussing here.

               stp80pf55
        fuse               diodes
 batt --X----+--- S   D ----|>|----+--- bus
             |      G              |
            2.7k    |              :
             |      |             etc
             +------'
             |
  TTL        C
   ------- B
             E  2n5551
             |
            2.2k
             |
            gnd

 Note: paralleled mosfets should have individual
 gate resistors, say 100 ohms each.

Caveat: In that mosfet's datasheet, figure 6 "static drain-source on
resistance" has the wrong graph.
It's actually the graph for diode forward behavior, same as figure 11.

 

[Winfield]

Caveat: In that mosfet's datasheet, figure 6 "static
drain-source on resistance" has the wrong graph.
It's actually the graph for diode forward behavior,
same as figure 11.

Yes, good catch. I used figure 10 to get the Rds(on)
multiplier value for 100C. Even so, a 1.2 multiplier
at 100C seems a little low, IIRC.

Still, the stp80pf55 is an amazing P-channel part.
http://www.st.com/stonline/products/literature/ds/8177/stp80pf55.pdf