How can I switch 100 Amps at VERY low voltage?

[Peter A Forbes]

On Sun, 04 Feb 2007 08:20:10 -0600, jdworley@hiwaay.net wrote:

I found reference to them last night but no prices (yet) - and have the telephone
number for their U.S. distributor so I can call them tomorrow. I do not know if I
can afford their product line until then, but also, I have no idea what a "double
contact break" is either (was planning on asking them that question). Same with
the magnetic blowout (don't know what that is, didn't know to ask them, however ...
thanks).

Double contact break is where there is a contact bar across the input and output
terminals, so you actually have two contacts that break together, increasing the
gap very quickly and preventing an arc forming. Magnetic Blowouts pull the arc
out of the gap and extinguish it.

What I've been seeing, is many of the manufacturers (I've talked to so far) are
stating increased DC ratings of their lower current contactor relays. However, by
the time you get to their higher current product lines (where they would support
the 100 Amps I need, mostly with all contacts in parallel), they are actually
de-rating the contacts at DC. Not sure why.

They do a specific motor-reversing relay that would be up your street.

My primary goal, cost aside, is to avoid facing a mechanical failure at an
inopportune moment - so I've been trying to focus on a robust solid state
solution. ANY solution I choose, however, must not generate too much heat (the
biggest stumbling block to a solid state solution, no?).

It's protecting the switching device against back-emf that is the biggest
problem.

Does Albright offer a DPDT relay that can switch at least 100 ADC 450 times per
minute through each contact (or 4PDT at over 50 ADC per contact)? If so, do you
have any idea of the price range?

I have their current data sheets at the office (which we are currently moving
premises with) but I can look at the folder in the morning and pass your email
details to them if you want?

We use the SPDT SW201A and the SPST SW200 as well, both/all are double contact
break types. We are using one of their new range as a PCB mount contactor on a
small 110V DC 10A charger, we use the contactor to disconnect the charger in the
event of a fault or over-volt condition.

Thanks Peter - your input is appreciated!

We've had good responses in the past from the guys here, happy to return the
favour when we can )

Albright only build to order, so if you get a few 'funnies' with the US
distributor, give me a shout and we can probably get something moving this side.
There are also people in the fork lift truck market who carry piece parts and
who can make these up quicker than the Albright people, costs you more though.

Peter

--
Peter & Rita Forbes
Email: diesel@easynet.co.uk
Web: http://www.oldengine.org/members/diesel

 

[Gary Reichlinger]

On Sun, 04 Feb 2007 12:54:23 -0600, jdworley@hiwaay.net wrote:

I struck out on searching for info on "multi-phase buck regulators" ...
do you have an info link?

Try a google search using keywords:

multiphase buck converter (no quotes)

or else try searching the sites at www.irf.com or www.ti.com.
The goggle search even turned up a wikipedia entry for general
information on the concept.

 

You can get mosfets in the milli-ohm range.

Saturated switching is where these things can be used safely and
effectively in parallel. The more you use in your full bridge
arrangement, the lower the loss will be in the switch.

http://www.infineon.com/upload/Document/BSC020N025S_Rev1.0_G.pdf

This looks like a perfect match for my needs and a very affordable price -
thank you! The only difficulty I can see is the package itself is not
something I am used to where it comes to making connections and heat
sinking. I don't suppose I could impose on you to ask if there is a
similarly handy device in a more easily mounted package (something more
"hobbyist class")?

With four fets per switch arm, switch loss would be half the loss in
the 1mOhm load.

I didn't look at (scrutinize) the spec sheet yet, but I do wonder how many
amps these will support without any heat sink ... perhaps with 2 to 4 in
each leg I could get by with a cooling fan?

A full wave bridge at this voltage level could be driven directly with
15volt logic - though you'd want to keep the change-over slow, to
allow the current to reverse without resorting to body diode paths.

Yes, the next hurdle after choosing a device ... the driving circuit(s).
As much work as getting to that point has been, I think I'm looking
forward to that <s>.

THANK YOU RL!!!!!!! I SUSPECT MY SEARCH FOR THE RIGHT DEVICE IS COMING
TO A CLOSE.

Now to concentrate on the other aspects ... making a circuit board,
finding the high amp capable hardware, and, perhaps, canceling on the two
SCR's (? --- simpler solution, but I bet the on resistance is much higher
than can be achieved this way).

Thanks again,
Steve K.

> RL

 

[Mook Johnson]

Try the IRF2204PB from digikey for ~$3.50 each.
16 will cost you about $50.

this is an H bridge circuit. Since your supply voltage is so small you
don't need T5, T6 R1, or R2 to drive the high side gates. You can drive them
directly.

A good quad mosfet driver for this would be the TC4467COE also from digikey.
http://www.coledd.com/electronics/robot/h-bridge_large.png

To operate you turn on T4 then T1 for positive voltage across the motor.
Then turn them both off. Then turn on T3 then T4.

If you turn on the top and bottom fet at the same time you effectively short
your supply t ground through the mosfets. Usually the would blow the FETs or
the supply but in your case it might not be so critical.

Each of the 4 mosfets in the H bridge should actually be a parelleled est of
4 mosfets to reduct teh RDS to below 1mohm.

http://www.powerdesigners.com/InfoWeb/design_center/articles/MOSFETs/mosfets.shtm

when dealing with 100 amps, you had better have very heavy wires and PC
board traces.

You also need a decent heat sink to mount them too. 10-20 watts will be
dissipated in the H-bridge mosfets.


good luck




<jdworley@hiwaay.net> wrote in message news:45C62E26.CC5BE512@hiwaay.net...

Hi,

Sent reply by e-mail but it bounced - not edited for ng...

Thanks for the reply ... running out of time today (gotta run some
errands)
- do you have any links on the parts involved? I don't know if the
"TO-247
Mosfet H bridge" you mention below is a circuit reference or part (?),
will be
very interested if you have the time to respond - thanks you for your
input!

Best,
Steve K.

p.s. HP power supply in cc mode set to perhaps 1 VDC compliance (?) is
the
plan. Got lots of suggestions, but no part numbers for any semiconductor
solutions that I can afford ... I do have to wonder if the original design
using
a relay was that out of date after all!

Mook Johnson wrote:

jdworley@hiwaay.net> wrote in message
news:45C571D1.B1480424@hiwaay.net...
Hi all,

After a half dozen re-writes, I realize I may be asking
the impossible, but here goes...

What I need to do is supply 100 ADC to a .001 ohm load
and reverse the polarity to this load every four seconds.
I'm seeking a (least expensive) solution that can be used in
lieu of multiple relays because I believe most contactor
relays within my budget would be short-lived.

The most desirable characteristics in order of
importance are:

1) Robust design (failure intolerant situation)
2) tolerate 50% duty cycle @ 100 Amps (100% if devices
share the same case/heat sink?)
3) operate at as low a voltage as possible
4) Low thermal loss (lowest switching device saturation
voltage)
5) easily mounted device(s) to facilitate heat sinking
(i.e. TO-220 case style or chassis mountable)
6) easiest and/or most convenient drive requirements

How much loss can the switch have? is the supply a current source or a
0.1
V voltage source?

My first thought would be a TO-247 Mosfet H bridge with 2 - 4 0.002Rds
FETs
per switch (8 to 16 total FETs.)

with 4 FETs per switch your looking at .5mohms per switch so thats
.001ohms
through both the top and bottom switch in the bridge.

At 100 amps your looking at .1V of drop and 10 watts of Pd. That would
mean
your source would need to provide .2V to get 100 amps through the load.

You will need to drive fets with mofet drivers located close to the FET
gates. I'd use at least 12V for hard saturatinon.

If you need lower contact resistance than 0.001 ohms you will have a
diffiult time.

 

[Ross Herbert]

On Sat, 03 Feb 2007 23:40:33 -0600, jdworley@hiwaay.net wrote:

Hi all,

After a half dozen re-writes, I realize I may be asking
the impossible, but here goes...

What I need to do is supply 100 ADC to a .001 ohm load
and reverse the polarity to this load every four seconds.
I'm seeking a (least expensive) solution that can be used in
lieu of multiple relays because I believe most contactor
relays within my budget would be short-lived.

The most desirable characteristics in order of
importance are:

1) Robust design (failure intolerant situation)
2) tolerate 50% duty cycle @ 100 Amps (100% if devices
share the same case/heat sink?)
3) operate at as low a voltage as possible
4) Low thermal loss (lowest switching device saturation
voltage)
5) easily mounted device(s) to facilitate heat sinking
(i.e. TO-220 case style or chassis mountable)
6) easiest and/or most convenient drive requirements

Thus far, considerations have included; 4 3PST contactor
relays, 2 contactor relays + 2 power sources, 4 SCR's, 2
SCR's + 2 power sources, MOSFET(s), and power
transistor(s). IGBT devices were abandoned because I could
not find one with low enough saturation voltage, and DC
SSR's due to cost.

Optional control scenarios included programming the power
source to zero current to turn off SCR's, and in the case of
relay's, additionally firing an SCR across the p/s output to
circumvent contact arcing (might not be possible if o/p
voltage is under 1 VDC?). Obviously, an additional goal is
a favorable MTBF of p/s('s) and component choice(s)... .

I have some junk box parts, consisting of used
contactors that will support 100 ADC with paralleled
contacts (coils are 24VAC, high wattage...), 4 SCR's are on
order to at least try (MITSUBISHI TM200DZ-H's X2, but I
don't expect these to meet characteristic #4 above), a few
SSR's suitable for p/s input or contactor drivers, and
several large heat sinks.

Can anyone recommend alternative device(s)
(manufacturers & part numbers?!) or even a full
switching/control scenario utilizing as much of my on-hand
items as practical - that will generate the least amount of
excess heat without eating up my two used power supplies
that would cost me $4000+ each to replace if I had to do it
during a test cycle... .

Thanks in advance!!!!
Stephen Kurzban

R Legg has suggested the most practical method, ie. the use of the
H-Bridge using parallelled MOSFETS.

What you seem to be trying to achieve is akin to the H-bridge inverter
used in sinewave (or modified sine) DC - AC inverters for 50/60Hz
power generation, except that you are switching at a much lower
frequency.

University of Texas Prof H. Mack Grady has some interesting course
material on-line here
http://users.ece.utexas.edu/~grady/

Look under Course Materials/Materials for EE362L... for stuff on
H-bridge inverter, MOSFET firing etc.

 

[Homer J Simpson]

<jdworley@hiwaay.net> wrote in message news:45C5EB9A.99CD2ABF@hiwaay.net...

Does Albright offer a DPDT relay that can switch at least 100 ADC 450
times per
minute

Yikes!

 

[Homer J Simpson]

<jdworley@hiwaay.net> wrote in message news:45C5E784.13559CF6@hiwaay.net...

Why? An explanation would help to focus the answer.

Resistance measurement similar to those taken with the galvanometer in a
current comparitor bridge - the system is currently missing the
accessories,
one of which is a DPDT 100A high current switch. I'm designing a
replacement
for it.

I'd be inclined to look at a rotary switch for the sort of rates you are
talking about. But it will be a very special rotary switch!

 

[Homer J Simpson]

<jdworley@hiwaay.net> wrote in message news:45C634CE.3E119B3E@hiwaay.net...

http://www.infineon.com/upload/Document/BSC020N025S_Rev1.0_G.pdf

This looks like a perfect match for my needs and a very affordable price -
thank you! The only difficulty I can see is the package itself is not
something I am used to where it comes to making connections and heat
sinking.

I'd be trying to think in a 'plumbing' mode - where will the heat be and how
can I get rid of it?

 

[Jamie]

Homer J Simpson wrote:

jdworley@hiwaay.net> wrote in message news:45C5EB9A.99CD2ABF@hiwaay.net...


Does Albright offer a DPDT relay that can switch at least 100 ADC 450
times per
minute


Yikes!




DC drive of around 150 HP will work just fine. one for each

direction operating in Armature feed back..

Or just get a DC amplifier.


--
"I'm never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

 

Switching control comes from the bridge...

I've got a good start with all the info this thread has provided - thanks to
all!

I'd be inclined to look at a rotary switch for the sort of rates you are
talking about. But it will be a very special rotary switch!

 

[Gary Reichlinger]

On Sat, 03 Feb 2007 23:40:33 -0600, jdworley@hiwaay.net wrote:

What I need to do is supply 100 ADC to a .001 ohm load
and reverse the polarity to this load every four seconds.
I'm seeking a (least expensive) solution that can be used in
lieu of multiple relays because I believe most contactor
relays within my budget would be short-lived.

You never mentioned just what you plan to use for your original
power supply to get the 100 ADC. If you can allow some dead time in
between transitions, turning the power supply on and off might be the
easiest way to go. This is essentially what I was talking about with
the buck converter multiphase approach. The converter would actually
be converting the power from a higher voltage/lower amperage source.

 

[Mook Johnson]

I would recommend having the driver IC very close (within 1 - 2 inches) of
hte modfet gates. I would also recommend putting a small 10 ohm resistor
directly to the gate and a 16V zener between the gate and source of each
mosfet. This will kill and oscillation between the wiring undctance ang the
gte capacitance and clamp and spikes that might make it get on the gate.

You don't want long wires between the driver and the mosfets.

I would strongly recommend twisting the source and return wires for the 100A
to reduce the amount of magnetic field generated.

 

[John]

"Gary Reichlinger" <reichln@navix.net> wrote in message
news:1s9ds29hbkpc1tb9hhi6l56omp5cp5jpra@4ax.com...

You never mentioned just what you plan to use for your original
power supply to get the 100 ADC.

Yeah, he did. See his post in this thread for Sunday, February 04, 2007 1:04
PM.

 

I struck out on searching for info on "multi-phase buck regulators" ...
do you have an info link?

Thanks!

Take a look at some application notes for multi-phase buck
regulators. International Rectifier (www.irf.com) and others offer
parts for this approach. I would envision that you could use 2
regulator configurations (one for each polarity). By leaving the
ground side mosfets on during the time the other regulator array is
active, you would complete the circuit. The multiphase approach
allows the use of multiple mosfets without the need to parallel them.

 

Mook Johnson wrote:

Try the IRF2204PB from digikey for ~$3.50 each.
16 will cost you about $50.

There was another device suggested earlier in the thread too, a BSC020N025S claiming
2 mOhms on its data sheet - but it has a package type I'm not confident I can heat
sink (never saw it before). The one you list is easier to mount ... thank you.

this is an H bridge circuit. Since your supply voltage is so small you
don't need T5, T6 R1, or R2 to drive the high side gates. You can drive them
directly.

Thanks to the tutorial below on these devices (I didn't know about the high input
capacitance), I do follow now <s> thanks!

A good quad mosfet driver for this would be the TC4467COE also from digikey.
http://www.coledd.com/electronics/robot/h-bridge_large.png

Very convenient, sounds easier than designing my own, but for this application I may
not have to be concerned with switching speed - and - if all devices are on during
transition, well, it might actually work better that way (especially if there is a
delay in the cycle for the p/s to achieve regulation or if it has overshoot issues).
I guess I can experiment if need be, but it is also cheap insurance to have the
suggested driver IC handy... .

To operate you turn on T4 then T1 for positive voltage across the motor.
Then turn them both off. Then turn on T3 then T4.

If you turn on the top and bottom fet at the same time you effectively short
your supply t ground through the mosfets. Usually the would blow the FETs or
the supply but in your case it might not be so critical.

In any event the p/s circuitry should limit current to 100 amps. Minimizing heat is
an issue more to do with both controlling thermal voltages and insuring room
temperature remains stable as possible during the measurement cycle(s).

Each of the 4 mosfets in the H bridge should actually be a parelleled est of
4 mosfets to reduct teh RDS to below 1mohm.

Yes, I do understand that, but until I read from your tutorial link I was unaware
there was no need for the equivalent of an emitter resistor like BJT's use. (Even)
if one device could handle the power, the leads need to be able to handle it too -
and of course the lower the on-resistance, the less heat - absolutely concur on the
parallel devices (thanks)!

http://www.powerdesigners.com/InfoWeb/design_center/articles/MOSFETs/mosfets.shtm

when dealing with 100 amps, you had better have very heavy wires and PC
board traces.

Wiring is planned to be 2 ga. copper with crimp terminals. I guess I will have to
have multiple 12 or 10 Ga. copper wire runs - one from each FET to a common block for
cable assembly hook up.

You also need a decent heat sink to mount them too. 10-20 watts will be
dissipated in the H-bridge mosfets.

I was hoping the on-resistance looked low enough to use chassis mounting or small
heat sinks and a fan (was going to test with a thermocouple on one FET). Oh well,
too much is better than starting over due to burned up parts, but to satisfy
curiosity, I could always set the current limit low and monitor device temperature
rise? If so, I suppose the data sheets tell me where I've gone too far, but do you
recommend a certain temperature to run these at (ambient will be a controlled 74.3
degrees)?

good luck

Thanks to yours and others good advice on this 'project', hopefully my task won't
totally be dependent upon just luck.

I appreciate your input and all the references, recommendations and pointers!

Thanks again.

Best,
Stephen Kurzban