Pulse Amplifier

I have an Allen Automotive Distributor Machine that was built in 1958. The
distributors then were all points. I need a circuit that will take the input
from a modern electronic distributor(small magnet/coil generated signal, or
Hall switch signal) and convert it to a square wave signal like points would
do. There were pulse amplifiers available way back when, but are hard to find
and very expensive. I would imagine the circuit would be quite basic. Thank
You, Paul

 

Err, the output from points was anything but a square wave. Huge
voltage spike (over 100V as I recall) and a lot of ringing. After
all, the points were opening a big inductive circuit with a capacitor
(err, make that "condenser") across them.

The question is what your machine is actually doing. If it has a
dummy ignition circuit in it, it may have its own coil and be
expecting a contact-closure with a condenser across it, to
generate a spark in the old-fashioned way. This might be
tricky to emulate exactly, but you might start with an ordinary
junction power transistor to replace the points. Get the
drive for that with a simple low-voltage amp on the output
of your distributor pickup. This is pretty much what the
early "transistor ignition" circuits did, just used the points
to fire the transistor. They didn't have MOSFETS then,
but one might work. The reverse spikes from the coil
kick-back will require a reversed diode across the
transistor to keep it out of breakdown.

Hope this helps!



Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Bob Masta wrote:
(snip)

A diode across the transistor will clamp reverse voltage, not excess
normal voltage. If you put a diode across the coil ot prevent the 100
volts or so normal spike, the coil won't produce a spark. The
transistorized points I have seen just use a high voltage capable
transistor so it can survive the high voltage the coil produces.

 

I was going to point you toward a web page that showed how to do this,
but found lots of crap. For instance, this site has so many errors
that we could have a course based on only what it gets wrong:
http://hyperphysics.phy-astr.gsu.edu/hbase/magnetic/ignition.html
Here is an application note that describes the difficulties:
http://home.swipnet.se/~w-29552/files/igbt.pdf
Here is the data sheet for a chip that ties a hall sensor to a
transistor ignition that gives you a good idea what all is involved:
http://www.st.com/stonline/books/pdf/docs/1360.pdf

 

John, it's the reverse voltage I was talking about. I'd think
I'd want the transistor to be rated for several hundred volts.
I was particularly thinking of the diode for use with a MOSFET,
since their reverse breakdowns are pathetically low. Don't know
if they would take well to a bolt of reverse current in breakdown.
(Anyone know about this?) I think junction transistors may be
less of a problem. (Can't seem to find any reverse breakdown
data at the moment.)

Best regards....


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Mosfets have an internal diode junction between source and drain that
may or may not have a particularly fast reverse recovery when the
drain voltage again is in the correct direction.

Junction transistors have a reverse current gain of beta that allows
them to conduct somewhat in the backwards direction. The real danger
with a junction transistor is that they will not have enough reverse
beta to carry the reverse current before the emitter to base junction
gets reverse biased to its breakdown voltage which is only 5 or 6
volts. If that happens, the transistor will steadily loose gain.
Another problem with reversed collector voltage is that the forward
biased base to collector junction will tie this reverse voltage into
the circuit driving the transistor, and may cause damage or
misoperation of that circuit. But that depends on what that circuit
consists of. If the driving circuit strongly clamps the base to the
battery rail during the off time, the collector to base junction will
perform the reverse diode function.

 

John Popelish wrote...

I think what Bob is talking about is the voltage rating of the
MOSFET before avalanche breakdown, which is one thing limiting
how far an inductor can flyback. You can get low-cost MOSFETs
up to 1200V, so that's not much of a limitation. As far as an
avalanche is concerned, FETs can handle much more than BJTs,
with their second-breakdown, and SOA safe-operating-area limits.
For those concerned, there are several easy ways to prevent
breakdown of the switching transistor, whether FET or BJT.

Another attractive choice these days is high-voltage IGBTs.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

Actually, it was the reverse conduction that I was talking
about. I have been thinking about how to use MOSFETS
for high-power AC control, and this is the problem. It's
easy to find high-voltage, high-current devices, but they
don't block the opposite polarity at all. I was hoping to
take advantage of low MOSFET losses and easy l
oad-sharing, but it seems you need a rectifier in series
with the device... so there goes all your power savings
into the diode drop!

In the case of the OPs problem, I was concerned that
the transistor might not take kindly to the reverse spike
from the coil transient, hence the reverse shunt diode
suggestion.

Regards,


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Bob Masta wrote...

Correct, all vertical MOSFETs (often called DMOS, VMOS, etc)
have an intrinsic diode in parallel with the FET.

If you wish to use FETs, the solution to this is to use two
N-channel parts back-to-back (connect their sources together
and their gates together). FETs happily conduct current in
both directions, so when you apply gate voltage, the FET that
has reverse current will be on, shunting its intrinsic diode
with a low Ron, so you won't suffer any diode drop.

..
.. --------+-|<|-+-+-+-|>|-+------- +/-400V and 2A rms max
.. | | | | | (w/o any heatsinks)
.. IRF740A ', ,-,' | ',-, ,'
.. (2) | V |s | s| V | the diodes shown across each
.. ----- | ----- FET's drain-source leads are
.. PVI5033R ---, | ,--- intrinsic to all vertical FETs
.. ------, | | |
.. gate |-----+--|--'
.. drive |--------' isolated gate-drive voltage
.. ------'
..

Excuse the poor ASCII drawing. One thing that's evident from
the drawing is that for AC-line switching, the gate drive will
have to be floating. That's the problem faced in a solid-state
relay, or SSR, that uses FETs, and this is solved by using a
stack of photodiodes driven by an IR LED. It's not fast (200us,
etc) but it works well and gives 2500V of optical isolation.

If you want to make an optocoupler SSR with your own choice of
FETs, you can use IR's PVI5033R isolated 10V 5uA gate-driver,
which includes a special fast shut-off circuit (see figure 4).
http://www.irf.com/product-info/datasheets/data/pvi5033r.pdf
DigiKey (and Newark) has these in stock at $5.75 ($5.45) each.

For example, using a PVI5033R and two IRF740A MOSFETs you can
make a +/- 400V, 1-ohm switch that will turn on within 12 ms
and off in about 0.5 ms, using 5mA drive to the PVI5033's LED.


Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

Win:

Thanks for the detailed reply. I wasn't aware that the
FET would work in reverse... good thing to know about!

My application (heating a low-ohm resistive load from
the mains) requires an average current that could run
to 40A. I'd be using zero-crossing control of the
duty cycle to get this, but the peaks would still be 80A
or more. Can I safely parallel devices to get the
total current handling, and expect them to share
nicely? Any other tips about applying your dual-N-FET
method at these high currents?

Thanks!






Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Bob Masta wrote...

You can use larger FETs and parallel them, too, but you'd
be better off with opto-coupler SSRs that employ triacs
rather than MOSFETs. Such high-current solid-state relays
are sometimes inexpensively found on eBay, e.g.,
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3821362659
You may need to mount them on heat sinks, like this:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3821135841

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

Win, are you implying that I can parallel 4 of these
25-amp SSRs to get 100A handling? I didn't
think there were any good ways to parallel triacs
because they don't share current like MOSFETS
do. Also, I need to zero-cross duty-cycle modulate the
load, so the devices need to be fairly fast. I need
proportional control, not the sort of bang-bang
control on (say) conventional ovens. Are SSRs
fast enough for proportional control?

One advantage of parallel MOSFETS is that
they are way easier to heat-sink than one monster
device. Especially since the "monsters" nowadays
come in little power-tab packages... I have some
40A triacs in TO-218 cases, and suspect
it would be quite difficult to heat sink them enough
to justify their ratings.

Thanks for your advice...



Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Bob Masta wrote...

Not triacs, and if you re-read what I wrote you'll see
I said you can parallel switched MOSFETs. But that I
recommended triacs instead.

Certainly they can be, you have to read the datasheet.

I think we can't give you futher advice unless you can
give us much more detail about your load. E.g., a 400V
MOSFET may have little problem handling high currents
for a few tens of microseconds, but if it's a 40A peak
pulse lasting for most of an ac half-cycle , etc., that's
likely going to be too much for ordinary FETs, even if
they are in parallel. If you're seeking ease of use,
easier gate drive, etc., use IGBTs instead of FETs.

You can get nice IGBT power blocks on eBay.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3821356668

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

On 18 Jun 2004 05:22:52 -0700, Winfield Hill
<snip>

Win, thanks very much for your advice.

The application is driving silicon carbide heating elements
for a small electric kiln. The elements (2 in series) have
a nominal resistance of 2.36 ohms at temperature, increasing
to maybe twice that as they age (months to years of use).
SiC doesn't like the surges of on-off thermostat-type control
(as used in home ovens and most pottery type electric kilns),
but is happier with proportional control.

I have a monster tapped transformer rated to 25 amps
for driving these now, but would like to replace manual tap
adjustment with computerized proportional control. And
I'm thinking it might be nice to run off the mains directly,
to avoid the current limits of the transformer. (The elements
are good for 40A.)

WOW... 600V at 200A!!! This definitely looks like it could handle
anything I could throw at it. But I'm still curious about heat
dissipation. I'm assuming that just about any device is going to
have a volt or two of drop at 25A. Even the MOSFETs would
seem to have this on the reverse phase where the body diode
is conducting. Do these IGBTs have subtantially lower drops, or do
they need fancy cooling schemes?

Thanks again for your help.


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

If you have zero-crossing duty cycle control, why not
use triacs? You can get them to practically arbitrary
sizes.

http://www.mouser.com/catalog/618/335.pdf

I notice a 40A 800V unit has a Vtm of 1.55; hmm - that's
almost 50 watts - you might need a fan.

Have Fun!
Rich

 

Rich, this is exactly the issue: You can get triacs with
monster specs, but how to get the heat out of the
tiny packages? That's why I was thinking about
ways to have multiple devices, simply to spread out the
heat. And as far as I know, only MOSFETs can be
paralleled and share load current happily without
hogging and thermal runaway.

Thanks for the thought!


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

Bob Masta wrote...

The datasheet says these 600V IGBTs drop less than 1.5V at 40A,
or 60W max. The IGBT's thermal resistance spec is 0.14C/W,
which means 60W dissipation will increase its junction temp by
less than 10C above the insulated (!) thermal mounting plate.

Compare this to the FDH44N50, a high-performance 500V FET. It
has R_on = 0.25 ohms max (warmed up) for a huge 10V drop at 40A.
You'd have to parallel six FETs to match Toshiba's MG200 IGBT.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

I'm honored to bow to Mr. Hill's experience and stuff here.

Thanks!
Rich