[SI-LIST] : Clamp diodes in models

About this list Date view Thread view Subject view Author view

From: WAUGH,RAY (HP-SanJose,ex1) (ray_waugh@agilent.com)
Date: Mon Jan 10 2000 - 16:03:45 PST


Why not use a fast diode (Schottky) with a low parastic series resistance
(0.7 ohms) in a low inductance package (2nH)? I'm just an RF guy, so maybe
I'm missing something.

I have an unpublished note on this topic, complete with diode SPICE
parameters, which is available from me in PDF format. All you have to do is
drop me a note at:

        ray_waugh@agilent.com

Ray
------------------------------------------------------
      Raymond W. Waugh - WSD Diode Applications
      E-mail: ray_waugh@agilent.com

          USPS: Agilent Technologies
                Wireless Semiconductor Division
              39201 Cherry Street, MS NK20
              Newark, California 94560
------------------------------------------------------

-----Original Message-----
From: jrbarnes@lexmark.com [mailto:jrbarnes@lexmark.com]
Sent: Monday, January 10, 2000 11:51 AM
To: si-list@silab.eng.sun.com
Subject: Re: [SI-LIST] : Clamp diodes in models (was Input
switchingthreshold & CPCI)

> I modeled a circuit using vendor provided SPICE models and noticed that
the
> undershoot in the SPICE *never* got below -.7 volts, but that my board
> measurement was showing -2.5 to -3 volts for 1 or 2 nanoseconds, then
leveled
> out at -.7 volts.

David,
The turn-on spikes seen with PN-diodes are not due to capacitance, but due
to
transit time. When a PN-diode is reversed-biased the electric field pulls
charge carriers away from the junction, forming a depletion region. This
depletion region may have a few pF of capacitance associated with it. If
you
suddenly try to forward bias the PN-diode it takes some time (roughly the
width
of the depletion region / carrier drift velocity) for charge carriers to
cross
the gap. During this turn-on time only the parasitic
capacitance/inductance/resistance of the circuit limit the voltage across
the
diode. The duration of the turn-on time is only a weak inverse function of
the
voltage that you are trying to put across the PN-diode. The turn-on time
ends
once a reasonably steady flow of charge carriers flows across the junction,
and
the PN-diode once again functions "normally".

About twelve years ago we ran into a problem on the IBM 4224 dot-matrix
printer,
where we were finding blown N-channel MOSFET's in printhead drivers at final
test. We traced the problem to a change of vendors for the 1N400x (one of
the
1N4000 to 1N4007 family) commutating diodes across the N-channel MOSFET's.
The
scenario was as follows:
1. Turn on a printhead solenoid, driving it with approximately 30V through
a
P-channel MOSFET and an N-channel
      MOSFET.
2. Pulse-width-modulate (PWM) the P-channel MOSFET to limit the current to
about 1A (this MOSFET had a very-fast
     commutating diode).
3. Turn off the N-channel MOSFET to turn off the printhead-- the MOSFET
turned
off in about 300ns, and the solenoid
     tried to generate about a 250V kickback spike on its drain.
4. The original 1N400x diodes across the N-channel MOSFET's turned on in a
fraction of a microsecond, limiting the
     drain voltage to 40-50V or so, which was safe. A few percent of the
new
1N400x diodes took up to 5us to turn on--
     the 250V kickback spikes from the printhead solenoids caused
drain-source
shorts in the N-channel MOSFET's.

Looking at different vendor's datasheets for 1N400x diodes, they never
mentioned
turn-on or turn-off characteristics. The parts were advertised as
"rectifiers"
and intended for use at relatively-slow dV/dt and dI/dt rates. We
considered
changing to the very-fast commutating diodes for the N-channel MOSFET's, but
that clobbered us for radiated emissions. We wound up going back to our
original source of 1N400x diodes, and testing their turn-on time as part of
the
GenRad in-circuit test for the printer cards. Driving the diodes with 5V
from a
current-limited digital driver, and checking the forward voltage 0.6us after
a
rising edge was enough to distinguish good diodes from too-slow diodes.
(This
was based on characterizing a bunch of known-good and known-bad diodes, then
monitoring the first few weeks of production after we put the test in
place.)

With the edge-rates we've seen with our last couple of generations of
products,
I am not surprised that you are seeing 2-3V overshoot on falling edges
before
clamp diodes have a chance to take effect. I've gotten to the point that,
if I
expect that a fast digital net will be longer than 1-2 inches total on the
printed-circuit board (PCB), I'll almost automatically source-terminate it
with
a series resistor at each driver.

                                              John Barnes Advisory Engineer
                                              Lexmark International

**** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****

**** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com.
In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:34:33 PDT