# intentional Slew rate limiting

Discussion in 'Electronic Design' started by Mook Johnson, Mar 16, 2007.

1. ### Mook JohnsonGuest

I need to make a high curent driver (say 500mA peak) that I can adjust the
slew from say 50nS to 500nS in both directings. The input is 3.3V digital
and the ouput is a 0- 10V square wave driving about 200ft of wire. I figure
the lump capacitance is about 1000pF. I need to adjust he slew rate to
reduce the amount of cross coupling from this wire to its parellel neighbor
but I need the edges as sharp as I an get them.

I'm looking at lower end FET drivers with resistors between the output and
the line. Is there a way to make these have a linear slew from 0 - 10Vover
the 25 - 250nS range mentioned?

Any other circuit ideas for intentionally limiting the slew?

2. ### Winfield HillGuest

I've made multiple similar circuits requiring precision slewing
ramp rates, using high-speed amplifiers and programmable opamp
slew-rate circuits. I doubt you need a high level of accuracy,
so a more simple scheme could work for you. There's a concept
called a Miller integrator, which you can implement by adding a
feedback capacitor from the output MOSFET's drain to gate, plus
use a series gate resistor to set the dV/dt integration rate,
where Rgate = Vin-Vgs / dVo/dt determines the gate resistor.

3. ### LionelGuest

Maybe a second FET with the drain & source between the driver output &
a cap to ground?

4. ### Phil AllisonGuest

"Mook Johnson"
** Why ?

** That would be a massive source a radio interference on AM and SW bands.

** 5 pF per ft - so it sure ain't co-ax.

** The solutions to crosstalk are well known.

Co-axial cable, metal shielding, twisted pairs etc.

Use one of them.

...... Phil

5. ### Fred BloggsGuest

View in a fixed-width font such as Courier.

6. ### Tony WilliamsGuest

Below is the downmarket, belt 'n braces suggestion.

+15V---+---+-----+
| |0.1? |
| === \220
\_|_ | /
Zener,5V1 /_\ | |
| | |/e
+---+---|pnp
23mA\|/ _|_ |\
| /_\ |
\ | | Linear
430 / '-----+ +1 amp.
\ _|_
| \_/ -+- +15V
0V-+- | _ |
20mA\|/Constant. | \|
+------------+--|+1>-->Out
38mA\|/Switched. | |_/|
3.3V-+-----+------. _|_ | |
| | | \_/ C=== -+- -5V
|\| |/ | | |
---| >---|npn | +--------. | Tr/f = 10*C/19mA
|/| |\e | | _|_ |
| | |/ \| /_\ |
0V--+- +----| npn |--. | |
| |\e e/| | | |
| | | '---+---+-0V
| '-+-'
| | Diodes are Schottky.
\ \
680/ /110
\ \
| |
-5V --+--------+-----

If C = Cstray = 50pF then Tr/f(min) = 26nS.

If C = (470+50)pF then Tr/f = 274nS.

7. ### whit3rdGuest

Well, there's an OLD idea that might work: an increductor.
The inductance of a simple choke will be the slew rate limiter,
and the inductance responds to the field in the core, so you
make a high-impedance winding with some DC current in
it and at high DC current it has low inductance (50 nS slew into
load) and at low DC current it has high inductance (resulting in
500 nS slew into load). The signal is sent through a low-impedance
winding in the same core.

Because it's current-programmed, it's easy to adjust. Capacitors
with voltage-adjust are generally not of such size as to work
for 500 nS delay at 500 mA drive.

8. ### Ken SmithGuest

Its too bad the CA3080 isn't with us today. There was a nifty circuit
that used one of them to make slew rate limiting.

But here's an idea

ASCII art:

---------------------------
! !
! r-r opamp ---!!---+
---!+\ ! !
! >-+-/\/\----+--!-\ !
---/\/\--+-!-/ ! ! >--+------- Output
! ! GND-----!+/
--!!-- Power op-amp

The r-r opamp limits the slew rate by hitting its power supply rails.
Finding a r-r opamp that recovers well from this may be the biggest
problem with this.

9. ### John LarkinGuest

The classic slew limiter is a diode bridge...

i+
|
|
+
/ \
a a
k k
/ \
in---------+ +-------+------out
\ / |
a a |
k k c
\ / |
+ |
| gnd
|
i-

where current i into the cap sets the slew rate.

John

10. ### Fred BartoliGuest

Ken Smith a écrit :

11. ### mook JohnsonGuest

The input is 3.3V digital and the ouput is a 0- 10V square wave driving

You are correct the capacitance is actually 10,000pF or 50pf/foot. The
cable is twisted shielded quad (4 wires inside a single shield). Wire to
wire coupling capacitance is about 10p/ft.

My opinion is that the situatuin is near hopeless for high speed
communication. As soon as you get the edges steep enough to look like
digital its going to crosstalk like a SOB.

12. ### Fred BartoliGuest

mook Johnson a écrit :
Is it quad or dual pairs?
Do you have a ref?

For long cables, quad is worse than twisted pairs since it's coupling
the same way all along the path.
Twisted pairs cancel much of this.