Microchip's _MCLR pull-up cct

[MarkMc]

Hi

I've a relative newbie to electronics and PIC microcontrollers in
particular. I'm trying to design a cct using PIC16F628A.

I'm looking at the Microchip mid-range PIC document
(DS31028A-page28-4), and this shows a diagram of how to isolate VDD
from VPP when the ICSP connector is connected.

The diagram consists of VDD connected to a resistor (R1) and a
capacitor (C1). Between the capacitor is a diode(D1) which is
connected (normal bias) to _MCLR/VPP and the ICSP connector. A second
connection is made from VDD to the VDD pin via resistor(R2) on the PIC
and ICSPconnector.

(anybody care to show me some ASCII cct drawing tools?)

I understand that D1 protects/isolates the main cct from the +13v VPP
when programming.

My questions are;

1) D1 is referred to as a Schotky diode in the document. Can somebody
explain what one is, and recommend a model number for me to use.
2) Also, no values are given for R1, R2 and C1. I assume 'cos they
can't possibly know the capacitance of a third party cct, but can
anybody recomend good starting values for these? I can give you a
pointer to my circuit on the web if you'd like to see the cct. Should
C1 be a tantalum cap or electrolytic, ceramic or what?
3) I don't use any of the other programming pins for my application to
make life easy, but my cct has another component which fires off a data
burst to my PIC periodically. I assume this won't affect the PIC while
it's being programmed? Or should I isolate the entire cct from VDD via
a dip switch or something to be sure?

Thanks
Mark

 

[PeteS]

A Schottky diode is one of a class of 'hot-barrier' diodes and they are
used because their forward voltage is typically 0.2-0.3V for small
signal type devices. This is to make sure MCLR discharges quickly.
There's no protection required if MCLR is connected only to a reset
signal.

You should not need to isolate this pin beyond the typical schematic
shown in the manual.

Typical devices for this sort of application might be the MMDL770T1
from On Semi (nee Motorola SPS) [http://www.onsemi.com/]
Get a parametric table of devices by searching for schottky diodes. You
want low Vf, Low If (forward voltage, forward current) and fast
recovery is preferable.

You only need the full RC circuit for a slow rising Vcc system.

This is from DS31003A (Mid range PIC reference manual)
**Figure 3-3 shows a possible POR circuit for a slow power supply ramp
up. The external Power-on Reset circuit is only required if VDD
power-up time is too slow. The diode, D, helps discharge the capacitor
quickly when VDD powers down.**

A slow rising input is defined in that manual as < 0.05V/millisec.

If you have a slow rising Vcc, then MCLR should stay low until Vdd is
valid. That occurs at around 2.5V, but I would suggest making sure it
holds off until Vdd is about 3V.

You need to calculate the RC pair, and for this you first need to
define your power. Let's say it's 1V/millisec. That means we don't want
MCLR to go above ~75% Vdd until at least 3 milliseconds later. A delay
of 10msec would work for a good conservative circuit. (It also
simplifies the mathematics).

1xRC is about 63% Vdd, and we'll use that as it's close enough if we
extend the time a little. The listed VIH for MCLR (the input voltage at
which a high is guaranteed to be detected) is 0.8VDD.

So RC >= 0.01
Choose a resistor (Microchip suggests <40k to minimise voltage drop due
to input current on the pin).
So let's go with 20k
The C = 0.01 / 20k = 0.05uF

Note that this gets you about 63% Vdd after 10msec, but that's ok.
A typical value to use would be 0.47uF

Cheers

PeteS

 

[MarkMc]

Thanks very much for enlightening me Pete.

Very much appreciated.

I'm using the _MCLR pin straight to VDD (supplied by a 7805 vreg), but
also it connects to an ICSP +13v supply for programming. Sorry for
being dense here, would this mean that I do or don't need the diode
set-up?

If not, should I still have, say a 0.1uF tant. cap between VDD and gnd,
right next to the PIC IC?

Cheers,
Mark

 

[Anthony Fremont]

Thanks very much for enlightening me Pete.

Very much appreciated.

I'm using the _MCLR pin straight to VDD (supplied by a 7805 vreg), but
also it connects to an ICSP +13v supply for programming. Sorry for
being dense here, would this mean that I do or don't need the diode
set-up?

You need the diode to block the Vpp(+13V) from powering up the rest of
your circuit and potentially killing some parts by applying excessive
voltage to them. I suspect that they used a Schotky(sp?) or "hot
carrier" diode just for the low voltage drop feature (.2 - .3V) just to
keep MCLR as close as possible to Vcc.

If not, should I still have, say a 0.1uF tant. cap between VDD and gnd,
right next to the PIC IC?

If you'd like, it shouldn't harm anything. You would probably only need
that if glitches/spikes were likely on the Vcc that might result in a
device reset.

 

[PeteS]

Well, the diode is there to make sure that the MCLR pin doesn't exceed
Vcc during power down. If you have a fast falling Vcc, then the cap/res
pair will hold MCLR high without the diode, exceeding the rating of the
device and powering the rest of the chip through the input protection
diodes (this is generally not a good thing [tm]).
I would normally have the diode there for protection against transients
anyway, but ICSP requires that you do not have the diode that way.

Look at the application diagram on page 28-4 of the previously
referenced manual. There is a diode there (make it a schottky again so
that when MCLR is powered by Vcc you aren't 0.6 - 0.7V below Vcc, but
0.2 - 0.3V below) but this time to isolate the power supplies (we call
this diode OR'ing, by the way). In this particular circuit, assuming
Vpp is not present, then the diode will still discharge the cap at
normal Vcc power down, not exceeding about Vcc + 0.3V at the MCLR pin,
so you still get the device protection.

You should bypass the device, but I would suggest a 1-10uF ceramic as
esr dominates ripple performance, and ceramics have outstanding esr
ratings (typically below 10 milli ohm - I have seen them as low as 1
miili ohm).
I don't see a need for bulk bypass beyond that - the maximum current is
Icc + I (IO), so you should be in the order of 10s of milliamps. Note
that using a ceramic like this means you don't have to also have a
small value ceramic (which you would need with a tant, unless it's one
of the later ones), and tantalum has it's own problems (highly toxic if
they overheat for whatever reason). During programming, the current is
higher, of course, but not that high.

Hope that helps

Cheers
PeteS

 

[Bill Carson]

2) Also, no values are given for R1, R2 and C1. I assume 'cos they

can't possibly know the capacitance of a third party cct, but can
anybody recomend good starting values for these? I can give you a
pointer to my circuit on the web if you'd like to see the cct. Should
C1 be a tantalum cap or electrolytic, ceramic or what?

You might find the information in this free PIC book useful:
Getting started with PIC microcontrollers, by Fred Stevens
a PDF can be downloaded from here:
http://www.the-electronics-project.com/

see page 18 for a disussion of component sizes for the RC circuit.

HTH,

B.C.

 

[MarkMc]

The diagram on page 28-4 was the diagram that first prompted my post.

Forgive me for asking, I'm not an electronics professional, but what
does "BYPASS" mean? is this the normal capacitor for protecting
against ripples and spikes going in to an IC? I've always been told to
use a 0.1uF tant. capacitor for this due to their speed to react. I'm
pretty sure I read this in a PIC book once too. I'll happily learn
some other way, but I'm not sure I fully understand.

Cheers,
Mark

 

[PeteS]

Bypass is, as you assume, to deal with local voltage variations due to
the chip itself. As the chip operates and switches it's outputs (and
its inputs are changed), the device uses more or less current, and this
shows up as noise on it's power pins (there are lots of reasons for
this, including the nductance of the bond wires from the die to the
pins of the device).

Bypassing is used to combat this noise, and we usually talk of it in
two ways:

Bulk bypass. This is a fairly large capacitor used as energy storage
for those sudden increases (and decreases) in current. Large variations
in current a long way (electrically) from the regulator sense point can
have dramatic voltage fluctuations - bulk bypass is used for this.

Signal bypass: This is to eliminate (well, minimise) the noise on the
power lines due to high speed switching of the device.

Note that these two do overlap considerably; high speed switching can
also have large current variations associated with it, although we try
to minimise that in really high speed stuff by using (almost) constant
current devices (and that's another story entirely).

I am sure tantalums were recommended because their performance is
superior, electrically, to electrolytic capacitors. As I noted above,
tantalums have their own issues though. Modern ceramic capacitors have
up to 100uF at 6.3V in a 1210 package (I know, I've used them - check
out this panasonic part : ECJ4YB0J107M), have superior performance to a
tantalum (lower esr, better ripple current performance, better
temperature characteristics) for general bypassing.

This is not to say tantalums are not widely used - I use them myself as
part of the loop filter in some LDO supplies which require my output
ESR not be *too* low (Look carefully at the datasheet for the National
LP3961 for an example).

That's the quick explanation.

Cheers

PeteS

 

[MarkMc]

Thanks Pete, that helps clear things up. I'm enjotying learning about
this stuff. Let me see if I'm getting this. I should use ceramics
(100uF?) in close proximity to all of the IC's (PIC being one of them)
in my cct.

Should I put a tant. (0.1uF) on the 5v regulator too? (78L05)?

Cheers,
Mark

 

[PeteS]

You should have a bypass cap for the 7805 (I assume it's powering a few
things, such as the I/O devices) - about 10uF is a good rule of thumb
for light loads.

Use a 0.1 - 1uF cap for the PIC.

If you use ceramics, that's all you should need.

Cheers

PeteS

 

[MarkMc]

That's great, thanks for all your help Pete

 

[niftydog]

(anybody care to show me some ASCII cct drawing tools?)


dis one is da bomb. http://www.tech-chat.de/aacircuit.html

 

[MarkMc]

That's perfect, thanks niftydog.

Cheers,
Mark