Maker Pro
Maker Pro

78xx output cap

P

Pimpom

Jan 1, 1970
0
I know that, to ensure stability when using a standard 3-terminal
regulator, the output capacitor should be placed as close as
possible to the regulator. Usually this is no problem. But I'm in
the middle of a PCB design in which it's more convenient to have
slightly over an inch of track between the two. This is on both
the hot and the ground sides. The tracks are shared by load
currents of a few mAs. Part of the load is a CMOS counter IC
operating at a few kHz while another section has further RC
filtering on the supply. Is this likely to cause problems?

I'm asking here instead of simply trying it out because I have to
make a dozen of this circuit. Even if the test unit works, I
won't know if it's only marginally stable and I don't want to
risk having some units or future replacements go unstable.
 
P

Pimpom

Jan 1, 1970
0
David said:
IIRC the national semi app notes said within 3 inches. I have a
little
experience with using the caps over 1 inch without problems
(but
wouldn't call that comprehensive). If its PTH then if it fails
you can
always tack a cap directly to the IC.


Thanks. I have the NatSemi Linear Application Handbook but
haven't seen where they cited the 3-inch figure (haven't looked).
Yes, it's TH.
 
J

Jeff Johnson

Jan 1, 1970
0
Pimpom said:
I know that, to ensure stability when using a standard 3-terminal
regulator, the output capacitor should be placed as close as possible to
the regulator. Usually this is no problem. But I'm in the middle of a PCB
design in which it's more convenient to have slightly over an inch of
track between the two. This is on both the hot and the ground sides. The
tracks are shared by load currents of a few mAs. Part of the load is a
CMOS counter IC operating at a few kHz while another section has further
RC filtering on the supply. Is this likely to cause problems?

I'm asking here instead of simply trying it out because I have to make a
dozen of this circuit. Even if the test unit works, I won't know if it's
only marginally stable and I don't want to risk having some units or
future replacements go unstable.


Is it possible to add a smaller cap that near the regulator? I doubt you'll
have any serious problems but make sure you bypass that counter. Increase
track thickness if you can and properly bypass(low ESR caps). The track
length isn't going to be a huge issue if done right.

If your not sure and are worried how bout including some SMT caps near the
regulator and near the bypassing just in case? These can be quite small,
have low ESR, and you don't have to add them until there is an issue.
 
N

Nemo

Jan 1, 1970
0
From National's 78XX data sheet at:
http://www.national.com/ds/LM/LM7512C.pdf

"Considerable effort was expanded to make the LM78XX series
of regulators easy to use and minimize the number of
external components. It is not necessary to bypass the output

I tried that with a NatSemi 7805 last year.

Went unstable. The next respin included a 22uF OSCON and some 1uF
ceramics, and it's fine (a bit ripply perhaps but usable).

I also found an old Analog Devices app note suggesting that to minimise
noise, put a ferrite bead between the output of the 7805 and its
decoupler. That might have been valid many years ago but it made the
regulator oscillate until I replaced the ferrite with a zero ohm resistor.

Nemo
 
S

Spehro Pefhany

Jan 1, 1970
0
I tried that with a NatSemi 7805 last year.

Went unstable. The next respin included a 22uF OSCON and some 1uF
ceramics, and it's fine (a bit ripply perhaps but usable).

I also found an old Analog Devices app note suggesting that to minimise
noise, put a ferrite bead between the output of the 7805 and its
decoupler. That might have been valid many years ago but it made the
regulator oscillate until I replaced the ferrite with a zero ohm resistor.

Nemo

I've never been able to make a 78x05 oscillate under sensible
conditions. We shipped >10K units with zero output bypass cap (benign
load of a few mA fairly steady).

Putting an inductor on the output without a bypass right at the
regulator would not be in my "sensible" category, though.


Best regards,
Spehro Pefhany
 
M

Michael Karas

Jan 1, 1970
0
Thanks. I have the NatSemi Linear Application Handbook but
haven't seen where they cited the 3-inch figure (haven't looked).
Yes, it's TH.


Since you are using TH mounted components you could consider making some
room on the side opposite the 78XX part to place a 0805 or even 0603
sized set of SM pads between the OUT and GND pins of the regulator.
These would disrupt the layout in a minimal way and then if it proved
necessary to have some cap installed there it can be added at the last
minute. For example if the cap that you plan for being ~1 inch away was
a 22uF cap the one placed on the SM pads could be maybe 0.1uF. In the
0805 and 0603 sizes there are a whole range of values of capacitance
that you can choose.
 
T

TheGlimmerMan

Jan 1, 1970
0
CO is not needed for stability; however,
it does improve transient response.

Providing a bit of forward feed?

Making it respond to regulatory stimulus faster.
 
I

I AM THAT I AM

Jan 1, 1970
0
put a ferrite bead between the output of the 7805 and its
decoupler.

Put it ON the lead of the regulator where it mounts.
 
R

Rich Grise

Jan 1, 1970
0
John said:
From National's 78XX data sheet at:

http://www.national.com/ds/LM/LM7512C.pdf

"Considerable effort was expanded to make the LM78XX series
of regulators easy to use and minimize the number of
external components. It is not necessary to bypass the output,
although this does improve transient response. Input bypassing
is needed only if the regulator is located far from the
filter capacitor of the power supply."

Also, Motorola's (ON Semiconductor) data sheet at:

http://www.onsemi.com/pub_link/Collateral/MC7800-D.PDF

states, on page 1:

"Cin is required if regulator is located an
appreciable distance from power supply
filter.

CO is not needed for stability; however,
it does improve transient response. Values
of less than 0.1
F could cause instability."

So, the bottom lines are:

1. Save yourself some rupees and blow off the output caps if the
regulator's transient response is fast enough to detect a large
change in load current and keep its output voltage stiff.

2. Save yourself some rupeess and blow off the input caps if the
inductance of the input leads/traces isn't high enough to cause the
input of the regulator to drop below its regulating threshold
voltage when a large change in load current invokes E = Ldi/dt.

I concur here, if your F above was meant to be microF. (uF, µF ;-)

The last time I used a 7812, it was on a hobbyist prototype, and
I hung a 0.1 uF cap on both the input and the output, tightly snugged
up against the regulator, with a lead length of about 0.1". But that
was when I was young and afraid of everything. :-D

It _did_ serve its purpose, however, of providing 12 VDC to the 9"
monitor I used for my "TV Typewriter," distortion-free. :)

Cheers!
Rich
 
P

Pimpom

Jan 1, 1970
0
Pimpom said:
I know that, to ensure stability when using a standard
3-terminal
regulator, the output capacitor should be placed as close as
possible to the regulator. Usually this is no problem. But I'm
in
the middle of a PCB design in which it's more convenient to
have
slightly over an inch of track between the two. This is on both
the hot and the ground sides. The tracks are shared by load
currents of a few mAs. Part of the load is a CMOS counter IC
operating at a few kHz while another section has further RC
filtering on the supply. Is this likely to cause problems?

I'm asking here instead of simply trying it out because I have
to
make a dozen of this circuit. Even if the test unit works, I
won't know if it's only marginally stable and I don't want to
risk having some units or future replacements go unstable.

Thanks for all the replies. I always had the impression that
weekends were slow times in technical NGs, but I was pleasantly
surprised to see that the two threads I started received quite a
bit of attention.To reply to the various posts, some with similar
contents:

No problem adding an SMD or a ceramic disc close to the regulator
on the copper side. I just don't like 'afterthought' additions if
they can be avoided. And yes, there's plenty of room to increase
the track sizes which are already 30 and 40 mils wide.

The output cap is a 0.1uF ceramic disc which I've always used as
a standard output cap for 78XXes except where the nature of the
load makes heavier buffering desireable. Here's the design as it
is now:
http://img148.imageshack.us/i/outputcap.png/
 
S

Spehro Pefhany

Jan 1, 1970
0
Thanks for all the replies. I always had the impression that
weekends were slow times in technical NGs, but I was pleasantly
surprised to see that the two threads I started received quite a
bit of attention.To reply to the various posts, some with similar
contents:

No problem adding an SMD or a ceramic disc close to the regulator
on the copper side. I just don't like 'afterthought' additions if
they can be avoided. And yes, there's plenty of room to increase
the track sizes which are already 30 and 40 mils wide.

The output cap is a 0.1uF ceramic disc which I've always used as
a standard output cap for 78XXes except where the nature of the
load makes heavier buffering desireable. Here's the design as it
is now:
http://img148.imageshack.us/i/outputcap.png/

I don't think you'll have any trouble with that, although personally I
would have tried to fit it between the 78xx and the chip, maybe by
sliding those power input pads up and down, or by moving the chip
left a bit.

It's been years since I've seen a PCB design done from the "bottom
view"-- used to do it all the time that way for single-layer boards.


Best regards,
Spehro Pefhany
 
P

Pimpom

Jan 1, 1970
0
Spehro said:
I don't think you'll have any trouble with that, although
personally I
would have tried to fit it between the 78xx and the chip, maybe
by
sliding those power input pads up and down, or by moving the
chip
left a bit.

It's been years since I've seen a PCB design done from the
"bottom
view"-- used to do it all the time that way for single-layer
boards.
I do most of my designs as single-layer TH. It's only in recent
years that we've had online PCB services here in India and the
nearest one is literally more than 2000 km away. To avoid weeks
of delay and 20x the expense, I still make most of my own PCBs
using iron-on transfers. I'm reluctant to use smaller than 15 or
20 mil tracks and I haven't yet found a source for carbide drill
bits. So I use single-sided phenolic boards with TH components
most of the time.
 
J

Jeff Johnson

Jan 1, 1970
0
Pimpom said:
I do most of my designs as single-layer TH. It's only in recent years that
we've had online PCB services here in India and the nearest one is
literally more than 2000 km away. To avoid weeks of delay and 20x the
expense, I still make most of my own PCBs using iron-on transfers. I'm
reluctant to use smaller than 15 or 20 mil tracks and I haven't yet found
a source for carbide drill bits. So I use single-sided phenolic boards
with TH components most of the time.

Why not attempt to do SMT? I have found that it's much easier and faster.
 
S

Sjouke Burry

Jan 1, 1970
0
Pimpom said:
Thanks for all the replies. I always had the impression that
weekends were slow times in technical NGs, but I was pleasantly
surprised to see that the two threads I started received quite a
bit of attention.To reply to the various posts, some with similar
contents:

No problem adding an SMD or a ceramic disc close to the regulator
on the copper side. I just don't like 'afterthought' additions if
they can be avoided. And yes, there's plenty of room to increase
the track sizes which are already 30 and 40 mils wide.

The output cap is a 0.1uF ceramic disc which I've always used as
a standard output cap for 78XXes except where the nature of the
load makes heavier buffering desireable. Here's the design as it
is now:
http://img148.imageshack.us/i/outputcap.png/
Should the output cap not be as close as possible to the
regulator?
 
E

ehsjr

Jan 1, 1970
0
Pimpom wrote:

I do most of my designs as single-layer TH. It's only in recent
years that we've had online PCB services here in India and the
nearest one is literally more than 2000 km away. To avoid weeks
of delay and 20x the expense, I still make most of my own PCBs
using iron-on transfers. I'm reluctant to use smaller than 15 or
20 mil tracks and I haven't yet found a source for carbide drill
bits.

Dental burs work well for drillng PCBs. Maybe your dentist will
give you some used burs to try.

Ed
 
P

Pimpom

Jan 1, 1970
0
Sjouke said:
Should the output cap not be as close as possible to the
regulator?

That's the whole point of the thread. Please read the first few
sentences of my opening post again.
 
P

Pimpom

Jan 1, 1970
0
ehsjr said:
Pimpom wrote:



Dental burs work well for drillng PCBs. Maybe your dentist will
give you some used burs to try.

Thanks for the suggestion. I work a lot with members of the
medical profession including dental surgeons, but the idea of
using a dental bit never occurred to me. If it's good enough to
drill teeth, it should be good enough to drill FR4.
 
S

Sjouke Burry

Jan 1, 1970
0
Pimpom said:
That's the whole point of the thread. Please read the first few
sentences of my opening post again.
Well, it sounds a bit like gambling to me, and maybe risking
the occasional oscillator.
But, if it works for you....
 
P

Pimpom

Jan 1, 1970
0
Jeff said:
Why not attempt to do SMT? I have found that it's much easier
and
faster.

I've been thinking about it for quite some time now. Maybe this
is a good time to start.

I've accumulated a wide range of parts over the years, most of
them bought on visits to the major cities and courtesy of
friends. It's almost impossible to deal with Indian retailers
from a distance and the bigger distributors ignore orders smaller
than production quantities. The biggest obstacle to switching to
SMT will be buying the parts.
 
S

Sjouke Burry

Jan 1, 1970
0
Pimpom said:
I've been thinking about it for quite some time now. Maybe this
is a good time to start.

I've accumulated a wide range of parts over the years, most of
them bought on visits to the major cities and courtesy of
friends. It's almost impossible to deal with Indian retailers
from a distance and the bigger distributors ignore orders smaller
than production quantities. The biggest obstacle to switching to
SMT will be buying the parts.
In the past I have seen some suppliers offer boxes with assorted
values surface mount components.
Try and find that.
 
Top