Noise problem I think

[Chris W]

I have 4 of these 16 bit serial LED drivers hooked together on
solderless proto boards.
http://www.allegromicro.com/datafile/6276.pdf

I have a RCM3110 module sending data to them and everything works great
till I have just over half of the LEDs on at a time. If I just turn
them all on and leave them on it's fine but if over half are on and some
are flashing, I get a few others near the ones that are supposed to be
flashing also flashing in what appears to be a random manor. I have
tried slowing down the clock pulses and increasing the time between data
changes and the clock going high, but that has had no effect. I don't
think it is the clock speed though since with less than half of them on
at a time I can have flashing sequences going VERY fast with no
problems. When I say fast I mean the data is clocked into the chips as
fast as the RCM3110 can send them. The scope meter tells me that with
no code to slow anything down, the clock is high for 2uS and then low
for about 15uS, giving a total of about 17uS per cycle or 58 kHz. I
don't know for sure how slow I have taken it, but I would guess that it
was far less than 1 kHz, increasing the high and low time as well as the
time the data was set before the clock goes high. I have done extensive
testing and debugging to confirm that the code is not the cause of the
problem. .

Is this something that will just go away when I put it on a PCB and the
lines between the pins are a lot shorter than the wires I am using now
to hook 4 of the together on 2 different proto boards? Or do I need to
add something to help prevent noise? I think this is only happening
when there are several (more than 32) 1s in a row but I'm not sure. Any
ideas?

--
Chris W

Gift Giving Made Easy
Get the gifts you want &
give the gifts they want
http://thewishzone.com

 

[Larry Brasfield]

I have 4 of these 16 bit serial LED drivers hooked together on solderless proto boards.
http://www.allegromicro.com/datafile/6276.pdf

I have a RCM3110 module sending data to them and everything works great till I have just over half of the LEDs on at a time. If I
just turn them all on and leave them on it's fine but if over half are on and some are flashing, I get a few others near the ones
that are supposed to be flashing also flashing in what appears to be a random manor. I have tried slowing down the clock pulses
and increasing the time between data changes and the clock going high, but that has had no effect. I don't think it is the clock
speed though since with less than half of them on at a time I can have flashing sequences going VERY fast with no problems. When
I say fast I mean the data is clocked into the chips as fast as the RCM3110 can send them. The scope meter tells me that with no
code to slow anything down, the clock is high for 2uS and then low for about 15uS, giving a total of about 17uS per cycle or 58
kHz. I don't know for sure how slow I have taken it, but I would guess that it was far less than 1 kHz, increasing the high and
low time as well as the time the data was set before the clock goes high. I have done extensive testing and debugging to confirm
that the code is not the cause of the problem. .

Is this something that will just go away when I put it on a PCB and the lines between the pins are a lot shorter than the wires I
am using now to hook 4 of the together on 2 different proto boards?

It could go away, but that would depend on your
PCB layout. The same effects can be brought on
by poor ground (and power) distribution.

Or do I need to add something to help prevent noise? I think this is only happening when there are several (more than 32) 1s in a
row but I'm not sure. Any ideas?

You could add redundant ground wires and a gob
of extra bypass capacitors to your proto board.
Given your description, I would expect to see the
threshold number of driven LEDs go up. If that
happens, you will know "ground bounce" is hard
at work bedeviling your circuit.

The solderless proto boards are often a waste of
time. They induce or force poor layout, introduce
a few pF between adjacent pins, and inevitably
have excessive coupling due to all the air wires
in proximity to each other. When I was starting
my electronics education, it went faster once I
realized that my proto boards should be shelved
for use only in DC or near DC circuits.

As ICs have become faster, those proto boards
have become all the more useless for the kinds
of circuits that are interesting by being busy.

 

[Chris W]

You could add redundant ground wires and a gob
of extra bypass capacitors to your proto board.
Given your description, I would expect to see the
threshold number of driven LEDs go up. If that
happens, you will know "ground bounce" is hard
at work bedeviling your circuit.

I don't have much in the way of capacitors around so I added 2, 2.2uFs
that I had around, as close to the Vcc and GND pins on 2 of the 4 LED
drivers as I could. That solved the problem, I could have almost all
of the LEDs on and just a few flashing (so I could detect a errors) and
didn't see any.

After that, it occurred to me to check the voltage drop from the PS.
The prototype board for the RCM3100 has a 5V source I was using to power
the LED drivers. After some testing I found that if I had more than
about 6 LED's on at once, the voltage from the PS started dropping. By
the time I had all the LEDs on the voltage was down to 4.8V and the
regulator chip was starting to get fairly warm. So I disconnected the
LED drivers from the prototype board of the RCM3100 and used a 4 cell
NiCd pack to power the LED drivers and LEDs. With that set up even with
out the capacitors it worked fine.

Makes for a good reminder why I never skimp when buying a power supply
for my computer. A bad power supply, or one that just isn't regulating
as well as it should, can cause all kind of random errors and it's one
of the last things most people think to check.

--
Chris W

Gift Giving Made Easy
Get the gifts you want &
give the gifts they want
http://thewishzone.com