Panel meter glitches

[johnt]

I'm connecting to a LED panel meter through a 16" 14-conductor ribbon cable.
The meter is set so that 2.5VDC will display as "100". Unfortunately any
electrical noise seems to make the display glitch for an instant. I suspect
that using twisted pair would greatly minimize this but I have to stick with
ribbon cable. Is there a way to remove the noise from the signal...perhaps
an inductor or capacitor?
thx, newbie

 

[CFoley1064]

Subject: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/20/2004 7:35 PM Central Standard Time
Message-id: <wj67c.31175$fg7.113@fe2.columbus.rr.com

I'm connecting to a LED panel meter through a 16" 14-conductor ribbon cable.
The meter is set so that 2.5VDC will display as "100". Unfortunately any
electrical noise seems to make the display glitch for an instant. I suspect
that using twisted pair would greatly minimize this but I have to stick with
ribbon cable. Is there a way to remove the noise from the signal...perhaps
an inductor or capacitor?
thx, newbie

Hi, John. It might be a good idea to create a low pass filter with a resistor
and cap, like this (read in fixed font, or M$ Notepad):

.---------------------.
+ ___ | |
o----|___|-----o------o + |
R | | |
--- | |
Voltage Source C --- | DPM |
| | |
o--------------o------o - |
- | |
| |
'---------------------'

Try 100 ohms for the resistor, and a 1000 pF (.001uF) ceramic cap for C. The
closer the R and C are to the DPM input, the better. Let us know how this
works.

Good luck
Chris

 

[johnt]

Once I get the DPM signal cleaned up I want to read it with the ADC of a
microcontroller (ATmega). Does this fact change the R and C that I should
try? The signal is essentially DC and I want to be able to read it once a
second.
john


"CFoley1064" <cfoley1064@aol.com> wrote in message
news:20040321002141.29951.00000109@mb-m29.aol.com...

Subject: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/20/2004 7:35 PM Central Standard Time
Message-id: <wj67c.31175$fg7.113@fe2.columbus.rr.com

I'm connecting to a LED panel meter through a 16" 14-conductor ribbon
cable.
The meter is set so that 2.5VDC will display as "100". Unfortunately
any
electrical noise seems to make the display glitch for an instant. I
suspect
that using twisted pair would greatly minimize this but I have to stick
with
ribbon cable. Is there a way to remove the noise from the
signal...perhaps
an inductor or capacitor?
thx, newbie

Hi, John. It might be a good idea to create a low pass filter with a
resistor
and cap, like this (read in fixed font, or M$ Notepad):

.---------------------.
+ ___ | |
o----|___|-----o------o + |
R | | |
--- | |
Voltage Source C --- | DPM |
| | |
o--------------o------o - |
- | |
| |
'---------------------'

Try 100 ohms for the resistor, and a 1000 pF (.001uF) ceramic cap for C.
The
closer the R and C are to the DPM input, the better. Let us know how this
works.

Good luck
Chris

 

[johnt]

I was able to get the 'noise' down to an acceptable level using a 100 ohm
resistor and a .22uf cap. I was able to look at it on a scope and the
spikes were huge.
Thx


"CFoley1064" <cfoley1064@aol.com> wrote in message
news:20040321002141.29951.00000109@mb-m29.aol.com...

Subject: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/20/2004 7:35 PM Central Standard Time
Message-id: <wj67c.31175$fg7.113@fe2.columbus.rr.com

I'm connecting to a LED panel meter through a 16" 14-conductor ribbon
cable.
The meter is set so that 2.5VDC will display as "100". Unfortunately
any
electrical noise seems to make the display glitch for an instant. I
suspect
that using twisted pair would greatly minimize this but I have to stick
with
ribbon cable. Is there a way to remove the noise from the
signal...perhaps
an inductor or capacitor?
thx, newbie

Hi, John. It might be a good idea to create a low pass filter with a
resistor
and cap, like this (read in fixed font, or M$ Notepad):

.---------------------.
+ ___ | |
o----|___|-----o------o + |
R | | |
--- | |
Voltage Source C --- | DPM |
| | |
o--------------o------o - |
- | |
| |
'---------------------'

Try 100 ohms for the resistor, and a 1000 pF (.001uF) ceramic cap for C.
The
closer the R and C are to the DPM input, the better. Let us know how this
works.

Good luck
Chris

 

[CFoley1064]

Subject: Re: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/24/2004 7:05 AM Central Standard Time
Message-id: <2If8c.1017$I61.293@fe2.columbus.rr.com

< from prior post>
Once I get the DPM signal cleaned up I want to read it with the ADC of a
microcontroller (ATmega). Does this fact change the R and C that I should
try? The signal is essentially DC and I want to be able to read it once a
second.
john
<from prior post>

I was able to get the 'noise' down to an acceptable level using a 100 ohm
resistor and a .22uf cap. I was able to look at it on a scope and the
spikes were huge.
Thx

Hi, John. I read your first message and assumed your noise was coming from
capacitive/inductive coupling from adjacent high voltage/high current wiring.
If you needed an 0.22uF cap to knock it down, I'd guess it's probably part of
your signal, which would give you another type of problem.

A lot of cheapie DPMs don't have much in the way of input protection for out of
range voltage spikes. Also, no MCU ADC input has much of any protection for
input voltages that exceed the power supply rails. If your input signal
exceeds the MCU power supply, you might cause the MCU to spit up and get lost,
or damage/destroy the device, depending on the amplitude of the spike.

You're going to need some real input protection. I'm going to assume you've
got a 5VDC power supply for the DPM and the microcontroller, that the voltage
spikes are pretty significant, and that you're only looking for 6-bit
resolution (100 digits full scale). That would mean you could live with, say
10 mV error. (View in fixed font or M$ Notepad):

DPM_ADC Buffer .------------.
| 1/2 LM358 |
| VCC |
Input Signal | |\| |
'-----|-\ |
___ ___ | >---o----------o
o---|___|--o---o---|___|------|+/ | Output Signal
1K | | 2.2K |/| .-.2.2 K
--- - | | | .---o
--- ^ 1N4732 === | | |
.22uF | | Vz=4.7V GND '-' |
| | | |
=== === === ===
GND GND GND GND

Try feeding your misbehavin' signal into this, and you should be OK. At least
you won't fry the ATmega.

Good luck
Chris

 

[CFoley1064]

Subject: Re: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/24/2004 7:05 AM Central Standard Time
Message-id: <2If8c.1017$I61.293@fe2.columbus.rr.com

< from prior post>
Once I get the DPM signal cleaned up I want to read it with the ADC of a
microcontroller (ATmega). Does this fact change the R and C that I should
try? The signal is essentially DC and I want to be able to read it once a
second.
john
<from prior post>

I was able to get the 'noise' down to an acceptable level using a 100 ohm
resistor and a .22uf cap. I was able to look at it on a scope and the
spikes were huge.
Thx

Hi, John. I read your first message and assumed your noise was coming from
capacitive/inductive coupling from adjacent high voltage/high current wiring.
If you needed an 0.22uF cap to knock it down, I'd guess it's probably part of
your signal, which would give you another type of problem.

A lot of cheapie DPMs don't have much in the way of input protection for out of
range voltage spikes. Also, no MCU ADC input has much of any protection for
input voltages that exceed the power supply rails. If your input signal
exceeds the MCU power supply, you might cause the MCU to spit up and get lost,
or damage/destroy the device, depending on the amplitude of the spike.

You're going to need some real input protection. I'm going to assume you've
got a 5VDC power supply for the DPM and the microcontroller, that the voltage
spikes are pretty significant, and that you're only looking for 6-bit
resolution (100 digits full scale). That would mean you could live with, say
10 mV error. It also assumes your maximum DC signal isn't going to be too much
more than 2.5V (if so, choose a rail-to-rail single supply op amp that can
handle
inputs below the negative rail (View in fixed font or M$ Notepad):

DPM_ADC Buffer .------------.
| 1/2 LM358 |
| VCC |
Input Signal | |\| |
'-----|-\ |
___ ___ | >---o----------o
o---|___|--o---o---|___|------|+/ | Output Signal
1K | | 2.2K |/| .-.2.2 K
--- - | | | .---o
--- ^ 1N4732 === | | |
.22uF | | Vz=4.7V GND '-' |
| | | |
=== === === ===
GND GND GND GND

Try feeding your misbehavin' signal into this, and you should be OK. At least
you won't fry the ATmega this way.

Good luck
Chris

 

[johnt]

What other type of problem can this 100-.22uf filter cause that I should
look out for? This seems to be working fine including reading it from the
ATmega. The remaining small glitch I can 'ignore' with software.

I do see another problem that I'll also need to resolve. If I have over
2VDC applied to the ADC input of the ATmega when it is powered up then the
digital inputs will be at an odd state. I read a rotary switch on power-up
and if >2VDC is going to the ADC then I get the wrong switch values. This
occurs whether the low-pass filter is in place or not. The ATmega has
pull-up resistors on the digital inputs but they are perhaps not strong
enough (swag)?

John

I was able to get the 'noise' down to an acceptable level using a 100 ohm
resistor and a .22uf cap. I was able to look at it on a scope and the
spikes were huge.
Thx

Hi, John. I read your first message and assumed your noise was coming
from
capacitive/inductive coupling from adjacent high voltage/high current
wiring.
If you needed an 0.22uF cap to knock it down, I'd guess it's probably part
of
your signal, which would give you another type of problem.

A lot of cheapie DPMs don't have much in the way of input protection for
out of
range voltage spikes. Also, no MCU ADC input has much of any protection
for
input voltages that exceed the power supply rails. If your input signal
exceeds the MCU power supply, you might cause the MCU to spit up and get
lost,
or damage/destroy the device, depending on the amplitude of the spike.

You're going to need some real input protection. I'm going to assume
you've
got a 5VDC power supply for the DPM and the microcontroller, that the
voltage
spikes are pretty significant, and that you're only looking for 6-bit
resolution (100 digits full scale). That would mean you could live with,
say
10 mV error. (View in fixed font or M$ Notepad):

DPM_ADC Buffer .------------.
| 1/2 LM358 |
| VCC |
Input Signal | |\| |
'-----|-\ |
___ ___ | >---o----------o
o---|___|--o---o---|___|------|+/ | Output Signal
1K | | 2.2K |/| .-.2.2 K
--- - | | | .---o
--- ^ 1N4732 === | | |
.22uF | | Vz=4.7V GND '-' |
| | | |
=== === === ===
GND GND GND GND

Try feeding your misbehavin' signal into this, and you should be OK. At
least
you won't fry the ATmega.

Good luck
Chris

 

[CFoley1064]

Subject: Re: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/24/2004 1:32 PM Central Standard Time
Message-id: <5nl8c.1843$9f6.1739@fe1.columbus.rr.com

What other type of problem can this 100-.22uf filter cause that I should
look out for? This seems to be working fine including reading it from the
ATmega. The remaining small glitch I can 'ignore' with software.

I do see another problem that I'll also need to resolve. If I have over
2VDC applied to the ADC input of the ATmega when it is powered up then the
digital inputs will be at an odd state. I read a rotary switch on power-up
and if >2VDC is going to the ADC then I get the wrong switch values. This
occurs whether the low-pass filter is in place or not. The ATmega has
pull-up resistors on the digital inputs but they are perhaps not strong
enough (swag)?

Microcontrollers can be frustrating sometimes. Generally, if they're
physically not doing what they're supposed to be doing, you have to wonder what
you're putting into the chip. Any digital or analog input which exceeds either
power supply rail (even for nanoseconds) can cause odd behavior, degrade the
reliability of the chip, or flat out destroy it. In addition, if the input is
not made to handle non-logic level inputs, applying an intermediate voltage
(say, 1.5V) can cause excess power dissipation and possible erratic results.

You have to back up the truck, and protect your processor. If you want to nuke
it, use a 74C914 buffer for all digital inputs with series 1K resistors between
your "real world" signals and the C914 inputs. Also, buffer your digital
outputs by tying the uC output pins to a digital buffer.

There are a lot of things you can/should do, but which ones are dependent on
the issues you have to address. Possibly this is the point where you might
start thinking if there's another way to do this.

Good luck
Chris

 

[johnt]

"CFoley1064" <cfoley1064@aol.com> wrote in message
news:20040325101726.03590.00000077@mb-m01.aol.com...

Subject: Re: Panel meter glitches
From: "johnt" noemail@yahoo.com
Date: 3/24/2004 1:32 PM Central Standard Time
Message-id: <5nl8c.1843$9f6.1739@fe1.columbus.rr.com

What other type of problem can this 100-.22uf filter cause that I should
look out for? This seems to be working fine including reading it from
the
ATmega. The remaining small glitch I can 'ignore' with software.

I do see another problem that I'll also need to resolve. If I have over
2VDC applied to the ADC input of the ATmega when it is powered up then
the
digital inputs will be at an odd state. I read a rotary switch on
power-up
and if >2VDC is going to the ADC then I get the wrong switch values.
This
occurs whether the low-pass filter is in place or not. The ATmega has
pull-up resistors on the digital inputs but they are perhaps not strong
enough (swag)?

Microcontrollers can be frustrating sometimes. Generally, if they're
physically not doing what they're supposed to be doing, you have to wonder
what
you're putting into the chip. Any digital or analog input which exceeds
either
power supply rail (even for nanoseconds) can cause odd behavior, degrade
the
reliability of the chip, or flat out destroy it. In addition, if the
input is
not made to handle non-logic level inputs, applying an intermediate
voltage
(say, 1.5V) can cause excess power dissipation and possible erratic
results.

You have to back up the truck, and protect your processor. If you want to
nuke
it, use a 74C914 buffer for all digital inputs with series 1K resistors
between
your "real world" signals and the C914 inputs. Also, buffer your digital
outputs by tying the uC output pins to a digital buffer.

There are a lot of things you can/should do, but which ones are dependent
on
the issues you have to address. Possibly this is the point where you
might
start thinking if there's another way to do this.

Good luck
Chris

It helps if you actually 'enable' the pull ups on the input! : )
Chris, thanks for the great information and feedback.
johnT