Strange quirk of the LM3914-1n

[Colin Dawson]

I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to 100%.
My circuit modelling software says that this should light the LED labelled
D11 (it's on pin 10 on the chip). However, on my breadboard version the
last LED, only gets enough power to come on, but not to full power like the
others.

How can I sort this out?

Regards

Colin Dawson.
www.cjdawson.com

 

[John Larkin]

On Sat, 27 Mar 2004 18:27:28 +0000 (UTC), "Colin Dawson"
<news@cjdawson.com> wrote:

I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to 100%.
My circuit modelling software says that this should light the LED labelled
D11 (it's on pin 10 on the chip). However, on my breadboard version the
last LED, only gets enough power to come on, but not to full power like the
others.

How can I sort this out?

Regards

Colin Dawson.
www.cjdawson.com

It's obvious what's wrong with this circuit:

IC1 should be U1

VR1 should be R1

SW1 should be S1

John

 

[John Woodgate]

I read in sci.electronics.design that Colin Dawson <news@cjdawson.com>
wrote (in <c44h2g$cg8$1@sparta.btinternet.com&gt about 'Strange quirk of
the LM3914-1n', on Sat, 27 Mar 2004:

I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to
100%. My circuit modelling software says that this should light the LED
labelled D11 (it's on pin 10 on the chip). However, on my breadboard
version the last LED, only gets enough power to come on, but not to full
power like the others.

How can I sort this out?

The chip is probably oscillating; they are prone to do this. Put a 100
nF directly across the supplies to the chip. The connections from RL,
OUT and ADJ to the 0 V pin need to be very short.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[Walter Harley]

"Colin Dawson" <news@cjdawson.com> wrote in message
news:c44h2g$cg8$1@sparta.btinternet.com...

I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to 100%.
My circuit modelling software says that this should light the LED labelled
D11 (it's on pin 10 on the chip). However, on my breadboard version the
last LED, only gets enough power to come on, but not to full power like
the
others.

How can I sort this out?

Read the datasheet, at http://www.national.com/ds/LM/LM3914.pdf,
particularly page 14, to find out more about dealing with oscillations.

As a side comment, I think your note "R2 needs to be 1/10th the value of
VR2" is wrong. There is no relationship between the values of R2 and VR2.
The only important thing about R2 is that it is somewhat less than the value
of the resistors in the internal voltage divider chain. Coincidentally,
those values are also 1k. But even if VR2 were 100k, the circuit behavior
would not be changed. The point of R2 is to make the voltage at the "top"
of the resistor chain (that is, the voltage at the RH pin) be slightly less
than the 10v reference, so that when the pot is at its maximum, the voltage
at the SIG pin is greater than the voltage at the RH pin. In fact, you
might want to increase R2 to 1k, and see if that helps your problem.

 

[Colin Dawson]

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:jtGNdCFq1eZAFw0O@jmwa.demon.co.uk...

I read in sci.electronics.design that Colin Dawson <news@cjdawson.com
wrote (in <c44h2g$cg8$1@sparta.btinternet.com&gt about 'Strange quirk of
the LM3914-1n', on Sat, 27 Mar 2004:
I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to
100%. My circuit modelling software says that this should light the LED
labelled D11 (it's on pin 10 on the chip). However, on my breadboard
version the last LED, only gets enough power to come on, but not to full
power like the others.

How can I sort this out?

The chip is probably oscillating; they are prone to do this. Put a 100
nF directly across the supplies to the chip. The connections from RL,
OUT and ADJ to the 0 V pin need to be very short.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Thanks john.

I've made the adjustment that you've said...

100 nf between pins 2 & 3 (v+ and v-)

Also I've shortened the wire from OUT and ADJ as much as I can. But it
doesn't apeared to make any difference.

Any more ideas?

Regards
Col.

 

[Fred Bloggs]

Colin Dawson wrote:

I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to 100%.
My circuit modelling software says that this should light the LED labelled
D11 (it's on pin 10 on the chip). However, on my breadboard version the
last LED, only gets enough power to come on, but not to full power like the
others.

How can I sort this out?

Regards

Colin Dawson.
www.cjdawson.com

Here are the standout observations:
1) The LM3914 does not like SIG inputs exceeding V+-1.5. In your case
V+=12-0.7=11.3 so that V+-1.5=9.8V. Coincidentally, this is about the
point where you are supposed to trip the top tap LED- which is roughly
10*(1-100/10.1K)=9.9V.
2) The LM3914 LED drivers are 10x current mirrors of the REF OUT
current. Your schematic shows REF OUT to be shorted to GND, and ADJ
grounded through a 4.7K. This should be changed.
3)V+=~12V is a bit rough on the thermal comfort of the IC when you draw
any significant current through the LEDs in bargraph mode. If the LED Vf
is 1.8V then the LED driver voltage must be 11.3-1.8=9.5, so that for a
LED current of 10mA, the driver dissipates 10mA x 9.5V=95mW, then in
bargraph mode that would 10 x 95mW=950mW total at maximum input-which is
a bit much.

Your application is a ratio measurement which means you want the
bargraph to display the ratio of VR2 pot setting to full scale. This
means there is nothing critical about the 10V. You should place a 220R
between the 10V source and the top junction of VR2 and R2. This will
lower the voltage at the junction to ~ 8/10 *10=8V and is not critical,
but it does prevent you from exceeding V+-1.5 by a healthy margin. A
0.47u from this junction to ground would be a nice touch also.
Moving on to the power dissipation problem, placing a resistor in series
with each LED is a possible alternative. Assuming you want 10mA per LED
with Vf at 1.8V, then a resistor that drops the driver voltage to 1.5V
would be R=(11.3-1.8-1.5)V/10mA=820R 1/4Watt power rating. Now the
maximum dissipation is 1.5V x 10mA x 10LEDs=150mW, which is much better.
The 10mA LED operating level is set by loading REF OUT= 10mA/10=1mA.
Since REF OUT=1.28V typically, a 1.2K resistor from REF OUT to GND is
required, and GND the ADJ pin. Finally, it would be better to increase
R2 to at least 1/4 of Rhi-lo/10=270 ohms typically.
See if this doesn't work better.

 

[Robert Baer]

Fred Bloggs wrote:

Colin Dawson wrote:
I've built a bar graph circuit on a bread board. The circuit diagram is
available at

http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

Everything seems to work OK, until set the 1k Pot (Labelled VR2) to 100%.
My circuit modelling software says that this should light the LED labelled
D11 (it's on pin 10 on the chip). However, on my breadboard version the
last LED, only gets enough power to come on, but not to full power like the
others.

How can I sort this out?

Regards

Colin Dawson.
www.cjdawson.com



Here are the standout observations:
1) The LM3914 does not like SIG inputs exceeding V+-1.5. In your case
V+=12-0.7=11.3 so that V+-1.5=9.8V. Coincidentally, this is about the
point where you are supposed to trip the top tap LED- which is roughly
10*(1-100/10.1K)=9.9V.
2) The LM3914 LED drivers are 10x current mirrors of the REF OUT
current. Your schematic shows REF OUT to be shorted to GND, and ADJ
grounded through a 4.7K. This should be changed.
3)V+=~12V is a bit rough on the thermal comfort of the IC when you draw
any significant current through the LEDs in bargraph mode. If the LED Vf
is 1.8V then the LED driver voltage must be 11.3-1.8=9.5, so that for a
LED current of 10mA, the driver dissipates 10mA x 9.5V=95mW, then in
bargraph mode that would 10 x 95mW=950mW total at maximum input-which is
a bit much.

Your application is a ratio measurement which means you want the
bargraph to display the ratio of VR2 pot setting to full scale. This
means there is nothing critical about the 10V. You should place a 220R
between the 10V source and the top junction of VR2 and R2. This will
lower the voltage at the junction to ~ 8/10 *10=8V and is not critical,
but it does prevent you from exceeding V+-1.5 by a healthy margin. A
0.47u from this junction to ground would be a nice touch also.
Moving on to the power dissipation problem, placing a resistor in series
with each LED is a possible alternative. Assuming you want 10mA per LED
with Vf at 1.8V, then a resistor that drops the driver voltage to 1.5V
would be R=(11.3-1.8-1.5)V/10mA=820R 1/4Watt power rating. Now the
maximum dissipation is 1.5V x 10mA x 10LEDs=150mW, which is much better.
The 10mA LED operating level is set by loading REF OUT= 10mA/10=1mA.
Since REF OUT=1.28V typically, a 1.2K resistor from REF OUT to GND is
required, and GND the ADJ pin. Finally, it would be better to increase
R2 to at least 1/4 of Rhi-lo/10=270 ohms typically.
See if this doesn't work better.

To further reduce that dissipation significantly, increase the LED
resistors to get only 5mA current; there are many bright LEDs that are
easily visible at currents as low as 2mA.
An alernate way is to keep the current at 10mA, but pulse/modulate the
LED supply on and off.

 

[John Woodgate]

I read in sci.electronics.design that Colin Dawson <news@cjdawson.com>
wrote (in <c44vk0$rkh$1@sparta.btinternet.com&gt about 'Strange quirk of
the LM3914-1n', on Sat, 27 Mar 2004:

I've made the adjustment that you've said...

100 nf between pins 2 & 3 (v+ and v-)

Also I've shortened the wire from OUT and ADJ as much as I can. But it
doesn't apeared to make any difference.

Any more ideas?

Do you have, or can you borrow, an oscilloscope to look for r.f.
oscillation?

Have you tried another LED in the 'dim' position?

Can you measure the current in the 'dim' LED and in a bright one? You
will need a 100 nF cap across the meter leads at the LED end, otherwise
you may get oscillation.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[Colin Dawson]

----- Original Message -----
From: "John Woodgate" <jmw@jmwa.demon.contraspam.yuk>
Newsgroups: sci.electronics.design
Sent: Sunday, March 28, 2004 10:21 AM
Subject: Re: Strange quirk of the LM3914-1n

I read in sci.electronics.design that Colin Dawson <news@cjdawson.com
wrote (in <c44vk0$rkh$1@sparta.btinternet.com&gt about 'Strange quirk of
the LM3914-1n', on Sat, 27 Mar 2004:
I've made the adjustment that you've said...

100 nf between pins 2 & 3 (v+ and v-)

Also I've shortened the wire from OUT and ADJ as much as I can. But it
doesn't apeared to make any difference.

Any more ideas?

Do you have, or can you borrow, an oscilloscope to look for r.f.
oscillation?

Have you tried another LED in the 'dim' position?

Can you measure the current in the 'dim' LED and in a bright one? You
will need a 100 nF cap across the meter leads at the LED end, otherwise
you may get oscillation.

No chance of me gaining access to an oscilloscope.

I've tried another LED, and that also show the same symtoms.

That brings me on to the next problem.... I just blew the chip. I don't
quite know what happened, but I'm getting really frustrated with this
circuit now. I'm beginning to regret trying to do this. Back to the
drawing board I think. I think it would be wize for me to start again from
scratch.

I'm trying to make a circuit that I can hook up to a Velleman K8004 Pulse
Width Modulator...

http://www.velleman.be/Downloads/0/Manual_K8004.pdf

I've build this kit and it works perfectly. I'm adding a 1K Pot as shown
in Fig C so that I can manually control the pusle width, and thus the time
that the load is turned on. I want to measure the position of the 1k pot and
display a bar graph, of the power setting of the PWM circuit.

Having looked again at the circuit diagram, I've just realized that I'd
designed the wrong circuit anyway. My circuit will need only 3 inputs,
and not the 5 as I've previous though. (the two that I can drop are a
"clean" power source) My 3 inputs will be the VREF, IN and GND as supplied
from the K8004 kit. I think this will simplify things alot


Referring back to the lm3913 datasheet.

http://www.national.com/ds/LM/LM3914.pdf

The circuit that I need to build is similar to the typical application that
they give on page 2. I'll take a close look at this and try to mod my
diagram, to fit the bill. (if I'm quick I can get to the shop and get the
components) Hopefully I'll get it right this time. I'll keep you all
posted on this, and hopefully it'll be useful for someone else whose trying
the same thing.

Regards

Colin.

 

[Colin Dawson]

That brings me on to the next problem.... I just blew the chip. I don't
quite know what happened, but I'm getting really frustrated with this
circuit now. I'm beginning to regret trying to do this. Back to the
drawing board I think. I think it would be wize for me to start again
from
scratch.

I'm trying to make a circuit that I can hook up to a Velleman K8004 Pulse
Width Modulator...

http://www.velleman.be/Downloads/0/Manual_K8004.pdf

I've build this kit and it works perfectly. I'm adding a 1K Pot as shown
in Fig C so that I can manually control the pusle width, and thus the time
that the load is turned on. I want to measure the position of the 1k pot
and
display a bar graph, of the power setting of the PWM circuit.

Having looked again at the circuit diagram, I've just realized that I'd
designed the wrong circuit anyway. My circuit will need only 3 inputs,
and not the 5 as I've previous though. (the two that I can drop are a
"clean" power source) My 3 inputs will be the VREF, IN and GND as supplied
from the K8004 kit. I think this will simplify things alot


Referring back to the lm3913 datasheet.

http://www.national.com/ds/LM/LM3914.pdf

The circuit that I need to build is similar to the typical application
that
they give on page 2. I'll take a close look at this and try to mod my
diagram, to fit the bill. (if I'm quick I can get to the shop and get the
components) Hopefully I'll get it right this time. I'll keep you all
posted on this, and hopefully it'll be useful for someone else whose
trying
the same thing.

Have read and re-read the LM3914 pdf file, I think that I've got a circuit
that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

My simulation software (newly bought) shows that all the parts on the chip
runs within the tolerances
set by the LM3914.pdf. The only exception is that the Ref Adj is using
250uA when the datasheet
says it should be 120uA max. But altering the circuit to do that will mean
that none of the LED's will get enough power.

At the moment they'll be running about 8.18mA each. Which I think is
reasonable.

Can someone confirm that I've done the math correctly this time? I'm
already running out of money

Regards

Col.

 

[Fred Bloggs]

Colin Dawson wrote:

Have read and re-read the LM3914 pdf file, I think that I've got a
circuit that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

You're blowing the IC with excessive power dissipation- your circuit
should look like this:
Please view in a fixed-width font such as Courier.



1N4007 1K
12V ------|>|-+---------------------------+----|>|--/\/\-------+
| |10x o |
| |LEDs o |
| | o 1K |
| +----|>|--/\/\-------+
| 10u | 1K |
| +----+------||---+----|>|--/\/\-------+
| | | + | |
| -+- | | |
K8004 | /// | 100n | LM3914 |
+-------------|--+ +------||---+ +-------------+ |
| V+ o | | +-+V+ | |
| | === +-+MODE | |
| | 100 |470n 270 | | |
| VREF o ---/\/\--+----+-----/\/\----+RH | /10
| | | | | |
| | | 10K | | |
| | 1K / +-----/\/\--+-+SIG | |
| | \ | | | | |
| DC IN o----+--+> / | +-------+ | | |
| | | | \ | | | | |
| | | | | | === | | |
| | | | | | |100n | |----+
| | | +--|----+ | | |
| |=== | +--/\/\---+REF |
| | |100n | | 1.5K | |
| | | | | |
| | | | +---------+RLO |
| | | | | | |
| GND o----+-----+--------+---------+ADJ |
| | | | | |
+----------------+ | | | |
| +---------+V- |
| | | |
--- --- +-------------+
/// ///

 

[Walter Harley]

"Colin Dawson" <news@cjdawson.com> wrote in message
news:c46vh5$qfl$1@sparta.btinternet.com...

I'm trying to make a circuit that I can hook up to a Velleman K8004
Pulse
Width Modulator...

http://www.velleman.be/Downloads/0/Manual_K8004.pdf

[...]

Have read and re-read the LM3914 pdf file, I think that I've got a circuit
that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

It seems to me like you're not thinking quite clearly about the problem. Or
maybe it's me that doesn't understand. From what I can tell, you shouldn't
be thinking about trying to "measure a pot"; what you're trying to do is
measure a *voltage*. Specifically, you're trying to measure the voltage at
the "DC in" connection of the Velleman circuit; and you're trying to compare
it to the voltage at the "Vref" connection.

You should be driving the LM3914 circuit from the same power supply as the
Velleman circuit (otherwise, you're liable to blow more chips). So, the
connections between the Velleman circuit and the LM3914 circuit will be V+,
GND, "Vref", and "DC in".

At the LM3914, "Vref" will go to a 100 ohm resistor and the other end of the
resistor will go to the LM3914's "RHI" input. "DC in" will go to the
LM3914's "SIG IN" input. V+ and GND go to V+ and V-. "REF ADJ" and "RLO"
are both grounded. "REF OUT" goes to a resistor and then to ground; the
value of the resistor determines the LED brightness. Mode pin gets
connected to V+, for bar-graph mode. Other than the LED connections, that
should be all you need.

IN PARTICULAR: you should NOT connect "REF OUT" to "RHI". That's only
needed when you want to use the internal reference; but in this case, you
don't want that, because you have an external reference (namely, Vref).

 

[Colin Dawson]

"Fred Bloggs" <nospam@nospam.com> wrote in message
news:40671A23.4000305@nospam.com...

Colin Dawson wrote:
Have read and re-read the LM3914 pdf file, I think that I've got a
circuit that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

You're blowing the IC with excessive power dissipation- your circuit
should look like this:

So that's why my chip just died.

Please view in a fixed-width font such as Courier.



1N4007 1K
12V ------|>|-+---------------------------+----|>|--/\/\-------+
| |10x o |
| |LEDs o |
| | o 1K |
| +----|>|--/\/\-------+
| 10u | 1K |
| +----+------||---+----|>|--/\/\-------+
| | | + | |
| -+- | | |
K8004 | /// | 100n | LM3914 |
+-------------|--+ +------||---+ +-------------+ |
| V+ o | | +-+V+ | |
| | === +-+MODE | |
| | 100 |470n 270 | | |
| VREF o ---/\/\--+----+-----/\/\----+RH | /10
| | | | | |
| | | 10K | | |
| | 1K / +-----/\/\--+-+SIG | |
| | \ | | | | |
| DC IN o----+--+> / | +-------+ | | |
| | | | \ | | | | |
| | | | | | === | | |
| | | | | | |100n | |----+
| | | +--|----+ | | |
| |=== | +--/\/\---+REF |
| | |100n | | 1.5K | |
| | | | | |
| | | | +---------+RLO |
| | | | | | |
| GND o----+-----+--------+---------+ADJ |
| | | | | |
+----------------+ | | | |
| +---------+V- |
| | | |
--- --- +-------------+
/// ///

I'd put this into my circuit program, and everything seems to be OK. Looks
like another trip to maplins tomorrow for parts.

The simulation that I've got hits the 100% mark dead on, which is great, but
as I turn down the Pot, it get so 30% power and the last led goes out....
how can I trim the range that the circuit is monitoring?

Thanks to everyone for the help so far.

Regards

Colin.

 

[Colin Dawson]

"Fred Bloggs" <nospam@nospam.com> wrote in message
news:40671A23.4000305@nospam.com...

Colin Dawson wrote:
Have read and re-read the LM3914 pdf file, I think that I've got a
circuit that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

You're blowing the IC with excessive power dissipation- your circuit
should look like this:
Please view in a fixed-width font such as Courier.



1N4007 1K
12V ------|>|-+---------------------------+----|>|--/\/\-------+
| |10x o |
| |LEDs o |
| | o 1K |
| +----|>|--/\/\-------+
| 10u | 1K |
| +----+------||---+----|>|--/\/\-------+
| | | + | |
| -+- | | |
K8004 | /// | 100n | LM3914 |
+-------------|--+ +------||---+ +-------------+ |
| V+ o | | +-+V+ | |
| | === +-+MODE | |
| | 100 |470n 270 | | |
| VREF o ---/\/\--+----+-----/\/\----+RH | /10
| | | | | |
| | | 10K | | |
| | 1K / +-----/\/\--+-+SIG | |
| | \ | | | | |
| DC IN o----+--+> / | +-------+ | | |
| | | | \ | | | | |
| | | | | | === | | |
| | | | | | |100n | |----+
| | | +--|----+ | | |
| |=== | +--/\/\---+REF |
| | |100n | | 1.5K | |
| | | | | |
| | | | +---------+RLO |
| | | | | | |
| GND o----+-----+--------+---------+ADJ |
| | | | | |
+----------------+ | | | |
| +---------+V- |
| | | |
--- --- +-------------+
/// ///

I've just done some measurements on this circuit, and found a couple of
errors in my translation from the ASCII. I see there's no need for a low
end trim, as it's already done.

I'm sill a little concerned though, having read the data sheet it says that
there shouldn't be more than 120uA at the Reg Adj pin. According to the
simulation, it says that there is 832.23uA going into this pin. Won't this
damage the chip? I've already broken several of them and don't really want
to kill another.

Regards

Colin.

 

[Colin Dawson]

The simulation that I've got hits the 100% mark dead on, which is great,
but
as I turn down the Pot, it get so 30% power and the last led goes out....
how can I trim the range that the circuit is monitoring?

I've just updated the
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg image with
this circuit.

Having played with the simulation for a little while. I'm really impressed
with the both the design of this circuit and the simulation software... the
program shows that the LED will be brighter or about 1/4 of a second when it
first lights, then settle into the same intensity as the rest of the LEDs.

Also having corrected a couple of errors when I was copying in from the
ASCII above, the LED's follow the Pot over the full range. My mistake -
I've been making lots lately.

I just want to make sure that the both my translation of the circuit is 100%
and the power disipation problem has gone.

Thanks again for the help.

Colin.

 

[Colin Dawson]

It seems to me like you're not thinking quite clearly about the problem.
Or
maybe it's me that doesn't understand. From what I can tell, you
shouldn't
be thinking about trying to "measure a pot"; what you're trying to do is
measure a *voltage*. Specifically, you're trying to measure the voltage
at
the "DC in" connection of the Velleman circuit; and you're trying to
compare
it to the voltage at the "Vref" connection.

That's exactly want I'm trying to do. It's been about 10 years since I did
any serious
electronics. I know that the description was rather crude, but from this
you've obviously
understood exactly what I'm trying to accomplish.

You should be driving the LM3914 circuit from the same power supply as the
Velleman circuit (otherwise, you're liable to blow more chips). So, the
connections between the Velleman circuit and the LM3914 circuit will be
V+,
GND, "Vref", and "DC in".

That is the intention. The Vellman kit will be used as a Heater controller,
connected to an
85Ah Deep Cycle Lead Acid Battery. It'll be controlling a little 60W
heater that will be wrapped
around my telescope. I know that the K8004 is up to the job as I'm
currently using one at the moment
but it doesn't have the nice bar graph display that I'm trying to get
together.

At the LM3914, "Vref" will go to a 100 ohm resistor and the other end of
the
resistor will go to the LM3914's "RHI" input. "DC in" will go to the
LM3914's "SIG IN" input. V+ and GND go to V+ and V-. "REF ADJ" and "RLO"
are both grounded. "REF OUT" goes to a resistor and then to ground; the
value of the resistor determines the LED brightness. Mode pin gets
connected to V+, for bar-graph mode. Other than the LED connections, that
should be all you need.

IN PARTICULAR: you should NOT connect "REF OUT" to "RHI". That's only
needed when you want to use the internal reference; but in this case, you
don't want that, because you have an external reference (namely, Vref).

Another kind member of the newsgroup has posted up a circuit diagram, I've
copied this out
and it seems to do the trick. All of the points you mention are in the
circuit. My copied out version is at
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg It's
amazing how great minds think alike.

Regards

Colin Dawson.

 

[Walter Harley]

"Colin Dawson" <news@cjdawson.com> wrote in message
news:c47h28$2b1$1@sparta.btinternet.com...

I'm sill a little concerned though, having read the data sheet it says
that
there shouldn't be more than 120uA at the Reg Adj pin. According to the
simulation, it says that there is 832.23uA going into this pin. Won't
this
damage the chip? I've already broken several of them and don't really
want
to kill another.

Three principles to understand here.

1. The "120uA max" is not a limit that *you* need to worry about; it's a
promise that the manufacturer is making *to* you. The manufacturer is
promising that the chip won't send more than 120uA out that pin, assuming
you don't do anything unusual. Sort of like input bias current for an
op-amp. How can you tell? Because in the datasheet, it's listed under
"electrical characteristics" rather than "absolute maximum ratings". The
AMR's are the things that you need to observe in order to not blow up the
chip; the EC's are the things that the manufacturer does for you.

The reason why they specify this is that, if you are using a voltage divider
on the output, this current results in an error term. See "Internal Voltage
Reference" on page 7 of the datasheet for details.


2. Generally speaking, putting <1mA through anything won't hurt it. There
are exceptions, but as a rule of thumb, that amount of current is too small
to worry about.


3. Simulation software is something you have to learn to work with. It
tends to be a bit alarmist. It's worthwhile to pay attention to what it
says; but particularly when it comes to models of IC's, the answers you get
tend to be a bit different than what happens in the real world. This is for
two reasons: first, the models are incomplete and imperfect; second, the
software assumes perfect components by default, and in the real world
components are never quite perfect. For instance, a voltage source in
simulation software has zero series resistance unless you specify otherwise;
but real voltage sources always have some series resistance. And so on.

So, you should look at that 832.23uA and first off read it as "around 800uA"
(because five significant digits is way more than you really have - don't
tell me your power source is exactly 12.000V and you're using .001%
precision resistors); and second off, say "hmm, I wonder why it thinks
that". Sometimes the answer will be "there's something wrong with the
circuit"; sometimes the answer will be "there's something wrong with the
simulation".

 

[Fred Bloggs]

Colin Dawson wrote:

The simulation that I've got hits the 100% mark dead on, which is great,

but

as I turn down the Pot, it get so 30% power and the last led goes out....
how can I trim the range that the circuit is monitoring?



I've just updated the
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg image with
this circuit.

Having played with the simulation for a little while. I'm really impressed
with the both the design of this circuit and the simulation software... the
program shows that the LED will be brighter or about 1/4 of a second when it
first lights, then settle into the same intensity as the rest of the LEDs.

Also having corrected a couple of errors when I was copying in from the
ASCII above, the LED's follow the Pot over the full range. My mistake -
I've been making lots lately.

I just want to make sure that the both my translation of the circuit is 100%
and the power disipation problem has gone.

Thanks again for the help.

Colin.

Yep- that circuit translation is good. Now that I know what you're
trying to do- a better circuit is one that measures the actual power
level into the heater rather than make inference from the potentiometer
setting with the unknown gains and offsets in the K8004. As long as you
are going to the trouble, it is worth adding just a few more components
to make it perfect. The ASCII art explains the program:
Please view in a fixed-width font such as Courier.



FIRST IDEA:
VOLTAGE AT HTR (-)
|
|
VBATT >--------------------+ VBATT| ---+ +----+ +----+ HTR OFF
GND >--------------------|-+ | | | | | |
| | | | | | | |
K8004 | | | | | | | |
+--------------+ | | 0V| +--+ +--+ +-HTR ON
| | | | -+--------------------------
| V+ o-------+ | time->
| | | |
| GND o---------+ |
| | HTR |
| + o---------o-/\/\/\/\/\/\-o-----+ <--+
| LOAD | + - |
| - o------------------------------+
| |
| |
| | +--------------> to LM3914 PWR
| | | *
| | 100 | 1K POT bargraph ckt
| VREF o------/\/\-+-/\/\/\--+
| | ^ |
| | | 100n|
| DC IN o---------------+--||-+
| | |
| GND o---------------------+
| |
+--------------+
* POT adjusts ratio of heater
ON/(ON+OFF) or avg pwr to HTR




SECOND IDEA:



2
Vbatt
HTR Maximum PWR= Pmax = ------
HTR R


ON
Duty cycle= -------- adjusted by 1K POT and,
ON + OFF


HTR PWR
Average HTR PWR= D x Pmax so that D= ------- or,
Pmax


D= fraction of full power. Therefore it is better


to use LM3914 to display D in place of 1K POT tap.


Q. How to use LM3914 ckt to display D?

A. Use circuit below.




K8004
-----------+
|
|
| HTR
+ o--------------+-----------------o-/\/\/\/\/\/\-o-----+
LOAD | | + - |
G - o----------------+------------------------------------+
N I | | |
D N VREF | | +-------+
-+--+----+----+ 1U | |
| | | +---||---+----+------------+----+-----+
| | | | 0.1U | | | | / |
| v 1K | +---||---+ | | | 10K |
+-/\/\--+ | | / / / / c
| D ADJ | | 470 2.2K 20K0 \ |/
+--------------+ | / / / +---|BC547
--- | \ \ \ e |\
/// e | | | |/ e
\| | | +---+--|BC556 |
BC556 |--+----+ | | |\ +-->LM3914
/| | / c RHI
c === 20K0 |
| 330n / |
/ | \ |
1K0 | | |
/ +---+----+
\ |
| 100K |
+--/\/\--+----------------------> LM3914
| RF | | SIG
/ | CF |
1K0 === |
/ 330n |
\ | |
| | |
+--------+--------+
|
---
///

Vbatt
Then LM3914 SIG = D x ----- as averaged by RF/CF and
2
Vbatt
LM3914 RHI= ----- so that LM3914 bargraphs D.
2




1K
VBATT>-----------+----|>|--/\/\------\
|10x o |
|LEDs o |
| o 1K |
+----|>|--/\/\------\|
10u | 1K |
+---+------||---+----|>|--/\/\------\|
| | + | |
--- | | |
/// | 100n | LM3914 |
+------||---+ +-------------+ |
| +-+V+ | |
== +-+MODE | |
|470n 270 | | |
RHI >------+-----/\/\----+RHI | /10
| | |
| | |
SIG >--------------------+SIG | |
| | |
| | |
| | |
| | |
| |----+
| |
+--/\/\---+REF |
| 1.5K | |
| | |
+---------+RLO |
| | |
+---------+ADJ |
| | |
| | |
+---------+V- |
| | |
--- +-------------+
///

 

[Fred Bloggs]

Fred Bloggs wrote:
[...snip...]

Note: K8004 frequency should be adjusted > 1KHz, and internal RV1/RV2
should be adjusted so that 1K pot range will cover 0-100% duty cycle
with maximum turns.

 

[Fred Bloggs]

Colin Dawson wrote:

"Fred Bloggs" <nospam@nospam.com> wrote in message
news:40671A23.4000305@nospam.com...


Colin Dawson wrote:
Have read and re-read the LM3914 pdf file, I think that I've got a
circuit that will do the job...
http://www.cjdawson76.btinternet.co.uk/circuits/potdrivenbar.jpg

You're blowing the IC with excessive power dissipation- your circuit
should look like this:
Please view in a fixed-width font such as Courier.



1N4007 1K
12V ------|>|-+---------------------------+----|>|--/\/\-------+
| |10x o |
| |LEDs o |
| | o 1K |
| +----|>|--/\/\-------+
| 10u | 1K |
| +----+------||---+----|>|--/\/\-------+
| | | + | |
| -+- | | |
K8004 | /// | 100n | LM3914 |
+-------------|--+ +------||---+ +-------------+ |
| V+ o | | +-+V+ | |
| | === +-+MODE | |
| | 100 |470n 270 | | |
| VREF o ---/\/\--+----+-----/\/\----+RH | /10
| | | | | |
| | | 10K | | |
| | 1K / +-----/\/\--+-+SIG | |
| | \ | | | | |
| DC IN o----+--+> / | +-------+ | | |
| | | | \ | | | | |
| | | | | | === | | |
| | | | | | |100n | |----+
| | | +--|----+ | | |
| |=== | +--/\/\---+REF |
| | |100n | | 1.5K | |
| | | | | |
| | | | +---------+RLO |
| | | | | | |
| GND o----+-----+--------+---------+ADJ |
| | | | | |
+----------------+ | | | |
| +---------+V- |
| | | |
--- --- +-------------+
/// ///




I've just done some measurements on this circuit, and found a couple of
errors in my translation from the ASCII. I see there's no need for a low
end trim, as it's already done.

I'm sill a little concerned though, having read the data sheet it says that
there shouldn't be more than 120uA at the Reg Adj pin. According to the
simulation, it says that there is 832.23uA going into this pin. Won't this
damage the chip? I've already broken several of them and don't really want
to kill another.

Regards

Colin.

I am mystified over your fixation with this ADJ pin current. Internally
this is the reference voltage for REF OUT. This pin should be GND'd and
it supplies about 80uA typically into GND. That 800uA+ current you
measure sounds suspiciously like a factor 10 error in the simulation
program model- it is nothing to worry about- those people are nothing
but mistakes.