# Strange Problem with LM3914

#### PFebby

Joined Feb 19, 2018
3
Hello all,

I am building a phase correlation meter using an LM3914 to drive an LED bar display. Everything is working correctly besides the LM3914. The LEDs light but they slowly cycle one at a time all the way up and all the way down the bar as if there’s a step time. I have no clue why this is happening or where the step time would be coming from considering the circuit was working completely fine a few days ago. I am using two equal sine waves as the inputs for the phase meter and as increase the frequency of the sine waves the step time of the LEDs seems to increase as well. Why is this happening and how can I fix it?

#### LesJones

Joined Jan 8, 2017
2,599
Are you sure the "two equal sine waves" are realy EXACTLY the same frequency.

Les.

#### crutschow

Joined Mar 14, 2008
24,709
Feed the two inputs from the same sinewave.

#### ebeowulf17

Joined Aug 12, 2014
3,274
Hello all,

I am building a phase correlation meter using an LM3914 to drive an LED bar display. Everything is working correctly besides the LM3914. The LEDs light but they slowly cycle one at a time all the way up and all the way down the bar as if there’s a step time. I have no clue why this is happening or where the step time would be coming from considering the circuit was working completely fine a few days ago. I am using two equal sine waves as the inputs for the phase meter and as increase the frequency of the sine waves the step time of the LEDs seems to increase as well. Why is this happening and how can I fix it?
Could you share a schematic of how you've got things right now? Otherwise anything we say will be uninformed guesses.

You say everything else is working - how do you know? Are you measuring the voltage at the input to the LM3914? Are you sure the LM3914 isn't working properly?

I see @LesJones posted while I was typing. He raises a good point - maybe the whole circuit is working properly and just showing you an unexpected result!

#### PFebby

Joined Feb 19, 2018
3
I see @LesJones posted while I was typing. He raises a good point - maybe the whole circuit is working properly and just showing you an unexpected result!
Ahhhh interesting this result would actually make some sense if the two are slightly different frequencies. They are from two separate function generators so it's possible. I will look into it and report back.

This is the design I'm using:

However, I used the LM3914 design in the picture attached instead since i had a ten bar LED display and wasn't sure why the 1N4148's were required coming from the 9th LED.

Last edited:

#### Alec_t

Joined Sep 17, 2013
11,119
wasn't sure why the 1N4148's were required coming from the 9th LED.
Because for some reason the circuit designer wanted both output 9 and output 10 to drive the 9th LED. The diodes perform an OR function.

#### ebp

Joined Feb 8, 2018
2,332
The diodes just "OR" the top two outputs so the LED is ON if either of them is active. This makes sense only when the circuit is used in dot mode, not bar mode.

#### ebp

Joined Feb 8, 2018
2,332
He probably had a display module with 9 LEDs. When using it in dot mode, all LEDs would be off at full-scale without the ORing diodes.

#### ElectricSpidey

Joined Dec 2, 2017
1,055
I don't think you even need the diodes, considering the outputs are just open collector.

I admit I have never tied the outputs of the same 3914 together, but I have tied the outputs of 2 different 3914s together to create a crisscross display.

#### ebp

Joined Feb 8, 2018
2,332
Spidey is quite correct. The outputs are open collector NPNs with emitter resistors to make crude but adequate constant-current sinks.

I haven't used an LM3914 in many many years. I remember it from when it was a new product. I'm a bit surprised they are still in production. I see TI offers it in SM in a PLCC, which ain't exactly a modern package. That typical circuit drawing sure looks like a copy from a National Semi data sheet from auld lang syne.

Unless two signal generators are DDS types with decent crystals, the probability that they will deliver exactly the same frequency is pretty much zero. 0.1% difference at 1 kHz is 1 Hz, or 2π radians per second.

#### PFebby

Joined Feb 19, 2018
3
I have an update as well as another small problem!

It turns out LesJones was correct! I monitored the sine waves and they weren't in phase. I adjusted this and the circuit works perfectly! The only problem now is that my indicator is reading the exact opposite of what it should. When the sine waves are completely in phase I have no indication and when the waves are out of phase I have full indication. I thought this might have been occurring earlier on in testing and now I realize that the design has the LEDs placed in the opposite order (my 3914 pin 10 is green and my pin 1 is red. I have a bar LED so i can't simply change the order of LEDs and wiring the opposite way would get messy. How can I invert the signal coming out of the CD4070? Would an XNOR instead of an XOR chip have the result I'm looking for? I guess I don't want to INVERT the signal out of the 4070 but just switch the results of the logic to be opposite. Is this correct?

#### LesJones

Joined Jan 8, 2017
2,599
Try swapping over the inverting and non inverting inputs on ONE of the two comparitors to produce a 180 degree phase shift in one channel.

Les.