Pic Micro - LED Driver interference ?

Hi,

Have a long time used Pic 4520 acting as a time and temperature controller with a NTC sensor to its Analogue port, all powered from its own 12v linear transformer.
Today we added a led driver as below to an array of 10 x 3w leds , the driver needs a control voltage of 0 -10 v DC signal ( not pwm) , so we took a tap from the Pics 12v dc power supply and use a little buck converter as the variable control.

The leds work fine, but the Pics temp sensor control goes haywire, sometimes too high, next read too low ?! ( x16 averaged readings)
The rest of the Pics functions /lcd display seem to work as normal.

Found that if we used a little mains to12v power brick to power the control voltage/buck converter instead of the Pics 12v supply, then things run ok.

To us this suggests it not as we first thought interference on the cable from the NTC , more something on the power lines ??

Any suggestions as to exactly whats causing the interference and if there is a simple way we can eliminate the problem and run the control voltage from the Pics power supply.

That means your DC power supply does not have enough output current to power up your whole system. If you want to power up the microcontroller and all other components from a single power supply, you can choose a DC power supply with a higher current rating.

Good point, will do a current check on the drivers control circuit, assumed / would have thought it would be very, very small ?

Always over rate our psus by at least 50%, unless something in the Pic controller has started to fail /partialy short, though not noticed it running hot or warm .

If you have a DSO use it on infinite persistence, look at PIC supply rail
to see what pk-pk noise is, if it stays within spec.

If no scope use a comparator, set it to trip Vdd out of spec low, then high,
and trigger a D F-F to light LED as a means of detecting out of spec supply.

What bypass caps do you have on PIC Vdd pin ? Can you post a schematic ?

The NTC, what R range is it. If very hi Z its subject to pickup, possibly use a
cap on analog in pin to suppress noise. Look at that pin with s 10 Mohm probe
to see what noise level looks like. Again infinite persistence setting on scope.

What noise looks like on infinite persistence on a supply rail or input -



Ultimately over time it will be a solid band of pk-pk noise. I see normally ~ 200 mV
typical, get worried if I see 400+mV. Check data sheet specs on power rail tolerance
and bypass recommendations.

Capitor effectiveness for bypassing :



Normally a bulk and a ceramic on a processor supply pin.


Regards, Dana.

Hi dana,

Afraid we do not have a good enough scope these days to be able to monitor the power rail.
The Pic controller is an old build, though with decent caps along the rails of the linear psu, and a 100nf across the analogue input to the 10k ntc.

Seems both you and tepalia02 feel it is a power rail problem, possibly brought on / tipped over the edge by the extra load of the control voltage line to the led driver via the Lm2596 which used 8ma though the actual control line uses just 0.25ma.

The Pic controller ( old assembly code) was just to hand to get a project up and running and as its been successful we will be building a proper unit with an ardunio or esp, so for now, knowing where the problem lies, we will just continue to use the separate 12v power brick for the control voltage.

thanks both

The LM2596 is powering what ? Maybe a block diagram of power distribution.

Using a solid ground distribution between all elements ?

You have read all the precautions in the datasheet associated with its application ?
That regulator has a mountain of considerations in its use, and Cap requirements
as well, their ESR, etc....including some fast transient spikes.

Regards, Dana.

The 12v dc taken from the Pics power supply is fed into the LM2596 buck converter to provide the 0 -10v for the led drivers control voltage, not actually detailed in the datasheet.

As said we will just continue to power that line from a separate power brick until we get a new system built, do not want to start messing with the old Pic controllers boards which might end with a dead board.

Thanks.