PDA

View Full Version : Noise issue related to Curtis 1238



mpipes
30 August 2018, 0638
Currently having an issue with what I'd guess are particularly sensitive electronic items picking up noise.

I have an Arduino controlled relay module that I built, which switches a few circuits like the headlamps, horn, accessory, as well as the charge/discharge mode state.
I also have a light control module, one of those setups for converting incandescent park/brake/turn lamps to LED as well as allowing multiple functions with one lamp assembly.

When I power up the Curtis 1238, after the KSI relay closes all is still good, but once the Curtis closes the contactor (LEV200, 48v coil, 50% PWM signal from the Curtis) the two modules noted above go haywire. The brake lights turn on although very dimly, and even though there is no hardwire connection between the Curtis and braking system/12V side of things. The relays cycle rapidly, firing the horn, headlights, turning the radio and Andromeda display on/off.

As soon as the contactor shuts off, all the issues go away.

I have not tried moving the KSI relay away from the contactor yet. It is pretty close, just a little over 2" from the body of the contactor so I suspect the 12V coil is picking up noise from the contactor because of the PWM signal.

I was also thinking of ordering a different contactor with a 96V coil so the PWM from the Curtis can be minimized. Not sure if that really solves the issue.

I'm also working on a more detailed wiring diagram I can post later if anyone wants to take a look for anything else I might be missing. One thing that sticks out to me is I have a 200Amp diode immediately after the 12V battery, and then a 40amp diode immediately after the DC/DC converter in order to isolate them from each other, and these items are basically at opposite ends of the 12V power bus from each other. There are two 12V distribution blocks also near opposite ends. I'm wondering if the diodes could be creating a resonant loop?

Any ideas are appreciated! If I have any noise in that 12V system, it's gonna be bad news when I get the audio system powered up (about 2,000 watts, YEAHH! haha)

frodus
30 August 2018, 0902
Post a schematic when you get it. Include all ground connections you've got.

It could be EMI, or could be that something is getting grounded from the Curtis side, to your 12V side. You could start by enclosing the modules in a metal box. If that doesn't help, it's likely coming in through something else.

Are you using a DC-DC converter? if so, which one?

mpipes
30 August 2018, 0918
I'm using a Mean Well 500W DC/DC converter. However this issue is happening whether that unit is powered or not.

I'll put some more time and detail into the wiring diagram.

The Arduino/Relay module and the LED Lamp module are right next to each other but both are about 4 ft away from the Curtis/Contactor.

I had a similar issue earlier with only the 12V circuit when the horn was activated. The horn itself caused enough noise to trip the other relays so I got a different horn and issue solved.

It's been a fun project. LOL.

Stevo
30 August 2018, 0950
Maybe EMI noise between the 2 arduino modules? Any wires of these two systems running side by side in a harness?

frodus
30 August 2018, 1028
I'm using a Mean Well 500W DC/DC converter. However this issue is happening whether that unit is powered or not.

I'll put some more time and detail into the wiring diagram.

The Arduino/Relay module and the LED Lamp module are right next to each other but both are about 4 ft away from the Curtis/Contactor.

I had a similar issue earlier with only the 12V circuit when the horn was activated. The horn itself caused enough noise to trip the other relays so I got a different horn and issue solved.

It's been a fun project. LOL.

Sometimes on or off doesn't matter with a DC-DC. Sometimes the noise gets through regardless, if they're coupled somehow. What Meanwell is it?

mpipes
30 August 2018, 2205
Sometimes on or off doesn't matter with a DC-DC. Sometimes the noise gets through regardless, if they're coupled somehow. What Meanwell is it?

It's a SD-500H-12

Also have an on-board Thunderstruck charger connected to the 96V side, with the EVCC on the 12V side and CAN bus between them.
Also using an Orion BMS which closes the 12V ground to the KSI relay coil.

frodus
31 August 2018, 0823
OK, that DC-DC should be isolated (looks like a switching power supply).


I think I'd start looking for ground loops somewhere. Check to see if you can measure any resistance between B- and the chassis, or any of your Grounds or Negative wires on your 12V system.

mpipes
31 August 2018, 1957
Thanks Travis, will measure and see what I find out.

In the meantime, here's a wiring diagram. It's not 100% complete as I haven't added all the connections at the front of the vehicle nor the complete charging circuit at the rear, which is basically just an incoming A/C power and a relay to keep it disconnected unless charging is actually happening.

Not sure why the images keep getting resized down when posted here, the original is 2200px wide. Try the SVG file inside the ZIP archive.

7908

7909

frodus
31 August 2018, 2054
What is the B&B Canop? Is it a Can opto isolator?

How does the 12V battery charge if there's a diode?

Try this: Try disconnecting the + and - wires from the main 12V distro. Just have the 12V battery power the Arduino and Display (keep the CAN). Start there. This might help diagnose since you have a way to power things on the DC-DC side and the Battery side. So you can split into smaller pieces.

mpipes
31 August 2018, 2140
Yes, the CANOP is the opto isolator for the CAN.

RE: the diode, the idea was to add a switched bypass around that diode to allow control of the charging of that battery. The more I think about it though, I think I'd rather not have the voltage drop from using it. I can dial the voltage on the DC-DC so it would not exceed the specs for the battery and it basically could never overcharge.

Yes, I have previously disconnected the 12V wires at the front of the vehicle, basically the battery ran the front and the DC-DC powered the rear. This produced no issues.

I did just go out to the garage and measured resistance from B- at the controller, to the front distribution (where all the devices are tapped at the same spot) and it measured at roughly 2 MegaOhms.

frodus
01 September 2018, 0927
OK, so no issues with the 12V wires being disconnected, correct? Does the battery power the left side devices and the DC-DC powers the right side devices?

Was the same 12V connected when you measured the resistance from 12V- to B-?

I'm thinking the noise is coming from the DC-DC converter. Even though it is isolated, it can be a source of noise.

mpipes
01 September 2018, 1900
OK, so no issues with the 12V wires being disconnected, correct? Does the battery power the left side devices and the DC-DC powers the right side devices?

Correct.


Was the same 12V connected when you measured the resistance from 12V- to B-?

Yes.


I'm thinking the noise is coming from the DC-DC converter. Even though it is isolated, it can be a source of noise.

OK. I can also try disconnecting the DC-DC from the system to see if that isolates the issue.

Thanks for your time, Travis!

I've updated the wiring diagram to include the AC input for charging and also the cell tap wires for the BMS... Hmmm it's amazing how obvious these possible paths become when you draw them out! :D

7910

mpipes
01 September 2018, 2200
Alright, making some headway I think.

Went out to take some more resistance measurements and at first I was stumped because the numbers kept changing and were literally all over the map. Every time I pulled the probes and re-set them, I got a different number. 50kOhm, 30MegOhm, 300kOhm, 8MegOhm...

Then it hit me like a flash... literally... I had fired up the system and saw a bright glow from inside the DC-DC..... carbon fiber dust igniting. D'OH!

Seems like I have the constant rapid-fire relays mostly under control. There's still an occasional random relay fire which could just be residual dust but the tail/brake lights are still on only while the Curtis is powered up. Not sure if that's because of the DC-DC or other issue. I'll find out tomorrow when I'm off work and have more time to investigate.

frodus
02 September 2018, 0949
Oh no! Yeah, those enclosed DC-DC's may have issues with dust getting in.

Do you have room for Sevcon DC-DC's? I sell 500W versions, not sure if that is enough to run some of the 12V system.

mpipes
02 September 2018, 1313
I checked the dimensions on the Sevcon DC-DCs, I definitely have room for two. :)

Did a little more testing today. This time I turned off the PWM driver for the main contactor to see if it's picking up noise from the PWM signal, and looks like just the Curtis logic alone is noisy enough to make that LED lamp module pass a weak signal. When I pull the ground off the B- post on the Curtis, yet leaving the rest of the system running, the parking lights switch off as they're supposed to.

frodus
03 September 2018, 1101
When you say PWM driver for the main contactor, what do you mean?

mpipes
03 September 2018, 1345
When you say PWM driver for the main contactor, what do you mean?

Under the Main Contactor Menu in the programmer, I set the MAIN_ENABLE parameter to OFF and then cycle power to the controller so it turns on but doesnt close the contactor. This way I can take the contactor noise out of the equation.

Today, I have power disconnected at the 12V battery (B+ and B-), the DC-DC converter is physically disconnected from both +B(96v) and -B(96v) so there's not even power connected to the 12V bus. The BMS is also 100% disconnected from power, cell taps, and all other I/O. I run the blue KSI wire on the controller directly to B+(96v) to power up the Curtis and the parking lights still power up.

Unplug the 96VDC charging cables running to the Thunderstruck Charger, and now the parking lights go out. I plug the 96V side back into the charger, parking lights go on. Unplug the A/C mains input from the charger, parking lights go out. The A/C mains input is grounded to the 12V side, as per instructions from Thunderstruck: http://www.thunderstruck-ev.com/images/EV%20Charger%20Controller%20v2.3.2b.pdf

frodus
04 September 2018, 1113
OK, gotcha. That way it wouldn't power on.

Sounds like there's a noise coupling somewhere. Maybe one of the chassis's of the devices is grounding to the chassis of the vehicle.

mpipes
04 September 2018, 1907
OK, I will double check that. The chassis is all carbon fiber so it won't conduct very well, but I still used fiberglass or plastic enclosures at the 12V distribution points for safety. The mounting brackets for the controller and charger should have a solid film of a filled structural adhesive isolating them from the chassis but are worth a look.

How about the motor? It's bolted directly to a carbon fiber mount on the swing arm which technically should be isolated from the main chassis via polyurethane suspension bushings but there's still an aluminum shock/air spring.

frodus
04 September 2018, 1959
I just mean that something, like charger or DC-DC is injecting noise back into the system. When removed, things work. When put back, things stop. There's likely some sort of either EMI, or a ground loop.

flo
04 September 2018, 2200
OK, I will double check that. The chassis is all carbon fiber so it won't conduct very well, but I still used fiberglass or plastic enclosures at the 12V distribution points for safety. The mounting brackets for the controller and charger should have a solid film of a filled structural adhesive isolating them from the chassis but are worth a look.

How about the motor? It's bolted directly to a carbon fiber mount on the swing arm which technically should be isolated from the main chassis via polyurethane suspension bushings but there's still an aluminum shock/air spring.

Hi
just to point out: carbon fiber conducts quite well!!!

flo

mpipes
05 September 2018, 0923
Measuring 50kOhm+ from the case of both the charger and the Curtis, to nearby points on the carbon chassis so looks like there is a path through the mounts. Easy enough to fix with a reciprocating saw. :)


Hi
just to point out: carbon fiber conducts quite well!!!

flo

carbon fiber does conduct but I wouldnt say quite well, or at least maybe not reliably/consistently enough depending what the intent is. Measuring resistance of a raw 12k fiber bundle I get 25ohms over a 6" length.

flo
05 September 2018, 1036
Well - guite well might be misleading.
But with some rc-plane designs we regulary ran into trouble with twitching servos because the steal linkages (touching the wings) would transfere the impuls from one to another..
and that is on about 5-8 volts depending on tech at the time...
Also cf planes regulary burst into fire when accidentally hitting a land line while wooden or glass-fiber planes simply get torn appart :p

greets

flo

mpipes
16 September 2018, 2059
Still sorting through things here. So far have swapped out a 2-pole relay for a 3-pole relay on the AC Mains input power for the Thunderstruck charger so I can completely isolate it (and its ground) from the system while it's not being used. The BMS and EVCC charge controller work together to control that relay for both dedicated charge state and emergency shut-off.

Also made some isolating spacers to insert into the holes in the mounting brackets for both the Curtis and the T/S Charger, to separate the device chassis and mounting bolts from the carbon mounts. Havent had a chance to get them installed yet.

Spoonman
17 September 2018, 0220
Isolating the casing of the charger isn't going to fix anything is the charger isn't plugged in.

...I had a far longer post than this almost written but as I was finishing it something occurred to me.
Are you using appropriate logic fixing resistors at the inputs to your power loads in the absence of a control signal?

In other words have you a pull-down resistor on the Anode of those brake LED's so that the pin isn't simply left floating when the relay is disengaged?
...or are you using double throw relays to avoid leaving the pin float?

If you aren't then any passing cat could dimly light up your parking lights on account of the fact that the LED will present an infinite impedance on that line up to the point of reaching the linear component of it's IV characteristic curve.


Now the fibrillating relays are a bit more of a mystery - those are power components and not nearly a susceptible to direct excitation in such a fashion, but if they're switched by signal amplifying circuits then those could be. That being said, it's far more likely that there's a power line stability issue where they're concerned. Certainly the dust issue mentioned could cause that. Have you access to an oscilloscope that you can take a proper look at the integrity of the rails?

mpipes
18 September 2018, 2236
Are you using appropriate logic fixing resistors at the inputs to your power loads in the absence of a control signal?

The lights are controlled by a commercially available LED module manufactured by Penta-Star. The inputs are 12V normally tapped directly from switched lighting circuits, and they also provide the main power for the module. There is no constant power to the unit so in absence of a signal, the unit is powered off.

Currently, there is no 12V source attached to the system at all while I'm testing and isolating issues, but I can measure 20mV on the 12V power bus when the Curtis motor control is powered on, so it makes sense that that weak voltage is causing the input to float. That voltage shouldnt be there at all though, and while dim lights arent a big deal, the future audio system picking up any noise is.

I now have the isolating mounts installed at both the Curtis and charger cases to ensure they are not coupled together through the carbon fiber chassis, and the 3-pole relay on the AC Mains input to the charger is also setup so it breaks the ground to the 12V bus while the dedicated Charge Mode is inactive. I did this because the charger internals ground the case and the 96V side together but there's no point having it connected to the system while it's not needed.

The lights are much much dimmer now after doing this so I'm getting somewhere, but still have some other ground loops to take care of.

Stevo
19 September 2018, 0900
That voltage shouldnt be there at all though, and while dim lights arent a big deal, the future audio system picking up any noise is.

Maybe if you wire a filter into the audio circuit input, there wont be any issues with noise. Audio is an awesome idea ... it's got me thinking!!!...:rolleyes:

mpipes
19 September 2018, 1756
Progress. Snooped inside the headlight housings (metal) and saw that the park/turn lamp socket assembly is screwed directly to the housing and has the ground wire attached at one of the screws. Simple fix there with relocating the socket mount into a plastic case and running the ground into it instead of the metal housing. Looks like it's a similar situation inside the rear tail lights, those are tougher to get access to so I havent torn into it yet. I also bumped the drive pulley while navigating around the shop and noticed that the brightness of the tail lights fluctuated. Looks like I get to isolate the traction motor from the chassis now too. :)

mpipes
03 October 2018, 2046
OK.. so far here's all the ground loops I've been able to isolate. :D I still have a weak one back at the drive motor even though I put isolating washers/sleeves at the bolts on the motor face. I either broke a sleeve while installing (tight tolerance) or part of the motor can or even part of the terminals are rubbing somewhere at the mount. Havent looked that closely at it yet. However, the lights are no longer powering on when the controller is powered so moving the right direction. I am still measuring about 400mV being induced at the 12volt bus though. Today I cut a section out of the structural core support for the dashboard because it's also all carbon fiber, and is where the audio system head unit and Andromeda display are mounted (thus grounded to the carbon). I grafted that back together with fiberglass. Tomorrow morning that work should be cured and I can start re-connecting all the systems to see if the arduino is still getting any interference.

8037

Stevo
04 October 2018, 0947
#$@%& pesky gremlins!:mad:

mpipes
10 October 2018, 1443
Powered things up and much improved but still a little glitchy with the relays controlling the accessories.

After a little more investigation it looks like floating inputs on the Arduino. I have one of the Adafruit 5-pad capacitive touch sensors and the output leads from it, to the arduino input are 8ft long. Been so long since I set it up that I forget if I have the internal pull-ups enabled but I doubt it. Might need to add external resistors.

Fun stuff :D

Stevo
11 October 2018, 1317
Resistors?
If a cable is 8 feet long, maybe a select capacitor on one end? Not sure.
I was thinking of the alarm signal from a Cellog to control a relay... especially if the wire is of significant length, a capacitor helps the signal stay stabilized.

mpipes
11 October 2018, 1923
Unmounted the motor from the chassis but kept it connected to the controller, one last ground loop confirmed! The glitching went away with the motor pulled from the frame and controller powered up.

I had built the motor mount to be a tight fit but it's also pretty tough to wrangle into place. The insulation on one of the motor terminals got snarled up on a sharp edge during install so there was contact to the chassis.