A tale of 2 grounds

Go To Last Post
12 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi all,

I am designing a device for an automotive application.

I have put in 2 separate grounds for my device:

-uC ground, this is for the uC and other ICs, the ground current should be about 150ma. Having a solid ground is very important.

-Heater ground. Part of the device function is to control a heater element to maintain a certain temperature. The heater is pulsed on/off @ 190khz, and the max instantaneous current is about 5A. I do not really care how noisy the heater ground is.

I decided to split the ground, because in the current design I tie both grounds together on the pcb and the heater seems to be causing too much noise on the ground.

My question is where is the best location to hook up the 2 grounds for a car:

- Tie the uC ground closest to the battery and tie the heater ground farther down.

-Tie the heater ground closest to the battery and tie the uC ground farther down

-Tie them both together closest to the battery.

-Other option, please enlighten me.

Keep in mind the primary concern is that the uC ground is nice and clean.

My gut feeling is that tieing the uC ground closest to the battery and the heater ground farther down is the correct solution.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

"clean ground" and automotive don't really go together, AFAIK.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ok, how about "cleaner ground"

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

IMHO - the largest 'SINK' to noise in a car is the BATTERY. Therefore, the most noisy thing you have should be there (closest to the battery). All else should be further away.

But that *IS*, just my opinion... not biased by any real FACTS!! :)
Steve

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do you have to run the wires through the firewall?

I'd opt for two separate wires, both back to the battery terminal.

Running the 5 A heater through a long wire may not be desirable do due the voltage drop along the wire, power loss in the wire, and therefore less voltage and less heat from the heater. If this is an issue you can tie the heater ground into the frame, with a large bolt, and washers and lock washers. You need to scrape the paint off the frame to do this, however, and be on shiny metal frame. The battery, (US vehicles, anyways), will usually have a heavy cable going from the battery negative terminal to the frame in close proximity to the battery.

All that said, it may not be your ground that is causing a problem at all.

Have you scoped the V+ to the uC to see what it looks like? Automotive electrical systems are incredibly noisy. When you really want to ruin your day google "load dump".

Make sure your uC has appropriate by-pass caps across the +V and ground pins, as close to the pins as you can get them. Typically a 0.1 uf and either a 1 or 10 uf cap. Filter your power supply heavily. If you are using a linear regulator be sure your input caps are after your reverse polarity protection diode.

Noise can also be coupled into the uC through any other connections to the outside world. This means any sensors connected to the uC need filtering on their inputs, too.

Using shielded cabling to any sensors will also help decrease the noise coupled into the wires, and thence into the uC.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Keep in mind connecting a device directly to the ground on an automotive battery is risky (assuming a negative ground vehicle setup). If there is corrosion or a failure in the main battery ground to frame cable and your device contacts the vehicle ground anywhere else, then the battery will dump all its current draw through your battery ground connection. In other words, your device battery ground connection can possibly become a substitute ground path for a failed main ground cable. This includes engine starter current loads. Its a great way to unintentionally weld metals, vaporise metals/circuits or start fires. The usual sure cures are an in-line fuse between the vehicle battery ground and your device ground, or you connect your device ground at/after the main battery ground to frame connection (for example the vehicle body).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Mike,

Agreed, but still you are describing a dual fault failure mode. The project has to have a secondary path to ground, AND there has to be a primary failure within the vehicle's electrical system, (its frame ground).

Your degree of confidence in the vehicle's electrical system makes my paranoia look normal :)

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How are the grounds tied together on the pcb?
Can you post a pic of the grounding on the circuit?
Do you have any large capacitors on the circuit close to the heater?

There is a chance adding an extra return won't fix the problem- you may just make a big ground loop and make things worse. 190khz at 5A is a fair bit of current/frequency, the inductance of the power wiring is going to make its effects well known I would expect.

oddbudman

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'd put the electronics GND directly at the battery. I'd also fuse the main power and GND running to the PCB - if connected directly to the battery.

Maybe there is a spare GND post in the fuse box, along with a switched accessory post that can be used.

And I'd also use some sort of noise suppression (Ferrite beads, dual winding inductors, etc.) to block any vehicle noise from entering or leaving the PCB power distribution system - in both the VCC and GND power connections on the PCB.

And don't forget... The main GND connection on the PCB isn't the only place that electrical noise will enter your control circuit. Any signal inputs (and their GND reference lines) should be properly filtered to keep any electrical noise from entering the control circuit.

Use heavy (read, appropriately sized) wiring for the heater. Use appropriately sized wiring for the control electronics. Run both to the battery, as close as possible - using separate power & GDN lines - each according to their specific purpose and current carrying requirements.

Consider & add these things during the initial design/PCB layout phase of the project.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why does your heater gnd have to come to your board? If you switch the high side, the heater ground can be totally separate from your board. This means the mosfet drive is a little more complex, but there are specific chips to solve the problem or use a p-chan mosfet. You'll find in many ECUs that there is not much ground current as all the high current stuff is switched on the high side.

Something else to consider is inductive coupling with your wiring - 190khz at 5A spells a bit of RF energy floating around - 60W in fact.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:
Your degree of confidence in the vehicle's electrical system makes my paranoia look normal :)
Actually this is experience from vehicle mobile radio two-way communication installations. If you keep a lead acid wet battery type vehicle enough years, the battery cable crimps are bound to go bad (the battery fumes just corrode the crimps over time). I've seen impressive damage from incorrect ground wire installations.

Since the 1980s Motorola has designed microprocessor based radios that work very reliably in vehicles. They always connect the un-fused ground power wire to the vehicle chassis, not the battery negative terminal (for negative ground vehicles of course). The fused positive power wire always connects to the battery positive terminal. The engineering magic that overcomes the nasty vehicle power environment is in the power regulator design.

Many Japanese radios are designed to connect directly to the battery terminals, but both ground and positive power wires are fused. I suspect they don't have a power regulator design that is good enough to deal with a vehicle chassis ground connection.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

All good points.

Thanks,

JC