I have been looking at this issue for a while and I would like to get feedback from the more knowledgeable members.

I have a PCB with a few FET drivers ( amongts other things) driving a number of external solenoids/valves. (see attached pdf file). The file just show a generic high side driver drawing so you will get the idea. The L would be one of the solenoids and for simplicity sake I just put one of them there.
D1 is placed in the PCB and the solenoids are located somewhere else, with the ground to them wired directly rather than through the board.
Power and GND for the board are wired as shown. This arrangement is pretty standard and worked fine for years and still works.

However, here is the fun fact. If the GND wire to the PCB breaks or the connection fails in any way, then the main ground path will be open. When this happens the actual flyback diode will provide a path to ground through the solenoid and back to the solenoid ground. As a result the ground in the PCB now is floating with respect to the main ground and whenever any of the solenoids is activated the ground voltage spikes just go through the roof, eventually killing either the switching power supply or the controller.
I have tried to battle this problem, which is not huge, but just bothers me, using TVS and other concoctions, but little results, so have anybody seen/faced/solved this issue before?
Thanks

 

looks to me that when the board ground is open, the flyback diode is basically just sitting there, hence the spikes.

I've always mounted tvs or diodes directly to the relay. You could try that, or run a dedicated ground from the diode to the coil...

 

When ground is open the diode is useless and the Back EMF will kill the FET, solution is hard wire an appropriate diode directly at the solenoid terminal

 

When ground is open the diode is useless and the Back EMF will kill the FET, solution is hard wire an appropriate diode directly at the solenoid terminal

That would be easy if we were supplying the solenoids, which we aren't, so we need to rely in whatever is in place already. In the past we supplied the solenoids with the systems and everything was fine, but now we have no control on that end, hence my situation.

 

You need a "virtual ground" of some sort?

So, you have access to the PCB's but not the solenoids? or You are supplying the PCBs and the end user is attaching and suppling the solenoids?

 

You need a "virtual ground" of some sort?

So, you have access to the PCB's but not the solenoids? or You are supplying the PCBs and the end user is attaching and suppling the solenoids?

I want to find an easy way to prevent the system from working in a case of a loss of main ground to the PCB, when the main ground wire is either broken, not connected or just poorly connected. As it is today it will work as the flyback diodes, on the PCB, will provide a return-to-ground path through whatever solenoids are connected and not in use.

We supply the PCBs and the end user supplies and connect the solenoids, so I have little control on what they use and how they wire it.

Thanks all for your replies.

 

Preferably, add power terminals to the board, so the power supply connects in and the solenoid ground connects out.

Failing that, add VDRs (Zenamics etc) across the output devices so the flywheel current can be drawn back from the positive supply.

You want parts with a threshold voltage not far above the supply, so they should have an AC rating about 2/3 your DC supply.

It would not hurt to connect one across the board DC input as well, to limit spikes in case of faults.

(Using zenamics in place of diodes also gives much faster turnoff response with solenoids etc. - you may find you can eliminate the diodes and then use a simple voltage sensor or relay to shut the unit down in case of a fault.)

 

Preferably, add power terminals to the board, so the power supply connects in and the solenoid ground connects out.

Failing that, add VDRs (Zenamics etc) across the output devices so the flywheel current can be drawn back from the positive supply.

You want parts with a threshold voltage not far above the supply, so they should have an AC rating about 2/3 your DC supply.

It would not hurt to connect one across the board DC input as well, to limit spikes in case of faults.

(Using zenamics in place of diodes also gives much faster turnoff response with solenoids etc. - you may find you can eliminate the diodes and then use a simple voltage sensor or relay to shut the unit down in case of a fault.)

Hello Robert
Do you mean something like shown in the file attached here, with a MOV/VDR across the source and the drain?
Thanks

 

Yes, that looks OK.

We've been using this method for fifteen years without problems.
As long as you pick an adequately sized VDR, so it does not get hot when the solenoid switches, it should have a good service life.

You still need to protect the DC supply rails on your board, as the effect of the flyback current (if the ground link is lost) is push it's negative rail below 0V, possibly overvoltaging components.

Another VDR or better a silicon surge protector / large zener diode would limit this effect.

 

Yes, that looks OK.

We've been using this method for fifteen years without problems.
As long as you pick an adequately sized VDR, so it does not get hot when the solenoid switches, it should have a good service life.

You still need to protect the DC supply rails on your board, as the effect of the flyback current (if the ground link is lost) is push it's negative rail below 0V, possibly overvoltaging components.

Another VDR or better a silicon surge protector / large zener diode would limit this effect.

Robert
Thanks for the reply.
Have you used this solution on any automotive application?
I am concerned with a load dumping situation ( don't know if you are familiar with it) where the actual rail could go up to 60V (compared to regular 12V) for a few hundred ms before any protection will kick-in in the car. In this case if the VDR conduction voltage is reached then it will dump the voltage into the load therefore turning the solenoid on, which is not what you would like in a failure situation.
If the VDR voltage is chosen too high ( to prevent this from happening), then the spikes would be high. It is kind of Catch-22 situation. This situation would never happen in non-automotive situations as it is caused by an alternator failure. There are TVSs and other surge protection devices all over the device and in other sections of the car as well, but still the VDR will have to be at least 35V. I really don't see a way around it.
Thanks
Eduardo

 

Hi Eduardo,
no, not automotive, we build industrial control gear. No load dumping, just large motors switching & overhead crane tracks arcing etc.

We have the VDRs connected directly across the 'solenoid' outputs (actually electromagnetic clutch/brake units), but our C/B controllers have separate ground in & out so the output is connected solely through the controller.

It sounds like you are back to the VDR or silicon surge protector across the control board supply. That will clamp the voltage so the VDRs can never conduct other than from the solenoid back emf.
You could also add a polyswitch in the main supply; if you get a fault that will effectively go open circuit until the fault is cleared.

I'm guessing this is a fuel or nox injection controller?

 

Hi Eduardo,
no, not automotive, we build industrial control gear. No load dumping, just large motors switching & overhead crane tracks arcing etc.

We have the VDRs connected directly across the 'solenoid' outputs (actually electromagnetic clutch/brake units), but our C/B controllers have separate ground in & out so the output is connected solely through the controller.

It sounds like you are back to the VDR or silicon surge protector across the control board supply. That will clamp the voltage so the VDRs can never conduct other than from the solenoid back emf.
You could also add a polyswitch in the main supply; if you get a fault that will effectively go open circuit until the fault is cleared.

I'm guessing this is a fuel or nox injection controller?

Robert
Sorry for the delay and thanks for the reply. I wish it was a fuel injection system but we are dealing with fairly high currents instead for hydraulic operation control, so the spikes are fairly large and heavy.
I will try to experiment with a solution using VDRs and see what happens.
Thanks

 

in such case, You can have the output line pulled up to Vcc (using a very high resistance) instead of the VDR to keep the flyback reverse biased. This will stop current flow through the diode during loss of ground into the coil. Make sure the resistance is high enough not to turn on the solenoid on its own.

 

in such case, You can have the output line pulled up to Vcc (using a very high resistance) instead of the VDR to keep the flyback reverse biased. This will stop current flow through the diode during loss of ground into the coil. Make sure the resistance is high enough not to turn on the solenoid on its own.

Do you think OP will come back after 4 Years