propagation of back EMF

[Simon]

So as some of you probably know I have been crazy enough to take my precious Rigol oscilloscope to work and go "spike hunting" with it on the air con system we are producing. Well I captured spikes of up to +/- 200 V !!! on a 24 volt supply. now the thing is that I cannot find what is is producing these spikes.

So the situation is that I have 3 24 volt 8ish amp fans running under the control of relays (so should not cause a problem even if they have no diode) and a thermostat controls a compressor clutch (that does have a back diode) and 2 8 amp fans also controlled by a relay. When the clutch + 2 fans are switched off via the relay in the thermostat and at the same time another relay is turned off and this turns on a solenoid valve a +/-200V spike is generated on the power in pin of the control box which is not directly connected to anything but the solenoid that has a diode in it.

Basically the question would be can back EMF propagate up the negative ? I am getting oscillatory bursts that start with a + and - peak of 200 V and dies down over maybe 2 uS

Of course we should have back diodes on EVERYTHING but at the moment I'm trying to state the case to our customer that in it's current state it is not our gear causing the thermostat to blow which is still working miraculously

[Kiriakos-GR]

You got a large puzzle to solve .

Lets start with the small things,  is this 24V power source an transformer or inverter ?   

[Simon]

two car batteries in series, not quite a replica of the customers military vehicle but as close as I can get with no budget, basically if I'm powering a fan from a relay when it is switched off it is isolated from the positive, so the only way the spikes can work their way around is by coming up through the negatives (In effect revering the power polarity for a short time ?)

[ModernRonin]

> So the situation is that I have 3 24 volt 8ish amp fans running under the control of relays (so should not cause a problem even if they have no diode) and a thermostat controls a compressor clutch (that does have a back diode) and 2 8 amp fans also controlled by a relay. When the clutch + 2 fans are switched off via the relay in the thermostat and at the same time another relay is turned off and this turns on a solenoid valve a +/-200V spike is generated on the power in pin of the control box which is not directly connected to anything but the solenoid that has a diode in it.

I would go device by device. Unhook all but one device, turn it on and off, see if you get the spike. Repeat for all devices. If you're lucky, it's only one of them. If you're really lucky, it's just a bad diode, which you can replace and problem solved.

If you're not lucky, well... let's not go there. Test one device at a time in isolation and see what happens.

> Basically the question would be can back EMF propagate up the negative ?

In theory, yes.

[Simon]

Yes we have been thinking of that (so I'll have to more more leads)

and Yes basically the question is can back EMF propagate up the negative ?

[Kiriakos-GR]

Ok , round No2 ..

Are those relays new ?
Did you check the voltage in the coils , if it stays " in specs " when those fans are operating.  

[Simon]

all parts are new the system runs at about 24 volts as it's a lot of gear to run on car batteries I would expect re relay's to have 24 volts on them, all is fine until something switches

[Kiriakos-GR]

Well if it was planed by me , I would had use  12V relays or  18V , with  some circuit to stabilize it from 24V .
Because those relays works at their limits as coil voltage, any sudden load , could cause an significant voltage drop,  good enough so to cause on the relays loss of tension = Spikes . 

[alm]

I don't understand how a signal would propagate through the negative lead with the positive lead broken, is there some third connection somewhere, like a ground at a different potential than the negative? You need a closed loop for current to flow. The way I would expect it to work is that as soon as the relay opens, the voltage across the terminals increases (like inductive kick), until it arcs. The back diode shorts this current. Maybe it would be useful if you could draw diagram of how everything is connected (is the solenoid somehow connected to the fans?), since that's not really clear to me.

[ModernRonin]

> I don't understand how a signal would propagate through the negative lead with the positive lead broken

The negative rail could go more negative, thus leading to a larger gap between + and -. That would look just like a voltage spike on the positive rail. In order for that to happen, a large coil would need to be conducting at full reverse current when the relay popped open. The coil would then suck current out of the ground rail.

This scenario seems incredibly unlikely to me. But, you know, I don't really know how the motor drivers in these devices work. So, in theory, it is possible.

I still say the problem is something else, though.

[jimmc]

I'm not to clear about the circuit, any chance of a diagram?

A few quick thoughts:

1/ Could the spike be caused by the back EMF from the fans causing the relay contacts to arc over as they break?

2/ Could stray capacity from the motor windings to ground provide a return path for the current in the -ve rail?

3/ Could the motor insulation be breaking down from the motor +ve line to ground when the spike occurs?


Jim

[Simon]

I'm not too sure I can post a diagram (company + military stuff) and don't have one handy right now as I'm not at work.

So could the fans (that pull 8 amps) be sending back EMF out the relay contact just as they start to open ? that might explain why I have one 200 V burst and then a series of 100 V bursts

[desolatordan]

Could this be due to inductance in the wires leading from the batteries leading to the relays?

[tecman]

The voltage you are seeing is not uncommon.  With many 24 volt systems I have seen 400 volts of "flyback" on a contact break.  The voltage will "ring" or oscillate at the resonance of the inductor and stray capacitances.  A coil will have enough distributed capacitance to generally self-resonate.  We have seen common 24 volt solenoids ring at a few KHz to 10's of KHz.

You can remedy this a number of ways.  A flyback diode across the coil offers the best surpression.  The main downside is a slower drop out of the solenoid/clutch.  The slow drop out could result in some additional clutch wear as it slowly releases. A series diode and zener is the next best choice.  Fast drop out but the zener can get big and expensive if the on-off duty cycle is high enough to require more zener power handling.  MOV surpressors also work but will yield higher voltage on the coil at deenergization, and again watch the power and thermals in the MOV.  Last is an RC network.  The highest voltage, more ringing and the least effective of the options.

Leaving it unsurpressed will kill contacts on the relay and endanger other devices on the same power source.

Paul

[alm]

> I don't understand how a signal would propagate through the negative lead with the positive lead broken

The negative rail could go more negative, thus leading to a larger gap between + and -. That would look just like a voltage spike on the positive rail. In order for that to happen, a large coil would need to be conducting at full reverse current when the relay popped open. The coil would then suck current out of the ground rail.

The way I understood the question, it was 'the positive lead is broken, so could it come from the negative lead'. If there is some other connection, than sure, the negative lead is just as good at conducting spikes. It wouldn't really be back EMF in that case, though.

So could the fans (that pull 8 amps) be sending back EMF out the relay contact just as they start to open ? that might explain why I have one 200 V burst and then a series of 100 V bursts

That's the way inductive kick works, and arcing could explain the high voltages (the voltage will increase until something gives and conducts the current that was flowing before the contacts opened). I'm not sure if this applies to motors, though.

[Simon]

Could this be due to inductance in the wires leading from the batteries leading to the relays?

I have actually wondered what the difference is between my replica of the vehicle loom (the specifics of which I do not know) and the real vehicle, I would think lengths will come into play as they would alter the inductance and having the wires bunched up and sat on the floor on top of each other instead of running separately through the vehicle will cause capacitance I expect, however it won't be my loom per se that is causing the spikes

[Kiriakos-GR]

Does it makes any difference, if the engine is running or not .. speaking about the spikes ?

[Simon]

I don't know and I'm sure the vehicle does not run on 2 45 Ah batteries so I'm not really recreating the vehicles battery supply either never mind the alternator. the first aim of the exercise is to establish if we are generating spikes and how big, next is, is this breaking the thermostat and next what is causing the spikes and how can we stop them

[djsb]

Hi,
Are the diodes you use standard rectifier diodes or schottky. It may help if you fit faster acting schottky diodes across any coils as a first step.

http://www.microsemi.com/micnotes/401.pdf

David.

[Simon]

I think they are standard as they are 1.2 V forward voltage, that may be something else I'll have to educate them in. for the record I produced a spike of -400 and +200 V (600 Vpp)

[jahonen]

Have you verified that spikes actually exist, i.e. the observed spike is not just a common-mode/magnetic field pickup measurement error? You can verify this by short circuiting the scope probe at the measurement point, scope probe ground and signal connected to measurement point ground. If big spikes still appear, then they are result of common mode interference or  on the measurement and not real.

It shouldn't make a big difference whether schottkys or standard rectifiers are used as free-wheeling diodes, when the switching frequency is low (no reverse recovery situations).

Regards,
Janne

[Simon]

Have you verified that spikes actually exist, i.e. the observed spike is not just a common-mode/magnetic field pickup measurement error? You can verify this by short circuiting the scope probe at the measurement point, scope probe ground and signal connected to measurement point ground. If big spikes still appear, then they are result of common mode interference or  on the measurement and not real.

It shouldn't make a big difference whether schottkys or standard rectifiers are used as free-wheeling diodes, when the switching frequency is low (no reverse recovery situations).

Regards,
Janne

Can you explain further, if I short the scope probe surely it will pick up nothing as the cable is shielded and by shorting nothing will happen. I will try this as it could explain how a common comparator chip is surviving -400 V

[jahonen]

Can you explain further, if I short the scope probe surely it will pick up nothing as the cable is shielded and by shorting nothing will happen. I will try this as it could explain how a common comparator chip is surviving -400 V

It can pickup the noise through the shield current (common-mode noise causes this), see Dave's blog on the strange oscilloscope phenomena-episode. No matter how well he short-circuited the probe, the effect was still visible. You can imagine that there is no field inside the shielded conductor. Now, the inside conductor stays at 0 volts, but the shield voltage changes. Result is that you'll have signal at the scope input amplifier. Note that amplitude of this signal is not changed even if you change the probe attenuation. So larger your probe attenuation, the bigger the spikes will become. Which is logical since the noise doesn't pass through the attenuation resistors inside the probe.

Or alternatively, the ground clip (are you using that?) forms a loop where a rapidly changing magnetic field can couple quite effectively, giving rise to weird results.

Regards,
Janne

[Simon]

aha well I see your point and really what can I do if this is the case ? The oscilloscope probe is grounded to the supply ground (wall socket) ? could I shield the cable with an independent shield ?

one small comfort is that changing the range does make a difference, when at 50V/div it was going out of range setting it to 100V/div produve the expected 2+ a little bit divs measurement

[Kiriakos-GR]

I don't know and I'm sure the vehicle does not run on 2 45 Ah batteries so I'm not really recreating the vehicles battery supply either never mind the alternator. the first aim of the exercise is to establish if we are generating spikes and how big, next is, is this breaking the thermostat and next what is causing the spikes and how can we stop them

The only widely known way about filtering , are adding capacitors , here and there , in parallel .
Like 0.1uF / 1000v  

[Simon]

Can you explain further, if I short the scope probe surely it will pick up nothing as the cable is shielded and by shorting nothing will happen. I will try this as it could explain how a common comparator chip is surviving -400 V

It can pickup the noise through the shield current (common-mode noise causes this), see Dave's blog on the strange oscilloscope phenomena-episode. No matter how well he short-circuited the probe, the effect was still visible. You can imagine that there is no field inside the shielded conductor. Now, the inside conductor stays at 0 volts, but the shield voltage changes. Result is that you'll have signal at the scope input amplifier. Note that amplitude of this signal is not changed even if you change the probe attenuation. So larger your probe attenuation, the bigger the spikes will become. Which is logical since the noise doesn't pass through the attenuation resistors inside the probe.

Or alternatively, the ground clip (are you using that?) forms a loop where a rapidly changing magnetic field can couple quite effectively, giving rise to weird results.

Regards,
Janne

what sort of waveform am I looking for ? I'm getting a typical back EMF spike that just shoots straight down to -400V and then back up and dies sown after a bit of oscillating which is no surprise (the oscillating) as there are 8 8amp fans in the circuit

The other thing is if it is picking it up "over the air" it is still coming from the setup

[Simon]

well the plot thickens. I am now quite certain that a substantial voltage is flying around the system based on a very simple fact: I got a shock off of it. I got a shock off the PCB of the thermostat that I was holding in my hand as I had opened it up to probe the comparator IC itself. Now spikes of up to -400V were observed on the power supply of the thermostat. I then opened it and probed the power in to the stat and the power pin of the comparator IC.

See image 52.png now this was all done using the battery neg as ground, the stat got -400V (blue) on it but the IC got -200V (yellow)

Then I used the IC's ground pin as the scope ground and this is where things start to go wonky: see 55.png well yes the IC now only has -15 volts on it but the whole stat still has -400V. It explains how the IC may be surviving but I can't understand how the relationship has changed

BEWARE that image 55 have different V/div scales

[tecman]

Simon:

What you are seeing seems to violate some laws of nature, so:

Where is the flyback diode mounted ?  If it is some distance from the coil, how much wire is involved ?

A flyback diode should contain all of the circulation current between the coil and the diode.  Are there any other inductors ?  How about wire lengths in that "loop" ?

Are there any "sneak" paths that current gets into another coil/inductor ?

What is the inductance of the coil ?

Is there an alternator (dynamo) in the circuit ?  If yes you may be seeing an alternator dump issue, common in battery/alternator systems.

Lastly is your measurement methodology clean ?  Grounds and things can make bad problems look much worse, your shock acknowledged.

Paul

[Simon]

Ah yes violate laws of nature it does, my one certainty is that with respect to ground in the sense of the floor there is a sizeable voltage as I got a shock off it and was not even touching a PCB pin directly.

There are no flyback diodes, there are diodes on the relays but there is a total of 8 fans that pull 8 amps each with no flyback diodes.

All wires to and from parts are about 1.75 metres.

There is no alternator in the circuit but there will be in the vehicle (which I cannot work on as it's over 250 miles away)

as far as i know I've done my best on the measurement side, I do have a 1.5 metre length of wire from the negative of the battery that I use for earth when i have to stray far from the battery with the probe and I am using this as a common ground for all measurements so that I'm working on the same reference for all of the circuit.

The puzzling thing is that although the voltage on the IC was less when i used the stats negative as measuring earth the voltage on the power in was still the same. my only offering there is that there is a diode on the power in (so in theory I should not even see the -15V)

[alm]

As I posted in the other thread, the negative voltage seems really counter-intuitive. There seems to be a lot of ringing going on, but that might be in the setup (sounds like you have tons of inductance and capacitance. That would also be the case in a real car, but might be significantly different, and hard to replicate. I would try to keep ground leads really short to prevent ringing and magnetic pickup.

One simple experiment would be to install a freewheeling diode and see if it changes anything, I'd just use whatever you have in hand, even 1N400x/540x. They might be able to handle the current since it's a short peak, and if you blow up a $0.01 diode, who cares? A freewheeling diode wouldn't help against negative peaks (it would only be reversed biased or breakdown), but should help against EMF that behaves properly according to textbooks .

The peak lasts about 30ns, that's way too fast for a 50/60Hz rectifier diode, so a schottky might work. The frequency seems close to the bandwidth of the scope, so the peak might actually go even lower. I'd be careful with the duty cycle, since a ~30MHz 400V peak is likely beyond the specifications of your probe/scope (even if it's for only 100ns).

Does setting the CH1 vertical setting to 100V/div change anything in the second situation? In that case the voltage might be induced by a current into the probe (magnetic pickup by the ground lead?). Maybe the ~15pF of the scope probe over the IC shunts the spikes?

[Simon]

well as I said there are 6 8 amp fans running all the time, they have built in speed control resistors and indeed things are worse when they run at speed 3 (no limiting resistor) so ringing I expect is natural, what i never expected was the fact that I would get bursts of ringing, basically what you see happens maybe 20+ times but each burst is quite spaced out (like 10 times the length of the main burst) and only the first burst is so large but I guess that is because it is the initial energy that jumped the relay contact gaps. As this relates to a 2 piece military vehicle things certainly will be more hectic than with my little set-up and I've already warned them that these phenomenons are quite a physical thing and can only be faithfully replicated by using the actual installation which is likely to never happen

[Rhythmtech]

Usually when I've gotten to this stage I start thinking Heisenberg's Observation effect - Which abstracted on to what you are doing says that you effect your measurement by simply making a measurement.  Is it possible that some of the voltages you are seeing are being incurred by probe impedance or other measurement variables?