advice on capturing spikes with a Rigol scope

[Simon]

The company i work for has developed an airconditioning system for another company who build military vehicles. We have some issues along with a dispute over blame as it seems that the vehicles system has some spike/back EMF spike going around and damaging some of the more sensitive things (like a thermostat that has no inbuilt protection). Having established the existance of the spikes the blaming game has begun and so I have been tasked to recreate the setup of just our air conditioning equipment (by rigging up the same harnessing that the vehicle uses to power and control the various parts) and put it to test to see if our equipment is the cause of any power spikes when things like fans/blowers and other high power inductive devices are switch on and off. Clearly this means non repetitive signals so i need to "capture" them with my digital scope. what is the best way of doing this ? as far as i know the Rigol scope can be set up to capture anything that occurs over a certain voltage. alternatively I could replace the manually operated switches with relays but that would only generate low frequency at which point i might as well sit there and press pause when a spike shows.

I'm also having trouble saving waveforms on my scope. the instructions are a bit vague and i'm getting rather lost, I take it that bitmaps are the best format to save in

[marianoapp]

set the trigger to single shot mode and the scope will stop as soon as the signal meets the trigger condition.

anyway be careful because those back EMF spikes can reach very high voltages and you don't want to damage your scope..

[saturation]

I don't have a Rigol, but I know this is a job for a triggered event recorder. 
The Fluke 289 can do this, and its portable for on the road use.



The lowly Velleman handheld HPS10 can do it too, it can monitor up to 36 hours. 

On the Fluke 85/87 series, to capture if such an event occurs, I use the min-max setting.  It won't tell you when it happened, or what caused it, but if you're measuring 12V and there's a spike of 50V DC, then something is clearly wrong, and specs claims it can catch transients within 250 uS duration.

http://us.fluke.com/fluke/caen/Digital-Multimeters/Fluke-80-Series-V.htm?PID=56135

What matters between them all is the speed of capture, which is ~ to its frequency response.

[Simon]

set the trigger to single shot mode and the scope will stop as soon as the signal meets the trigger condition.

anyway be careful because those back EMF spikes can reach very high voltages and you don't want to damage your scope..

hm good point, will 400 V cope ? I could try asking for some X100 probes ? make it 4000 V

[Simon]

I don't have a Rigol, but I know this is a job for a triggered event recorder. 
The Fluke 289 can do this, and its portable for on the road use.



The lowly Velleman handheld HPS10 can do it too, it can monitor up to 36 hours. 

On the Fluke 85/87 series, to capture if such an event occurs, I use the min-max setting.  It won't tell you when it happened, or what caused it, but if you're measuring 12V and there's a spike of 50V DC, then something is clearly wrong, and specs claims it can catch transients within 250 uS duration.

http://us.fluke.com/fluke/caen/Digital-Multimeters/Fluke-80-Series-V.htm?PID=56135

What matters between them all is the speed of capture, which is ~ to its frequency response.

Well it is down to the equipment i have already

[wd5gnr]

I'm also having trouble saving waveforms on my scope. the instructions are a bit vague and i'm getting rather lost, I take it that bitmaps are the best format to save in

Best depends on what you want to use it for. Here's how to save waveforms:

1) Press Storage
2) Press soft key 1 and select one of:
    Waveform (data file)
    Setups (the current scope settings)
    Bit Map (Picture of the scope screen)
    CSV (Comma delimited file)
    Factory (reset to factory defaults, not what you want)

3) Select any options (not sure what Para Save does on bitmaps? Maybe does or does not save the parameters?)
4) Choose internal or external
5) If Internal use the multipurpose knob to pick one of the 10 slots and then hit load/save/delete. I think (S) means set and (N) means empty. As far as I know you can't change the names.
6) If external it takes awhile on a big drive for the file browser to come up.
7) Pick New File (unless you really want to save over an existing file)
Use the X soft key and the multi knob to get the file name looking like you want.
9) Hit save.

Does that help?

Hey I noticed something. On 2.02SP2 or whatever if you held a button down for a few seconds it would pop up a help screen on that button's feature. In 2.04 that seems to be gone? Or is there some other reason that quit working I wonder? It still looks like it is trying to load something the first time but it doesn't work.

[wd5gnr]

Hey I noticed something. On 2.02SP2 or whatever if you held a button down for a few seconds it would pop up a help screen on that button's feature. In 2.04 that seems to be gone? Or is there some other reason that quit working I wonder? It still looks like it is trying to load something the first time but it doesn't work.

Reverting to 2.02 did bring back the help. So if you don't get a button and you didn't upgrade to 2.04 just press and hold and the scope will tell you what the button means (sort of, some of the descriptions are a little hazy).

[Simon]

where can I get cheap 100:1 probes ? (or even 20 or 50) on RS they seem horribly exspensive like £100, they will never pay that for test probes at work if I ask. is there any way I can rig something up myself ?

[alm]

where can I get cheap 100:1 probes ? (or even 20 or 50) on RS they seem horribly exspensive like £100, they will never pay that for test probes at work if I ask. is there any way I can rig something up myself ?

£100 new is actually quite cheap for a 100:1 scope probe, although it's probably not great quality. A regular 10x probe from A-brands like Agilent, Lecroy or Tektronix will usually be much more than £100, 100:1 probes will be even more expensive (Tek P5100 is $400 list). The higher the bandwidth, the more critical probe quality becomes. Think of it as a sensor, as opposed to just a piece of cable.

If you only need very low bandwidth (maybe up to a few hundred Hertz), a high-voltage probe designed for a DMM might work. Make sure to compensate for the lower input impedance of the scope compared to a DMM (hook it to a 10x probe?). But spikes are usually fairly high-frequency. I'd probably look for a used name-brand 100x probe. There are a few cheapy made-in-china 100x probes on ebay for $50 or so, although I wouldn't trust them not to distort the signal and actually meet their voltage rating.

I wouldn't try to build something yourself, building something that at high voltages and high bandwidth is actually quite hard. What if your jury-rigged probe fails to detect the spikes, and you lose a customer because of that? Was it worth saving a few hundred quid? Are you going to tell your customer that you base your decisions on DIY/cheap crap probes, when you could've easily bought something better?

I would use something like peak detect or glitch capture on the scope. Make sure your timebase setting is fast enough to capture the spike properly (a slow timebase might indicate a lower amplitude, unless your scope has a good peak detect feature). Note the dead time between triggers, it's likely that your scope is blind like 97% of the time, especially in roll mode. It's usually also blind for a certain amount of time after each trigger, so it will miss two events in a short time span. A regular level trigger should be able to pick up the spikes.

[Simon]

well this is more about me convincing the company that my equipment needs protecting. they will not pay the price tag for a one off test, I've told the technical director that I'd like something that will safe guard for spikes over 300 V like a varistor or whatever they are called, I'll probably grab a few myself.

the spikes should occur at the flick of a switch if they are to occur (switching fans on and off)

[alm]

If it's a one-off test, is rental of a scope + probes an option? You could rent probes that were actually designed for this job, and it would be the rental company's equipment that's destroyed, which they can easily charge your company (since they have a rental agreement). I wouldn't risk my personal equipment if the company didn't want to invest anything to protect it, unless you get in writing that they'll pay for repair/replacement if anything breaks.

If you're going to use an active protection device, make sure it reacts fast enough. It should be faster than the components in the probe and the front-end. Something designed for ESD protection (TVS diodes? not sure if they can handle the energy though). With 10x probes, the probe is usually the limiting factor (scope inputs are often rated at CAT I 150V or so, 10x probes at CAT II 300-400V, the expensive ones are rated for higher voltages with low duty cycles), although it can blow your scope's front-end if the attenuation resistor/capacitor get fried.

[Simon]

Well as always they are trying this on the cheap which starts with me doing the testing versus an external competent person with his own equipment, I know my own equipment best so I'd hope to stick to it. To be honest this is not too different from what I might be doing on my friends car so I'd best be prepared either way. the scope says on the from=nt that it is CAT I to 300 V RMS, can i assume that this is with a 1X probe ? will introducing the 10X atenuator on the probe mean that the scope will see 1/10 of the input taking it to 3000 V ? I ask because with the 10X probe the scope will accept up to 400 V (and in it's "natural" state will take up to 40 V - 10V/Div X 4 Div's)

[saturation]

Sorry, you are right.  

As a side issue, as you note in other replies, a problem with taking a bench quality scope into the field with an unknown electrical environment is the risk of overloading the input amps and damaging the device, or remotely harming the tester.  

As I read in the manual, the Rigol is at best CAT II rated.  I don't think you'll encounter voltages enough to damage the Rigol with this rating but there is a small risk.

However, portable field equipment are often rated CAT III or IV, far above anything you'll find in a car.  Its just shows why Flukes are popular, because even their portable scopes and the 289 DMM are at least rated CAT III and IV respectively, and can take being dropped while in the field.

If you attenuate the signal 100:1 make sure you trigger the event at 4V+ and up, a value that perturbs most input logic.

I presume you'll also need to use a DC-AC converter to power the scope, and switching supplies do put a lot of noise on a DC line.

Lastly, in troubleshooting mobile devices like cars, its best to monitor the vehicle in motion over a period of time while various devices switch on and off rather than a static test while at rest.

Given this is something at 'work', I'd expect work to pay for all the equipment you need and only the best should be available, especially if there is a small risk of damaging it.

I don't have a Rigol, but I know this is a job for a triggered event recorder.  
The Fluke 289 can do this, and its portable for on the road use.



The lowly Velleman handheld HPS10 can do it too, it can monitor up to 36 hours.  

On the Fluke 85/87 series, to capture if such an event occurs, I use the min-max setting.  It won't tell you when it happened, or what caused it, but if you're measuring 12V and there's a spike of 50V DC, then something is clearly wrong, and specs claims it can catch transients within 250 uS duration.

http://us.fluke.com/fluke/caen/Digital-Multimeters/Fluke-80-Series-V.htm?PID=56135

What matters between them all is the speed of capture, which is ~ to its frequency response.

Well it is down to the equipment i have already

[alm]

the scope says on the from=nt that it is CAT I to 300 V RMS, can i assume that this is with a 1X probe ? will introducing the 10X atenuator on the probe mean that the scope will see 1/10 of the input taking it to 3000 V ? I ask because with the 10X probe the scope will accept up to 400 V (and in it's "natural" state will take up to 40 V - 10V/Div X 4 Div's)

The CAT I 300V is indeed without any attenuation (i.e. 1x probe or piece of cable), so in theory your scope's input will survive 3000V with a 10x probe. But the probe also has a maximum rating, which is usually about 300Vp-600Vp. This is the rating of the 9Mohm resistor and the parallel compensation capacitor, and the voltage rating of the insulation. I'm not aware of any 10x probe that's actually rated for 3000V (possibly some industrial probes by PMK designed for ScopeMeters and similar equipment). If the high voltage fries the probe, the full voltage will be on the scope input. I'd limit myself to the probe ratings, and especially with cheap off-brand probes, I wouldn't expect them to exceed this rating.

[Simon]

so your saying that I'm safe to 3000 V because the 10:1 atenuation will take that to 300 V. My rigol says CAT I out of interest. The stuff under test is not in a mobile vehicle so I'll have mains power available, in the case of my friends car it's in his garage.

At work we are basically setting up the working scenario of our air con system WITHOUT the vehilce, the idea being to see if our air con equipment is making any of the spikes. if it does we loose the case, if we cannot find the spikes then we can say that it is the customers vehicle that is causing the spikes and being a military vehicle it may have some hefty stuff,i know they say the power system is 450 amps, I assume it is the alternator. There have been rumors that the issue was caused by the alternator and we hope this to be the case. The strange thing being that our thermostat was being damaged at initial engine start up, the thermostats were replaced and there are no more problems after that. so it looks like that on initial start up something has to setup parameters and in the mean time some nasty spikes happen, after it is set, no more problems. Our air con system has no devices that will "store" things so we do not think it is us.

As for the cheapscate aproach, well thats my company I'm afraid, we do not normally deal in electrics and electrical testing and this is a one off. I'm hoping more stuff like this comes my way as then I may have the oportunity of pushing myself towards design or at least showing them that I'm worth teaching to design aircon systems. we obviously won't be using a 100:1 probe as neither me or the company will go to such an exspense.

[Simon]

the scope says on the from=nt that it is CAT I to 300 V RMS, can i assume that this is with a 1X probe ? will introducing the 10X atenuator on the probe mean that the scope will see 1/10 of the input taking it to 3000 V ? I ask because with the 10X probe the scope will accept up to 400 V (and in it's "natural" state will take up to 40 V - 10V/Div X 4 Div's)

The CAT I 300V is indeed without any attenuation (i.e. 1x probe or piece of cable), so in theory your scope's input will survive 3000V with a 10x probe. But the probe also has a maximum rating, which is usually about 300Vp-600Vp. This is the rating of the 9Mohm resistor and the parallel compensation capacitor, and the voltage rating of the insulation. I'm not aware of any 10x probe that's actually rated for 3000V (possibly some industrial probes by PMK designed for ScopeMeters and similar equipment). If the high voltage fries the probe, the full voltage will be on the scope input. I'd limit myself to the probe ratings, and especially with cheap off-brand probes, I wouldn't expect them to exceed this rating.

I have a set of probes rated at 600 V and we are hoping that the spike is no higher, I was told that we are looking for a 48 V spike but then I don't trust anyone anymore, the customer keep changing their minds and my lot don't know a lot about electronics anyhow.

what if I put a 10 Mohm resistor in series with the probe ? would I alter the probes specs to the point that it's just not going to work ?

[saturation]

If you put series resistance with the probe, it will alter the impedance but you should be able to see a waveform, albeit it could be distorted.

To protect against damaging transients, the simplest thing is to put a metal oxide varistor with the right clamping voltage between the test point and ground.  It costs about $US0.30 if you can find them nearby.  They have nanosecond response times to transients, and 1nF-1pF capacitance.

Once you have an idea what the type of transient looks like in what you describe as more laboratory conditions, you can then take measurements without the MOV and the series resistance to see if it did add distortion.

Keep us posted with your results, and if you have a pic of the transient it should be interesting!

Happy hunting.

[Simon]

yes I think I'll go for some protection, can't hurt. I'll have to do some experimenting over the weekend beforehand. I'm hoping to "catch it" and save it an image as no doubt we will need the "evidence". Our technical manager did mention a spike peak and duration which I can't remember but I don't think it is fast if I remember rightly he said 250 uS but again I'm not trusting anyone

[alm]

If you put a resistor in series, you should place a cap (with a similar voltage rating) in parallel, otherwise you get increased attenuation at higher frequencies (where higher means >2kHz or so). It's the same effect as when you compensate your probe by turning that little trimmer capacitor in the probe.

One other issue: the maximum allowed voltage is usually derated with the frequency, your probe won't be rated for 600V at 50MHz. There should be a curve of max. voltage versus frequency in the probe's datasheet. The frequency depends on the rise time of the spike.

This maximum voltage is limited by two things: maximum voltage rating of individual components (this is the initial flat part), and the maximum dissipation (the linear decrease with the frequency). The impedance at higher frequencies will be much lower than 10Mohm, probably about 200ohm or so at 50Mhz (this should be the second curve in the datasheet), caused by the 15pF or so in parallel with the 10Mohm. The dissipation at 600V would be 1800W! No way that the probe will survive that. This effect does depend somewhat on the duty cycle. I've no idea what the frequency content of the spike would be, I wouldn't expect it to be anywhere near 50MHz.

[Simon]

well as these are single shotr spikes the dissipation should not be an issue, but if the probe is 10 Mohm series resistance, how can it at any frequency come down to 200 ohms ?

[jahonen]

well as these are single shotr spikes the dissipation should not be an issue, but if the probe is 10 Mohm series resistance, how can it at any frequency come down to 200 ohms ?

Common 1:10 probes also typically have about 10 pF of capacitance, so at about 80 MHz, the probe input impedance has dropped to 200 ohms.

Regards,
Janne

[Simon]

so the capacitor is first in line and then the resistor so the full signal goes accross the capacitor, i was thinking it was round the other way

[alm]

The capacitor and resistor are in parallel. At low frequencies, the resistor is the dominant part in the impedance, at high frequencies, it's the capacitor. This capacitor is there to compensate for the capacitance in the scope and the probe cable, and is the one you adjust when you compensate your probe with the 1kHz square wave. Some probes use more complex compensation schemes, but this is the basic idea.

[rf-loop]

Here some very simple educational basics: http://www.box.net/shared/de5b3paxxn

By Andy Frost, Don Whiteman, and Jason Tsai, Hewlett-Packard
Are you measuring your circuit or your scope probe?

[Simon]

so if I read that right having two seperate wires may prove better ?

another thought came to mind, isn't back EMF spikes negative ?, how do I set up the trigger in this case, if i'm after a negative peak the scope will run non stop at the power supply will be over the trigger voltage