Noise Levels between Scope Models 3243 vs 4425

Ask any questions relating to the PicoScope hardware or kit contents here.
If you have any questions prior to purchasing the kit post them here.
User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Noise Levels between Scope Models 3243 vs 4425

Post by FioranoCars »

While doing some Can-Bus work (yes the No-Gear Bull), we noticed that the 2 scopes we have gave remarkably different results, in terms of noise!

3 Series
3 Series full screen
3 Series full screen
4 Series
4 Series

Up until doing these screen shots, I thought it was just noise, and to that extent the 2 captures (and screen shots) are taken directly using the same leads, same USB2 cable, same laptop, with least wiggle during swap over possible. All 4 channels were earthed for both in the same place (as we have different earth needs for the 2 scopes - so needed to rule that out) ... As near identical as we could manage.
3 Series Zoom
3 Series Zoom
4 Series Zoom
4 Series Zoom
However, sample settings did change between swapping of scope, although only noticed this now doing the screen shots! ...

We have had this noise issue before (well noticed it), both from using Pico Templates and our own templates/existing captures, but also from scratch set up when switching devices during tests.

In this scenario we started with the 3243 and swapped over to the 4425 (manual reconnected and device selection needed as the auto connected can't swap device upon USB unplug - replug ... can this be fixed, even with a warning of device type swap?, so we don't loose the settings? :twisted: )

So, now I do have some possible logic as to the noise levels being differnet (a 20fold higher sampling rate), so the 3series is just missing the extreme peaks, although surely there would be a 1 in 20 chance of catching the peaks at the higher level? :idea:

The 3423 is using 200ns Sample interval, 5MS/s sample rate = sample size of 10,000, while the 4425 then uses (based on same starting template) Sample interval 10ns, Rate 100MS/s = Sample size 200,000) ?!

This all came about as we were concerned at noise levels leading to corrupt messages (as we still have one ECU being problematic in this regard), but based on the Diagnostics all talking to each ECU and otainign valid payload data, without errors or DTC, then the noise was finally rule out (or at least not primary to our issue - although certainly wasted a couple of man days chasing it :evil: :evil: :evil: )

So a few questions:
- Why does the sample rates change between device changes, as we started with the 3 series (the less capable), surely the lower setting could be used on the 4 series, why would they change?
- Why does the 3series never pick up the extreme's if it only about 20times better sampling surely the peaks are still peaks and once in a while we'd see them (and we don't)?
- Is the noise a factor? Should we be concerned?

Ok, I know very few users have a 3series and 4 series, but swapping devices and device side by side comparisions should be a little better.

I still have an alergic reaction tot he usb cable discconection loosing all the data that's unsaved (please fix this, just disable the start/record button, as viewing and setting up should be possible without the scope plugged in)

Keen to hear other observations and learn a little more :oops:

Thanks
Richard
Attachments
USB2 Pico4series.psdata
4 series capture
(4.44 MiB) Downloaded 585 times
USB2 Pico3series.psdata
3 series capture
(592.49 KiB) Downloaded 584 times

Alan
Pico Staff Member
Pico Staff Member
Posts: 264
Joined: Thu May 25, 2006 8:02 am

Re: Noise Levels between Scope Models 3243 vs 4425

Post by Alan »

Hi,

We do have issues at time when faster hardware reveals things that were already there (ie revealing previously hidden problems) but not visible before but having said that the noise levels look rather high with the 4425.

If you are still connected can you only collect one channel on your 3423 - this way the sampling rate on the 3423 should jump to 20MS/s, the 5MS/s is not really enough to see the detail on CAN. This may make the signal look closer to that of the 4425 but bear in mind the 3423 has a lower analog bandwidth as well.

I suspect that if you also enable a 10MHz filter for the matching channel on the 4425 then the waveforms will look pretty similar. So to recap try 1 channel on the 3423 to get the 20MS/s vs collecting on all 4 channels on the 4425 but have a 10MHz filter set on the 4425. If you can post both of those it would be great.

At the moment I think this noise really is there in the vehicles wiring - its a sharp pulse of around 20MHz that repeats every 35us or so - its odd.

As for the USB cable disconnection I agree its annoying but its not as simple to overcome as it first seems - to ensure you get a fast screen update rate when using deep memory the scope hardware effectively creates the image to be shown on screen and transfers that. In other words until you select file / save most of the data is actually only in the scope not on the PC. When you unplug the USB cable to the scope the power and data has gone.

Regards
Alan

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: Noise Levels between Scope Models 3243 vs 4425

Post by FioranoCars »

Thanks Alan

Ok, will try this later.

Noise:
The sharp pulse, it's not mains to the laptop or a trickle charger, as we've already ruled out external power, removed the charger, and left the latop on battery only. Removed all OBD and breakout boxes (using the Pico OBD breakout), so the only non-production car element connected is the probes from the Scope. All Pico leads :-)
The noise continues once the car goes to sleep.

Any thoughts?

USB Cable I'll post a separate thread under software!

Thanks
Richard

Alan
Pico Staff Member
Pico Staff Member
Posts: 264
Joined: Thu May 25, 2006 8:02 am

Re: Noise Levels between Scope Models 3243 vs 4425

Post by Alan »

Hi Richard,

You ask for suggestions, a bit of a long shot but if the noise is still there when the vehicle goes to sleep then we are already in the twilight zone of diagnostics.

How about making sure the noise is still there on the scope screen then disconnecting the main battery. If it goes away then we have something not going properly to sleep.

If its still there then either the signal is still coming from within the vehicle so must be something with its own battery (tracker unit etc?) or its somehow the length of can bus wiring is acting as an antenna picking up a signal from the surrounding area. No idea what the latter might be so try the rest first!

Regards
Alan

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: Noise Levels between Scope Models 3243 vs 4425

Post by FioranoCars »

Thanks Alan
All makes sense, and I'll run tests tomorrow, was a little tied up today!

Thanks
Richard

e2e4
Newbie
Posts: 8
Joined: Mon May 22, 2017 9:23 pm

Re: Noise Levels between Scope Models 3243 vs 4425

Post by e2e4 »

I wonder if the floating ground on the 4425 is the culprit. The input with a floating ground could be a nice antenna. The higher frequency and the longer wires, the better such antenna reception. A wider frequency band and the floating ground don't get along, especially in such environment as an OBD2 vehicle is.
OBD2 vehicle compresses a lot of electric parts, and also every fast-enough moving metal part creates additional noise on the existing signals (pistons, piston rings, camshaft, crankshaft, alternator etc).
In the circumstances of the garages, the things become worse because there could be several vehicles which are repaired at the same time, several two-post vehicle lifts, lots of other electric equipment, and of course a lady with her vacuum cleaner...
The CAN bus has different impedance at every nod, at every part of its wires. Such complex impedance, together with the oscilloscope impedance with the floating ground, might cause the floating ground input circuits to become a bit unstable when it comes to a bit higher frequencies and higher sample rates.

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: Noise Levels between Scope Models 3243 vs 4425

Post by FioranoCars »

Hi e2e4

Sadly it's not just the floating ground versions of the Pico Automotive scopes, even the old 3series has the same issues.

some of this is down to excessive sampling, so that can be dialed back, but it's not completely down to that alone, and much of what you say will indeed have an influence, but we've tried very short leads (15cm) and making all earths common at source (tied all 4 bnc to common ground) none cure it, and the common ground or leads length made little difference on the job we had back then.

All views welcome, as would like to get to the bottom of what can be done to reduce the issue.

Thanks
Richard

e2e4
Newbie
Posts: 8
Joined: Mon May 22, 2017 9:23 pm

Re: Noise Levels between Scope Models 3243 vs 4425

Post by e2e4 »

Unfortunately, when it comes to internal combustion engines, there are a few indispensable things: pulses for injectors, and pulses for spark plugs on petrol engines. These very strong pulses have pretty steep edges so their harmonics cover a pretty wide frequency band. The average power is pretty low but during the pulse, the electromagnetic field under the bonnet is rather strong.

Using a 15cm input cables with the oscilloscope doesn't help because the CAN buses are still not 15cm short. Together with that 15cm long input cable, there is the whole CAN bus with its two wires to which the oscilloscope is connected. Those wires of which the CAN buses consist, are very good antennas too; those cables can catch the existing noise from the present electromagnetic field easily so one who uses a fast, wide band oscilloscope see the high-pitched spikes on the signal. Another one who uses slower oscilloscope can't see them because they can not be seen with such oscilloscope and also, the input impedance of such oscilloscope plus the long wires together become an LF filter what actually eats the transients, in other words they don't exist with such intensity on the bus like in the first case.

There are also another paths through which the unwanted spikes pass to the CAN bus wires, I wonder if the car makers take care of it.

User avatar
FioranoCars
TwoWaves
TwoWaves
Posts: 386
Joined: Thu Dec 13, 2012 11:26 am
Location: London, UK

Re: Noise Levels between Scope Models 3243 vs 4425

Post by FioranoCars »

Hi
Yes, I agree completely and had the engine been running for the captures that we had the issue with I might have gone further to see if the timing matched spark or injector or indeed any other coil based device that might be EMF back feeding ... but this was an KOEO (ignition only) set of tests.

Fundamentally the car designers have put some thought into the CAN bus wiring, by using twisted pairs, they have built in a limited amount of self correction, as if 1 wire is going to pick up interference then both are likely to, and can then be cancelled to some degree. I would like to think that the routing was also not randomly picked, but that might be overly kind!

Shielded cables add weight, and sadly that means it's a big no no, and verse twisted pair in the car environment probably is not goig to offer big improvements, but I'm no expert on this, just installed a lot of RJ45 / Cat 5e/6 network cable, and before that in days gone by, tons of BNC ethernet which was of course shielded, and suffered more than the twisted pairs in terms of routing and noise pick up arounf factories/offices etc.

We used the short cables BNC scope to confirm we were not creating an external issue, just to be sure, as the 5m cables that we had out were coiled, and it's was a case of trying to rule out all user created problems before asking for help here!

Thanks
Richard

e2e4
Newbie
Posts: 8
Joined: Mon May 22, 2017 9:23 pm

Re: Noise Levels between Scope Models 3243 vs 4425

Post by e2e4 »

Hi,

have you ever used a similar but standalone oscilloscope to do the same measurements?
I wonder if your notebook is able to supply the current asked by the used PC based oscilloscopes. The much higher sampling rate asks for much higher current. So, you can try using your scope with some other notebook so you'll see what happens.
I also wonder if you used a too long USB cable consisting of thin wires. If so, try to use not longer than 1.5m but a thick, high quality USB cable between the notebook and the PC based oscilloscope.


Regards
e2e4

Post Reply