Noise on Log Channels

[DJETH]

Martyn...
Sorry for the slow response. I am currently do not have access to the hardware but I will the middle of next week. I will post the results then.
What do you want me to record on the 4 channels? How many of each item...
Capped
12 volt battery
temp sensor voltages
ECT...
Thanks DJETH

[Martyn]

A selection will be fine, similar to the Picolog traces.

[DJETH]

All,
Below is page 1 of the attached PDF document that I had to create because of the amount of data and files sizes... I am happy to get the originals to you if you can find a way for me to send you six 17MB files (although you requested only one...I am also attaching the Picolog file that I reference and in the next post I will attach the 8MB Picoscope file I also reference.


Martyn and Design Team,

I have recorded several Picoscope waveforms at the requested 500Sec per division, 5MS, 14bits and 1 buffer. I did get a sample interval of 1ms and 1000 S/sec. Each buffer is 1hr23min long and the file size is a little over 17MB each. Of course your forum website will not allow me to upload this size of file. (design team needs to be made aware of this limitation) I did get a partial that is at the end of the Picolog capture of section 7 (see next page) but is only 40 minutes long and 8MB file size that I will upload if possible.
The total Picolog is a little over 9 hours long and I have divided it to correlate with the Picoscope 6 waveforms taken at the same time the Picolog (200ms sample interval—all 4 channels) was being recorded. This ensures that there will be no mistake when the noise in the Picolog comes and goes ...“The Pattern”. I have adjusted the scaling to 20X on all the captures except section 4 capture I did screenshots of 20X, 1X, 2X, and 10X because section 4 has half noise and half “quiet” in the same capture. The 20X scaling simulates how picolog graphs (magnifies) a trace when the voltage does not change very much. You can also compare the Picoscope section 4 captures (20X, 1X, 2X, 10X) and see that without the 20X setting you cannot see the (red and blue traces) temperature voltage change hardly at all.
When you look at all the Picoscope 6 waveforms you can see there is no pattern in the Picoscope captures only in the Picolog…. Somethings wrong with Picolog. An interesting discovery… Look at the last 3 pages (12, 13, and 14) and you will see the partial capture of section 7 (which as stated above I will try to upload). I accidentally dropped my phone on the scope probe going to the red trace (temp voltage and caused a 1 second series of spikes 2.9 volts Peak to peak. You can see when this event happened in Picolog (which I indicate on the next page). Picolog shows a 37mv spike Peak to Peak for the Picoscope 2.9 volt series of spikes. This tells me that the 15mv –20 mv spikes I see in the noisy section of “The Pattern” I should see 1 volt to 2 volts of noise or spikes in the two temp sensor voltage waveforms...which can be seen is not there… Again a Picolog problem… Is there a way to correct this or filter it out….
Thanks DJETH

[DJETH]

Section 7 Pico Scope file

[DJETH]

Martyn… Just an offer…

I would like to offer to go to Pico Technologies in Tyler Texas and take my desktop, laptop and both 5000 series scopes to test there. I could run my hardware at your facility for a day along with your team there running a their 5000 series scopes on PicoLog (recording the same items) and see if we get the same resulting pattern at 200ms – 1 sec sample intervals. Then if your team does not get the same pattern we can switch hardware and record a second day so I can narrow down the culprit. Then we can post the results here.
I am available the week of September 21 – 25. Let me know who the contact person will be and what days so I can make plans…. DJETH … I’ll bring a mask if I need too.

[DJETH]

Bump... Just checking On the August 18th post and my offer...DJETH

[Martyn]

Can you email in to support@picotech.com and then we can look at whether this is possible.

[DJETH]

Martyn,

I emailed them and have been working with them for a couple of days... I will Keep the forum posted

[DJETH]

THIS IS THE FINAL POST ….I have come to a conclusion to this matter and the 5000 series is a fantastic scope but an ok datalogger…Read further to get my results and closure.

Well I FINALLY got someone at Pico to test a 5000 series with caps on the inputs and they got “similar” results to my “pattern” complaint. See “FROM PICO Sep 24, 2020 8-11-56 AM - Channels set apart”. Channels A and B are set to latest sample and don’t show the pattern (and never did) and C and D are set to duplicate my settings. Because their 5000 scope has the pattern I don’t need to send my two “across Seas” to France for repair as they had suggested (before finally duplicating my tests). The tech did say he would forward this to the PicoLog guru for analysis.

CONCLUSION: My contention is that a scope/logger must be consistent in the readings I see. Although the noise is well within their specs, the noise level changes so how am I going to tell the difference between a problem in the hardware I am testing or a change induced by the scope. There are two parts to the pattern… The fuzzy or noisy part which Pico indicates is well within their specs (I agree) and the smooth part of the pattern (the quiet part). I was expecting the entire waveform to be quiet but that is not actually the case. The quiet sections are the anomaly (at least this is what I understand from Pico). During our latest tests I redid some of my previous logs and when you remove the power supply from the scope it will ONLY run on two channels. The noisy part stayed consistent and there was never a quiet part. I made a 5V power supply that ran off of a 12V battery so I could power the other two channels and not use the A/C Mains and the pattern returned. So whenever the scope has all 4 channels running in PicoLog the pattern exists (200ms – 1sec sample rate and set to averaging). Let me restate… when you use all 4 channels there is a substantial decrease in the noise level inside the scope that comes and goes…and a pattern is created.
After the above tests I talked tech support into testing a ADC-24 Logger with their breakout board to see if its traces were clean and they were. I bought the unit (received it last week) and…. Well pretty nice… Very clear… See “Oct 5, 2020 6-18-22 AM”. Here I am comparing one of the scopes (pattern apparent) with the ADC-24 Logger (Very Clean) reading on the same 12 volt battery. The ADC-24 has a maximum input of 2.5V input so I have to use a voltage divider to read the 12 volt battery without creating smoke. I was using two different dividers using different top end resistance values to see if there as any noise difference in the ADC. Axis 1 is connected to the same divider as is axis B (different divider but all connected to the same point). Axis 3 is the scope on the 12 volt input. Well… Picture is worth a thousand words.

BOTTOM LINE… If you want to log I would get the ADC-24. It has a maximum sample rate of 16 samples per second per channel (please verify this yourself) and you get 16 single ended channels or 8 differentials. Its noise specs are WAY BETTER than my 5000 scopes. If you want a scope and do both datalogging and scope work then test your system and know the limitations. You can use the 5000’s to log but if you want pattern free log set the sample rate for less than 100 ms or over 2 seconds.

Sorry for the long delay in keeping the forum up to date but as you can imagine it took a while to convince someone to run the tests.

Peace Out…