Table of contents

RoadTest: PicoLog Data Logger + Raspberry Pi 4 (included)

Author: kmikemoo

Creation date: 10 Apr 2020

Evaluation Type: Test Equipment

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?:

What were the biggest problems encountered?:

Detailed Review:

Introduction:

I work with electrical generators. I was excited about this road test because the PicoLog CM3 is an AC current data logger. AC current is the primary value that is sensed to determine the amount of load on the generator or being used by a piece of equipment, distribution panel or building. Not every application has a control system that can measure the AC current. In many applications, current is only sensed by a circuit breaker that opens once the current draw is too high. Being able to log the AC current answers so many questions.

How do you know that an emergency generator will perform when called upon?  You have to test it.  To be sure that it will carry the required load, you have to connect it to a load bank (essentially a giant toaster).  And you have to document it - because many of the systems are required by public law.  Data loggers are the de facto gold standard for proof of performance.  "Did you test this?"  "Why, yes.  Here is the data log."

Is your generator big enough?  How big does it need to be?  If we measure your current load, we can make a reasonable decision.

Is your generator too big?  Does such a problem exist?  Absolutely!  If a diesel generator is run lightly loaded (<30%), it will develop soot deposits in the exhaust system that can choke it off and cause oscillations in speed (frequency).  We call it wet-stacking.  It's the reason that all required standby generator systems have a minimum load and/or load bank requirement.  Are you running underloaded?

I could go on, but I think you get the point.  An AC current data logger answers these questions - and more.

Bottom Line Up Front: I think the PicoLog CM3 AC Current Data Logger is a solid performer and a good value for the money. I say this based on whether I would buy this as a companion logger when I need one. Yes. I could not find another unit that I felt was comparable. It was either a single clamp meter or a full blown power meter at twice (or more) the price. The software is fairly intuitive, especially in direct connect / stand alone mode. The CM3's performance on a Raspberry Pi 4 is excellent.

What I love most about the PicoLog CM3 is its application versatility. This unit will run plugged into a laptop. It will run PoE or regular Ethernet powered by a USB phone charger as a remote recording node. Windows is required for this mode. It will run as an independent unit from the Raspberry Pi; wired, wireless, PoE on the RPi4, stand alone or headless via VNC. I did have to use a dummy HDMI connector to get RealVNC to display on my Windows 10 computer when running headless.

Back to the road test... 

My road test application proposed four procedures.

Procedure 1: Install PicoLog data logger on output conductors of 70kW generator connected to load bank with Shark 200 Power Meter. Connect calibrated AEMC 8336 PowerPad III to same conductors. Run generator for one hour, changing loads every three to five minutes. Compare readings of PicoLog to Shark and AEMC meters. Use PicoLog in standalone mode for this procedure.

Procedure 2: Connect PicoLog to output conductors of generator to load bank. Perform standard NFPA 110 and JCAHO four hour load bank test. Record data. Create pdf chart of test. Use PicoLog 6 software to downsample readings to create output chart for integration into final load bank test report. Use WiFi VNC connection to monitor load bank test from adjacent building.

Procedure 3: Connect PicoLog to load conductors of paralleling switchgear feeding the main building. Use PoE hat to power unit and monitor load. Use PicoLog IP Sockets to store data on Training Center computer next door. Record data for one week. Use PicoLog 6 software to generate pdf of building load profile.

Procedure 4: Connect PicoLog data Logger to feed conductors of building air compressor. Record cycle times of air compressor, including how many times it cycles needlessly during the evenings and weekends. Record for one week. Use data to determine the cost of not having the compressor on a timer.

In The Box:

The PicoLog CM3 Current Data Logger came with everything I expected that it would have and everything that was needed to put it to work.

I was a bit taken aback by the 200A current clamps.  In their defense, they will fit on some 2/0 welding cable (Class H) - with the help of a wire tie to close all the way.  We use extra flexible cable for load bank testing because it is much easier to handle and move.  The clamps will fit correctly on Class B stranded without coercion.  Class B is what is normally used in a building,

And... of course... the Raspberry Pi 4.  Slick.  Fast.  It's awesome.  I also got two 16GB SD cards with NOOBS.

Setting Up The Pi:

Of course, I was excited to be selected for this road test so I immediately bought a few accessories.  I had also just received a RPi4 from Hackster so I purchased two 32GB micro SD cards, a case/heat sink/power supply kit and a case/PoE hat kit.

I had a little time so I read all of the documentation on the Pico Technology website and was ready when the package finally arrived.  I installed Raspbian Buster on both cards.  For the first one, I used the Raspberry Pi Imager.  For the second, I used Etcher and loaded the full Raspbian Buster image.  Both worked well and were incredibly fast in Raspberry Pi terms.  Then it was time to load the PicoLog 6.  From the Pico Technology Downloads page, select the software for Raspbian for Raspberry Pi and you get the page below.

Use the button to download the package.  Copy / Paste the commands to load the software - except that it errors.  It did take me a few moments to notice the file name difference.

May 18, 2020 - PicoLog 6.1.16 is now available.

There were errors on the install but one has to love Raspbian / Linux for helping you through.

A little "apt --fix-broken install" later, I was ready to rock.  Truth be told, I made the same mistakes on both SD cards.

The PicoLog 6 software loads into the Accessories menu on the Raspberry Pi.  During the initial launch of the program, the software found some discrepancies and allowed me to correct them.

From here, the PicoLog 6 software was ready to go.  I would call the install experience Very Good, almost Excellent.

A super nice feature of the software was the Demo mode.  I was able to play with the software before the actual PicoLog CM3 showed up.  I found the software very intuitive.

When the PicLog CM3 with Raspberry Pi arrived, I also played with the provided 16GB SD cards.  For whatever reason, NOOBS would not turn on my monitor on the RPi4.  I put the card in an RPi3 and everything displayed correctly for the installation.  This was enough to remind me of the keystrokes required to do a NOOBS install without a functional monitor, so I did this with the second SD card on the RPi4.  WOW!  The speed difference between the RPi3 and RPi4 is unbelievable - especially during the upgrade process.  The RPi4 is "have a cup of tea".  The RPi3 is "go have dinner".

Back to the road test...

Procedure 4:

Connect PicoLog data Logger to feed conductors of building air compressor. Record cycle times of air compressor, including how many times it cycles needlessly during the evenings and weekends. Record for one week. Use data to determine the cost of not having the compressor on a timer.

Yes, normal people start at one and proceed from there.  With the current clamps not exactly fitting on the 2/0 welding cable, I was conflicted.  I only had two #2 AWG cables and really needed four.  So, I opted to install the data logger on the air compressor first since it was reasonably accessible.

The air compressor is a 480V, 10HP across-the-line start system.  The box is a custom fabricated sound shield.  In my mind, this thing cycles quite a bit.  The data logger will prove or disprove this.

The hook-up is simple.  Set up the Raspberry Pi.  Plug the PicoLog CM3 into the Pi.  I used a USB 3.0 port for the PicoLog.  The PicoLog 6 software will auto-discover the connected unit and place it on the desktop.  You can then add your channels.  The software warns you with a yellow triangle if it cannot detect the current clamp.

Plug the current clamps into the PicoLog CM3.  When you initially do this, you will see that the system reports a current value.  Depending on the clamp, it can be six or seven Amps.  Given time, the value will slowly decrease but my unit only zeroed on Channel 2.  With the TA138 clamps provided in the kit, they would start at just over an Amp and decrease to around 50mA over the course of 10 minutes.  I did read that at low current levels, the CM3 dampens it's response.  Given the ultra low flux density at these current levels, this is understandable.  The time that it takes to do this needs to be cut SIGNIFICANTLY.  Dampen over a few seconds, not minutes.

Put a clamp on each conductor.  For the PicoLog CM3, the clamp orientation makes no difference as we are only measuring magnitude.  Per best practices, all of the clamps are oriented in the same direction.

For my 480V application:  Brown = Phase A = Channel 1.  Orange = Phase B = Channel 2.  Yellow = Phase C = Channel 3

Press Record.

I really like the next pop-up.  PicoLog 6 estimates how long you will be able to record at your current settings.

This is where it gets a bit boring.  For the purposes of the test, I opened an air valve to force the air compressor to cycle to mark the start of the data logging.  I also used this time to play with some of the controls in the Graph tab.  The All Data button on the bottom left side was my friend.  The Pan / Zoom were a bit clumsy although any area on the graph could be displayed with great detail.  It was just not smooth to get there.  It reminded me of a CAD program.  It gets better with practice, but it isn't as intuitive as the rest of the program.

This system recorded data every second for each of the three phases for one full week.  Three days in and because there was some discrepancy with Channel 2 as compared to the other channels, I decided to add a Math Channel that would average the readings.

I was warned that the system would only be able to provide the new value from this point forward.  I also discovered a couple of quirks.  One is that the system makes no assumption as to the units you want even if all of the inputs are of the same type.  This is probably intentional.  The drop down menu to select the units has a typo.  The bottom (highlighted) value is supposed to be pA.  The most important thing I learned is that the expected unit value when selecting current is mA - even if the values being displayed are reading Amps.  The inserted channels display their units when you add them.  If you select the wrong units, your Math Channels will be off and not correct themselves - because it is a user defined operation.

The actual graph of the power consumed during the week is below.

So left on its own, my air compressor runs about once every twelve hours.  It's just an odd coincidence that I happen to be in the shop right about the time that it runs.  Even when use causes it to cycle on outside of the twelve hour window, it oddly falls back into approximately the same run times.  The software allows for easy export of the graph (without Channels & Axes) to a pdf.  I have included this as an attachment - Air Compressor.pdf.

Exporting the CSV file took quite a bit longer to generate than the pdf.  25MB and over 600,000 lines on the spreadsheet.  With the data exported, I felt comfortable playing with the down sampling capability of the software.  It's awesome.  Instead of manipulating the spreadsheet data, one can down sample and average the values right in the software.  Once every 10 seconds would only be 60,000 lines on the spreadsheet.  How precise do you really need to be?  When it came to doing the cost analysis on my cycling air compressor, not as much as I thought.

To complete the cost analysis of not having the compressor on a timer, I fought my way through the 25MB file.  There are actually 19 cycling events.  Two are back-to-back.  To lessen the impact of the software dampening of the readings, I treated everything below two amps as zero.  There could be up to 30 seconds of dithering (dampening) on the tail end of each event, but in the end, it makes almost no difference.  The average run time of a compressor cycle was 2:03 minutes with just under 10 amps of average current draw.  At slightly under 8kW at 9.9 cents per kilowatt-hour... 2.7 cents per cycle.  The compressor costs less than 60 cents per week to run.  I wondered why we weren't concerned.  This helps explain why.  The cost of adding a timer to the air compressor would far exceed any potential electrical savings that the timer might provide.  I initially started with trying to use the graph for time marks and current levels, but the export CSV file was just so much easier to manipulate.  Still... the picture is proof positive that the compressor does not cycle as often as I thought it did.

Procedure 1:

Install PicoLog data logger on output conductors of 70kW generator connected to load bank with Shark 200 Power Meter. Connect calibrated AEMC 8336 PowerPad III to same conductors. Run generator for one hour, changing loads every three to five minutes. Compare readings of PicoLog to Shark and AEMC meters. Use PicoLog in standalone mode for this procedure.

For this procedure, I used the current clamps (TA138) that came with the PicoLog,  Because I was using extra flexible welding cable, I did have to use wire ties to insure that the clamps closed correctly around the cable.  Again, Channel 2 zeroed eventually, but Channel 1 and 3 did not.

Setting up for this test also revealed something that I did not expect.  My 70kW generator can really only carry 62kW!!!  This IS the reason we load bank.  Quite disappointing, however.

I chose to start small and continue to add load.  Early on, I could see a deviation in the readings.  With two other meters connected to the same loads, I figured that I was safe to continue.  I suspected an issue in Channel 2 and I had switched the clamps around to see if it was the channel.  The anomaly stayed with Channel 2.

When the test was complete and the load removed, the PicoLog 6 software again ran its algorithm that slowly tapers the reading to zero instead of going to zero. I decided to see if the response was different if the load went to a low value instead of a zero value. The first load swing is 60kW to 0 kW. The second is 60kW to 2kW. The difference in response is insignificant. All low current (<10A) readings get the dampening. I do have to remind myself that analog current gauges have a no-response band as well.  Again, I feel the delay in this dampening needs to be shortened significantly.

The next task was to compare the data gathered from the three devices.  Since the AEMC PowerPad III was a calibrated power meter, that will be considered to be accurate and true reading.  The three sets of reading were copied into one spreadsheet and coordinated for similar times.  I say "similar" because  the Shark 200 meter only records data at one minute intervals and the PicoLog dampens the reading response when below 10A or so.  Also, the PowerPad III doesn't do readings 10A and lower at all.  The Shark 200 and the PowerPad III read similar values.  The greatest deviation of there average readings was less than 1.5%.  "Good" field meters are accurate to 2%, so I thought this was excellent.  Even with the deviation of Channel 2, the average current on the PicoLog CM3 did not deviate more than 5% from the average value of the PowerPad III.

* Okay, there were a couple of times that it did HOWEVER two of the instances were caused by the dampening algorithm.  Speed that up and the difference goes away.  The other one was load change lag.  The PicoLog CM3 does tell us in the setup that it takes 3000ms to perform a read "cycle".

If I take out the deviation of Channel 2 on the CM3, the PicoLog agrees with the PowerPad III along the lines of the Shark 200 - well within the percent error of either meter.  Although the deviation looks like it gets greater as the current increases, statistically it becomes less impactful, decreasing to less than 1% deviation.  As I have other procedures to complete, I will investigate the deviation once they are complete.

Comparing the PicoLog CM3 to the other meters, I would say that it is accurate.  Channels 1 and 3 performed flawlessly.  The CM3 provides a real time graphical depiction of the device under test that the others do not.  Both of the other meters require that the data be downloaded into their software for display.

The chart generated from the PowerPad III software is below.

The charts generated by the Shark 200 are below.  The first graph includes the pre-test operation that revealed the load limit of the generator.  The actual test starts in the middle of the graph.  The second graph is a screen shot from the PDF output report generated by the software.

Procedure 2:

Connect PicoLog to output conductors of generator to load bank. Perform standard NFPA 110 and JCAHO four hour load bank test. Record data. Create pdf chart of test. Use PicoLog 6 software to downsample readings to create output chart for integration into final load bank test report. Use WiFi VNC connection to monitor load bank test from adjacent building.

As mentioned earlier, legally required emergency standby generators have a requirement to be tested. The base standard in the United States is NFPA 110, Standard for Emergency and Standby Power Systems from the National Fire Protection Association. Even if a State has not fully adopted the standard, most healthcare certifying agencies (including Medicare) require like testing. Each month, these facilities must exercise their life safety and critical care transfer switches and they must run their generators for 30 minutes. Diesel generators must carry at least 30% of their rated load each month or they must have an annual 1.5 hour load bank. It used to be two hours. The Joint Commission for the Accreditation of Healthcare Organizations (JCAHO) requires the annual load bank for all units regardless of the 30% monthly load. And... every three years, these units must run for four hours. In the absence of specific guidance, we follow the JCAHO requirement of 30% load for one hour, 50% load for one hour and 75% load for two hours. This was Procedure 2.

There are additional readings that must be taken while load banking, but it is as unexciting as one can imagine. The unit cannot be left unattended because load banks periodically catch fire. To fight the fire, one has to first remove the source of electricity. Additionally, if the generator is wet-stacked (carbon build-up in the exhaust), you can sprinkle burning embers downwind. Exciting load banks are as bad as one can imagine. And there's always the potential that the unit will not perform as expected or develop problems during the test. It's a bit easier to fix things that you were there to see fail.

Given the potential for fire, I modified the test plan slightly. I did not go to an adjacent building but ran the generator and load bank in the back of the building and monitored it in the classroom up front. I intentionally placed the PicoLog CM3 and Raspberry Pi on the load bank to see if either was sensitive to the EMF of the load bank. The EMF is slight and they didn't notice a bit.

At this point in the road test, I did not have a solution for running fully headless on the Raspberry Pi. I still needed a monitor connected to the Raspberry Pi before my Windows laptop would display anything from the Pi. The monitor actually proved handy as I could use the clock on the Pi for load changes. It also allowed me to verify that the test was still going as planned when the VNC Viewer disconnected due to inactivity (twice). I used the Raspberry Pi wireless connection for monitoring and control. All PicoLog functions were performed via VNC over the internet. I also checked the system from my iPhone from the parking lot.

The PicoLog CM3 manual states that the logger can function with any standard 1mV/A clamp. Yes, and no. The CM3 does not use the typical Fluke spacing on the plugs so it would not work with my i410 clamps. It worked well with some third party clamps that I purchased that had independent and leads. Oddly enough, the current reading had less deviation between the channels during this test. All of the readings agreed with my Fluke clamp meter.

I had wanted to use a larger generator for the test, but the 36kW in the yard was the generator that needed to four hour test. So... four hours of making heat it is.

Printing a pdf chart of the test was simple. It's one of the export selections. I exported the data to a CSV file in both one second intervals and 10-second intervals. 15K lines versus 1500 lines. I really like that down sampling feature. I also exported whole Amps, no decimals, one-minute interval, averaging. Blam! Everything I needed for my load bank test report – including time stamps.  I will include the final Load Bank Test Report as an attachment for the curious.

Also interesting to note is the lack of deviation between the phases of the current readings with the third party current clamps.  Two hand-held clamp meters provided similar reading to what the PicoLog was reporting.  This consistency of readings was what I expected but did not get during Procedure 1.  I will have to investigate the TA138 current clamps after completing Procedure 3.  Using the third party clamps, the channel setup used 1mv/A scaling.  When using the provided clamps, TA138 was selected.

Procedure 3:

Connect PicoLog to load conductors of paralleling switchgear feeding the main building. Use PoE hat to power unit and monitor load. Use PicoLog IP Sockets to store data on Training Center computer next door. Record data for one week. Use PicoLog 6 software to generate pdf of building load profile.

This procedure wound up being the fourth load test performed.  Given that I had already used the USB connection from the Raspberry Pi on the three previous procedures, I decided to focus on the IP capabilities of the PicoLog CM3.  The CM3 can operated as a PoE remote monitoring device when paired with a Windows computer.  It does require that a static IP address be entered into the device.  The setup is initially done via the USB cable.  It really helps if you read the manual.  I did not.

I also opted for this route because the User's Guide recommends not to powering the CM3 with both the USB and the PoE (Page 8).  The USB connection had already proven itself so this would be a good test of the PoE.

I did also run the CM3 from the RPi4 with PoE Hat for a few minutes just so I knew that it would work.  It looked like all of the previous tests.

I went a little astray here.  If you've played with the software, you would have already discovered the Settings icon on the left toolbar.  From there, one can select the Ethernet Devices tab and add a device – except that I was never successful in getting any device added from here to talk.

After some considerable frustration, I clicked on the graphic of the CM3 and opened Device Information.  Of course, the Windows version is different from the Raspbian version and the Windows version contains the IP setup that the system will actually recognize.  This is in the manual.  With the software being fairly intuitive to this point... I got lazy.

The one with the serial number works.  The other doesn't.

With this finally complete, I installed the CM3 with the larger, third-party current clamps on the paralleling switchgear that feeds our main building.  The CM3 was powered by PoE from a drop on our building network.  It did take a couple of minutes for the CM3 to be recognized on a computer in an adjacent building (on the same network), but once the connection was established, there were no communication issues for the seven days that it was connected.

The data logging was uneventful, especially given the reduced workload during the COVID-19 pandemic.  The infrequent large spikes are the 30-ton gantry crane.  On Tuesday, we got a large generator in.  You can definitely see when the workday starts and ends.  You can even see when the security lights and illuminated signs come on in the evening.

Aside from my self-inflicted struggle, this procedure was simple and fast.  If there was no PoE available, the monitoring could have been performed with an unpowered Ethernet drop and a phone charger.

Yes, I connected the CM3 through an unpowered hub and connected the USB to a phone charger and monitored it from a wireless laptop in another room.  I love it’s flexibility of application.

Conclusion:

As stated at the beginning, I feel that the PicoLog CM3 AC Current Data Logger is a solid performer.  It is a good value for the money and I would buy another.  It performed well when the load was above about 10 Amps and its flexibility of application was excellent.  The system reads amperage below 10 amps - which is better than the expensive power meter - but the dampening caused the readings to be less accurate than they could have been.  For almost every application, this feature wouldn't have an impact - except when the logging was tracking ON/OFF states.  It robs us of the ability to trigger alarms.

The only two complaints I have are the dampening algorithm is excessive and the IP connection (when using this mode) is slow.  I can live with waiting the couple of minutes for the PicoLog 6 software to find the CM3.  I'm really hoping the dampening algorithm has been improved in the latest release.  If not, please fix it.

For those that are interested, I am including the .picolog files for you to play with in the file Procedures.zip.  You can download the PicoLog 6 software (for free).  Copy the extracted .picolog files into your PicoLog directory and test drive the system.

https://www.picotech.com/downloads

The next steps for me are to figure out the Channel 2 anomaly with the TA138 current clamps.  It is unlikely to be in the CM3 as the third party clamps didn't exhibit the same issue.  I also want to set the PicoLog with Raspberry Pi up with it's own hotspot.  The vision is for a technician to install this on a power panel or automatic transfer switch and later read the data from the hotspot.

PicoLog CM3 Data Logger with RPi4 Hotspot

Thank you to Pico Technologies and element14 for allowing me to road test this data logger.  It was great fun and I will use it regularly.