In my past projects, I often needed my multimeter to log data for extended periods. I have a reliable Owen B35T that can connect to my phone and log data, but it only supports one channel and is somewhat buggy. After about half an hour, the app usually stops logging. Although I have an oscilloscope, it’s designed for measuring fractions of a second, and we know it’s not ideal to use oscilloscopes as multimeters.
Previously, I encountered issues with my 3D printer experiencing thermal runaway after about an hour of printing. I wanted to connect multiple multimeters to log the time frame in which it would fail, but I found this very frustrating to do. Given my experience with microcontroller programming, I decided to use a simple microcontroller to create a multimeter that could send serial data to my computer for logging. With just a console program, it worked beautifully.
About a week ago, I faced a situation where this setup would be useful again, and it occurred to me: why not turn this into a proper project? Thus, my 8-channel USB multimeter was born. While I am not a professional and there may be easier or more proper ways to achieve this, this was my journey through the project.
It felt right to start at the very beginning that started this whole ordeal.
04/22/2024 The First Glimpse
The first chance I had to debug my printer, I tried connecting my multimeter to my computer via Bluetooth. After about half an hour of frustration, I gave up. The multimeter doesn’t log data on its own; it requires software, and the only software I have is on my phone. So, I built a quick voltmeter that can measure up to 27 volts and used a standard serial plotter to record everything. While I could have just recorded the data, this setup provides a nice, clear graph.
I only had about half an hour to work on this, so I decided to call it a night. The thermistor is a 10K thermistor, so I’ll need to run it through a transistor to sense it properly. I’ll leave that task for tomorrow. My plan is to monitor the temperature using Marlin and the voltage using my voltmeter. When the printer crashes, I’ll check the logs to see if there’s any indication of what went wrong.
The first major issue is that the power supply runs at 24 volts until the heater kicks on, which pulls it down to almost 18 volts. When the Peltier coolers activate, it drops further to 17 volts. This indicates that my cooling system isn’t putting a significant load on the power supply, as it only draws about 3 amps of current. If this were truly an 8-amp power supply, you wouldn’t see much voltage dip until the current exceeded 8 amps.
When I first tested these power supplies, I could only draw just under two amps before the voltage started to drop from 24 volts. I’m hoping this is the root of the problem. Ideally, when the alarm goes off, I’ll check and find the voltage at around 12 volts or something similar.
Even though the power supply is subpar, it’s not the root of the problem. You can see that it was running quite steadily when the thermal runaway occurred.
Although 17 volts is low, it didn’t drop too significantly, and the temperature remained stable throughout. My next plan is to connect the thermistor to a transistor and monitor the temperature externally. I’ll also increase the polling frequency. Currently, the graph updates every 500 milliseconds, but I’ll reduce this interval to detect any quick breaks in the cable. If that doesn’t reveal the issue, I’ll reassemble everything and retune the PID. I prefer not to start replacing parts until I fully understand what’s happening.
04/23/2024
I think it’s official: my printer’s power supply is slowly dying. Today, I monitored the thermistor, the voltage to the head, the voltage on the supply, and the voltage to the piezo buzzer to know when it faulted. The main supply voltage dropped to about 16.5 volts when the head was on, and I believe it took too long for the temperature to start rising, which caused the fault.
I retuned the PID, and it’s been printing without any problems for the last two hours. I bet it won’t have any issues for the rest of the night. I’m going to order a new power supply and overdo it a bit by getting a 14-amp power supply. They’re almost the same price, and it’s the largest slim PSU that can still fit in my case. Since I’ll only be using about a third of its capacity, the supply should last much longer. Even if it’s a cheap Chinese one, I should still get 4 to 6 amps out of a 14-amp power supply. LOL
09/05/2024 The Project Starts
I found a new project. At work, I use equipment that takes batteries. When the batteries start to run low, the equipment shuts down until it detects new batteries. The voltage in the batteries when they are deemed dead is still around 1.2 volts for alkaline D batteries. The equipment uses two of these batteries, and it feels like it is prematurely claiming they are dead.
I considered installing a DC to DC converter to compensate for this, but as any electronics enthusiast knows, this is not a viable solution. Before making any modifications to the expensive work equipment, I thought it would be wise to test the concept on the bench first.
In the past, I struggled with monitoring voltages from multiple channels for extended periods. A while back, I had a problem with my 3D printer, and I used a microcontroller as a multimeter, sending the data to the computer for logging and monitoring. This approach worked well to help me debug my 3D printer’s issues. So, I thought I would apply the same method to this new project.
It then occurred to me that this could be a fun project. I have a couple of microcontrollers with built-in USB serial interfaces and 10-bit ADCs, which are ideal for a multimeter. Thus, the project was born: I am going to build an 8-channel USB multimeter.
Most of the project will involve programming software that runs on the computer to make the data loggable and useful. The multimeter part itself will be relatively simple. Currently, I am planning for six channels with a maximum of 16 volts and two additional channels with a maximum of 50 volts. I believe this setup will cover all my needs for building a multimeter.
09/06/2024
Just an update on my project. I’ve built the entire microcontroller on a breadboard, and it seems to be working well. I’ve decided not to include any kind of interface on the device itself. Instead, I’m adding a multi-color LED. I’m thinking of having it light up as yellow for power on, green for proper communication with the software on the computer, and red anytime the voltage at an ADC reaches its highest level, which could indicate overvolting of that particular system.
However, after attaching the LED, I’m noticing a lot of noise in the ADCs. I need to figure out what’s causing this.
09/07/2024
This is a quick breadboard setup of my 8-channel multimeter. The right oscilloscope shows the onboard power supply without using the LED, and the left one shows it with the LED. It’s a 4 kHz signal with a 100 mV deviation, which is significant. While taking these pictures and writing this, it occurred to me that I’m not running the LEDs at full power; I set them to run at half power so they wouldn’t be too bright. I wonder if this is causing a PWM (Pulse Width Modulation) effect. Time to investigate!
09/08/2024
I hope you don’t mind, but I have to share my progress. I was able to filter out about 90% of the noise from the noisy LED, so I’m back to having really stable measurements. My microcontroller is now fully communicating with my measurement software, both sending measurements and transferring debug information. The measurement software is receiving and parsing all the measurement data, and I’ve also started developing its settings window. This part is really fun because the settings for the device are sent over to the device, which then implements them. For example, the interval between measurements can now be set from the measurement software.
I’m planning on having the measurement software treat the data more like an image viewer. You don’t use every single pixel in a photo when you’re just viewing it on the computer, but sometimes you might want to zoom in on a spot, and that’s where those extra pixels become useful. When I put together the spreadsheet and graphing for displaying the information, it will have a resolution slider. So, when I look at a set of measurements that were a minute long, I might set it to one measurement per second. It will pull out one measurement per second from the data and display it, but if I decide I want to “zoom in,” I will be able to.
I also like the idea of having floating windows. I might make a separate window for each live voltage reading that you can resize and customize. The same goes for the graphing. I think I will also make the graphing able to view old data, automatically expand with new recorded data, or follow the new recorded data with a specific duration. The main window will be where I keep the spreadsheets, making the data easily copyable and selectable to view on a graph. This is turning into a really fun project!
09/15/2024
I switched over to a tri-colored LED instead of using the individually addressable WS2812D LED. Now, the noise level has completely dropped off for my entire project. I definitely think the PWM in the addressable LED was causing all the noise.
09/23/2024
I fixed the lag problem I was talking about. Apparently, if you don’t set a receive buffer size on the comm port, it’s considered quite high. I reduced the buffer size, and the reaction time was fixed. Tomorrow, I’ll start working on the calculations tab in the settings. That’s going to be a complicated one.
I’m not sure how high the buffer goes, but what was throwing me off was that I would pause the execution of my program, go in and code some, come back out, unpause, and during that time, the buffer would have just been filling up. My software running on this machine looks like it was processing about 20,000 samples a second when the buffer was full of that large amount of samples. It really made it feel like there was a delay while it chewed through the data.
Since my device’s light would turn yellow when it stopped receiving communication from my program because it was paused, in my mind, it wasn’t sending, but it still was. This will turn out to be a feature once compiled because that means if the program freezes at all, the buffer will still capture all the data. The microcontroller gives each sample its times Okay tamp, so when you’re looking at the data, you won’t miss a beat.
09/24/2024
I was thinking last night about how I want to handle the calculation settings. I’ve decided to have three separate calculation settings per channel: one for voltage, one for ohms, and one for other. The “other” setting would allow you to incorporate any kind of sensor and still plot natively with whatever values you calculate. For example, you could use a thermistor for monitoring temperature or a Hall sensor to calculate the distance to something magnetic or the strength of a magnetic field.
One thing I realized is that calculating ohms on one of the 16-volt channels using only the 5 volts I have on board as the supply voltage would only give me a resolution of 320 steps instead of 1,024. This means a standard 10K thermistor would only have a resolution of about 2° per step. So, instead of having six 16-volt channels, I’m going to have two 5-volt channels, four 16-volt channels, and two 50-volt channels.
The 5-volt channels would be more challenging to use because of their voltage limit, but for tasks like using sensors or measuring logic circuits, I would get the full 1,024 resolution from 0 to 5 volts. For the thermistor example, this would give me a resolution of 0.06°. Just some thoughts I had while staring at the ceiling before falling asleep.
09/28/2024
I just realized something awesome about my multimeter project. It has voltage limits. If the ADC experiences more than about 0.2 to 0.3 volts over the reference voltage (5V ±), it can release the “magical blue smoke.” I thought I could use a Zener diode to clamp the probe voltage to ground. My initial idea was to use a 16V Zener diode to ground for the 16V channel. This way, it wouldn’t conduct until the voltage exceeds 16V, protecting the pin from going over 5V through the voltage divider.
However, I realized that if you’re measuring a power supply that can push 2 amps, the Zener diode would need to dissipate 4 watts of heat. Most Zener diodes on circuit boards are only rated for a quarter watt to maybe five watts for the larger ones, so this wouldn’t work well. Then it occurred to me: why not put the Zener diode on the other side of the voltage divider? There’s practically no current going through there after it passes through the divider because I’m using a 10K ohm divider. The current required to clamp it to zero is so small that even the smallest Zener diodes I have will work. I can use the same 5.1V Zener diode on all the voltages right before they go into the ADC.
The picture shows my input voltage being 32V with near zero current draw, while the voltmeter shows the voltage through a 16V divider being clamped to 5V. The bottom image shows what happens without the Zener diode clamping it.
This has a bit of an artifact as we get close to the Zener diode’s clamping voltage. It starts to draw a minute amount of current at around 4.8 volts. This means that across the voltage divider, from about 14 to 16 volts, I’ll need to take more samples because the plot curve will drastically change for the equation that maps it to a voltage.
My variable power supply only goes up to 32 volts, but it was rising just a little bit at a time even at 32 volts. So, instead of having a voltage limit at 16 volts, it looks like about 80% of the resolution will be at 14 volts and under, and 20% of the resolution will be from 14 volts to about 50 volts.
09/30/2024
- I’ve made some advancements on my 8-channel multimeter project. As you can see here, I now have many of the settings completed for the calculations part of the software. All 8 channels are functioning and have three separate options for configuration. This allows you to set them up for measuring voltage, ohms, or just about anything else you can think of. There’s plenty of room for adding sensors, thermistors, or whatever else you want to monitor. I’m also adding a notes card feature, so you can add notes for each channel to explain what a particular reading is for future reference. There are three preset configurations that can be used at any time. Both two and three-degree polynomials are supported, and there’s complete integration with my settings equation fitting calculator. Each equation in the settings has a button that allows you to import all the settings from the calculator. The calculator also has a checkbox to automatically prefill the current raw values when you enter a reading from a known source, making data entry quick and easy. Whatever tab you have selected for a channel is where the information will come from.
It feels pretty intuitive. There’s still a lot left to do, but this was a significant jump today. Thanks for letting me share!
10/05/2024
OK, I know this doesn't look like much, but it was hard work under the hood.
It now records everything from the device, basing the recording time off the actual time, and then tracks the time from the device's timestamps. This means the measurement times will correlate perfectly despite the delay (currently about 50 to 100 ms). It's also recording everything dynamically in RAM, allowing indefinite expansion. However, it seems like every 10 minutes or so at 100 measurements per second only increases the memory usage by about 2 MB for the program, so it shouldn't be a problem.
Additionally, all the calculations are done in real-time, and when running at its top speed of about 400 samples per second, the program only uses about 2% of my CPU's single core. So, everything is working really well at this point. Next, I get to start designing how I'm going to display and save that data.
Thanks for letting me share.
10/11/2024
So, even though I'm not done with this project, I would like to take a moment and try something like PCBWay to print the circuit boards. It would be nice to know how to go about making professional circuit boards instead of the ones I've been making, which haven't been too bad but aren't professionally done. I'm going to use this project to get some professionally made boards. Since I haven't made any changes to my breadboard at all, I figured I would go ahead and run these off. While I wait for them to be printed, I can take the week or two to finish my software. Here is the wiring diagram. There are actually more components than I expected, but it's all very straightforward and simple.
10/13/2024
So, I think I've finished my scrub bar control. I have it working in many different ways. Just like the horizontal scroll bar, when you click the empty area, you can see the controls jump in 5% increments based on the minimum and maximum values. Additionally, clicking in between the two causes them to split the difference, which I thought was kind of fun. Also, the last control you used has keyboard control. You can use the left and right arrow keys to change it in increments of one. Holding the button down causes it to scroll quickly based on the min and max differences. The up and down buttons will cause the entire control to pan left and right. You can also see with these other horizontal scroll bars, which are for testing, that the control is connected programmatically to the values of the scrub bar. This scrub bar reacts as you would expect. Changing the minimum and maximum values causes it to react as well. Design-time properties are also working out properly, and being able to resize the control in any way functions as expected. When you size it small, it will reduce the font of the controls and try to make everything look pretty. I think I'm at the end of the control and ready to move back to the main program. This has been a pretty fun side quest!
10/16/2024
I was working on my multimeter project and adjusting some of the voltage splitter resistors when I noticed that I've only got 4.7 volts on this board. I know the USB 3.0 standard for default voltage is supposed to be between 4.0 and 5.25 volts, and that's what I'm using as the reference voltage for the ADCs. This means every time I hook this up to a different USB port with varying voltages, I would have to recalibrate it, which is not ideal. It's not very noisy, but it's still a little noisy. The microcontroller I'm using does have an internal voltage regulator I can switch to, but that's only 1.1 volts, and that gives me less error margin when it comes to my resistor choice. I do have some ultra-low noise 3.3-volt regulators for ICs, and I think I'm going to switch to that. This will give me the headroom for a bit of error, meaning the USB voltage, in the worst-case scenario, could be anywhere between 7 volts to about 3.6 volts and it would still function properly, providing the same readings. I haven't ordered the circuit boards yet, which is good. I think I'll hold off on ordering the new boards until I get the new mini fuses this weekend, which will give me time to make sure this is what I want to go with. In the video, you can see in the first half the noisy voltage directly from the USB port, and in the second half, how quiet that 3.3-volt regulator is.
10/18/2024
I've decided, instead of adding an entirely separate hand control, to just add it to the center area between the two slider controls. It makes sense—like in some video editing software where you grab between the two sliders to pan. I'm going to include that here so it works the same way, and it will eliminate the need for an entire additional slider, which will be nice. Now, I will have this multifunction scroll bar, a zoom slider, and a resolution slider. I think these will be all the controls required to scrub through the entire database quite easily.