Our hacker [John Duffy] wrote in to let us know about a video he put together to explain the design of his open-source multimeter, the HydraMeter.
If you’re interested in how the circuitry for a voltmeter, ohmmeter, or ammeter might work, this video is a masterclass. In this long and detailed video, [John] walks us through his solutions to various challenges he had while designing his own multimeter. We covered this multimeter last year, and this new video elaborates on the design of the HydraMeter which has been a work in progress for years now.
The basic design feeds voltage, current, and resistance front-ends into an Analog to Digital Converter (ADC), which then feeds into a microcontroller and out to the (detachable) display. You can find the KiCad design files on the GitHub page. There is also a write-up on hackaday.io.
The user interface for the meter is… opinionated, and perhaps not to everyone’s taste. In the video, [John] talks a little bit about why he made the UI work the way that it does, and he noted that adding a rotary range switch is a goal for version 2.0.
Thank you, [John], for putting this video together; it is an excellent resource. We look forward to seeing version 2.0 develop soon!
I’m not going to spend 1h40m trying fo find whether or not the design covers one important aspect of multimeter design: protection against gross overloads. TAoE and X-Chapters covers the HP34401A protection mechanisms in some detail.
Anybody care to comment about this design?
at 1m50s he wields a beige lunchbox that allegedly produces over a kilovolt for testing. also, the main pcb has cut outs for hv isolation, so he’s going about it the right way.
I gave up watching the video because the background music upset me. i guess i’m not hip enough like electrical engineers to be able to groove to it.
Fair enough, thanks for the feedback on that! Most other videos I found had some background music (also it helps keep edits from being as annoying/apparent), but I’ll definitely tone it down or remove for long talking sections in future.
The big lunchbox goes up to about 8KV ;)
triggered spark gap to discharge some caps into whatever is unlucky enough to be attached, once I’ve got measurement worked out I’ll post some details of that.
In the process now of seeing if those MOVs and slots and fuses actually do what they’re supposed to. Current input protection I’m not too worried about, voltage will likely be tweaked but is probably close, and I have no idea on resistance.
For those who don’t want to spend the time to go through 4000+ pages of TAoE and X-chapters, can you indicate which volume and page the discussion of the HP34401A appears? It’s not in the index or tables of contents of the Second, Third or X editions (sorry, I don’t have the 1st ed. handy to thumb through).
Actually, the 34401 info is in the ToC, viz:
5.12 Designs by the masters: Agilent’s accurate DMMs
5.12.1 It’s impossible!
5.12.2 Wrong – it is possible!
5.12.3 Block diagram: a simple plan
5.12.4 The 34401A 6.5-digit front end
5.12.5 The 34420A 7.5-digit frontend
and
4x.16 Designs by the masters: Bulletproof Input Protection
It only took <5mins to locate the books, locate the sections, and type this. Much less than1:40
Silly me, looking for “HP”, “protection”, “overload”.
As always, search terms are critical :)
Helps to have a PDF; fortunately Win Hill did upload TAoE3 (not x-Chapters) in 2017.
It is still there, and I’m not going to say where other than a dropbox account: I’ll let you search for it :)
It couldn’t be in the first or second edition (1989), since the HP34401 was introduced in 1994
And for those few who wonder what TAoE is, run (don’t walk) to https://artofelectronics.net/ and buy the books. (And do not view any of Horowitz’s interviews — he’s a much better human on paper.)
I think it’s important to know why his interviews (and views, and activities) are harmful, and make an informed decision as to whether to financially support him.
I made no claims of harm, no mention of views or activities, and did not counsel not supporting. Do your own research if you feel a person’s interview performance is an important factor in your buying decisions. I just shared my opinion that his contributions in printed form are better (i.e., more pleasant, more valuable use of time). I apologize if that turned out to be a trigger for something — That was not my intent.
Are you mistaking Paul Horowitz for the late David Horowitz. He received criticism for his SETI work with Carl Sagan as a waste of resources but that’s purely scientific.
I’m also not going to watch an 1h40m video right now, but the Hackaday article has a direct link to the github page with KiCad project, schematic and PCB.
So you can easily look for yourself whether the hardware suits your needs, and you can also add any extra protection you like. Also, on the project page of this project, you can see some decent sized fuses, and both MOV’s and gas discharge tubes, so at least some thought has been put into input protection.
For the rest. The github repository apparently has not been updated for 7 months, and that sort of lines up with the previous article from Hackaday about this DMM.
In due time I’m going to have a closer look at this project. I’m not looking for another handheld, but I am interested in a cheapish benchtop DMM with a quick update rate >100Hz and big display (LED or TFT). And a meter like this can be a decent start for a similar project.
“I gave up watching the video because the background music upset me. ”
Ah yes, that annoys me too. If you want to hear music, it’s so easy to mix in your own music. Either directly form your PC, or mix it acoustically from a completely separated stereo.
I asked notebookLM, the author doesn’t seem to be too sure about the safety:
The HydraMeter_0.4 open-source multimeter incorporates several safety mechanisms to protect against gross overloads, though it is crucially important to note that the designer explicitly states this meter is not designed, rated, or intended to be used with mains voltage and has no serious design analysis or testing [1]. It should not be used for anything with serious power [2]. The current version (0.4) also has some unverified tweaks and known issues with its protection circuits [3-5].
Here’s a breakdown of the intended and implemented overload protection mechanisms:
General Safety Warnings and Design Philosophy
The designer’s intent was to explore how to achieve certain protection levels (e.g., CAT3 600V for voltage measurement), but the meter itself is not rated for such use and should not be used on mains [6].
The goal of protection is to limit current and voltage during an overload to prevent component damage, arc flash (which can be deadly), and explosion [2, 6, 7].
Current Input Protection
The main danger in current measurement is overcurrent transients, which can cause arcing and explosion of the fuse or within the meter [1, 2].
* High Rupture Capacity (HRC) Fuse:
* This is the core solution [2]. HRC fuses are designed to contain high currents (tens of thousands of amps for milliseconds) and snuff out arcs by filling the gap with sand as the metal melts, preventing continuous conduction [2].
* The meter uses smaller HRC fuses, rated for similar braking current but a lower voltage (400V instead of 1000V found in standard meters) [2].
* Diodes Across Shunt Resistors:
* To protect the shunt resistors (low-value resistors used to measure current) from overheating and to limit overvoltage downstream, pairs of diodes are connected in reverse parallel across each shunt resistor [8].
* These diodes conduct very little current at low voltages (sub-nanoamps below ~100 millivolts) but conduct very strongly (amps or tens of amps) above a few hundred millivolts (e.g., 200mV burden voltage) [8]. This shunts excess current around the resistor, saving it from power dissipation [8].
* Current Limiting Resistor: A 1kΩ resistor is placed before the Analog-to-Digital Converter (ADC) to limit current if a high voltage makes it past the initial stages and feeds into the ADC [6].
* Metal Oxide Varistor (MOV): A MOV is used to clamp the voltage to a manageable level. The specific MOV chosen (nominally 3.3V) will turn on around 5V and might allow up to 8V across it with significant current [6].
* Internal ESD Protection Diodes: The final stage of protection relies on the internal electrostatic discharge (ESD) protection diodes present in the ADC chip [6].
Voltage Input Protection
Voltage protection aims to tolerate high voltages continuously and even higher voltage transients without damage or arc flash [6].
* First Stage – Resistor and MOV:
* A critical 1kΩ resistor is placed at the input [7]. It must withstand several kilovolts (e.g., 5kV) for microseconds, limiting the maximum current that passes through the meter and preventing arc flash [7].
* An MOV (rated for ~1000V) suppresses arc flash, limits the voltage to the next stage, and helps to smooth out sharp edges of fast transients [7].
* RC Filtering (Resistor-Capacitor Filter): A 100kΩ resistor followed by a ~200pF capacitor forms an RC filter, effectively suppressing sharp voltage edges with a time constant of about 20 microseconds [7, 9].
* Gas Discharge Tube (GDT):
* A GDT provides additional protection [9]. Unlike an MOV, which still maintains a standoff voltage, a GDT can “crowbar” the voltage to almost zero (a few volts) even at very high currents once it “strikes” [9, 10].
* It is slower than an MOV, so fast transients might initially pass it, but it will clamp the voltage to a tolerable level within a microsecond or two [9]. It helps contain the overload to the upstream part of the front end [9].
* Series Resistors (e.g., 10MΩ): High-value resistors, often composed of several smaller resistors in series, are used to withstand the kilovolt-level voltages that might appear across them during transients [10].
* BJTs (Bipolar Junction Transistors): A pair of BJTs configured to act as a low voltage, fast, low capacitance, and low leakage protection circuit. They clamp and absorb any smaller transients that make it through the earlier stages, typically turning on between 5V and 10V [11].
* Internal ESD Diodes: Similar to current protection, these serve as a final line of defense for small current amounts that reach the ADC [11].
Important Caveat for Voltage Protection: The designer notes that the actual implemented design (version 0.4) is “worse” than the theoretical ideal. It includes a large capacitor that would pass the entire overload voltage to subsequent stages, potentially blowing them up, and lacks sufficient downstream filtering, making it noisy [4].
Resistance Measurement Protection
Resistance measurement involves supplying current and measuring voltage, making it vulnerable to both high voltages and currents during overload [11, 12].
* Resistor and PTC (Positive Temperature Coefficient) Thermistor:
* A PTC is intended near the front of the circuit [11]. These components exhibit a dramatic increase in resistance (into megaohms) when their temperature rises above a certain “Curie temperature” (e.g., 70-100°C) [11, 13].
* During an overload, current flowing through the PTC heats it up, causing its resistance to skyrocket, which in turn limits the current and power dissipation, eventually equalizing to a tolerable amount [13].
* However, the current version of the meter does not include a PTC [11].
* Metal Oxide Varistor (MOV): Similar to voltage protection, an MOV blocks current flow at low voltages but conducts strongly above a threshold (e.g., 200V) to protect against voltage overloads [13].
* Diodes for Positive Voltage: A simple diode is used to allow current to flow out but not in, protecting against positive voltage overloads [14].
* PNP BJT (or Darlington Configuration) for Negative Voltage:
* This is a more complex circuit designed to handle negative voltage overloads [14]. It allows current to pass in one direction but prevents the output from being pulled too far below ground if a high negative voltage is applied [14].
* The circuit uses a PNP BJT (in a Darlington configuration for higher current gain), which effectively biases on during normal operation but acts as a constant current source during overload, protecting upstream components [15].
* This stage can be duplicated in series to extend the voltage range it can withstand, potentially up to 1000V [15, 16].
Important Caveat for Resistance Protection: The designer states that the protection circuit for the resistance current sourcing “does not work and has been thrown in bad idea jail” [5].
In essence, the multimeter’s overload protection is like a series of gates and shock absorbers along an electrical highway. Fuses are like emergency exits that completely shut down the flow in extreme situations, while MOVs, GDTs, and diodes are like speed bumps and crash barriers that absorb and divert excess energy, preventing it from damaging the sensitive parts of the circuit. The PTC is like a self-regulating valve that chokes off the flow when things get too hot. However, it’s crucial to remember that some of these safety features in the current design are theoretical, unverified, or known to be deficient, emphasizing the need to treat this as a proof of concept and not a fully protected commercial instrument.
Please do not pollute the comments with AI summaries. If I wanted to ask an AI, I would do so myself.
Video was done to cover all aspects, input protection starts around the 40 minute mark (current, then voltage, then ohms).
The ohms negative voltage protection is pretty much identical to the design you’re referring to, voltage needed a little different design because I can’t switch the HV side, everything has to be done low side, and I’m shooting for a higher rating than most bench meters are designed to.
Massive feature creep…
Yep :)
This entire project is scope creep starting from “I need to buy a new mulitmeter”.
I have cut like half the functionality I was initially planning on to get it working, but plan to add back on just as much eventually.
Most of the first hour is about “here is the obvious simple thing, why I can’t just do that, and what I added to get around it.” Usually, the reason was concerns about gross overloads.
I am often annoyed by the background music, but not so much here. I don’t know if that is a quirk of volume or a new upload or just different tastes.
Fluke makes a meter called the Hydra
crap.
It looks like it is discontinued, but they might still have copyright/trademark.
thanks for letting me know! I did look for anything else called hydra meter, but I guess it being classed as a DAQ made it hard to find.
Having done a module on Solid Edge in college, I will say that I had glowing things to say about it … glowing with anger at the way it was done
❤️
hahaha
Yes!
It seems like they wanted to keep it from stepping on NX/other sw sales by crippling the functionality for professional use, problem is they just made it so bad no one wants to use it. IMO NX is already kind of slow and annoying to use though, so making it substantially worse makes it basically unusable…
At least it’s not Catia 🙃