Open Source Multimeter

[Fox]

Oh i have a design for my own 5.5Digit Meter project almost finished (about 70% i guess, most of the digital stuff has to be done),
but i am not sure if meets the 5.5Digits resolution criteria and its still a bit in the flow, so i won't share it just yet.

[HLA-27b]

I didn't mean to offend you or something 

Hope it turns out good.

[Fox]

No offense taken

[Zad]

It is one thing to draw what you want, but another thing entirely for it to be physically possible. For example, consider how much volume a typical meter uses for voltage and current switching, including the precision film resistor packs and 10A current shunts. Add to that a data isolation and PSU, conditioning and ADC. Now consider that on the drawing shown so far, they have to fit in <10 cu. cm.

With ADCs you can always integrate readings and filter digitally, allowing the user the opportunity to offset speed for precision. If a user is looking for spikes, then they are highly unlikely to be interested in ultimate precision.

Here is a graph of some investigations I made into the performance of some fast 24-bit TI ADCs. Noise free 20 bits (6 digits) at 100SPS is perfectly achievable, but note that this is purely the performance of the ADC, without any conditioning, protection etc.

[HLA-27b]

It is one thing to draw what you want, but another thing entirely for it to be physically possible. For example, consider how much volume a typical meter uses for voltage and current switching, including the precision film resistor packs and 10A current shunts. Add to that a data isolation and PSU, conditioning and ADC. Now consider that on the drawing shown so far, they have to fit in <10 cu. cm.

Man if it takes space it takes space. Make it 300 cm3 and worry about miniaturizing it later.

With ADCs you can always integrate readings and filter digitally, allowing the user the opportunity to offset speed for precision. If a user is looking for spikes, then they are highly unlikely to be interested in ultimate precision.

Here is a graph of some investigations I made into the performance of some fast 24-bit TI ADCs. Noise free 20 bits (6 digits) at 100SPS is perfectly achievable, but note that this is purely the performance of the ADC, without any conditioning, protection etc.

If we can get 20 bit resolution out of a 24 bit ADC, does that mean that we have an error budget of 4 bits? Or maybe it is not that straightforward. I need to figure out how to stack up these errors...

[bobski]

Open Source multimeter will likely be an evolutionary project. There will be many iterations and refinements. By and by it will get better.

Another excellent reason for a modular design. Whether it's a eurocard form factor or something new, a modular design will allow for cheaper upgrades and modifications.

I gave SketchUp a try. It may actually be one of the easiest to use 3D apps I've ever encountered. Yeah, it feels tedious and clunky at times, but much less so than other software. Sorry... No buttons, LCDs or connectors yet.


That's an 18650 cell by the way.

[HLA-27b]

Glad to see SketchUp catching on 

What is the orange box made of, plastic or aluminium?

[bobski]

Plastic would be much lighter and provide better impact protection, so I guess I'll go with that. Some metal hardware is probably in order though... Those dovetail rails for the modules could be a weak point if they're just plastic.

[HLA-27b]

There are a few things to bear in mind when designing for plastics. If you want to get more serious about this better read  about "designing plastic parts" on google.

What remained in my memory from my design class is

- Uniform sections - Basically you want all your walls the same thickness, otherwise they warp when they cool.

- Mold draft - You need some angle of slope so that your part can come out of the mold.

- Parting line - Basically you pick two opposite sides of a part. All surfaces of that part must be clearly visible from one of these surfaces. You will need to change the orientation of your battery compartment. Otherwise the manufacturers will hate you (and bill you accordingly).

- Minimum part count - molds are obscenely expensive. No less than 10000$ a pop.

These things will improve your designs a lot.

[bobski]

Okay, I'll stop now. ^_^

[HLA-27b]

Good job. It looks a lot better.

[Zad]

Glad to see SketchUp catching on 

What is the orange box made of, plastic or aluminium?

Would you touch the case of a aluminium housed multimeter when it was measuring mains voltage? 

[HLA-27b]

Would you touch the case of a aluminium housed multimeter when it was measuring mains voltage? 

PXI stuff has metal panels and sits in a metal case. Apparently it is safe.

Amateur stuff? Not sure, it depends.

[alm]

Most commercial bench meters have a metal case. No safety issue as long as the case has a solid ground connection, and of course the front-end is fully floating and well isolated. I wouldn't use metal for a meter that can run on batteries, however, since you can't guarantee a ground connection (like those bench scopes with battery option that some people use without the mandatory ground connection for anything but SELV). Most commercial bench meters with battery option also used a plastic case I think. It is possible to make double isolated equipment with metal cases, plenty of consumer gear is. But you would need double the clearance and creapage distances as appropriate for the voltage rating.

[gregariz]

Most commercial bench meters have a metal case. No safety issue as long as the case has a solid ground connection, and of course the front-end is fully floating and well isolated. I wouldn't use metal for a meter that can run on batteries, however, since you can't guarantee a ground connection (like those bench scopes with battery option that some people use without the mandatory ground connection for anything but SELV). Most commercial bench meters with battery option also used a plastic case I think. It is possible to make double isolated equipment with metal cases, plenty of consumer gear is. But you would need double the clearance and creapage distances as appropriate for the voltage rating.

If your clumbsy like me you'll often be dropping your meter in the wrong place. Plastic is good when that place has some voltage.

[Zero999]

I don't see why there's lots of contaversy over a hand-held meter with a metal case. It's perfectly safe, as long as there's sufficient insulation between the live parts and case and it's mechanically robust enough to withstand being thrown around and stamped on. A good metal case should be stronger and safer than any plastic cause.

[Bored@Work]

I don't see why there's lots of contaversy over a hand-held meter with a metal case. It's perfectly safe, as long as there's sufficient insulation between the live parts and case and it's mechanically robust enough to withstand being thrown around and stamped on. A good metal case should be stronger and safer than any plastic cause.

And is a risk on a workbench or near / inside a dangerous machine, where the case could cause a short circuit. And if you hold that instrument the moment you cause the short circuit with the case, your body could become a significant part of the circuit.

[HLA-27b]

If your clumbsy like me you'll often be dropping your meter in the wrong place. Plastic is good when that place has some voltage.

I see your point. metal case is not good for hand held equipment.

[Conrad Hoffman]

My POV may not be anybody elses, but here it is anyway-

This is 2011. We have capabilities they didn't have back in the '60s. We might want to think not in terms of traditional dedicated test equipment like DVMs, that have been around forever, but something much more flexible. I collect used test equipment and even with all the stuff I've got, I can't make the measurements I (or my customers) need at a price I can afford. The world is awash in DVMs and even the junk ones are surprisingly good. 6 1/2 digit meters with HPIB are no problem, new or used. Ditto frequency counters. What I'd really like is a black box having a synthesizer chip (Analog devices has some great ones), several DVM type measurement channels (at least 2), a phase detector in hardware, some internal impedance standards and a USB interface. Maybe a counter, but not sure.

Why?

Because all that stuff can be configured not only as a DVM, but as a VAW meter, an LCR meter, a network analyzer, an audio analyzer, a signal generator and perform pretty much any bench test or component test you might want. Under PC control (IMO it has to have open software) I could configure it for whatever needed to be done, like swept measurements. I have no interest in a display, other than the PC, but a lot of interest in data-logging. Note that almost all design work in the commercial world requires data-logging.

Before you get the idea that this is a huge or complicated effort, look at the high level parts you can get today. The synthesizer chip solves that problem. There are phase detector solutions. Instantaneous voltage and current measurements give you the VAW meter function. It's a big design, but not nearly as difficult as it would have been a decade ago.

Take a look at the $16,000 Agilent E4980A LCR meter. Once you lose the display there's no reason that can't be done for a hobbyist price. With the right software you have a network analyzer (I've done this). Now look at something like the watt meters and stuff from Clarke-Hess. That functionallity shouldn't be tough to include.

I realize I've gone completely OT here, but IMO pursuing another DVM, regardless of features, will never give ROI commercially, and I can't imagine any DIY builder bothering with it. OTOH, I've been wrong before- just ask my wife!

Best,
Conrad

[HLA-27b]

Welcome aboard Mr. Hoffman,

I am so happy to see you here. Your ideas are most most definitely welcome.
In fact it never dawned on me that this thing may include a synthesizer.



PS: I received your mail. But since you are here how would you feel about posting your magazine articles (or links thereof) under your own name?

[bobski]

From what I've read so far, the circuits for measuring inductance and capacitance values are both based on applying a sine or other fluctuating signal to the component and measuring its reaction. Making that signal source more dynamic and available for other functions makes perfect sense to me.

[Conrad Hoffman]

Hi,

Sure, these are ancient but everyone is certainly welcome to read my Mini-Metrology Lab articles here. That's how things were done in the good old days. Feel free to wander the rest of the site- should be something there for just about anybody.

On the multi-function meter gadget, I don't know why nobody has done this before. All the hardware is practical, but the big players keep making their very expensive traditional instruments. Heck, the most desirable low frequency network analyzer is about $40k. Put the right package together and you might have a very desirable moderate volume product.

BTW, the key to LCR is the phase detector. Phase is your loss, and converts to esr, dissipation factor or however else you want to express it.

Best,
Conrad

[nullsmack]

Have you guys ever seen this: http://mondo-technology.com/super.html? It squeezes quite a bit of functionality into something handheld size.

[bobski]

It squeezes quite a bit of functionality into something handheld size.

Now you've got me thinking about gutting my Mastech pen multimeter again.

[KuchateK]

Ideas presented by Mr. Hoffman are great.

I'm looking around for something above multimeter capabilities and much smaller and cheaper than existing alternatives for labs/professionals. Generator, LCR, capabilities that super probe has. Small, flexible multichannel replacement for three multimeters that are sometimes required and with logging capability that would allow automatic testing or graphing what is really happening over time. Such programmable device with open firmware would be great and I think there is huge market for that. Consider cost of two/three good multimeters with PC connection.

Hobbyist working on project can't really justify spending thousands of dollars on lab equipment and professionals would love lab on the go.

I was recently looking for LCR meter with ESR and this is just amazing how expensive they are. I can get Rigol scope for that price. It was much cheaper to replace all caps in the device than to hunt broken ones.

Bobski's modular project looks great, but that amount of plastic would require significant tooling cost and is not really hobby friendly. I can't imagine someone making modules for this design at home.

From hacking/modding point of view I would more likely see something that HAL-42b presented. Big PCBs for all sorts of crazy modules, big components, flexible backplane, generic off the shelf enclosures. I don't like sticking to industry standard though. Pursuing compatibility that fraction of users would utilize is huge waste of time and resources. This device should have as big microcontroller as possible with as many possible connections on the backplane as possible. Look at arduino. Minimal design compared to some alternatives. KISS rule really showed. Make it flexible and easy to duplicate.

Something to extend idea presented by HAL-42b to make it easier to get and duplicate...



I recommend Mike's videos. He is showing some nice stuff and ideas and has very little views.

Start designing circuits guys. The rest will come later. Remember! First generation doesn't have to be perfect.