Open Source Multimeter

[sync]

Also the resistors must dissipate about 9W for 1000V input. Spreading it over several resistors make sense.

[mrflibble]

Doh! Extra voltage standoff and dissipation capability makes a lot more sense for those extra 2 resistors. Thanks!

[mahjongg]

two comments:

sorry if discussed before.

first, don't make the mistake to make them yellow!
https://www.sparkfun.com/news/1428

second, for reasons of compatibility of adapters and such space the two 4mm banana jacks that make up a signal pair exactly 0.75 inch, (19.05mm) apart!
measure any multimeter (or other meters with 4mm inputs and you will note the trend of them being 0.75 inch apart, there is a reason for that, its called a pomona adapter.

[miguelvp]

Yellow is fine, is the "dark-colored body or face and a contrasting yellow border, frame, molding, overlay, holster or perimeter." will get you in trouble.

You could do a red face yellow border and you'll be fine. (unless someone else has that trade dress)

[mrflibble]

Or the ever popular alternative: not care about any of that for a DIY project.

[abaxas]

Yellow is fine, is the "dark-colored body or face and a contrasting yellow border, frame, molding, overlay, holster or perimeter." will get you in trouble.

You could do a red face yellow border and you'll be fine. (unless someone else has that trade dress)

How about calling it the "fruke u".

[branadic]

OSHW-Multimeter
https://github.com/Stoney49th/OSHW-Multimeter

Seems like another never finished OSHW Multimeter project or are there any news about your project?

[farsi]

On the topic of open-source oscilloscope, here is a nice a blog post:

http://blog.farsinotare.com/2014/07/04/my-new-lab-with-red-pitaya/

[ludzinc]

While not a true multimeter, I'd always wanted to build my own volt / amp meter with programmable gain and volts / current mode switching.

Without using relays!

Read about it here:  http://ludzinc.blogspot.com.au/2014/07/multi-channel-meter.html

[Felicitus]

While not a true multimeter, I'd always wanted to build my own volt / amp meter with programmable gain and volts / current mode switching.

Without using relays!

Read about it here:  http://ludzinc.blogspot.com.au/2014/07/multi-channel-meter.html

Looks neat, but do you have specs? How low can it go for current measurements?

[ludzinc]

http://ludzinc.blogspot.com.au/2014/08/multi-channel-meter-part-2.html

Specs as requested in the link.

[Mickle T.]

Recently abandoned project of the 7 1/2 digits DVM:
http://radiokot.ru/forum/viewtopic.php?f=10&t=105564

[fcb]

Without reading all 27 pages of this thread...

Has someone successfully made an OSHW (high accuracy, at least 5.5 digits) multimeter yet???

[charlespax]

Sometimes the best way to make a complex device is to do one part very well then expand on it. What would be the minimal viable product for an open source meter? Would it be useful to make a device that measures a single thing? Maybe the best place to start is an open source voltmeter. Thoughts?

[Bloch]

Yep the voltmeter is the core of any multimeter.


So that is a great start 


Quote from: charlespax on Yesterday at 12:16:55 PM

Sometimes the best way to make a complex device is to to one part very well then expand on it. What would be the minimal viable product for an open source meter? Would it be useful to make a device that measures a single thing? Maybe the best place to start is an open source voltmeter. Thoughts?

[Piedtyper]

WHEW!  27 pages…

I’m new to this so please be gentle.  I’m an Electronic Engineering Tech student in my second year and I’m planning on a relatively simple meter for my cap stone project. 

I’m almost as old and dirt and was a computer programmer/IT type for about 20 years.  It’s been very enjoyable competing with the significantly younger minds!  I’m confident in any of the programming needed but, of course, I’ll be challenged on the electronic design portion.

I’ve tried to keep the specs simple to start with the hope I can add to them in the future.  This will be completely open source and since it’s for school I’m not looking for anyone to do the work on at least the first cut. 

As to the future…a hand held with much the same capabilities outlined by Dave in the video would be a long term goal but based on the many comments it might not be reasonable.  Time will tell.

Initial specs

1.   High input impedance in the range of 10’s of M Ohms for voltage and low impedance … less than 10 Ohm for current.
2.   Auto-ranging between 200 to 0 AC/DC V and input protection for > 200 V
3.   Accuracy in the range of 4 ½ digits
4.   Two channel with fully isolated grounds
5.   Channel 1 measures V as stated above
6.   Channel 2 measures current up to 1 A with auto ranging and maximum resolution range of 0-100 mA
7.   Simultaneous measurement of V and I for power calculations


Design challenges and possible solutions:

1.   Auto-Ranging
a.   Start with 0-20V AC/DC only and design for a further attenuation stage before this stage.  All switching in this state would be solid state.  The stage in front would probably need to be a relay of some sort (digital I’d think) and would be designed to simply knock down the input voltage to the 20V range.
2.   Channel Isolation
a.   Combination of buffers and opto-isolators for I2C communication between the micro on channel 1 and the ADC/measurement circuit on channel 2.  Channel 2 will need to auto range the current up to 1 amp.
3.   User Interface/Output
a.   VERY simple to start.  Probably just an LCD displaying Voltage, Current and Power.  No user input except perhaps to display or not display power if both channels are in use.  It’s possible the measurements are discrete and shouldn’t be used for any power measurement.
4.   Input Protection
a.   Probably just start with Zeners (two 100V “ish” in series) through a high value resistor to ground.
5.   AC Voltage
a.   Use a RMS-DC chip.  It looks like this is possible.  Just feed in the appropriate level of AC and it spits out a DC value for RMS.

Questions: 
1.   Is there any way around having two distinct power supplies (even if batteries) between these two channels?  If not (and I suspect this is the case) then at least for a start would 2 AA’s per channel be appropriate as a proof of concept?
2.   What’s the best way to handle negative voltage?  The ADC’s I’m considering have a range or 0-2.048V with an internal reference.  How should I handle this?  I’d much rather not have to create a negative supply.

I wrestled with whether or not to start a new thread for this.  If anyone figgers it would be better in another forum or thread just let me know.

Thanks in advance for any input!

[David Hess]

4.   Input Protection
a.   Probably just start with Zeners (two 100V “ish” in series) through a high value resistor to ground.

This is similar to how it is normally done but low leakage diodes are used and a high input impedance buffer will be needed.

1.   Is there any way around having two distinct power supplies (even if batteries) between these two channels?  If not (and I suspect this is the case) then at least for a start would 2 AA’s per channel be appropriate as a proof of concept?

Batteries can work but normally some type of isolated power supply is used.  Small high isolation inverters are common and typically quieter than switching power supplies.

2.   What’s the best way to handle negative voltage?  The ADC’s I’m considering have a range or 0-2.048V with an internal reference.  How should I handle this?  I’d much rather not have to create a negative supply.

If you are using isolated supplies so the inputs are galvanically isolated, then adding a negative supply is simple.

Instrumentation ADCs usually have no trouble handing negative inputs but the input can always be offset by half of the ADC input range.

I wrestled with whether or not to start a new thread for this.  If anyone figgers it would be better in another forum or thread just let me know.

You should start a new discussion thread.

[Marco]

4.   Input Protection
a.   Probably just start with Zeners (two 100V “ish” in series) through a high value resistor to ground.

This is similar to how it is normally done but low leakage diodes are used and a high input impedance buffer will be needed.

LM662 will do both those things (not from -20 to +20 V though).

[David Hess]

4.   Input Protection
a.   Probably just start with Zeners (two 100V “ish” in series) through a high value resistor to ground.

This is similar to how it is normally done but low leakage diodes are used and a high input impedance buffer will be needed.

LM662 will do both those things (not from -20 to +20 V though).

The input buffer voltage range is usually quite small which you can see from the specifications of common voltmeters which support >1 Gohm input resistance on their low voltage ranges.  For instance on an HP3478A, it is only +/- 3 volts and on a DM502, it is only +/- 2 volts.  Specialized designs can extend it considerably by bootstrapping the input buffer which as a side effect also reduces the common mode error present in a non-inverting buffer.  A Keithley 917 for instance has an input voltage range of +/- 200 volts while maintaining an input resistance >200 Tohms.

It is generally a bad idea to rely on the input protection diodes of an operational amplifier in this sort of application.  Inexpensive low leakage diodes may be made from transistor junctions or JFETs.  The Keithley 917 mentioned above uses transistor base-emitter junctions for this.

[Michael Rempel]

This is about Dave's multimeter:

...

For Bluetooth / Zigbee / whatever you can use a removable module to save some money on the initial cost.

I'd recommend the RFM12B by HopeRF (http://www.hoperf.com/rf_fsk/cob/RFM12B.htm). They're low cost compared to ZigBee, at £6.17 individually from Farnell (UK), are available in smaller (Yet still easy to solder by hand; XBees are pretty large), more convenient SMD modules and, if I recall correctly, they consume less power. There's quite a lot of info about this module spread around this blog http://jeelabs.org/ (I'd recommend you take a read of this anyway, it's quite good).

Then there is the NRF24L01+ 2.4GHz Wireless Transceiver. From China to the world for a buck a piece in low quantities. You could set up a whole measurement universe with these, some cheap arduinos and a fair bit of software. Each function could be its own device if you want easily enough. You could even make different ones for different ranges and applications.

[Mickle T.]

A Keithley 917 for instance has an input voltage range of +/- 200 volts while maintaining an input resistance >200 Tohms.

Keithley 917 - what's this?

[David Hess]

A Keithley 917 for instance has an input voltage range of +/- 200 volts while maintaining an input resistance >200 Tohms.

Keithley 917 - what's this?

Sorry.  I meant Keithley 617.

[Marco]

Specialized designs can extend it considerably by bootstrapping the input buffer which as a side effect also reduces the common mode error present in a non-inverting buffer.

Huh, I never really thought about what would happen to leakage if you connected the input protection diodes to the bootstrapping voltage ... I guess Keithly could get away with two 1n4148s at that point (although they didn't, they used the base emitter junctions of a IT121).

[David Hess]

Specialized designs can extend it considerably by bootstrapping the input buffer which as a side effect also reduces the common mode error present in a non-inverting buffer.

Huh, I never really thought about what would happen to leakage if you connected the input protection diodes to the bootstrapping voltage ... I guess Keithly could get away with two 1n4148s at that point (although they didn't, they used the base emitter junctions of a IT121).

I am a little surprised Keithley did not reverse bias the base-emitter junction diodes by at least a volt or use two base-emitter junction diodes in series for four diodes total but I assume they tried those alternatives and found the configuration shown to work the best.  Considering that the input resistance is specified to be greater than 200 teraohms over a range of +/- 200 volts DC with an input bias current of less than 5 femtoamps, I am hardly in a position to second guess them.  I picked the Keithley 617 as an extreme example of what can be accomplished fairly easily and because documentation is available.

The leakage of a 1N4148 which is a gold doped switching diode can be high even at millivolts so there is no way to really use it in this application even with close to zero volts across it.  Bob Pease mentioned this very problem on page 69 of his book Troubleshooting Analog Circuits:

http://goo.gl/uVInjx

[Marco]

I wish he would comment in the text on why he used the base collector junction in the picture, the text only deals with the use of the base emitter junction.