As I can't afford , $100s to buy a bench meter I was wondering if there are any open source projects around ( and no not the CERN one thanks !!) I'm looking for one with good accuracy and resolution ( 5-6 digits )
dave
Not popular ones and not in that price at least. Many times open source ones can be even more expensive than alternatives. Then if you are looking diy solution how do you calibrate this meter. If you are looking accuracy and resolution then you just cannot do it yourself. With $100 you will have hard time to find 5 or more digits with bench multimeter but it can be done of course . You just have to try very hard and will take time. Example you can find 34401a $200 sometimes and that is really good deal. That probably will be over your budget but just giving you info.
Keep in mind that many times used bench multimeters do not come with test leads. If you want good ones look example probe master. Which multimeter(s) do you have now? Maybe you do not really need bench multimeter. There is many drawback but also many good things about them. Accuracy and resolution you can get also with handhelds.
Just some quick thoughts you will blow $50 on making a reference (lets pick an LM399 for the better end of 6-6.5 digits) for a start and very quickly run up $100+ on an ADC and miscellaneous bits so I cannot see you saving any $ whatever you do.
3468A, 3478A etc in HP will generally go less than than $150 as will some of the other older 19" sized ones.
Advantest 6871E is actually a 2 million count meter and tracks my 34401A (1.2 million count) really well often turn up for less than $200
Philips - assorted 51/2 less than $150
+ plenty of others
So unless you really want a project see what is around locally to you.
I don't know a ready open source bench meter in that range. I have seen a DIY meter from China, but it is probybly a little more expensive and AFAIK not open source.
I can use my high resolution multi-slope ADC board (tread under metrology) as a kind of limited DC voltmeter - though it still needs the PC for output. I have a simple LCD output version - but this still needs some work on the software. The BOM costs are still moderate but with input protection, case and display it would be really hard to stay under $100.
I have a not fully (but nearly) finished version at more like 4-5 digts (build around MCP3421 ADC with internal ref.), with manual range switching. It may be build relatively cheap, but even there the costs add up - especially the display and shunts. It includes current ranges and low grade true RMS, but no Ohms. A low cost version would likely need a different display.
A calibration for good accuracy alone may cost more than $100.
I do not know of any but that doesn't mean that they don't exist. I'm not one to go totally negative on the idea of DIY, if nothing else it can be well worth the learning exercise.
As I can't afford , $100s to buy a bench meter I was wondering if there are any open source projects around ( and no not the CERN one thanks !!) I'm looking for one with good accuracy and resolution ( 5-6 digits )
I don't really think that will happen. You end up having to achieve a very high level of performance and not having the hardware and design experience to get it done.
A better approach might be to build a meter in the 4 digit range out of a chip set specifically designed for multimeters. You should be able to source the components at reasonable prices and not be stretching your developing design skills. A meter with 4 to 4.5 digits should be plenty on many benches, remember the world started out with analog meters.
dave
I actually like the idea of more open source instruments. If you are going this route I'd certainly consider a meter design that is easy to achieve. I know many people are wowed by the latest high resolution meter with its great precision, however I'm going to upset a few people and say for most users the high resolution isn't needed. Obviously the end use dictates what you actually will need.
The stark reality is that with a low end budget you will get a low end meter. As such it is better to compromise on digits and spend more effort on software and getting as much functionality out of your chip set as is possible. Also a focus on MMI (man machine interface) and a really decent display might make sense.
You could (theoretically) build a $100 BOM basic 5.5/6.5digit multimeter. However it would require external calibration or you'll need to aquire external references and/or a 7.5/8.5 digit multimeter.
Don't know of any open-source multimeter designs, and it would take many 100's of hours for a skilled designer to create, can't see it happening anytime soon.
Probably better off saving for a good 34401A or similar if you need that degree of performance.
Those simple designs base on the LTC2400 have some weak points. The LTC2400 has the low side at ground and thus has a hard time with negative voltages.
Just a 5.1 V zener directly at the divider deserves a few
--> the zener leakage would ruin linearity. At least this is easy to fix with extra diodes and bootstrapping the zener from the buffer.
For a new design I would consider more like the relatively new ADS125H2 , as it allows a direct +-20 V but also 100 mV input range, and thus no external amplifier needed. Even the internal reference may be good enough (not that detailed specs), though there are better external ones. It is nice to have reference and input buffer internal, as with SD ADCs (especially the lower noise ones) the buffers driving the inputs and reference can effect the linearity. For a DIY design this adds another uncertainty : INL depends on more than just the ADC chip. Tesing the linearity is not easy.
Joromirs design for the handheld meter is not bad, though a bit expensive in some areas around the ADC.
I have not yet finished the DMM frontend part to my ADC (
https://www.eevblog.com/forum/metrology/diy-high-resolution-multi-slope-converter/). For a 5-6 digit level one could consider even leaving out external gain and just live with a 10 V input range and 1 higher voltage range with divider. Noise wise it does not absolutely need a 1 V or 100 mV range - it more like helps with linearity in some areas. Fewer ranges simplify calibration.
recently I just finished my 6.5 resolution voltage meter basic range debugging, and posted the HW in my blog. it's a low cost version, bom cost less than 100usd. It's based on ads1256, which is a very low noise adc, but the INL is not good, luckily it can be compensate. you can find the test result, schematic and Gerber file here:
https://aduecho.blogspot.com/2021/05/hw-open-source-65-digital-voltage-meter.html?m=1now the basic range (+-2.2v)works fine, with ODR2.5 Hz noise is about 1uV, most of time the last digit is not jumping even without sw filter.
in the following weeks I will make calibration for other ranges, after I finished all the debugging I will put it into open HW block here.
Thanks that looks remarkably like what I need