The concept is that the base instrument is an auto-ranging DC voltmeter, with a few expansion slots exposing a few signals from the meter.
To allow current measurement, a shunt resistor and a preamplifier is implemented on an expansion module feeding into the pre-scaler. To allow AC measurement some kind of AC-to-DC conversion (precision rectifier or RMS-to-DC conversion) is implemented on an expansion module as an filter. Resistance measurement is implemented as a constant current source on a module.
Reference modules are also possible: higher grade low voltage references for ADC can feed the voltage through the bus and higher voltage references used as internal calibration target can feed their reference voltage into the prescaler.
Sounds like it would be either fragile and flimsy, or huge and not portable and convenient to use.
It might work technically if it is a bench meter, but I can't see a single advantage over a decent meter and specialty equipment for extreme cases. Interchangeable references is probably the only thing that makes sense to be upgradeable.
The concept is not new and was popular some 40 years ago, for bench meters, with the availability of true RMS plug-in cards, for instance.
A modular construction on a portable DMM, while interesting from an engineering point of view, may not offer much benefits, commercially.
Sounds like it would be either fragile and flimsy, or huge and not portable and convenient to use.
It might work technically if it is a bench meter, but I can't see a single advantage over a decent meter and specialty equipment for extreme cases. Interchangeable references is probably the only thing that makes sense to be upgradeable.
So far the SATA connector I used as the module interface proved robust as those are designed to withstand the constant resonant vibration of server-grade hard drives spinning, and the expansion cards are surprisingly small - the biggest card, the RMS to DC converter implemented using AD637 and the LM399-based calibration target, are 1.5 inch tall and half an inch thick. The tiniest card, ADR03B-based enhanced ADC reference and TLC5615-based adjustable constant current source, are a little more than half an inch tall and a few millimeters thick.
My intention was to make the entire system upgradeable down to every subsystem (aka module) so I can tinker with different modules built down to different prices.
Add two AC modules to the chassis and you could measure the voltage across a load and the current consumed. Extra marks if you then calculate watts and (possibly) power factor.
Add an RF module and you could measure RF signals in dBm.
How big is the chassis
If you're the sort of person who'd add a better voltage reference then you'd just buy a good meter in the first place. You wouldn't buy a crappy meter then upgrade it.
Plus: It would never get a CAT rating, UL listing, etc. That's a deal breaker for many people.
And it would be very expensive so nobody would buy it.
Add two AC modules to the chassis and you could measure the voltage across a load and the current consumed. Extra marks if you then calculate watts and (possibly) power factor.
Add an RF module and you could measure RF signals in dBm.
How big is the chassis
If you want to measure any two thing at the same time you need a complete second analog front end. My board does not support this OOB but the related pins are broken out so a secondary, identical channel module can be added later should I need it. You will need two AC-to-DC modules though, one for each channel.
RF module is a good idea but just like the current module, the module will have to include the preamplifier itself before going to the prescaler.
Chassis size may vary as modules are installed and removed. Single-channel job with 6 analog slots will have roughly the same dimensions as one of those old bench-style portable meters. Dual-channel ones, or ones fitted with enhanced processor module (a fully digital module) will be significantly larger.
If you're the sort of person who'd add a better voltage reference then you'd just buy a good meter in the first place. You wouldn't buy a crappy meter then upgrade it.
Plus: It would never get a CAT rating, UL listing, etc. That's a deal breaker for many people.
And it would be very expensive so nobody would buy it.
The cost of getting a better meter is way, way more expensive than adding an enhanced reference module, especially when you already got the meter. At least you are not paying for a new main PCB, main ADC and main processor.
This, as of now, is a hobbyist project. Should I need to get UL and Category listing I can go there as long as I can afford it.
If you're the sort of person who'd add a better voltage reference then you'd just buy a good meter in the first place. You wouldn't buy a crappy meter then upgrade it.
The cost of getting a better meter is way, way more expensive than adding an enhanced reference module, especially when you already got the meter. At least you are not paying for a new main PCB, main ADC and main processor.
How much is the base model?
It doesn't matter if the voltage reference module is cheap. A big cost is the ADC. If you're going to overspecify the ADC just in case somebody adds a better voltage reference then the base model will be expensive.
Plus... the extra PCBs, connectors and fancy casing needed to make it modular will cost money. I don't really see how it can compete on price with an equivalent meter that uses a single PCB and has a simpler case design.
Plus ... if it's to be truly flexible it will also need a dot-matrix screen instead of 7-segment LCD. That needs an extra driver chip and beefier CPU, neither of which are free. The screen will maybe need a backlight, which eats batteries.
So: Nice idea but the price will make it unworkable in practice. People will still buy the meter they think they need at that moment.
SATA is not reliable. Servers use PCB mount backplane SATA connectors instead of floppy SATA cables, and HDDs are screwed and/or locked.
The smaller the connector is, the more reliable it is. Try DF12 connectors -- 0.5mm pitch, just enough to be hand soldered, and very tough, almost impossible to fail even in drop tests.
Or, the traditional 2.54mm headers, cheap and verified by the industry for decades.
When considering mechanical integrity, think of connectors (or any flexible but not elastic materials) inductance, and mass of your daughter card capacitance.
You want to minimize inductance and maximize capacitance for smallest vibration amplitude, and make sure they don't resonant with your expected vibration.
Hence, short and rigid connection (mezzanine rectangular connector) suits tiny boards the best.
What I used is exactly those backplane SATA connectors, and my personal experiences with those concludes that they are tight fits. And since the modules themselves are very light there won't be much inertia to allow relative vibration to happen in the first place.
If you're the sort of person who'd add a better voltage reference then you'd just buy a good meter in the first place. You wouldn't buy a crappy meter then upgrade it.
The cost of getting a better meter is way, way more expensive than adding an enhanced reference module, especially when you already got the meter. At least you are not paying for a new main PCB, main ADC and main processor.
How much is the base model?
It doesn't matter if the voltage reference module is cheap. A big cost is the ADC. If you're going to overspecify the ADC just in case somebody adds a better voltage reference then the base model will be expensive.
Plus... the extra PCBs, connectors and fancy casing needed to make it modular will cost money. I don't really see how it can compete on price with an equivalent meter that uses a single PCB and has a simpler case design.
Plus ... if it's to be truly flexible it will also need a dot-matrix screen instead of 7-segment LCD. That needs an extra driver chip and beefier CPU, neither of which are free. The screen will maybe need a backlight, which eats batteries.
So: Nice idea but the price will make it unworkable in practice. People will still buy the meter they think they need at that moment.
So yeah you have all the money to get your Keysight 3458A's and Fluke 87-V's. Even a used Keysight 34401A (cheaper than Fluke 87-V brand new) costs an arm and a leg here. This is becoming a "lesser of two evil" situation - this modular DMM is expensive, but it is cheaper than a second hand 34401A in the money and prices here.
And I don't need to pay the full price again if I do want to upgrade it one day - just bypass whatever is already on the base board with a module that implements the improved features. Everything on the board, down to the MCU and ADC themselves, can be bypassed should I want to. In fact, the MCU is programmed to allow being bypassed, and will happily initialise the add-on modules and shut down if so required.
You may have missed in one of my other threads that paints the picture of market here pretty well: 10ppm tempco resistors which probably are the norm in your country are not usually stocked by sellers here, and they are usually several-buck-a-pop order-start-with-quantity-of-100 expensive.
http://www.ebay.com/itm/Agilent-34401A-Multimeter-Broken-Real-Cover-/181411806535
Here you go, working Agilent (not old HP) 34401A for 2000 RMB, ships from Malaysia.
I am talking budget of CNY1500 or less here. By the way, CNY2000 is about half of the average salary of a college EE or CS graduate here in Shanghai if you are "clean slate". I can negotiate CNY16000-25000 just because I have some extra years of work experience.
So yeah you have all the money to get your Keysight 3458A's and Fluke 87-V's.
Not me. My best meter cost me about $100.
(I could probably get a Fluke 87V if I really saved up a few months but I simply don't need one for what I do).
I can negotiate CNY16000-25000 just because I have some extra years of work experience.
WOW, that's a lot. If you make 16k per month then why not get a new 34461A?
Cuz I need to take classes (not out of school yet) and the 16k-25k are the average I get on job offers that is coming my way. (yes I have a big backlog of rejected job offers)
What about categorize modules into 2 classes, class A is these ones you want to be there all the times, such as voltage ref, ADC and others, and class B is those optional, such as true RMS conversion, memory modules or input multiplexers.
You can implement half vias on class A modules, so you can develop them separately, while not paying the cost of reduced reliability.
For class B, use some easy-to-work-with connectors and make them user swappable.
The main ADC chip is MCP3911 which have two channels and a built-in reference. ADR03B are about an order of magnitude better in both initial accuracy and tempco so if the ADR03B module is attached the internal reference is bypassed. Similarly, if I have a better ADC the MCP3911 itself would too be bypassed. Only one of the two channels of MCP3911 is attached to a front end, the second channel is unused but broken out to allow expansion into a dual channel meter with a second identical front end module as the one included on the board.
On the digital side the main MCU is always there, and if an enhanced processor module is attached the main MCU will become a coprocessor that is controlled over I2C, allowing the processor on the module to take over after booting.
Have a look at this page:
http://www.amplifier.cd/Test_Equipment/other/8506a.html
It shows the Fluke 8506A
The base configuration for this meter was DC volts.
You could add: Resistance, Current and AC TRMS or averaging with plug in modules.
Regards,
Jay_Diddy_B
My concept is similar to this but thanks to SMT components it may be within the limits of "portable" gear. Also I have USB and Wi-Fi capability instead of GPIB for easier connectivity in a hobbyist environment.
Instead of looking at it as a modular DMM, what if you look at it as modular ATE. Your proof-of-concept is a chassis for a modular (semi) portable DMM.
Looked at this way, its a new entry into a relatively crowded market, which doesn't sound that great, but its better than being a new entry into a obsolete market
Instead of looking at it as a modular DMM, what if you look at it as modular ATE. Your proof-of-concept is a chassis for a modular (semi) portable DMM.
Looked at this way, its a new entry into a relatively crowded market, which doesn't sound that great, but its better than being a new entry into a obsolete market
I do have the option of allowing the meter to talk with other devices through USB, Ethernet, Bluetooth Low Energy or Wi-Fi, so the market section can be pretty interesting - IoT-friendly ATE.
Also And the just-released Raspberry Pi Zero module is interesting as the main MCU - I can get a high resolution touch interface for UI, the isolated analog section still works, Ethernet, BLE and Wi-Fi can work all at the same time, and the cheapest possible USB MCU can be used to bridge the Pi to the computer over USB, and the additional grunt (as well as the ability for the GPU to help some math) makes its possibilities very interesting.
Also And the just-released Raspberry Pi Zero module is interesting as the main MCU - I can get a high resolution touch interface for UI, the isolated analog section still works, Ethernet, BLE and Wi-Fi can work all at the same time, and the cheapest possible USB MCU can be used to bridge the Pi to the computer over USB, and the additional grunt (as well as the ability for the GPU to help some math) makes its possibilities very interesting.
...so long as you don't mind changing the batteries every ten hours or so.
Also And the just-released Raspberry Pi Zero module is interesting as the main MCU - I can get a high resolution touch interface for UI, the isolated analog section still works, Ethernet, BLE and Wi-Fi can work all at the same time, and the cheapest possible USB MCU can be used to bridge the Pi to the computer over USB, and the additional grunt (as well as the ability for the GPU to help some math) makes its possibilities very interesting.
...so long as you don't mind changing the batteries every ten hours or so.
It runs on 4 rechargeable 18650 Li-ion cells. I can get a battery life of a few days.
