Pax Instruments MultiLogger

[Marco]

High dynamic range, high temporal resolution current monitoring is a relatively specialized (but common) problem and doing it in the "normal" multimeter way requires too many compromises in my opinion.

I'd rather have a device which had two shunts in series, say 1 Ohm and 10K, with a very fast voltage clamp across the 10K resistor to limit burden voltage, low pass filters for the shunts and a couple MAX44252s to amplify the voltages by 100x (x10x10). That way you have two high bandwidth current measurements for two dynamic ranges which you can feed into a two channel scope, or a slower data-logger (with some extra low pass filtering most likely).

This kind of device makes more sense separate from a logger (there is more overlap with the uCurrent in fact).

[chickenHeadKnob]

I love the form factor of your T400! and having the probes come in at the top is very natural on the bench. If it is too tippy  as stmdude worries then perhaps make sure the stand hinge is at the very top of the case and have it open to an obtuse angle. You might have to modify the case but you should still be able to have a single case for both products. I don't particularly need or want another multimeter so I will let others guide you on specific features, however a few general comments come to mind:

• No point making an "also ran" that just copies other multimeter functionality. You already indicate you want to explore full time multi-channel. Great for measuring power,  if the other guys bench meters attempt this at all, they implement  it with relay kludge klackery.
• go for a high resolution ADC, 20 bits plus, because this is a bench machine. Select a delta sigma ADC for high resolution DC/low sample rate and offer two high speed 1Mhz sampling (or faster depending on processor) 12 bit channels connected  to the processor AtoD as stm32 have pretty good on chip converters, better than AVR I think. That way you can provide pocket DSO type inputs as well as exceed the typical multimeter specs. nobody else does exactly that, although to be fair the Keithley DMM7510 can run at high sample rate it also costs 4K$ If you provide fast inputs beware of the sustained through-put requirements for logging to USB storage or PC!
• As to processor selection I like the stm32 product line, however the stm32f103 does not come to mind as the first choice for a battery powered device. Entry level cortex M4F devices like Stm32f401 or '410 are  nearly as cheap and have very good power per mip especially when moderately underclocked at 32 to 20 Mhz. STM32l476 is intended for battery power and has nice power down/ wake-up modes but a cost premium
• color LCDs are becoming very cheap now, even cheaper than the older graphical mono displays, and I prefer them however I did not check on power consumption. Since you already have a display and the software to drive it probably doesn't make sense to upgrade.

While I am not in the market for a multimeter, I did look at your T400 and said "I wish Charley would make a 4 channel wheatstone bridge input version of this, with low voltage AC or pulse excitation and the requisite tables and formulas for resistive  temperature sensors like platinum RTDs and cheap ntc, ptcs"

[bitseeker]

Landscape is good for this especially for displaying recent samples in a scrolling graph, just like the T400. The enclosure might end up too small in the end, but the back could just be replaced with a deeper cabinet. That would provide space for more electronics, connectors, etc. as well as be more stable on the bench.

[EEVblog]

For hardware, I'd like to reuse the T400 enclosure if possible since I already have the tooling. The top panel is interchangeable and can be replaced with a 1.6 mm PCB. There are six buttons, a MicroUSB connector, MicroSD card connector, and a 132x64 pixel LCD.

I'm not sure you'd have enough room to mount the deep shrouded banana jacks required and have traditional protection components and spacing.
Certainly not for multiple channels which is what would differentiate a product in this space.
A low voltage multi channel voltage and current logger with screw/phoenix connectors, sure. A traditional multimeter, likely not.

An old video:

[charlespax]

Resolution of 0.1707mAh will be more than sufficient, but what about the accuracy?

Not sure, I haven't really looked down the coulomb counter road. I'm just thinking of ways for determining power consumption that are cheaper than brute forcing it with high speed current and voltage measurements.

[Marco]

Not sure, I haven't really looked down the coulomb counter road. I'm just thinking of ways for determining power consumption that are cheaper than brute forcing it with high speed current and voltage measurements.

You can low pass the signal across a shunt before sampling. Average value of a low passed signal is the same as the average value of the input.

I'm not sure you'd have enough room to mount the deep shrouded banana jacks required and have traditional protection components and spacing.

Without LowZ you don't need a PTC and you could use an external shunt for high current measurements. That saves a lot of room.

[TiN]

Perhaps one thing to consider is to make units networkable (e.g. have RS485 or BT coming out, so multiple units could talk to each other to combine data).
My use scenario for multiple channels are usually 2-3 channels for voltage monitoring (using bench 7.5-8.5d DMMs) + current monitoring (SMU) + 2-3 channels RTD sensors (other bench DMM and TEC SMU). There is not much you can do  in such a small case, so that's why the idea to make them expandable, with each meter have only one or two channels.
You can even have few versions of them, dedicated to more specific application (e.g. voltage high-resolution version or high-speed version).

As for ADC, TI ADS1262 is pretty capable cheap device. It have integrated current sources, temp sensor, multiple channels, easy to interface, and goes in solderable TSSOP. I had done a review on it's DK some time ago.

For energy meter there are some chips from AD in ADE series that can do power, factor and such measurements easily.

[charlespax]

I'm not sure you'd have enough room to mount the deep shrouded banana jacks required and have traditional protection components and spacing.
Certainly not for multiple channels which is what would differentiate a product in this space.
A low voltage multi channel voltage and current logger with screw/phoenix connectors, sure. A traditional multimeter, likely not.

A shrouded connector fits between the LCD and enclosure with just a few hairs to spare. With any luck, I can find a mating connector that mounts mid PCB. Another custom connector is probably not in the cards. I can say with high confidence that I will not be making any devices that measure mains voltage. Sticking to +/-12 V is fine with me. I'm not really interested in making a multimeter.



I hacked a PCB mount RF connector on one of my T400 devices. A screw on BNC adapter fits perfectly and looks pretty good. That could also be a connectivity option.

[EEVblog]

A shrouded connector fits between the LCD and enclosure with just a few hairs to spare. With any luck, I can find a mating connector that mounts mid PCB. Another custom connector is probably not in the cards. I can say with high confidence that I will not be making any devices that measure mains voltage. Sticking to +/-12 V is fine with me. I'm not really interested in making a multimeter.

In which case using shrouded meter leads is a bit pointless.
I don't think you can make this case into a multimeter, I don't see any market need for yet another (single channel) multimeter that can't be used on high voltage etc.
It should be a data logger, voltage and current, to match your temp logger. Logging usually means hooking up a semi-permanent test cable, so no point in using regular multimeter leads.

[EEVblog]

I hacked a PCB mount RF connector on one of my T400 devices. A screw on BNC adapter fits perfectly and looks pretty good. That could also be a connectivity option.

Bingo, a 4 channel high resolution scope type logger.

[Jeroen3]

Beware that you do not create fancy hardware with horrible software.

[charlespax]

stmdude whipped up a quick ARM PCB. We're working on basic hardware to get some momentum going and evaluate a few ideas. If you'd like to get your hands dirty working on the design, please join the discussion. This EEVblog thread will be for the more general discussion.

[sarepairman2]

I like the idea but choosing a spec to attain is the hard part.

I wanted to do this myself a few times, but I end up thinking.. do I want 2.5 or 4.5 digits on x-y-z range. what kind of voltage range is useful.... noise floor, update rate...

you can design 50 different systems all useful for something.. BUT HOW DO YOU CHOOSE A DESIGN??? 


and then.. how do you choose a design for sale? thats even harder....

this sounds useful for maybe green energy people... I think this would be a broader market then just electronics hobbyists. But then you might also want clip on measurement.

Alot of people are interested in temperature trends. Not so many people are interested in electrical trends. For AC power logging these is alot of options already. Not to mention the dangers of designing this stuff.

It seems that quite a bit of green sources make reasonably low voltage things appropriate for a meter that won't have CAT certifications and stuff. Also it would not make you crazy trying to design for nano ampers and stuff that hard core electronics people are interested in.

why can't it measure high frequency nanovolts???

This is assuming you want to make money. I have yet to design something to sell, only to satisfy personal curiosities.

From a hunch, if you want to sell near the price of the T400, which is 165$, with four channels, you won't be raising the bar in terms of any kind of measurements, making it difficult to appeal  to a volt/time/current nut, as cheap electronics have gotten precise enough that measuring drift is difficult.

A portable unit with its own batteries might be useful for drone power monitoring, green energy, MAYBE some automotive stuff (i don't know cars). Also maybe for monitoring the behavior of machinery like CNC machine/3d printer/etc, if it samples enough .

A bench top current logger without outstanding capability would probobly be relegated to measuring battery drain.. which you only need one channel for. And probobly a uCurrent to hook up to it  (business opportunity for a combo sale with dave jones lol). Plus lots of people are starting to get dummy loads now.

ANd selling something designed to hook into mains without the extensive safety testing just seems like a bad idea.

I have a FLUKE HYDRA DAQ.. I have yet to use it yet, and I have a fairly sophisticated interests... the main reason being that its resolution is not high enough to be useful to me.

Plus, for high speed measurement, there are tons of portable oscilloscopes out on the market. Not to mention regular bench top oscilloscopes are getting more and more channels cheaper, making it less and less appealing to have a separate unit.

What I don't see alot of is long term measurement devices that are light and have long lasting batteries.

I also think that the thing that people are asking for in this thread does not fit in with the marketing model that you have for your T400 (i.e. its useful for many many things, from bbq chicken to making moonshine vs its good for measuring power supply stability and battery drain). A big selling feature of your T400 is the data card.. a data card is not useful in a electronics laboratory when you are prototyping or repairing. It would just slow things down. It would only be useful for some kind of burn-in test but who is doing that? the mentality is "it works".

Using this forum for market research will contaminate your data with volt-nuts like myself

the last thing you want is "hackaday" to post a "hack" where you turn your oscilloscope time base wayyy down and plug a ucurrent into it to make a data logger instead of buying your product.

another interesting use for a data logger would be just general thing to connect transducers to, i.e. portable instrumentation package to put on a balloon/roof/mountain top/something

[charlespax]

A few of us have been hashing out a base design that will be used to experimenting with different measurement devices. The board has an STM32F103Rx microcontroller and all the essential components for interfacing with the T400 enclosure (LCD, buttons, MicroSD card, etc.).

There is a empty 32x66 mm space for soldering in a daughter board. The idea is to make a small batch of main boards, so we don't have to assemble the system core for each prototype. I think this will speed up instrument development.

I'm still laying out the board, but I thought you'd be interested in where things are. I'd love to have your feedback.

[Kean]

This looks great for some things that I had in mind for the extra enclosures I bought.  I was planning to probably use STM32F103 as well.
I don't have time right now to do my own layout as I'm in the middle of serveral other projects, but being able to use this with various daughter boards looks quite promising.
 

[charlespax]

I believe this first prototype revision is done. I'm going to sleep on it and order boards tomorrow. Please check out the code from https://github.com/PaxInstruments/labwiz-board and let me know if you have any feedback. You can post in this EEVblog thread, on the Pax Instruments forum thread or file an issue on Github. The schematic and layout are also attached.

[charlespax]

A lot of what we discussed is not all soldered up. Power is good. I still have to bring up the MCU.

I've going through the markets trying to figure out a good connector solution and I think I have one. In this photo you see a 12x2 half-height header with a 0.05 inch (1.27 mm) pitch. On the female side I can get custom bent legs that straddle a 1.6 mm PCB. The male connector fits nicely on a 1.0 mm PCB, but a 0.8 mm PCB might be better. I'll plug one of the holes, so the connector is keyed. Four of these will fit across the prototyping area with a one pin gap between and a bit on each side.

These two photos illustrate the size of a prototyping module. We could use up to four of these or just one big one. I'm really excited abouthis.

[danadak]

What sorts of things are you measuring during startup?


Basically power supply sequencing in embedded designs.


Regards, Dana.

[ataradov]

On the female side I can get custom bent legs that straddle a 1.6 mm PCB.

Where? If it is not a big secret. Those connectors are expensive as it is, I can't imagine customizations like that would be cheap.

[charlespax]

On the female side I can get custom bent legs that straddle a 1.6 mm PCB.

Where? If it is not a big secret. Those connectors are expensive as it is, I can't imagine customizations like that would be cheap.

Pretty much any factory that makes male/female pin headers can do it. Some things are really easy to do while other are hard. Things like pin length are pretty easy to change because it's a setting on the machine. Other things might be hard to do it the manufacturing process depends on that feature. I spoke with some factory people and they said this modification would not be expensive.

[charlespax]

Also, I'm pretty close to ordering the next batch of boards. Since these' some empty space I've added a few modules for testing. Three of them are for interfacing with the Seeed Studio Grove modules. This reduces the amount of time spent on hardware development and I can focus on writing code.

[Kean]

I'd be interested in getting a couple of the blank pcbs and assisting with debugging.  Please let me know if I can contribute.
I'm comfortable with doing my own assembly and not afraid of a bit of rework.  I probably also have many of the necessary parts in stock already, either from my other designs, or from the enclosure kits and mystery bags I got from you previously.

[charlespax]

I'd be interested in getting a couple of the blank pcbs and assisting with debugging.

That would be great! When I get boards back I will select one and solder down the minimum necessary components to verify the MCU can be programmed and there is no fundamental design error. I think it would be best to send you one in this configuration rather than a blank board. Otherwise, we run the risk of wasted money and time with shipping. How's that sound?

I can make a private item in the store and share a link here.

[Kean]

When I get boards back I will select one and solder down the minimum necessary components to verify the MCU can be programmed and there is no fundamental design error. I think it would be best to send you one in this configuration rather than a blank board. Otherwise, we run the risk of wasted money and time with shipping. How's that sound?

I'm happy with either bare or minimally populated boards.  Maybe just partially populate and quickly verify one for yourself, before shipping.
Even if there is a design error, I can probably work around it.  I've certainly screwed up my share of footprints in the past.

I can make a private item in the store and share a link here.

Good idea.  I'd love to grab two, but others here might also be interested depending on how many you're getting.

[charlespax]

I just ordered some boards from dirtypcbs.com. It's such a relief to get the hardware to this point. @protological has been crushing the firmware and has just about everything working in FreeRTOS.

Here's where we stand:

• STM32F103RGT6 microcontroller 1024 kb flash, 96 kb RAM
• Cortex debug has been working great with ST-Link and Eclipse
• Each module interface has CAN, i2C, USART, SPI, and GPIO
• There is extra space on the PCB, so I added three Seed Studio grove module adapters and one temperature module. The temperature module had an MCP9800 temperature sensor and an MCP3421 ADC for thermocouple input. This basically gives a single channel of what is in the T400. Having the grove module adapters will allow us to test hardware without having to develop and make boards right now.
• USB is configured as a virtual serial port and is working
• LCD is working
• The SD card is giving us a weird hard fault issue
• ESP8266 breakout header. Not tested.