Product & Circuit Design Brainstorming

[EEVblog]

Dave and David2 brainstorm new uCurrent and other product design options.
Autoranging systems, low current measurement design, noise, overload, USB isolation, ADC's, Gold vs Platinum, marketing, market direction, and a host of other stuff in this often tangential session.
And what common component name did Dave pronounce embarrassingly wrong for his entire childhood?
A video from the archives.

[f4eru]

from the "16 bit isn't enough" : nope, I don't think so.
You can use two converter inputs, and a take off before your final amplification, for example x10 and x1000
Effectively an after the fact software autoranging.

[f4eru]

Concerning current measuring dataloggers, the EFM8SB1-SLSTK2010A has one integrated, it can show the result in a log graph on the PC.
Part schematic is available in the toolchain, don't know if the current measurement is included.

Concerning the leaky relays, a solution is to use reed relays, to pot the result, and to stagger the matrix, so only to have a few in paralell...

[EEVblog]

from the "16 bit isn't enough" : nope, I don't think so.
You can use two converter inputs, and a take off before your final amplification, for example x10 and x1000
Effectively an after the fact software autoranging.

If you ignore burden voltage and noise floor, sure.

Assume 1A full scale and a 100mohm shunt, so 100mV burden voltage, acceptable.
10nA resolution = 1nV per bit. Good luck with that.

[FrankBuss]

Interesting discussion. A ucurrent with a digital output, maybe even with USB or one of these fancy cheap WiFi modules, would be cool.

BTW, have you seen the problem I had with the nA range? https://www.eevblog.com/forum/testgear/problem-with-ucurrent-gold/ Of course was no problem with the ucurrent, but just me being an analog noob, and maybe I missed a warning or information about this problem in the product description.

[TheRevva]

I'm not INTENTIONALLY being an a$$hole here but...
This video is something I can use as an example explaining the need of at least some rudimentary agenda from the outset.
90 mins of time discussing stuff, and the conclusions at the end are...?

[tpowell1830]

I like this discussion type video, very interesting to see the thought processes in action. But, I also want to hear the discussion that you had before this that involved the toilet seat that you erased. 

[bitwelder]

A video from the archives.

Can you tell how long ago have those brainstorming sessions been shot?

BTW they are very interesting.

I wonder if it would be feasible to make some 'interactive YT brainstorming' episodes:
- take a commercial product that could use some improvement
- publish plenty of photos and/or short intro clip of said product
- announce that on a certain date you are going to discuss it live
- go live and unchain the engineers 

[EEVblog]

I'm not INTENTIONALLY being an a$$hole here but...
This video is something I can use as an example explaining the need of at least some rudimentary agenda from the outset.
90 mins of time discussing stuff, and the conclusions at the end are...?

It was never really meant for release, it was kinda like just a live thing that we recorded a while ago. It is what it is.

[bktemp]

The video was both interesting and disappointing to me, because it looks like you tried to solve a problem that does not exist: You need to have a fast high dynamic range mesurement (>100kS/s, >1000000:1 dynamic range) for measuring both standby and operating current for calculating the total energy consumption.
Why not reduce the dynamic range range first before feeding it into the adc? If you add a capacitor across the input, this will low pass filter the current signal and avarage the low standby current and high operating current burst. If the burst is short enough, the peak signal is low enough to stay in the low current range, or if it is long, the slew rate will be rather slow, reducing the error due to a range switching deadtime. That is the easiest and cheapest way for measuring the power consumption of low power devices with high peak operating current.
If you want to record the current waveform, that is a completely different story, making the device much more complex.

Is there an analogue way of compressing the dynamic range first? I have seen a diy power supply using a dual power diode as a current shunt for driving an analogue meter: It wasn't accurate, but you could see if the current was in the single digit mA range or >1A. The other diode in the package was used for temperature compensation.
As you said in the video, 10% accuracy is good enough. If you could make a cheap device capable of measuring current over a large range without the problems associated with traditional range switching meters, it would be a nice and useful product.
The main problem is probably offset errors at the lower end of the range, because of thermal effects after high current bursts.

[jonovid]

for your ranging  did you think to try a binary switch Pushwheel or Thumbwheel Switches or dip switchs ? Daves good video . but a picture in picture for close-ups may had an improvement.
Tip- with more then one character on set you may need 3rd person off-set to help with script & props. also setting up the close-ups as you need them.  like what was the website you both was looking at? close-ups of your uCurrent™ ?  the circuit on the whiteboard , creative direction Fail.  little more professionalism goes a long way Dave & Dave.

[max_torque]

I agree there are two disparate aims here:

1) Accurately MEASURE the average power consumption of a load:  Here you want an accurate integrated current measurement, with enough dynamic range to capture all sorts of events.  A precision current source with a large capacitor parallel to the load will achieve this.  (Of course the system will need to auto cal to negate the capacitors self discharge if you want a really high precision at low power consumptions)

2) Accurately PROFILE the instantaneous power consumption of a load:  Here you either have to settle for a realtime analogue solution with a few fixed ranges (outputting to a scope etc) OR you have to move to a system that includes digitisation and non-real time recording to get a high bandwidth sample of the DUTs instantaneous power consumption.  That Digital system them dumps it's memory to a suitable bit of software for visualisation and further processing (RMS/Harmonic/Spectrum calcs etc)


These two approaches are necessarily very different, because the requirements in each case are very different, although if you develop the digital solution, then a software averaging process becomes viable to generate, er, average, power consumption data

[SydB]

Deciding the spec, as mentioned, is key here. Otherwise you end up designing a scope!

Absolutely agree with previous post of the two different (and useful) uses here - signal profile and power measurement.

I think the simplest single step forward for the ucurrrent is to have a fast low value shunt switch in very quickly when the current goes above range. This stops the micro from crashing when high current pulses are drawn and means you don't have to use a big supply cap downstream of the ucurrent and so you get to see more detail. While out of range, you don't need to measure, you just need the application circuit to keep working. Then ucurrent needs to then switch the low value shunt out of the way again quickly when the high current falls. The high current periods are effectively just 'blocks' of out of range data, which is fine, because you can measure them another time using a higher range. The important thing is that you have not throttled the current when the circuit needs it so the micro does not crash due to low volts, and you get to see the scope signal before and after the high current event.

[f4eru]

If you ignore burden voltage and noise floor, sure.

Assume 1A full scale and a 100mohm shunt, so 100mV burden voltage, acceptable.
10nA resolution = 1nV per bit. Good luck with that.

I don't see what a bigger resolution on your ADC brings to the table if the limitation is the offset voltage of the OP amp anyway. So 16 bit is plenty, and you absolutely need an autoranging to avoid getting stuck with your offset voltage, which is 1/25000 of your measurement anyway.

Now, there's a great way to do a soft autoranging with essentially a null recovery time : try this :


Calculating the measurement result is super easy : just add the two output values. That's it. No complicated processing involved.
Eventually discard the high range measurement 1-2 LSBs when the low range output is lower than 400mV

Now the same with 3 ranges :

Exercice for the reader : translate this to shunt style measurement. Yes it's possible, but you have to take a variable shunt in the OP loop

[twam]

Why not 2 outputs with different ranges, like Keysight's N2820A?

[FrankBuss]

Why not 2 outputs with different ranges, like Keysight's N2820A?

They have some demonstration videos, too, where they use this probe. Doesn't look too complicated: one fixed shunt, 20 mOhm, 100 mOhm and user-defined, and then two voltage outputs, one with a gain of 1.97 (strange factor) and a bandwith of 3 MHz, and one with a gain of 300 and a bandwidth of 500 kHz. Only 1% tolerance, but still very useful for many tasks like measuring how much power your microcontroller circuit needs. You can get it for a bargain of EUR 3,741.00. I would probably buy it, if someone sells such a thing for less than EUR 100.

[stefanL]

It was very interesting to follow this discussion, since I've recently worked on solving essentially the same problem. I designed the analog front-end for a high-dynamic range current/power logger as a semester project at my university. We are planning to release the design data in the next couple of weeks, so it might also be interesting for you to look at our design.

We wanted to design a portable device, that you can use to log the power consumption of a low power / energy harvesting device. We built a analog front-end cape (shield) for the BeagleBone to sample 2 current and 4 voltage channels at the same time. The BeagleBone then stores the data on a MicroSD card and allows the control of the measurement via a webserver. The maximum current it can measure is 500 mA and the RMS noise floor is about 1.5 nA (172dB SNR @1ksps). Our sample rates / bandwith are somewhat lower than what you're talking about: 1ksps nominal, 64 ksps maximum. On the other hand, we support longer-term measurements (hours, days).

We achieved this high dynamic range, by having two different ranges and automatically switching between the two. We have a high range, which covers 2-500mA and is a regular shunt current meter with a 50mOhm shunt, and a high-bandwidth instrumentation amplifier. The low range is a feedback current meter with a feefback resistor of ~1kOhm. We can use the high range for range switching, since it is always valid, using a window comparator. Then we just use two MosFETs two control the path of the current (the gate-source leakage was no problem with our MosFETs, but the gate charge was). We always sample both channels and then re-assemble the current waveform later.



There were some struggles with accuracy, drift & stability, but in the end, the result was very good, and we manufactured a small batch of these loggers (but the design is not really optimized for sale).

We have some additional ressources, if you're intersted in more details (or you can also contact me):

• DATE 2017 Publication
• Project Homepage
• Design data

Edit: Updated links

[FrankBuss]

We achieved this high dynamic range, by having two different ranges and automatically switching between the two. We have a high range, which covers 2-500mA and is a regular shunt current meter with a 50mOhm shunt, and a high-bandwidth instrumentation amplifier. The low range is a feedback current meter with a feefback resistor of ~1kOhm. We can use the high range for range switching, since it is always valid, using a window comparator.

Nice project. The feedback ammeter looks interesting. There is a NI low-current measurement device (PXI-4022), which uses this concept and has a burden voltage of 20 uV for the 100 nA range (so that's 200 ohm instead of 1 kOhm) and with a resolution of 0.5 pA or better, in combination with a good voltmeter:

http://www.ni.com/tutorial/5448/en/

[stefanL]

We achieved this high dynamic range, by having two different ranges and automatically switching between the two. We have a high range, which covers 2-500mA and is a regular shunt current meter with a 50mOhm shunt, and a high-bandwidth instrumentation amplifier. The low range is a feedback current meter with a feefback resistor of ~1kOhm. We can use the high range for range switching, since it is always valid, using a window comparator.

Nice project. The feedback ammeter looks interesting. There is a NI low-current measurement device (PXI-4022), which uses this concept and has a burden voltage of 20 uV for the 100 nA range (so that's 200 ohm instead of 1 kOhm) and with a resolution of 0.5 pA or better, in combination with a good voltmeter:

http://www.ni.com/tutorial/5448/en/

Thanks . The effect of the feedback resistor is scaled with the gain of the amplifier. Therefore the burden vultage would usually be determined by the op amp (with our ADA4522 typically 0.7 µV). At a very low current that's not an issue, but it's really handy at 2mA.

[thm_w]

For some reference, the R&S NGMO power supply uses 3 ranges (7A/0.5A/5mA) and has a specified resolution of 100nA (but will display <10nA). It is designed for measuring sleep current and high transmit current of a cell phone.
ADC used is AD7722 (http://www.digikey.ca/products/en?lang=en&site=ca&KeyWords=AD7722AS), 195ksps and 16-bit.
An a few DACs, possibly for automatic gain adjustment.

Thats not quite as good as Dave is envisioning, but it is an old design.

Why not 2 outputs with different ranges, like Keysight's N2820A?

Even better example, problem is how to get one to reverse engineer.

[f4eru]

no need to reverse engineer one, just design one. But this one is obviously simply a cdascaded amp, both outputs having a diff amp for allowing floating. Not very difficult to design. But the performance is not so good also.

[f4eru]

Hello StefanL,
Question : why not replacing the mosfet M2 by a simple protection resistor ? ( 100 ohm - 1kohm range)

My proposal up there for having a "soft autoranging" with a shunt ammeter would be for example to put a mos in linear mode as a protection as you did. But soft-switching, so regulating the max burden voltage over the transimpedance current measurement in case of overload.

Obviously, all these feedback ammeters have a drawback : speed in extremely fast overload condition. Protection diodes will help. How much do your protection diodes and mosfets slow down the small signal measurement by added capacity ?

[stefanL]

Hello StefanL,
Question : why not replacing the mosfet M2 by a simple protection resistor ? ( 100 ohm - 1kohm range)

My proposal up there for having a "soft autoranging" with a shunt ammeter would be for example to put a mos in linear mode as a protection as you did. But soft-switching, so regulating the max burden voltage over the transimpedance current measurement in case of overload.

Obviously, all these feedback ammeters have a drawback : speed in extremely fast overload condition. Protection diodes will help.

Tbh, I didn't think of that. It should also keep the feedback ammeter output voltage in check. It would, however, add to the burden voltage.

Sounds like a good idea, but I don't know how you would implement the details. IIRC Agilent is doing something similar in their SMUs ("seamless autoranging").

Yes, that is a drawback of the feedback topology.

Ediit: I didn't measure/calculate the impact , since the design has a rather low BW (10 kHz). It might be relevant for a faster design.

[KNSSoftware]

Nice to see Dave2 again.  Dave said he was potentially looking/in discussions, to bring on a new co-host... Warning us up, for a big reveal? I would say his odds are EEVen.

[Stephan_T]

Hi Dave,
you may take a look at this Video of the "Keysight Oscilloscopes" Youtube channel: