EEVBlog #87: The Electronics Design Merry-Go-Round

[ModernRonin]

I really liked this one. I haven't personally played the optimization / iteration game, but I've known many people who have.

From a technical perspective, I'm curious if a high-side current monitor based on an opamp is possible in this particular application.

I was looking into circuits to sense mains current a while back, and eventually concluded that I'd probably end up having to use an expensive isolation amplifier if I wanted to do high-side current measurement directly. (I assume that some kind of isolation amp is what's inside those expensive high-side current monitor ICs.) But there's got to be a better way, right?

A couple of ideas occurred to me on how to do it. One, is there some way to very accurately level-shift the high-side sense inputs down to lower voltages? (Resistive voltage dividers? Precision resistors are expensive...) Two, is there a way to power the op-amp from a mains-like voltage so the high-side inputs won't be far off common-mode voltage?

I like the EEVBlog because it make me think about stuff like this.

[rodneykeeling]

Can someone explain a high-side current monitor? I have a good idea as to what it is, as the name is quite descriptive in itself, but I'm not sure how it would be put into use.

[ModernRonin]

You can read the "Applications" section of the data sheet:

http://www.zetex.com/3.0/pdf/ZXCT1009.pdf

Seems to be mostly related to monitoring batteries. My guess is it's mainly about recording how much current they're presently discharging, or monitoring how much current a charger is putting into them.

[Zad]

Much appreciation from me for this Vblog. Much of electronic engineering is this "admin" kinda work. It is engineering because you are optimising your components for the given task. You need to know a lot about a lot in order to speed it up much, and they just don't teach it in Universities and Colleges, yet it is the part of the job that has the potential to save employers the most money (excepting extreme circumstances of course).

In the past I have generally used the resistive divider method of high-side current measurement. Looking at the Farnell site, you can get 0.1% for £0:05p. That sort of accuracy is unnecessary in most systems, so if you go to 1% and you find them at 0.4 of a penny. Of course the resistive divider isn't suitable for all systems, but that's engineering for you. "Time, Spec, Budget. Pick 2."

[hans]

A lot of read stuff for low and high side current sensing:

http://focus.ti.com/analog/docs/microsite.tsp?sectionId=560&tabId=2180&micrositeId=7

A big advantage of high-side current mode is you aren't measuring against ground, but just a voltage drop on a resistor. You can also wire up the load the way you want, the shunt resistor can be placed anywhere as long as it's in series with the current you want to measure.

[TheWelly888]

A lot of read stuff for low and high side current sensing:

http://focus.ti.com/analog/docs/microsite.tsp?sectionId=560&tabId=2180&micrositeId=7

A big advantage of high-side current mode is you aren't measuring against ground, but just a voltage drop on a resistor. You can also wire up the load the way you want, the shunt resistor can be placed anywhere as long as it's in series with the current you want to measure.

Oh dear! The polarity of the voltage developed across the R_shunt is incorrect ( assuming that the current is conventional rather than electron! )

Anyway back to Dave's blog - I would have thought that one useful parameter for microcontroller selection would be minimum ADC conversion time and maximum PWM output frequency?

[alm]

ADC speed can be an issue. The ATmega (and probably also the ATtiny) is really slow, max. 4kS/s or so without overclocking the ADC clock (which will reduce resolution). I believe PIC's are much better in that regard.

An issue I have with parametric search, especially at Mouser, is that often I find that only 20% of the devices I look at have specs beyond type/package/price, so that makes it almost useless. I usually use Digikey to search based on specs, and Mouser if price if my main concern, eg: 'give me the cheapest jellybean part you can find', for example if I need a DAC with at least 8-bit resolution and a 1kS/s sample rate. I also use a Greasemonkey script for Firefox that lets you sort based on price on Digikey (and also filter out results like 'call for price' or 1k minimum order). Digikey is usually much better at normalizing the specs, Mouser might offer you .1A, .10A and 100mA.

Another annoyance is that NXP uses package names different from anybody else, so I have to memorize that SOT-146 is DIP, SOT-163 is SOIC, and so on. Fortunately Mouser does include the standard names.

[TheDirty]

I also use a Greasemonkey script for Firefox that lets you sort based on price on Digikey (and also filter out results like 'call for price' or 1k minimum order).

I was going to mention this.  The Sort by Price script by Alex Leone is very nice.  You need Firefox, the GreaseMonkey Add-On and the script.  All the links are here:
http://www.sparkfun.com/commerce/news.php?id=355

[EEVblog]

Anyway back to Dave's blog - I would have thought that one useful parameter for microcontroller selection would be minimum ADC conversion time and maximum PWM output frequency?

Not for this application. A few Hz ADC sampling would suffice, and I can work around almost any PWM frequency.

Dave.

[TheDirty]

This is another blog that hasn't shown up in the standard RSS feed.  At least it's not showing up in my Miro player.

[RayJones]

High side current monitors come in very useful when you want to monitor 2 currents that share a nominally identical ground.

If sensing on the low (earthy) side, the multiple potential ground references makes for an interesting, if not impossible grounding of the ADC system without one current inducing biases into the the other's measurement.

One constant though, irrespective of high or low side sensing, is the current sense resistor itself.

Low burden voltage is often desirable, so the low milliohm valued shunt usually costs more than the semiconductor device!

[EEVblog]

This is another blog that hasn't shown up in the standard RSS feed.  At least it's not showing up in my Miro player.

Another Wordpress quirk.
I had to delete it and re-do, please try again.

Dave.

[hans]

Now the comments are gone

Oh well, not too bad. What I said there is that a lot of suppliers have bad parametric searches. Last week I looked for a DC-DC convertor. 5V out fine, like 300 left. Ranges that support 12V in also fine, still got like 80 found. Until I went over to the currents, I only counted like 25 converters that will match at all if I would require any current. In other words , only 30% or so of the products have full parameters. If you go to deep you potentially throw away a lot of products that could work, be cheaper and etcetera. Just like alm mentoined too, 100mA can be 0,1A 0,10A or 100.0mA etc.

I find Farnell and RS so bad for searching parts, I might go and try Digikey and Mouser more often, but I'm afraid that if I found my part and go back to find it on RS or Farnell it isn't there. Do you all recognize the same problems 

[alm]

If I recall, the comments had a fairly low signal to noise ratio anyway. Anyway you can somehow consolidate the comments? Having comments on three different places means many users don't get to see them all.

[EEVblog]

Sorry, a bug in Wordpress means I have to delete the post and start a new one to fix the podcast feed issue. That means all comments get lost in the process, no way to preserve them.

I don't know of any way to consolidate Youtube comments, Wordpress comments, and forum comments. If anyone does let me know.
The signal-to-noise ratio on Youtube is much lower than on wordpress!

Dave.

[alm]

I don't know of any way to consolidate Youtube comments, Wordpress comments, and forum comments. If anyone does let me know.

The only solution I can think of is to disable comments and post a link to the forum thread/Wordpress. I don't think there's any better way, short of writing your own script that copies comments (I don't think they're available as RSS feed, but I could be wrong).

People need a Youtube account to post comments on Youtube (I usually watch on Youtube because of the higher resolution, but don't have an account). And people need a forum account to post on the forums. It might lead to more registrations for the forum, although it's debatable whether this is a good thing.

[ModernRonin]

Looking into my mains current measuring problem again, I found this great article at EDN:

http://www.edn.com/article/CA434873.html

And the corresponding PDF:

http://www.edn.com/contents/images/434873f1.pdf

Basically, a zener diode and a large value resistor in series make a reasonable regulator to track the voltage of the mains. You can then tap off power and ground from around the zener.

There's also a couple of transistors involved to level-shift the output of the monitor IC down to ground.

[alm]

Make sure to treat everything as directly connected to live mains voltage, i.e. insulation, clearance, make sure no part can be touched. Since there's no galvanic isolation, there's no protection from transients and other dangerous stuff. There's a reason why they don't make power supplies like that anymore in consumer electronics.

[r]

it bin about 1 day but it has steel not came up in itunes

[tuta]

Hi

Does anybody else have trouble viewing episode 87 on YouTube? I tried it from at least 3 different computers and get a blurry picture every time. I'd love to watch it

Cheers

[ModernRonin]

I watched via and it's looking fine to me. Taking forever to load, as these usually do, but looks fine once it loads up.

[r]

it steel not on itunes 

[EEVblog]

Does anybody else have trouble viewing episode 87 on YouTube? I tried it from at least 3 different computers and get a blurry picture every time. I'd love to watch it

You must have watched it before it finished processing.
Either that or you are expecting a 1280x720 screen capture to be sharp and readable at 640x360 which is the minimum Youtube resolution.
Hit the 720p button bottom right and expand the window if you want to watch it in full resolution.

Dave.

[tuta]

That helped! Thanks a bunch. Why did'nt I try this by myself?:-P

Cheers
Tuta

[saturation]

Other that parametrics, one way to speed up your online work is get the biggest LCD or equivalent monitor you can, or get a video card that supports multiple monitors.  That way you can see and compare multiple sites simultaneously, thus faster, without flipping between pages.