EEVBlog #72....

[MTron]

hey, what happened to number 72....where'd it go?

I had only 20 min left, it was an awesome video, even if it was 50 some odd min long!!...very educational!

[EEVblog]

Sorry about that!
Stupid Youtube privacy setting thing again. No one was supposed to see the video until it finished processing (still going since last night!), but Youtube automatically made it public when I was asleep last night and I didn't catch it, so I reverted it to private mode this morning.
It'll be back up soon.
I'm not sure how anyone could watch it in that horrible pre-processed quality!

This "bug" in youtube is REALLY starting to piss me off!

Dave.

[EEVblog]

It's now back on line. Still processing HD, but it has finished medium quality processing, so is now good enough to at least watch.

Can't believe it turned out to be 51 minutes! Smashed the previous blog record 
I thought about trying to severely edit it down, but figured that was kinda pointless, as what advantage would say a hacked 40 minute version bring over the full 51 minutes? It is what it is.

No doubt many will complain about the length though!

Dave.

[MTron]

please! when i saw the length i was like....this is going to be 51 min of awesome...lol

For a beginner like me, this kind of video is exactly the kind of thing we need to get that conceptual framework of how to approach electronics design...step by step, very concise (and yea, i think in this case, 51 min is concise...at least compared to some of the uni lectures i have to sit through!) .

I look forward to the last 20 min!

[EEwannabe]

I was wondering what happened to the quality of the video, I still enjoyed it though. Looks much better now. : )

- Patrick

[Simon]

Well I've just watched the whole blog (have tree fellers next door my dad has insisted i keep and eye on  )

Well great work Dave I really enjoyed that and as I've done a similar project myself I found it very interesting. In my case I'm using a similar circuit as part of a larger project to control and regulate a dynamo. Because I could not put a sense resistor in the ground path (as it is the whole car body) I have to make a highside sense circuit, until i saw your blog about chopper amps I made the mistake of using a TL082 with a horrible 15 mv offset and got very poor results. I since discovered the MAX4172 chip which while not the cheapest (but no more exspensive than a chopper op amp at £ 1.74-2.95) provided a pretty accurate and easy solution and meant that apart from the sense resistor I only need a single resistor to determine the range (apart from choosing the sense resistor). I can see why you used the opamp versus the maximum chip because you needed a negative output as well.

Accuracy asside it is quite feasable to use a 1 milli ohm resitor for high currents and low power loss / burden voltage, I did not use a 4 pin resistor but got around the issue of the relative resistance of the rest of the board by taking the "taps" off the resistior (SMD) from the inside so that none of the carried current could be passing through the measurement path. This worked very well, I have not been able to accuratly test my values but from what i can tell from a rough test I have got a 100 mV/A measurement which is perfect for my application that does not require huge precision due to it being a current limit rather than a measurement and as the result is fed to a pic it is fine tunable in software.

I can appreciate all your considerations on parts I had to go throught the same process and your solution for assembling it all on the board and using it as a front panel is ingenius. In my case i was supplied with an original regulator box and told it had to all go in that with no holes (for ventilatilon) so my parts choice is orientated more around heat diddipation than precision parts.

Oh by the way i work with a guy that used towork in the C&K switch test lab, to think we actually used to make things in the UK !

[Ferroto]

It's now back on line. Still processing HD, but it has finished medium quality processing, so is now good enough to at least watch.

If youtube needs to convert it into some obscure format then why not release a free utility to do the conversion on your own PC. It would relieve alot of the load from the youtube server and you can instantly have HD content. It'd be a win for youtube (less server processing power required), it'd be a win for the users (more HD content) it'd be a win for bloggers (faster processing time).

It'd be the smartest thing Google has done with youtube since they acquired it 

[Simon]

well as far as google is concerned it's all down to how much they bother to control utube, they may just be sitting there taking in the cash and not worrying about how it is run

[EEVblog]

It's now back on line. Still processing HD, but it has finished medium quality processing, so is now good enough to at least watch.

If youtube needs to convert it into some obscure format then why not release a free utility to do the conversion on your own PC. It would relieve alot of the load from the youtube server and you can instantly have HD content. It'd be a win for youtube (less server processing power required), it'd be a win for the users (more HD content) it'd be a win for bloggers (faster processing time).
[/quote]

That won't really help, transcoding still needs to be done regardless of the source format you give Youtube, as not only do they store different versions for playback, but they also check every file using advanced algorithms to find stuff like copy protected and other content that violates the user agreement.
I'm already giving them the exact format they ask for.

Doesn't really matter to me how long it takes to transcode and process, but when I set the Privacy bit to "Private" I expect it stay private until I say otherwise!

Dave.

[desolatordan]

This topic is the sort of thing I really wish universities would cover. I love my school, and wouldn't trade it for any other, but at times I feel that being able to understand/calculate the  concentration gradient in a pn junction, or the electric field components in a rectangular waveguide is mostly useless. I believe the problems stems from the fact that professors are not industry experts most of the time, they often come from research labs, or even straight out of school, having completed cutting-edge research. But they have no experience designing a product for a customer, be it a business or individual. At the same time, with so many people receiving degrees, companies are wanting students to already have the more relevant design experience in school. If you want any real design experience in school, you either have to go out of your way to join student projects that do that sort of thing, spend money on your own projects, or wait until a senior design project sort of thing comes up. Proper soldering techniques and how to rapid prototype are never taught.

Thanks Dave.

/rant

[Simon]

when i did an A level course in electronics in Italy that taught nothing the state exam consisted of calculating the gain of a differential op amp configuration and the output given the input. Even with being able to use the electronics manual which covered everything from the basics of electrical current to anything but microcontrolers they could not even find the right page in the book and find the two simple sums to do (calling a gain calculation an equation is almost an insult to real maths). naturally being Italy 3 of us took a row at a time and handed out the solutions whilst our "professors" kept the external examiner board president busy in the school bar (who I'm sure knew what was going on)

[Ferroto]

when i did an A level course in electronics in Italy that taught nothing the state exam consisted of calculating the gain of a differential op amp configuration and the output given the input. Even with being able to use the electronics manual which covered everything from the basics of electrical current to anything but microcontrolers they could not even find the right page in the book and find the two simple sums to do (calling a gain calculation an equation is almost an insult to real maths). naturally being Italy 3 of us took a row at a time and handed out the solutions whilst our "professors" kept the external examiner board president busy in the school bar (who I'm sure knew what was going on)

Yea I'm taking a networking course and made damn sure that the instructor wasn't one of those "Higher then thoust" type of IT people (you know who I'm talking about)

He also maintains the school network, which thus far during my upgrading hasn't crashed yet, which is more then I can say of my former workplace.

[r]

HELP love the blog but eevblog 72 video not on itunes 

[TheDirty]

Ya, how long is the delay between posting it on YouTube and it being available on your RSS feed and iTunes?

Personally I watch on Miro using the RSS feed.

[tomvleeuwen]

Hi Dave,

I was just wondering, have you concerned some active current measurement?
I think there must be some way to actively control the voltage over the shunt resistor to 0 and measure the injected current.
Of course, you meet your own required specs also using this passive current shunt, but I was just wondering if you have concerned it and figured out what other challenges it would bring.

edit: I found one here: http://zone.ni.com/devzone/cda/tut/p/id/5448

Also, is your way of making a dual power supply not just dissipating the equivalent amount of power of the rail that is not in use, effectively doubling the power consumption compared to the switching one?

Thanks for this nice video!

Tom

PS: We already had the scientific calculator watch, and now the uCurrent, what's next?  

[migsantiago]

Loyal viewers will even view a 360p video if the Master Dave is on it hhehehe 

I watched it yesterday on 360p and it was not that bad. I also managed to survive the 51 minutes length... not even 1 minute wasted!

[desolatordan]

Hi Dave,

I was just wondering, have you concerned some active current measurement?
I think there must be some way to actively control the voltage over the shunt resistor to 0 and measure the injected current.
Of course, you meet your own required specs also using this passive current shunt, but I was just wondering if you have concerned it and figured out what other challenges it would bring.

edit: I found one here: http://zone.ni.com/devzone/cda/tut/p/id/5448

Also, is your way of making a dual power supply not just dissipating the equivalent amount of power of the rail that is not in use, effectively doubling the power consumption compared to the switching one?

Thanks for this nice video!

Tom

PS: We already had the scientific calculator watch, and now the uCurrent, what's next?  

http://en.wikipedia.org/wiki/Transimpedance_amplifier

Then I believe your burden voltage is simply the offset voltage of the opamp. And the opamp has to be big enough to sink the measured current across the Rf resistor, which is why it's not used for measuring larger currents. Battery life would be less.

[Zad]

Excellent as always Dave. In fact, dare I say even better than usual, because it gave a good insight into a typical engineering project. I agree with the comments above, colleges and Universities tend not to teach engineering, but what is essentially science. I was at uni with several people who were utterly brilliant and flew through the exams with Firsts, but none of them had a clue in the lab. I suspect many of them were "fast tracked" and are now managers.

I often describe engineering as the science and art of compromise. Many times it can be a trade off of price, noise, current, reliability etc, but it often boils down to: On time, On spec, On budget. Choose 2. What REALLY helps though, is to know as much as possible about what is available on the market. Up to the mid 90s or so, people like Farnell, RS (and I guess Mouser and Digikey) used to hand out free, printed, catalogues. When CD-ROMs and the web came along though, they became much harder for Joe Public to get hold of. Nowadays they are many thousands of pages and are only available to purchasers who spend a good amount with them. Even though the full database is available online from these companies, it is just so much more like hard work to 'browse' them (often browsing simply isn't possible anyway - a round of applause for Farnell though who have a virtual paper catalogue available!)

[EEVblog]

I was just wondering, have you concerned some active current measurement?
I think there must be some way to actively control the voltage over the shunt resistor to 0 and measure the injected current.
Of course, you meet your own required specs also using this passive current shunt, but I was just wondering if you have concerned it and figured out what other challenges it would bring.

edit: I found one here: http://zone.ni.com/devzone/cda/tut/p/id/5448

I only gave the feedback method a fleeting thought. There was just no need for it, as the 100 fold reduction in shunt resistor method already provides seriously low voltage drops, which is more than required for all but the most critically demanding applications. The feedback method also has upper current limit considerations that didn't suit what I wanted. And mixing different methods for the different ranges just didn't make sense.

Also, is your way of making a dual power supply not just dissipating the equivalent amount of power of the rail that is not in use, effectively doubling the power consumption compared to the switching one?

Not really. The 200K resistance used takes only 15uA + 130uA for the opamp. Low enough for this app.
A 7660 switched cap inverter is going to take a similar amount of active current, and it adds noise and extra cost.

PS: We already had the scientific calculator watch, and now the uCurrent, what's next?  

I already have the uCalc & the uSupply!


Dave.

[EEVblog]

I watched it yesterday on 360p and it was not that bad. I also managed to survive the 51 minutes length... not even 1 minute wasted!

Yeah, the final processed 360p version is just fine, and that's what most people watch I think. But when you upload a video they produce a really horrible version that must be like 100p or something, until it's finished doing all it's processing.
Why they would allow such ugly crap to become publicly visible is beyond me.

Dave.

[EEVblog]

Ya, how long is the delay between posting it on YouTube and it being available on your RSS feed and iTunes?

I don't use iTunes myself, but Wordpress has a setting called "Time to Live" for iTunes which can't be set to less than 24hrs. So I'm not sure if it takes 24hrs minimum or not?
On top of that iTunes has to ping my site and get the update, I don't know how often it does that. I can manually ping it if I want (just done now).

For those curious, my usual procedure when finished rendering the final video is:
1) Upload the HD version to YouTube (takes about an hour + at least several hours processing time once uploaded). It might even sit there overnight until I make it live.
2) Upload the podcast video to eevblog.com/video (takes 10 minutes or so)
3) Make the youtube video "live" and then create a new blog entry on the eevblog.com site and embed the youtube video.
4) iTunes et.al eventually pings my site and grabs the podcast. Other RSS readers should get it much quicker than iTunes.

So those who are absolutely desperate to get the latest video first will find it in eevblog.com/video
And it's not uncommon for the Youtube version to be up some time before it gets embedded in the blog sit, depends on how busy I am at the time.

Dave.

[tomvleeuwen]

Also, is your way of making a dual power supply not just dissipating the equivalent amount of power of the rail that is not in use, effectively doubling the power consumption compared to the switching one?

Not really. The 200K resistance used takes only 15uA + 130uA for the opamp. Low enough for this app.
A 7660 switched cap inverter is going to take a similar amount of active current, and it adds noise and extra cost.

I was more concerned about the current through the measurement feedback resistors. This current is 'mirrored' by your dual supply circuit to make sure ground stays in the middle. But considering the high input impedance of multimeter and 100K feedback,  the asymmetry will only be 2uA full scale, much less than the opamp supply. Could have done this calculation before i posted...

Tom

[jimmc]

Great idea and well explained. I think it is a well thought out design, but can I put my pedants hat on on raise a few points.

With 0.1% resistors in the feedback loop the maximum gain error is 0.2% (0.198% to be even more pedantic )
add this to the 0.1% tolerance of the nA shunt and the accuracy becomes 0.3% worst case.
For the uA range leaving the nA shunt (R2) in circuit adds another 0.1% possible error.
So worst case errors are nA 0.3%, uA 0.4% and mA 0.7%.

There is an interesting quirk on the MAX4239 data sheet, the common mode input voltage range (over the full temperature range) is quoted as  GND-0.05v to Vcc-1.4V; so using a mid point ground at the minimum supply voltage of 2.7v means that the common mode range is exceeded (by 50mV).
Possibly an asymmetric split supply (say +1.6v, -1.4v) would be better.

Finally, looking at the layout the positive supply is routed around the +ve input terminal and track to R12.
Since a leakage resistance of 1.5Gohm would generate a 1nA 'input' current, a grounded guard track run between these might be a reasonable precaution.

I do not think that any of these points will cause any practical problems, I'm just in a picky mood.

Jim

edit: I had mistyped the part number, thanks to tomvleeuwen for spotting it.

[hans]

I'm not sure how anyone could watch it in that horrible pre-processed quality!

It was not that bad, though I still would prefer 720p. But as I happen to saw it on there and the discussion point, I couldn't resist to watch it anyway.

Quite an interesting blog. It's too bad most universities don't teach this. Not even at the 'applied-sciences' which I currently am studying (2nd year). We generally say: "okay, we need to process this bit of audio in a band pass filter here, we just take a NE5534 which will be good enough". We never had to look really at how good the opamp would fit in the circuit as in terms of is it overkill (maybe a way too big bandwidth?) or underkill (maybe the GBW is too small for the input bandwidth and the given gain?). We just take it for granted that it will work properly.

I think you've shown that's its a lot of work to find exactly the right parts for the job. Now for us it's also a dilemma of having a very flexible part (i.e. possibly to use in a lot of applications) or order very specific parts for every circuit we would design.

[Simon]

I think most people that "have their heads screwed on" design wise will get to this sort of level on heir own. I was amazed to see how Daves design and part research process was so much like mine, at the end of the day it is common sense but yes schools should be a bit more practically minded not in the sense of just showing how stuff is done but the logistics of design.

I think pics (ok and other uC's) have become popular very fast because they are inexspensive particularly when bought in bulk and can be made to do many things