EEVblog #238 - Power Supply Design Part 7.

[FreeThinker]

Hi Dave
Just like to say I've enjoyed the videos so far mainly because of the none scripted approach as it gives an insight into the thought processes, and blind alleys you run into as a project 'matures'. However I think you would have saved yourself a lot of grief if you had explained to your viewers that this was the case ie an Adhoc project and not a step by step Tutorial (I hope I've explained myself clearly  ) also as commented elsewhere a dedicated feedback area would probably of helped too as people assumed that, because it was an open hardware project it was also a community design project, which it was not but they could suggest and discuss with other people and maybe have there ideas picked up, but at least feel involved and to have had there voice heard. Looking forward to the rest of the series keep it up

[FreeThinker]

Dave, I don't really care about the details of the final version of the PSU.

The real value is in being able to follow your design process and evolution of various ideas behind every little part of the PSU.


Now, when the final version comes back from the assembly line, we expect a "Don't turn it on, take it apart video"

Excellently put!! I have no need for such a comprehensive psu but find the design exciting and informative and am sure parts of it will be recycled in lesser projects in time to come.

[quarks]

Hello Dave,
the download link opens the older EEVblog #237 file.
Regards Marcus

PS: Now it works, thanks for the fix.

[firewalker]

Try this link .

http://www.eevblog.com/uploads/uSupplyBenchRevB.pdf

Alexander.

Hello Dave,
the download link opens the older EEVblog #237 file.
Regards Marcus

[EEVblog]

Here lies the difference between a greenhorn and a seasoned designer in any field, including electronics. The seasoned designer is capable of spotting his mistakes in due course. Watching over his shoulder and pointing a finger on mistakes is unlikely to improve things. Sooner or later the experienced designer will spot his mistake and correct it without anyone's help. Trying to correct him prematurely is more likely to annoy and distract. I am not saying that seasoned guys do not make mistakes, they do. However they are capable of spotting and fixing them. Greenhorns like you and me on the other hand can benefit from a more experienced person watching over their shoulder and pointing at the mistakes as they are made. It prevents us from building on flawed ideas, this is one way of learning.

Whilst it's true I'm going to eventually find my mistakes, I don't mind people pointing out genuine mistakes at all.

Dave.

[EEVblog]

Do you think Dave would have figured out the RESET blunder he is now playing down

In what way am I "playing down" the Reset oversight?
I just did a whole video and redesign that used that issue as the basis.

Dave.

[EEVblog]

Just want to give Dave a thumbs on the design.  This is a great looking open source project with lots of potential. I cannot wait to get one in my hands and see what others do with it.

I read a couple comments about an ethernet option. I have one of these and it is a quick way to get ur project on a network. Write a processing app and away one goes...

http://www.wiznet.co.kr/Sub_Modules/en/product/Product_Detail.asp?cate1=5&cate2=42&cate3=0&pid=1161

Yes, these Wiznet modules are nice, and better than the Arduino shield IMO (and cheaper, at least at the original version level).
I think I have one lying around somewhere.
Looks like they have Arduino software support too on the download page.
And a module of this size can maybe just fit into my existing case.

Dave.

[FrankT]

A couple of noob questions and observations...

* As asked a few times, is there any technical reason you chose not to multiplex the SPI  programming lines?  If not used as a shared SPI bus, the encoders could have been put on these lines.  From what you said in the video, you are going to support the serial bootloader, so it will probably not be used much after your development.

* You are using OC0A/OC0B/OC1A for the RGB which means you are using 2 of the 3 timers, leaving 1 8-bit timer.

[ejeffrey]

* As asked a few times, is there any technical reason you chose not to multiplex the SPI  programming lines?  If not used as a shared SPI bus, the encoders could have been put on these lines.  From what you said in the video, you are going to support the serial bootloader, so it will probably not be used much after your development.

You still wouldn't want to mutiplex the encoders with the hardware SPI/ISP lines, as that would interfere with programming -- the encoders provide an output signal and are not easily disabled / tristated during programming.  I don't see any technical reason Dave couldn't have used the hardware SPI for the DAC/ADC and shared those pins with the ISP.  All SPI slaves must leave all pins (including slave data out) high impedance when not selected.  However, I haven't read the data sheet in detail, possibly there is some issue with the programming voltages present.

[EEVblog]

* As asked a few times, is there any technical reason you chose not to multiplex the SPI  programming lines?  If not used as a shared SPI bus, the encoders could have been put on these lines.  From what you said in the video, you are going to support the serial bootloader, so it will probably not be used much after your development.

I left the hardware SPI lines lines free so they could be used by an external interface device, like Ethernet board, or other serial type comms device.

Dave.

[EEVblog]

I found my Wiznet module. It's the older one with the 2mm dual row header.
The (newer?) WIZ820io seems to be less popular, but it's the go, with 0.1" pitch and a much smaller footprint.
Will have to get one.
need to look into the details of exactly what pins are required though.
In either case it's not going to fit on the board (for back panel access), as it fouls one of the mounting screws.
I guess I could always lave that mounting screw out and rely on the 3 others.
Or mount elsewhere internally and have the run the cable inside...

Dave.

[baljemmett]

the download link opens the older EEVblog #237 file.

Argh - you're right - I just tried to watch 238 from my media server (which grabs the videos from RSS) and although the filename's right, the contents are the Makerbot unboxing video.  Time to find out if the new television's "play from YouTube" works, I guess!

[Psi]

* You are using OC0A/OC0B/OC1A for the RGB which means you are using 2 of the 3 timers, leaving 1 8-bit timer.

Anyone can modify the code to drive those pins as general outputs instead of output compare. The leds can easily be PWMed in software using just one timer if needed.
Using the output compare pins just means people can use OCA functionally if they want.
So using those pins gains optional OCA ability without any cost

For example, if someone decided they wanted to do IR communication and replace one of the leds with a IR led then they would be very glad that OCA was available.
Typical 36-44khz IR carrier frequency is much easier in hardware using an output compare.

Its all about future proofing

[reagle]

Re "Reset blunder". Worst case Dave would've cut a few traces and added a few blue wires, in the process figuring out that on that particular micro that was not a good idea. It's not really a requirement that everything comes up perfectly the first time- no learning happens then. When dealing with relatively simple boards and working on one's own schedule, that's very minor. Now, if that was a 12 layer monster and customers were screaming at the door for delivery ... 
Edit:  noticed another thing- ADC VIN seems to be missing a cap for low pass filtering before ADC. All the other signals are filtered

[ColinA]

The (newer?) WIZ820io seems to be less popular, but it's the go, with 0.1" pitch and a much smaller footprint.

Dave.

Yes it's new, I received as a sample. Was unable to find it for purchase back in November and was told it wasn't available for purchase yet.

Works great and easy to use.

[PeteInTexas]

I'm all for gilding the lily but can't help the feeling a shark has been jumped.    A "cascade of decisions" necessitated two more chips for want of a reset pin!?   Come on now.  Is ISP really necessary?  The micro is not tasked for general purpose i/o.  The pins are connected to specific functionality and so it is unlikely that programming the micro off-board will pose a great inconvenience.  Besides, its probably a good idea to put the micro in a socket seeing as its in a power supply.

And what's with capitulating on that protection diode?  There was a video done specifically to show it was not needed.

[Psi]

I'm all for gilding the lily but can't help the feeling a shark has been jumped.    A "cascade of decisions" necessitated two more chips for want of a reset pin!?  

It was more than that, adding the extra IO chips meant the lcd RGB backlight could be used, which will be useful.
It also added spare IO which is very important for people wanted to add extra features.

Come on now.  Is ISP really necessary?  The micro is not tasked for general purpose i/o.  The pins are connected to specific functionality and so it is unlikely that programming the micro off-board will pose a great inconvenience.

Yes. ISP is very important. It's a nightmare having to keep removing a MCU 20 times every hour for reprogramming while you code and test new features.

And what's with capitulating on that protection diode?  There was a video done specifically to show it was not needed.

Yeah, but some people wanted to have it anyway, might as well add space on the PCB for it.
Keeps everyone happy then.

[EEVblog]

I'm all for gilding the lily but can't help the feeling a shark has been jumped.    A "cascade of decisions" necessitated two more chips for want of a reset pin!?  

Yes, that's how it works. Basic math.
All the pins are used, and you need to free up a pin, what do you do? You either compromise on or-rejig features somehow, or you gain yourself extra I/O somehow.
That extra pin isn't going to magically appear.

Yes, I could have used an ADC input to monitor multiple switches. Yes, I could have forgot about the RGB LED's, or extra switch input capability, Ethernet/serial capability, or.... Yes, I could have done this or done that. But I didn't, I did what I wanted to do.
The only shark I've jumped is the one in your imagination.

Come on now.  Is ISP really necessary?  The micro is not tasked for general purpose i/o.  The pins are connected to specific functionality and so it is unlikely that programming the micro off-board will pose a great inconvenience.  Besides, its probably a good idea to put the micro in a socket seeing as its in a power supply.

This is an open source project, the ability to do ISP may be very handy for a lot of people, including myself.

And what's with capitulating on that protection diode?  There was a video done specifically to show it was not needed.

Some people may want it.
People might use this board for other purposes than intended.

Dave.

[Zad]

I think some people are missing the point. The point in this case is not the final product, but the development process. The journey rather than the destination. In electronics, the journey isn't a straight line, it frequently doubles back on itself and can entail returning to the start several times!

I wouldn't have done it the same way either, but then I'm not making it. Get two engineers together and you'll get two vigorously argued opinions about how to do something. Keep them together for more than five minutes and you'll get another two - probably more. It is tempting to think of electronics as a pure science. It just isn't. It is the art of compromise and judgement. Change one trade-off even slightly, and everything can be screwed up.

One thing that I would have done in this case would have been to prototype it on breadboard, and got some test software written. However, this wouldn't have been an issue if Itead had got their fingers out and got the PCBs shipped!

On a more constructive note - How about using 1 or more of the spare outputs (or reconfigured inputs) to strobe the key switches. This would open the opportunity to have more keys for a 0-9 keypad etc via additional strobes. It would complicate the software, but it can still be interrupt driven.

I wouldn't have specified Ethernet either, and would also have gone with RS232. Much easier to integrate with existing software and easier to hack some software together. But hey, if the SPI bus is available and there is a cheap module - why not? Leave it to the people in the open software community, it all adds to number of potential purchasers!

To clarify something I said in an earlier post; I wasn't intending that the PSU actually be an Arduino, but be a "shield", with the Arduino supplying the processing (and external brick power supply). I know this is ridiculous, but we have seen recently with the Quadcopter and Makerbot, that people just love buying "stuff" to plug into their Arduinos, even if it would be cheaper as a standalone unit with integrated controller. Places like Adafruit and Sparkfun will gladly stock a shield, whereas they would not want to stock yet another Arduino clone itself. Just a thought.

Of course, as it is an open design, a shield version could be designed at a later date anyway.

[PeteInTexas]

I'm all for gilding the lily but can't help the feeling a shark has been jumped.    A "cascade of decisions" necessitated two more chips for want of a reset pin!?  

...That extra pin isn't going to magically appear...

Did that reset pin really need to drive the LED? Is that LED really needed? Isn't there an LCD already?  Was finding screen real estate more costly than adding two more chips?

Come on now.  Is ISP really necessary?  The micro is not tasked for general purpose i/o.  The pins are connected to specific functionality and so it is unlikely that programming the micro off-board will pose a great inconvenience.  Besides, its probably a good idea to put the micro in a socket seeing as its in a power supply.

This is an open source project, the ability to do ISP may be very handy for a lot of people, including myself.

Did I catch it in the video that the micro will do the boot loader serial comm thing?  What's the difference.  Well, if I remember right you can't program fuses with the boot loader.  How often does fuse programming come into play?  20 times an hour seems ridiculous to me and not worth the pins.

And what's with capitulating on that protection diode?  There was a video done specifically to show it was not needed.

Some people may want it.
People might use this board for other purposes than intended.

Quite frankly I'm disappointed.  I know that means nothing to you and thats fine.  But I've seen a lot of your videos plus a lot of the Amp Hour and its always this "design by committees are bad", bureaucracy is bad, dilbert this and dilbert that and I ate it all up.  But then what did you go on and do?  Gave in to pointy headed bosses online even after you went to the trouble of video to show them wrong.  Maybe I was expecting too much.

And as far as "people might use this board for other purposes", thats a cop out.  Just because its open does not mean you have to put in the kitchen sink.

just my two measly cents

[Psi]

It's not being "designed by committee" at all.
It's being designed by Dave and he's free to accept or reject suggestions people have.

In "Designed by committee" no one person has final say.


The video only showed that the circuit appears to work without the diode.
Even after testing, any experienced engineer would include a footprint on the pcb just in case it turned out to be needed later on. After all, it doesn't cost anything to add 2 extra tracks/holes to a pcb.

[EEVblog]

Did that reset pin really need to drive the LED? Is that LED really needed? Isn't there an LCD already?  Was finding screen real estate more costly than adding two more chips?

It's want I WANTED. I know that might be hard for some people to understand, but try, please.
If you want something different, go design your own.

Did I catch it in the video that the micro will do the boot loader serial comm thing?  What's the difference.  Well, if I remember right you can't program fuses with the boot loader.  How often does fuse programming come into play?  20 times an hour seems ridiculous to me and not worth the pins.

It can do both.
Some people may want the bootloader, just like a regular Ardunio, to use their existing tools and the Arduino interface as they are used to. Others (like me) might like a direct ISP interface using an external programmer for development and debugging using any tool they like. In any case the ISP pins are already freed up for the Ethernet/serial interface, so you get that for FREE.
Both are useful for troubleshooting.
I wanted both, I catered for both, end of story.

Quite frankly I'm disappointed.  I know that means nothing to you and thats fine.

Correct. Because if I cared what every single person thought about every aspect of any one of my videos, I'd end "doing a photonic induction". Seriously, I would have to quit the whole damn thing.
So why don't you just quit it with the unproductive comments? I don't need it.

But I've seen a lot of your videos plus a lot of the Amp Hour and its always this "design by committees are bad", bureaucracy is bad, dilbert this and dilbert that and I ate it all up.  But then what did you go on and do?  Gave in to pointy headed bosses online even after you went to the trouble of video to show them wrong.  Maybe I was expecting too much.

And as far as "people might use this board for other purposes", thats a cop out.  Just because its open does not mean you have to put in the kitchen sink.

Adding a footprint for a diode that might very well be useful for some scenario's is not "putting in the kitchen sink", or giving in to anyone, or a "cop out".
My testing was not extensive, nor was my conclusion from what little testing I did. It's simply an additional footprint I put in because it might turn out to be worthwhile in some circumstances.

Dave.

[Rerouter]

great work as usual dave, to peteintexas, this kit has always been something with a purpose, where dave was nice enough to make it easy for us to mod, and by releasing the schematics able for us to completly redisign it if we wish,

love your choice of the MCP23008-E/P, means we any mix of up to 8 inputs / outputs, and that they can be used to run leds directly, being how they are clamped to 20mA

[quarks]

now it is fixed

the download link opens the older EEVblog #237 file.

Argh - you're right - I just tried to watch 238 from my media server (which grabs the videos from RSS) and although the filename's right, the contents are the Makerbot unboxing video.  Time to find out if the new television's "play from YouTube" works, I guess!
[/quote]

[djsb]

Will there be a revision C schematic? Is the design frozen at this stage? If so, I may volunteer to create a multi page schematic (Altium designer based) if that would be useful? It may take a few weeks though as I'm still learning how to use AD10. It will also give me a chance to get some practice. Please let me know if this would be a useful contribution.

David.