EEVblog #238 – Power Supply Design Part 7Posted on January 20th, 2012 34 comments
Dave has changed a few things in the Rev B schematic of his power supply design. Talk on opamp specifications, I2C I/O expansion and AVR microcontroller pin usability.
Download the RevB schematic HERE
Hey Dave , itunes podcast version shows the makerbot video.
Excellent evolutions ! The RGB LCD is a bit “too much” for my taste, but having extra I/Os is really nice.
Will you use a 16×2 or a 20×4 LCD ?
I’m blind or stupid (or both) Where’s the pdf ?
My incredible deduction skills led me to this conclusion:
The schematic is pretty unreadable! You probably tried to get it down in size, but good practice would dictate that a schematic should show a clear functional flow of the system.
As-is, the schematic looks like a mosaic. A random observer wouldn’t even know it’s a power supply.
Why didn’t you simply choose a larger micro? A 644PA probably would have saved you the extra IO ports, and there are additional bonuses like a second UART, SPI capability of hardware UARTs etc thown in.
I’d love to have one of these but I think your target design was for a maximum output of 10 volts and alot of times I work on stuff that is 12 and 13.8 volts, would it be possible to make it output at least 16V but preferably 20?
Yeah, that schematic’s kinda messy.
Probably time to put it on separate pages.
Any word on getting a kit put together for this?
Hi Dave, why not use the AVR’s SPI pins instead of bit bashing using the pins you have selected. It will be much easier. The atmel programming interface should cause no issue with this, and would only be used for setting fuses and programming a bootloader
I was wondering if it was possible to use an lm324… seems somewhat similar and available and more common
Thanks in advance.
Man, what’s wrong with people? Why is everyone bitching about this and that and asking questions that have been answered in previous blogs already?
Bit banging SPI or I2C is pretty simple, just have a look at the timing diagrams in the datasheets. It can be done with a few lines of code.
And for anyone who thinks that the schematic is messy, please, take a look at some hand drawn schematics of an old meter or something like that. Those can be pretty messy
I think Dave is doing a great job! Keep it up!
Its way easier to find small faults in someone else’s work and feel good about your self, then do all the hard work and take the chance you’ll be criticized yourself.
So the internal RC oscillator is not good enough, why not just use your FE-5680A Rubidium frequency reference ? – combine 2 projects in one, and have very stable clock for RS232 communication…
Holy crap, so many complaints. I love this series, the fact that you have constraints (like thru hole, price and so on) makes it more like real life.
In a perfect world a single chip would do everything (and it does happen), but where’s the fun in that?
If I was teaching some students and wanted them to get a glimpse of the design process this is what I’d have then look at. I learned some things myself. What more can you ask for?
Fantastic job, keep it up.
thanks for this series of tight technology teaching videos. They cover analog circuitry, SPI, I2C, rotary encoders, signal limiting, overload/reverse current testing and in the future possibly timing issues with I2C or software-SPI. Much we can learn from.
Go on with it!
Keep up the good work,
I like this series of videos, it gives many angles to a design process, even though I may not have made the same decisions as you, but again, I may have other preferences.
Dave, i see you use 10K on SDA/SCL . While that works fine on a 5 volt rail with a single device , on a 3.3 volts rail you are running marginally, especially with that many slaves. … i would drop those to 3k3 ( i typically use the rule 1k per volt of rail, it gives you 1mA if current )
With that many chips on the bus you will have quite some capacitance. If you want to maintain reasonably clean rising edges you will need a bit of current there.
When routing : use the dasiy chain constraint on SDa and SCL and place the pull up resistors at the last point (farthest away from the cpu) . If the bus is getting to be reasonably long ( let’s say over 10 inches) i typically split the pull ups and put a pair at both ends ( simply double the resistance value at this point. I put two sets of 4k7 giving me 2.35k effective per line ) from a 3.3 volt rail i get nice clean edges with a drive current of slightly over 1.2 milliamps.
just my 2 cents
Hi Dave, this PSU project is good stuff
But heads up, I think I see an error in the RevB schematic. Check it out: shouldn’t the opamp U3B (for current limiting) have its inputs the other way round? Unless I’m missing something, currently when measured current is lower than the setting, the opamp output goes high and the transistor clamps the voltage down, down, down…
I have been realy enjoying this series and looking forward to a kit release!
I like the whole design process and i think it is great to hear some of the other considerations, such as case size etc.
However to some extent i mirror DCflux’s comment about the higher voltage ability. Having said that though im looking forward to putting some theory into practice and looking at extending that capability myself, you can never have enough supplies…
EEVblog #238 – Power Supply Design Part 7 | EEVblog – The Electronics Engineering Video Blog – just great!
Can someone please help me understand how the differential amplifier (that senses current) has suddenly been disappeared from REV B to REV C and been replaced with a set of op amps? And why isnt this explained in the video?
I tried. failed to spot why.
Awesome series. Really interesting. Thank you very much for doing it.
At 21m43 secs you talk about the Vio grades of the TLC272 (A/B etc).
You were wondering why the TLC272CP has a higher Vio than the A or B.
It’s because the C here refers to the temperature grade (commercial) not the Vio.
So, in TI’s parlance:
TLC 272 x t p (x= Vio grade, t=temp grade, p=package)
x= nothing (std grade). A or B
t= C (commercial), I (industrial), M (Military) or Y (automotive).
Used to work there on Linear 15 years ago.
Can’t believe my head’s still full of this crap…it’s taking up valuable space.
1 Trackbacks / Pingbacks
[...] what a tiny heatsink he has. I wouldn't try to get 1.5A out of that at a low voltage… Check out this for an open source PSU design. This will be available in kit form almost certainly. Check out [...]
Leave a reply