EEVblog #259 - PSU Rev C Schematic - Part 12

[Shale]

I think Dave wants the micro to be socketed so it's easy to swap out if necessary.

I was thinking something like this
http://parts.digikey.com/1/parts/295966-ic-socket-32pos-plcc-smd-w-post-8432-21a1-rk-tp.html


http://search.digikey.com/us/en/products/ATMEGA328P-15MZ/ATMEGA328P-15MZCT-ND/2477178

couldn't you get the 328 chip to fit in something like this? would this be a solution? if not then why?

[m12lrpv]

The only pin difference offered by your suggestion is the two ADC pins (ADC6 and ADC7) which isn't much. The extra memory afforded by the 328 can be had in a dip replacement that the user can otherwise do themselves.

[FreeThinker]

I think Dave wants the micro to be socketed so it's easy to swap out if necessary.

I was thinking something like this
http://parts.digikey.com/1/parts/295966-ic-socket-32pos-plcc-smd-w-post-8432-21a1-rk-tp.html


http://search.digikey.com/us/en/products/ATMEGA328P-15MZ/ATMEGA328P-15MZCT-ND/2477178

couldn't you get the 328 chip to fit in something like this? would this be a solution? if not then why?

Well It would cost around $4 more for little extra functionality

[alm]

But, when the component manufacturer strongly suggests the use of an external low-pass filter for enhanced noise immunity in ther engineering application notes, I will ostentatiously disregard any quick-and-dirty hacks of non-engineering communities like the aforementioned one; of non-engineering communities that introduced that imbecile baby-talk by renaming the add-on PCBs to "shields" and the programs to "sketches"...

Do they really strongly suggest a low pas filter on the reset pin? I'm surprised, since that would interfere with debugWire.

From AVR042:

The recommended pull-up resistor is 4.7kOhm or larger when using STK500 for programming. For debugWIRE to function properly, the pull-up must not be smaller than 10kOhm.

To protect the RESET line further from noise, it is an advantage to connect a capacitor from the RESET pin to ground. This is not directly required since the AVR internally have a low-pass filter to eliminate spikes and noise that could cause reset. Applying an extra capacitor is thus an additional protection. However, note that this capacitor cannot be present if debugWIRE is used.

Not exactly a strong recommendation in my opinion, more a suggestion that it might be useful if your environment is very noisy. A lower value pull-up might be a good idea, but this would probably interfere with the capacitively coupled reset circuit.

[EEVblog]

Dave. why have you chosen to use the standard arduino 168 micro vs, the 328 or the one on the arduino Mega? After you changed the entire design to smd i would have thought you would change the micro as well, but i seem to be wrong. with all the people wanting to mod your design seems like the extra pins could just connect to headers and be left for people who might want them. as I am not yet an engineer I would like to know what drove those decisions.

The 168 and 328 are the same pinout, you can use either.
The 168 has plenty of code space and is cheaper.
The 328 might be handy if you want Ethernet though, but don't know, haven't tried Ethernet yet.

I just feel like having the micro in socketed through hole might be an advantage to some down the track. Just like the new UNO still has a DIP chip.
I don't need any extra I/O capability in this, it's feature-laden enough as it is.
The 328 means it looks like an Arduino Pro/Uno. To get something else "standard" you have to go up to an Arduino mega 2560, and that's a big price difference.
There are other ways and other chips you can use, but its in the standard Arduino build.

Dave.

[A Hellene]

[...]
Do they really strongly suggest a low pas filter on the reset pin? I'm surprised, since that would interfere with debugWire.
[...]
Not exactly a strong recommendation in my opinion, more a suggestion that it might be useful if your environment is very noisy. A lower value pull-up might be a good idea, but this would probably interfere with the capacitively coupled reset circuit.

Alm, the debugWire is a fairly new addition. When the reset pin did not have that functionality, the standard recommendation for the reset pin protection from external noise was a 100 nF capacitor to ground and a 10 kohm pull-up resistor.

Since I adopted the AVR family almost from their beginnings in '98, I can still remember another standard recommendation of that era, not to use the first EEPROM location (at address 0x00) in the absence of an external reset supervisor because the chip could execute random instructions when power was low during start up and power down that could (and did!) erase that memory location; this changed when the next AVR family with an internal brown-out detector and a more complex programming algorithm arrived.

I realize that the AVR nRESET input might now be hardened against external noise but I am still taking that simple precaution. I can remember like it was yesterday my first prototype without that 100 nF noise filter that worked flawlessly on the bench but it could not even boot properly up in automotive environment, where it was destined to operate. This is the reason why my opinion is so strong on this matter.


-George

[markus_b]

If we change the micro then I would go to an xMega like the xMega32A4U. It features an fast 12-bit ADC and DAC, making external ones superfluous. It also has USB on board and enough ports, so you need no i2c I/O. The Arduino support is unofficial, though.

An interesting application of the chip is the xprotolab MSO, Gabotronics have published the schematics and the source code.

Another thing: There was talk that the LM358 is not rail-to-rail. I propose to use the OPA4171, it is rail to rail and goes to 36V, like the LT3080.

Markus

[McMonster]

XMegas are a bit obscure comparing to other AVRs, much less resources available online on how to program this thing and they're not available in DIP packages. And it requires a different programming hardware (for programming the bootloader in at least), either Atmel's JTAG or something supporting interface called PDI. Last but not least, I think it's not Arduino compatible.

I think that if Dave wants Arduino-compatible PSU, the Mega168 is the best choice. And there's a whole range of pin compatible chips from 4/1 kB Flash/RAM to huge 128/16 kB with Mega1284, all in DIP packages.

[markus_b]

XMegas are a bit obscure comparing to other AVRs, much less resources available online on how to program this thing and they're not available in DIP packages. And it requires a different programming hardware (for programming the bootloader in at least), either Atmel's JTAG or something supporting interface called PDI. Last but not least, I think it's not Arduino compatible.

I think that if Dave wants Arduino-compatible PSU, the Mega168 is the best choice. And there's a whole range of pin compatible chips from 4/1 kB Flash/RAM to huge 128/16 kB with Mega1284, all in DIP packages.

Yes, xMegas are relatively new and they had a bad start to boot (plenty of errata, like a broken ADC). But they have matured and the second generation U-devices are good. Programming is using PDI, not ISP but even a dragon can do PDI these days. But using the USB bootloader is the easiest way.

There are Arduino libraries for it, but not supplied by the official Arduino people.

Personally I don't care much for a DIP package, they are getting mighty big if you need a bunch of ports. Yes, if you put is into a socket (you do ?) you can remove it to reprogram it using a stk500 (do you have one ?). Nah, is not worth the trouble and waste of board space.

[markus_b]

One more thing about microcontroller choice: This is Daves design and he has the say. I can see that taking a micro which is out of the box Arduino-compatible is interesting for him. The over all design as he presents it here is very interesting, but does not fulfill my needs. So I started to stray off, taking the things I like (the linear power and current regulation part is pure genius) and replacing the things who not fit my ideas.

However, maybe it is better if I shut up here on everything out of scope (and and another micros is) and discuss that in another thread.
Here: Dave's power supply mods.
or here: General Purpose Power Supply Design

Markus

[EEVblog]

One more thing about microcontroller choice: This is Daves design and he has the say. I can see that taking a micro which is out of the box Arduino-compatible is interesting for him. The over all design as he presents it here is very interesting, but does not fulfill my needs. So I started to stray off, taking the things I like (the linear power and current regulation part is pure genius) and replacing the things who not fit my ideas.

That's the thing, it can't be all things to all people, that's impossible for any design or any prodct. So I'm not even going to try, I'm building this for me.

Dave.

[Shale]

I have another question for you dave. Did you ever think about making your power supply like a shield for the arduino? if so why or why not and what would be disadvantages to that idea?

Thanks

[EEVblog]

I have another question for you dave. Did you ever think about making your power supply like a shield for the arduino? if so why or why not and what would be disadvantages to that idea?

Not even considered for a second.
Why?
Because it would be a very poor choice from a physical design perspective.
How do you fit the Arduino + shield in the case nicely? What about the heatsink and front panel switches?
It would be:
a) Ugly
b) more expensive
c) require more wiring
etc

You have to do a custom PCB either way, so doing a shield gives you absolutely no advantages what so ever.

Good product design can be mostly about (or driven by) the physical implementation.

Dave.

[Shale]

Good product design can be mostly about (or driven by) the physical implementation.

Thats something I hadn't thought about. Thank you for answering all my questions, and keep up the great work on the videos.

[king.oslo]

Dave, on multimeters, I know you do not like protruding range-switches. They are a weakness despite their low profile. If they fall on their face, damage may be inflicted.

Unfortunately, I spot a weakness on the PSU. The rotary encoders look nice, and I like them, but they are the first to sheer off when the power supply is transported. I would find some switches with a good feel.

Marius

[markus_b]

Unfortunately, I spot a weakness on the PSU. The rotary encoders look nice, and I like them, but they are the first to sheer off when the power supply is transported. I would find some momentary switches with a good feel.

I understand your concern about the robustness of these rotaries, but I find those an excellent feature. *Much* better than momentary switches. I have a digital power supply with momentary switches and I would very much prefer rotary switches to operate it.

[EEVblog]

Dave, on multimeters, I know you do not like protruding range-switches. They are a weakness despite their low profile. If they fall on their face, damage may be inflicted.
Unfortunately, I spot a weakness on the PSU. The rotary encoders look nice, and I like them, but they are the first to sheer off when the power supply is transported. I would find some switches with a good feel.

Yes, I have deliberately traded off robustness for the sake of having knobs. I like knobs.

However, I am working on another PSU design that has switches instead of knobs, as I consider ruggedness more important for this one.

Dave.

[king.oslo]

However, I am working on another PSU design that has switches instead of knobs, as I consider ruggedness more important for this one.

Where may we learn more about this?

M

[EEVblog]

However, I am working on another PSU design that has switches instead of knobs, as I consider ruggedness more important for this one.

Where may we learn more about this?

In due course.
At the rate the battery PSU is going, this one could come out first...

Dave.

[EEVblog]

Dave, on multimeters, I know you do not like protruding range-switches. They are a weakness despite their low profile. If they fall on their face, damage may be inflicted.
Unfortunately, I spot a weakness on the PSU. The rotary encoders look nice, and I like them, but they are the first to sheer off when the power supply is transported. I would find some switches with a good feel.

Yes, I have deliberately traded off robustness for the sake of having knobs. I like knobs.

And of course, as soon as I say that, I mock up the front panel, and the knob usability kinda sucks. Not as good as I thought.
Either top or bottom PCB orientation isn't working that great.
I may have to ditch my beloved knobs for switches  , to gain space and more sensible vertical mounting of the binding posts.



Dave.

[king.oslo]

I may have to ditch my beloved knobs for switches  , to gain space and more sensible vertical mounting of the binding posts.

Dave, I know the unpleasant feeling. I love knobs too, but on the contrary, I think switches are a mature decision. This is my suggestion:

Free up space, cost and ruggedness with switches in stead of knobs. With the new space it is possible to move the TO-220 to the back of an aluminium front panel (sexy aluminium where you can see it) and plastic back panel. I think stubby shorty binding posts which protrude less, rather then full size binding posts. The new aluminium front panel will reinforce the new binding posts. That way, the binding posts will no more look like a pointy protruding weakness which breaks off. You can get chamfer holes for the switches which will feel  luxurious for the finger

I hope you like the idea. I like knobs too, but these changes will move the sweet PSU into a new league. I am really enthusiastic about these changes

Kind regards,
Marius

[Andy]

Free up space, cost and ruggedness with switches in stead of knobs. With the new space it is possible to move the TO-220 to the back of an aluminium front panel (sexy aluminium where you can see it) and plastic back panel.

Is it really a good idea to make the front panel a heatsink. What if it gets hot...

[T4P]

Free up space, cost and ruggedness with switches in stead of knobs. With the new space it is possible to move the TO-220 to the back of an aluminium front panel (sexy aluminium where you can see it) and plastic back panel.

Is it really a good idea to make the front panel a heatsink. What if it gets hot...

That's why you know the thermal performance   
And Dave's using a tracking pre-regulator .

[jerry507]

What is the obsession with the analog tracking loop? It adds a lot more circuit complexity to the design, likely saves no money, and reduces the flexibility Dave has in writing the software to implement various features like voltage tracking current limits etc that he talks about in the video.

Analog dogma is usually a bad idea.

[amspire]

With the switching pre-regulator, think the worst case heat into the front panel is about 3W, and that is only the case when you have left the output shorted with the current limit at maximum. I really cannot see that as being a problem.

For "normal" voltages - ie above 2V out - the maximum power into the front panel will get under 2W, if Dave sets a switching regulator to supply out differential of 3V or less.

Richard.