Universal PI Micro

[sido]

Hi Pete, VR2 in the schematic is wrong, you have a 79L12 but the circuit looks like it needs to be a 79L05.

Hi Mick, have not compaired BW schematic, but good pickup on the "V" regulator specs.

Where is the single uC schematic design as the one PtB has posted? Is there a PIC sourse code provided already, or is it to be developed in due course?? according to the basic design parameters developed so far?....I am confused as to its forwarded format.

I must of missed something

Is it possible new threads can be created from the past to the NEW to aid the experimenter....instead of shifting through past documents to have a basic understanding of what is being forwarded now?

I may be missing some stuff with the whole project.........as i have may misunderstood the design changes to a single uC?

Sorry for my incompetance.

Am i the only one....maybe i am

Cheers Sid

[mickstv]

Hi Pete, VR2 in the schematic is wrong, you have a 79L12 but the circuit looks like it needs to be a 79L05.

[Pete the Builder]

A PCB to have and to hold...

OK, here's my first-pass attempt at a PCB.

I'll just post two images first off, to get everyone's feedback on size and layout.

This first attempt is a double-sided board, and it probably needs to be manufactured, unless you have experience with making double-sided boards. I've got a quote, for just one board on it's own is about US$70. That's the minimum, and I'll be making at least one for myself. Subsequent boards are much cheaper - for example, if 6 people want one, the full cost per person is US$25.00 That's for a solder-masked, silk-screened, through-hole plated, double-sided FR4 1/2oz copper board - better quality than anyone can hand-make. But it IS doable, if you've got experience. I've tried to minimise double-sided traces where possible.

I'd guess that we won't want the main board made up until the daughtercards are done, which is my next project.

If people prefer a single-sided board, I'm working on that, but it will be a bit larger and it will take some time, as I'm hand-routing everything.

This board is 3.75 by 3.25 inches. The main changes I've made (as we discussed earlier) are : a full 5V positive regulator, which means no additional regulators on each daughtercard; boxed DIL headers for the daughtercards, which can be converted to SIL headers if you prefer, just plug in the header to the appropriate holes. And the reset circuit has been added. I've also added some bypass caps, and the final release will be ground-planed top and bottom.

All the rest of the circuit is exactly as Bugs designed it, even the pin numbers are the same!

Here's a really crappy 3D image of the board...Who knew that Altium's $6,500 software can't print the 3D view for nuts?


... and here's the board layout showing the top (red) and bottom (blue) traces. Part overlays are green. And for some additional reason that escapes me, Altium won't show vias in Assembly view. But the vias are in the final output, so please ignore the bug.
The final silkscreen will have component values for caps and resistors, and connector names for the connections, so it's really easy to build without needing the schematic. All part of the service!
Note that the overlay view is the latest revision, and is different to the 3D view. The 3D view was an earlier revision, but it shows the parts in their correct places, and the mounting holes and so on. Just the traces have changed (for the better, I hope!). Sorry if that's confusing.


Let me know if you're OK with this, and I'll go on to the daughtercards.

Cheers,
PtB

P.S. I just realised that I forgot to mention something important : if you already have daughtercards made up, they will work perfectly in this motherboard with no changes. Just a thought.

[WM6]

Hi Pete ,
Very nice work,i hope more

Yes, good start PtB,it will be most on your own, but we all support you.

[moscadondres]

Hi Pete ,
Very nice work,i hope more

[Pete the Builder]

OK, here's the first attempt. I can't attach it as a bitmap or image, 'cause it's too dense. (A bit like me!). So it's in PDF form.

Please go over this, and just double-check that I haven't missed anything silly! Also, if you can, please verify that it matches the correct (i.e. latest) schematic revision.

I've made a few minor, and one major (but invisible) changes to the circuit. Here are the changes I've made :
- I've renumbered the parts. (I can undo that real easy if it's too much of a headache comparing. Most of the subcircuits are blindingly obvious, so it shouldn't be a brain-baker).
- I've added a standard cap and pullup to the reset input of the MPU to meet the Atmel specs. BW was extremely lucky that the floating reset input allowed the system to work at all! I'm assuming, therefore, that he had the programming header connected all the time.
- I've changed the 24x1 SIL headers to 50-pin (25x2) headers. You can use SIL, DIL, or boxed IDC headers as connectors with no changes.
- I've changed the main reg to a 7805 instead of an L05 (this has better line rejection and lower output impedance - both good things in this circuit)
- I've also added a VCC rail to the "spare" pin of the 50-pin header.

A bit of explanation is necessary about the last two changes : I really try to avoid multiple regulators on a single small system - they're just additional points of potential failure. There's no reason why the 7805 can't supply plenty of 5V to all the cards in the system. So by making the motherboard regulator a 7805, and adding the extra bus line, it allows two options for everyone : have all the boards powered exactly the same as the original design - in this case, you don't need to do anything. Just ignore the 25th pin, and the original boards will work perfectly. The second option (that I'll be using) is to use the supplied 5V rail to power the boards.

It's entirely up to you if you want to take advantage of the single power option, and you don't have to do anything either way. Just move a jumper on the daughtercards, and that's it.

I'll provide a design for the daughtercards that does the same thing - build it with the additional regulator on each board, or you can ignore the extra regulators and use the supplied 5V.

I hope this way to a) allow anyone who really, really, really wants to use the exact, same, identical-in-every-way design of BW to do so without even thinking, and b) let anyone who wants a more reliable supply and mounting system to be able to do so with almost no changes to anything - just a jumper on each daughtercard. I think that gives us the best of both worlds, but I live to be corrected. And of course, this needs to be tested thoroughly - this is something really basic that can catch up to you and ream you while you're not looking! So I'm not assuming anything with this change - not until it's been tested to death and verified that it's at least as reliable as the original. I hope that puts any nervousness at ease!

I hope this design looks right, and please let me know right away if you find anything wrong. As soon as I've heard back, I'll start the PCB design. In fact, I've started playing with the rat's nest already. It's going to be, er, fun to get it fitting the original PCB area...

Cheers everyone!
-PtB

Switched Cap Motherboard 3.pdf

[Pete the Builder]

OK, I've started. First thing : do we want to stick with the AtMega 644 as per the original design, or update it?

I was thinking, why not make this a truly "universal" board? I can create a universal template socket with all the required power and signal connections, and a pluggable chip carrier, that could take a PIC, Atmel, ARM, whatever. What do you all think of that idea?

The pros : Anyone can plug in their own chip, whatever they're familiar with. As processors improve faster than analogue stuff at the moment, it makes "chipping up" a heck of a lot easier. The pins could easily be changed to a different layout, if someone prefers. It could be used as a teaching device, too, with the appropriate signals sampled by whatever, and so on. It's a bit of a stretch, but stranger things have happened with my designs!

The cons : It's another set of connections to go wrong. I (or someone) has to design the boards to plug in (although that's a once-off thing). And even with the best engineering, it's going to be subject to vibration and extra movement forces, even if only slightly. A sharp knock could unseat the board, damaging the uP. (We could bolt down the board, though). And those types of thing look terrible, even with the best design (and this will be the best design I can do). Oh, yeah, it's also another set of connections the signals have to pass through.

I will be designing the first board with all the original CPU bits, as a first step anyway. And I think the only code we have will only run on the 644, at least until someone ports it. So this is more of a "what if" type of question for down the track. What are your thoughts about that?
-Pete

[moscadondres]

Hi Pete,
Thank you for this,i'm waiting.

[Pete the Builder]

An update!

G'day all...

Sorry for the delay, I've been getting completely confused by all the threads I'm trying to follow, and all the different designs I'm lining up. Whew. I've been offline a bit more than usual, new drugs are kicking in and my mind is very, um, cloudy. Anyway...

So. Looks like we're settling on a LiPo 14V supply (I'll also define components for 12V SLA, because I have a cupboard full of them and LiPos are a bit expensive for this pensioner... Anyway, I'm looking at a single supply input, with high-efficiency LDO regs (also nice and quiet), and a quiet and efficient inverter for -12 and -5V. Since we only need a few milliamps on the negative rail, it's a bit too much overkill with multi-output DC-DC inverters. We either need 2 inverters (one for low current rails, and one for + high current rail). The inverters I've seen either have small outputs on all the outputs (+ and -), which is great for the negative rails, but unusable for the + rail(s), or they have massive outputs on all rails and cost well over AUD$100. So a hand-carved dedicated supply is indicated. That's great, that's my speciality anyway.

Oh, yeah, there will be a link (or maybe links) to select between SLA, LiPo, and (maybe) AA cells. It's the discharge curve that defines the power supply input, so I'm going to look at whether or not good quality lithium cells might be usable with not too much fiddling around. Anyway..

The PSU will be shielded just in case. I like that idea, makes a lot of sense, since that's where most of the noise will be coming from! I hope that's acceptable to everyone? Is everyone OK with the battery/power supply options?

Next step is the motherboard. I've been playing around with a strain gauge from my neighbour, and I was surprised to see that DIL connectors are about 20% stronger than all the alternatives mentioned so far - they're less able to be moved from the seating position, and less likely to be moved out of plane by vibration or force. And we don't need to worry about securing nuts and washers and so on, even for the "piggyback" board. I'll do one SIL board layout, and an equivalent DIL layout as well. DIL also gives us the very nice option of extender cables for troubleshooting or initial testing, via IDC connectors. In fact, I can do a single mothercard (motherboard sounds a bit wanky ), using DIL footprints, and you can use SIL headers if you prefer. What do you reckon?

Oh, yeah, that's if the cap sample board ends up being a piggyback design. I'm looking at modular boards of 8, 16, and 32 caps each, what do you reckon? It would allow people to pick the sampling resolution they need, and gives us a more repeatable and consistent access time for each board, but if it's a bit too "user friendly", I'll go back to the piggyback board (which we know worked perfectly for BW). Your thoughts?

I can't wait to set up my main system and get the schematics done... Hmm... I have an old HP laptop... I'll see if I can get Altium running on that. I'll have to play with my licensing, but it might be doable. I'll try it out tomorrow and get back tomorrow evening (we're house-hunting tomorrow, so it's a write-off for me).

That's about it for now... Look, if anyone feels left out, I didn't deliberately ignore your ideas... I'd like to be able to incorporate every good suggestion into the design, but there's a lowest common denominator as far as complexity and the amount of work I have to put in, and there's a limit to both. But if I've overlooked something brilliant, let me know and we can vote on it or draw straws or something.

That's it for now, it's late, time for more drugs, man...

Cheers,
Pete

[vbeeeks]

I am currently surfing an interesting site (translated of course) for some educational insights not fights.
****http://phazitron.narod.ru***. Archives a la mention. You can also calculate the cost of some latest bogies and see the price of IP put into it , bla bla bla.

[sido]

I'll prefer 3-inline 90 degree bent berg strips for interconnecting daughter-boards that can be bonded with PVC spacers as one unit - for power 14+V Lipo and LDO SMPS in tin enclosed & output thru bypass caps as frequently seen in Satellite TV circuits, obviously low profile one, as seen in mobile phone. See my impression below.
[ATTACH]23083[/ATTACH]

vbeeks, agree with the bypass caps, was thinking of that while brainstorming the best possble RFI, EMI shielding for interconnecting modules. Similar to the analogue RF TV tuners where each input/output pins are connected to feedthrough bypass caps on metal enclosure.

Does anyone know where i can source these feedthrough Caps??

Low frequency Toroid/ferrite beads may also be used in critical interconnecting modules.

If you see how Military Spec gear is designed where RF and Audio is involved, you will see sub modules sealed in seperate metal enclosures, interconnected in some cases with feedthrough caps and high quality connectors and cables.
Might be a bit overkill, but at least the circuit designs will have the lowest possible chance of any internal/external interference to start with.

Sid

[vbeeeks]

preference

I'll prefer 3-inline 90 degree bent berg strips for interconnecting daughter-boards that can be bonded with PVC spacers as one unit - for power 14+V Lipo and LDO SMPS in tin enclosed & output thru bypass caps as frequently seen in Satellite TV circuits, obviously low profile one, as seen in mobile phone. See my impression below.

[sido]

Hi Pete.

The Sil connectors are ok if your stacking pcb's on top of each other and using hex spacers in between for added support between pcb's.

If stacking pcb boards in vertical position to the daughter board, Sil connectors can still be used provided each board is supported by also hex spacers placed horizontally to each vertical board if used, but the important addition would have to be angle corner supports or large diameter pins for each vertical board corners, supporting and facilitating extra strength for the sil connector. Similar to the way heat sinks are supported in standard UPS power supplies.

SIMM connectors are fine, but tends to populate the daughter board more and you still have the problem of vertical side to side movement if used with ones without corner retaining clips. The old AMD connectors i think had no retaining clips?

Forget Euro connectors for simplicity.

I would be leaning towards the Sil connector method with corner pins supports that can be soldered in place once your happy with design, testing and performance. This way all connection pins below the connector are strenghtened for field use and any testing that may be required can be done at ease before soldering the corner pins in place for final field use.

Not sure about power supply. Maybe a descrete low intereference type which is also a challenge in design or maybe consider a off the shelf SMPS with frequency sync in a self contained small metal packaged unit. Not sure if they come with multiple power outputs, + 5v, - 5v, + 12v -12v etc etc. Have seen single voltage units though.

Cheers Sid

Open Project Code of Conduct V3.0 Participant

[Pete the Builder]

Excellent. We're all old-fashioned scaredy cats. Wonderful! Actually, to be honest, I can only point hand-solder down to about 25 thou pitch, below that (QFN, TQFN, SOT) I have to over-solder and wick off. But my eyes aren't what they used to be, and I have severe nerve damage in my neck, so my hands have the "ta-tas", which makes even SOIC a challenge! But I like challenges... Anyway, through-hole it is for starters. Thanks for the feedback!

OK, I've been putting the project into DXP, and the first minor change we need to do is to sort out the daughterboard connectors to use. The SIL jobs BW used are just awful, in terms of reliability with larger boards (angular moments of inertia, I think, where a heavy or long board will act as a lever), moisture and dirt ingress, electrical reliability in a portable device, and mechanical retention. Great for the bench, lousy for the beach!

My personal preference would be SIMM clips (the old style 32-pin dual connector type, as used in the Dontronics PIC platform and old 486 motherboards), as they're available nearly everywhere, they retain the boards physically in place really well, they're not too big, and I have the patterns from a previous project! I've used them in autonomous robotic projects, and they can take an astonishing amount of violence without losing contact, so in a hand-held metal detector, they may even be overkill. But reliability is my touchword, so the SIMM option would be at the top of the list. At around $1 each, they make sense cost-wise too.

An alternative may be EURO type connectors, but the physically smaller receptacles can be hard to source, and using 32-pin receptacles would make a very large board necessary. Plus, they're not as physically rigid as the SIMM sockets. And they're much more expensive.

Does anyone have a better idea? (Don't worry about PCB footprints, I can design anything we need in terms of physical layout)

Ummm... Oh, yeah. Power supply. There are some excellent MAXIM devices that provide buck/boost and multiple outputs with super efficiency that I've used before with great success. Some will even work down to 1V, providing +/- 12, +5, and 3V3! The catch there is a little more complexity (one DIP, but about a dozen passives, including a couple each of FETs and inductors) plus the possibility of HF switching noise. What do you think? Stick with the linear/inverted negative rail option, or jump in the switching pool?

That's about it for now. Any other comments or suggestions to improve the basics?

Cheers,
PtB

[Davor]

I'm a "through-hole" kinda guy anyway!

Ditto. Actually most of us here are.