Community rework of standard maple and maple mini.

LeafLabs Garden » Support » Maple Hardware Support

Community rework of standard maple and maple mini.

(181 posts) (14 voices)

Started 2 years ago by feurig
Latest reply from crenn

12…13 Next »

feurig
Member

I have been looking at a maple ret6 based design for working on elua and was going to roll a design which fixes several of the flaws in the classic maple designs which leaflabs seems to have stopped development of. I thought the community could pick up the ball and move this forward and perhaps make a pull request or two for the legacy hardware.

We can call it maple-r6(community supported addition)

Changes suggested else where include:

1. ditch/rework the LiPo charger
2. Replace one of the voltage regulators with something a bit 'beefier' to support higher Vin voltages and more current out perhaps eliminate the dual supply altogether (i like the x1117 800ma ldo regulators in this case personally)
3. Add USB ESD and better USB filtering
4. Replace USB pull-up circuit with a variant of Siy's PNP digital transistor approach
5. Convert 0402 to 603 or 805 to make it easier to make
6. populate or at least pin out the 32khz cryatal
7. source/fab in the states.

2-7 could be added to the maple mini as well.

What do we think?

Posted 2 years ago #
crenn
Moderator

1. I haven't actually seen many projects use the LiPo charger, however I've used it before for powering some circuits. Maybe haven't a spot that could be left unpopulated?
2. Agreed providing the Vripple is similar to the Analog LDO Vripple it would be replacing. Another option is what Leaflabs was looking at with the Maple II and using an integrated LiPo/LDO IC.
5. For 'homebrew' variants, I'm fine with 603/805, but how much extra space will this require? Will the board size need to increased to support it?
6. Having a space (through hole or other wise) would be good for the 32kHz crystal.
7. Why not the UK or Australia? ;P

I'll suggest there should be a couple of things to make things mildly better and somewhat more Arduino compatible. The powerblock could be 'extended' to have IOREF like the newer Arduinos (this is not really needed as it could be done with a shield for the Maple, but something to think about). VBat should be available somewhere on the Maple. This one is going to sound strange, but 5V regulation would be good as I've had a few projects which need a source of 5V. I've been able to use Vin (because of USB power) but I can't do the same when I have an external 9V power pack. Lastly, automated power switching for the Maple. These are suggestions, and probably bad ones, but something I've thought about for some time.

One thing I don't want to lose off the Maple is the breadboard friendly pins on the 'digital' side. It's damn handy to have that!

Posted 2 years ago #
bubulindo
Member

Hello,

While I do understand and quite like the possibility of moving away from the Arduino form factor (and that horrendous spacing issue), that will also mean moving away from shield compatibility...

This being said, and after seeing this:
http://www.ebay.co.uk/itm/ENC28J60-Webserver-module-Ethernet-Shield-for-Arduino-Nano-/181050561847?pt=LH_DefaultDomain_0&hash=item2a27745537

I think it's possible to create some of these for the mini format, which would look rather nice. :)

1. I believe the Olimex board has a LiPo charger on it. Although I haven't used one yet, this is an interesting possibility.

7. Why not design it, check where the majority of the interest lies (geographically), source it there and ship it world wide?

Posted 2 years ago #
wdsim
Member

Hello,

This sounds like an interesting project. As it turns out, I developed a board based on the Maple RET6 last fall that addresses some, but not all of your ideas. The board is roughly the size of the Maple Native, but maintains the headers for Arduino style shields. The first prototype is shown here: http://www.spoatech.com . I am ready to spin the second rev to clean up a couple of things and I would be happy to incorporate your ideas if possible. The board allows input voltages up to 28 Volts and has a 6 volt switching regulator that provides the source voltage for the 5 Volt supplies and 3.3 Volt supplies.

The board incorporates onboard, a MAXIM RTC, Wiznet Ethernet module and Serial EEPROM over SPI. It has 16 channels of Op-Amp buffered analog inputs with the option to correct the AIN5 conflict issue with BUTTON. It also has 8 channels of PWM low side MOSFET switches capable of over 30 amps each.

I am also putting a connector for remote cabled reset and button control, so that the board can be re-flashed without having to touch the board directly. This came in handy with one of the installations where the board was hard to reach after installation, but could be re-flashed by going into bootloader mode by remote cabled interface.

Cheers

Posted 2 years ago #
gbulmer
Moderator

feurig - Ah, you've started a new thread. I was proposing changing the name of the other thread.

Anyway my $0.02 are at:
http://forums.leaflabs.com/topic.php?id=1303&page=10#post-22994
http://forums.leaflabs.com/topic.php?id=1303&page=10#post-23020

I haven't time at the moment (feeling ill), so I'll circle back later with a better explanation.

However "source/fab in the states" is not a goal for me.
I'd like people to be able to DIY the board, wherever they are, and that is part of the rationale for some of my changes.

I am working on two very simple Dual-In-Line (DIL) format boards for the LQFP48 and LQFP64.
The aim is to be low-cost 'cheap as chips', DIY-able, Maple-bootloader compatible, bread-board friendly, robust and small.

I want to fit into small 'products' and onto small-format robots. For example mini-sumo must fit within a 100mm cube, maze-solvers are typically well under 100mm x 80mm; half-size are about 50mm x 40mm. So the LQFP48 is a priority at the moment.

IMHO siy's boards at https://github.com/siy are a very good place to start. Siy's LQFP64 board is very feature rich, and they are single sided boards. However they use 0603 parts, which I find a real strain.

Edit: I should add, I don't like 're-inventing the wheel'.
IMHO for people who want an improved Maple board, look at Olimex's Olimexino-STM32
It is stuffed full of features, and fixes many of the defects of Maple. IMHO, it is very good value for money.
Further, it is already in production!

Folks could contact Olimex and ask if they are interested in making a RET6, or even better STM32F103RGT6 version - 1MiB Flash memory, 96 KiB of SRAM.

Posted 2 years ago #
tdc218
Member

My $0.02US:

I don't much like the Arduino form factor - I'd rather have the mini-like DIP layout.
More ram - RET6 or better.
More IO than the Mini.
Break out Vbat so the backup registers are useable.

On board EEPROM is cool but please consider using FeRAM instead - it's far faster.

Posted 2 years ago #
siy
Member

> On board EEPROM is cool but please consider using FeRAM instead - it's far faster.
FRAM is available in pin-to-pin compatible packages with standard EEPROM 24x (I2C) and 25x (SPI) chips, so there is no need to provide something specific.

By the way, since we're discussing form-factor (at least partially), I'd like opinions in regard to 'DIP48 0.8"' format. To provide pin count comparable to Maple/Maple RET6 regular DIP40 is not enough, it lacks too much pins. DIP48 is much better in this regard.

Posted 2 years ago #
siy
Member

@gbulmer:
> However they use 0603 parts, which I find a real strain.

Well, I'm feeling completely comfortable with these parts. In fact I feel no significant difference between 0603 and 0805 for assembling purposes. 0402 is much more painful. Also, size counts much less if to use reflow for soldering. I'm going to retrofit my reflow oven with STM32-based controller and hope it will not be collecting dust like existing AVR-based one :)

P.S. Also, I think many DIY-er will be interested in this http://www.youtube.com/watch?v=E8KG8cOiOOU&feature=youtu.be
(I'm not the author!)
Full description is provided here http://www.picping.lg.ua/SMD_PLACER/index.htm (in Russian). In short this component placing machine was build completely from the junk found in garage -couple of old printers, couple of old floppy disk drives and scraps of chipboard.

Posted 2 years ago #
gbulmer
Moderator

siy - "Well, I'm feeling completely comfortable with these parts. In fact I feel no significant difference between 0603 and 0805 for assembling purposes."

Yes, however I am getting old and decrepit, and you're young and healthy :-)

My local school used 1206 for most school projects, and IIRC the teacher rarely used 0805 and never used 0603 with his pupils. So I'm avoiding anything smaller than 0805 where practical. I'm trying to make this DIY-able by relatively new makers.

"I'd like opinions in regard to 'DIP48 0.8"' format. To provide pin count comparable to Maple/Maple RET6 regular DIP40 is not enough, it lacks too much pins. DIP48 is much better in this regard."

I strongly agree, 40pin is too few for a 64pin part.

I strongly agree Dual-In-Line (DIL) format is better for prototypes, experiments, and compact 'products' than the Arduino header format. If I want Arduino header format, I'd use an Arduino/Maple/Olimexino...

I'm redesigning my old LQFP-64 board; this is the old one:
https://github.com/gbulmer/openstm32hw/blob/master/piXe-2/Orone-piXe-2-v2r001.pdf
It has 56pins (2x28).
IIRC it surfaces all STM32F pins on the DIL headers except USB D+/D- and the main crystal OSC_IN OSC_OUT.

My current design removes the analogue regulator from the header pins.
One thought is maybe put USB D+/D- onto header pins, if the USB isn't used, there are two more I/O pins available.
Then it would be a small step to use STM32F051's, which don't have USB (and not fit the USB parts).

IMHO 40 pins is okay for LQFP-48 parts.
I'm reworking a Maple-mini format board too (that is actually my higher priority).
It's currently "Maple mini pin compatible" (though my board is 0.7" pitch). Though I'd like to ditch the AV+ pin, and get at the LED pin instead.

When I run Arduino workshops for beginners, we always start with a flashing LED on a breadboard, so having the on-board LED pin appear on a header would be nice. Also the flashing LED pin is a very helpful debugging aid.

Posted 2 years ago #
siy
Member

> I'm redesigning my old LQFP-64 board; this is the old one:
> https://github.com/gbulmer/openstm32hw/blob/master/piXe-2/Orone-piXe-2-v2r001.pdf
> It has 56pins (2x28).
> IIRC it surfaces all STM32F pins on the DIL headers except USB D+/D- and the main crystal
> OSC_IN OSC_OUT.

Seems I've designed very similar board :) I've also discovered that diagonal chip placement has no sense in this case (diagonal placement causes a lot of headache for layout of Maple Mini, for example).
One suggestion which might help: try to shift few pins from "christal side" of chip to bottom header and pins from "USB side" to top header. This will make all connections shorter and simplify layout for USB circuit.

Some other considerations: I've decided to move BUT from PC9 to PC13. PC13-PC15 pins are special - they are powered via internal power switch and have limited output capability and their clock is limited to 2MHz. While for PC14 and PC15 there is a useful application (OSC32 IN/OUT), PC13 just takes space in most cases. By moving BUT from PC9 to PC13 I'm enabling one more 5V-tolerant PWM capable pin (PC9) for use in user applications. This also significantly simplifies layout. One drawback of this approach is that changes in bootloader and board definition are necessary, but I've already done that (along with board definitions for IDE). I'm going to submit appropriate changes into main repo once finish prototyping and pin numbering finally settles.

Posted 2 years ago #
gbulmer
Moderator

siy - Sorry for brief reply, I'm not feeling well.

The power pins on my later LQFP64 boards are all swapped between top and bottom header in the way you've suggested.

As you have explained, moving BUT/BOOT0 from PC9 is an improvement. I'd like to move BUT/BOOT0 from PC9 too.

Moving BUT to PC13 has a lot going for it.
However PC13 has unique functions tied to it, i.e. there is no other pin with PC13's functionality:
- PC13 can output a Real Time Clock (RTC) alarm, RTC periodic alarm, or an RTC clock calibration signal. These might reduces its value as a BUT input,
- PC13 can be used as a 'tamper' input which can wake the chip, and wipe the 'backup registers'. Further in this mode, it can block software access to the backup registers too, though AFAIK that can be avoided through proper programming.

I'd like to understand if the community can reach agreement on changes.
Benefits of agreement could include: board pins and software can be compatible, and fewer variants of bootloader.
I think it's better to continue development than wait for agreement :-)

siy - http://forums.leaflabs.com/topic.php?id=1303&page=11#post-23468
I am hoping we'll move discussion of new boards off that thread (which is a different subject) to here. So I'll respond here.

I was told by ST support the STM32F3 USB is pretty much unchanged from STM32F103. So yes, LeafLabs bootloader should port with minimal effort.

However, the STM32F3 has a USB bootloader manufactured into System-ROM, so, AFAIK it doesn't need a LeafLabs bootloader. Further IIRC, using ala42's technique, once an 'ala42-style' program is loaded, booting from RAM can also be supported without changing the state of BOOT pins.

Posted 2 years ago #
ala42
Member

> I was told by ST support the STM32F3 USB is pretty much unchanged from STM32F103.
The leaflabs USB serial code worked on the F3 by just modifying the alternate function setting for the A11/A12 pins, so the USB engine is indeed unchanged.

About the bootloader, the posts are in this thread: http://forums.leaflabs.com/topic.php?id=9464&page=2

Posted 2 years ago #
gbulmer
Moderator

ala42 - Thank you for that information.

So, an STM32F3 board could be a nice improvement over STM32F103 :-)

One important point Siy has raised is boards can be smaller if some of the signals are moved around header pins to make routing simpler. I took this approach on my (unfinished) Dual-In-Line LQFP64 board, where there was no compatibility to maintain, and I also found it makes a significant difference. However moving pins would lose Maple-mini pin compatibility.

So I'd like to ask folks a question:
How important is Maple-mini pin-compatibility? If the choice were between:
a) 0.6" pitch (fits in a 40 pin socket) and losing Maple mini pin-order vs
b) 0.7" pitch and retain Maple mini pin-order,
which would folks prefer?
(NB there is no same size, same pins option, so I hope to learn something:-)

Posted 2 years ago #
tdc218
Member

I'm not married to compatibility with anything so I'm ok with anything that gets me more IO and more ram.

Posted 2 years ago #
feurig
Member

Agreed on the f3 and once the software architecture was worked out you could put it on the mini as well.
Also Crenn any non NIMBY country would do. I just live in the states and think its important to keep the manufacturing where the external costs are realized. Especially in countries where there are labor and environmental standards.

OTOH Give the USA a few more years of its current government (a sad repeat of its previous fraudulent government) I am not sure that fabbing in the states will matter.

Posted 2 years ago #

RSS feed for this topic

12…13 Next »

Reply »

You must log in to post.