Maple-Mini vs. Netburner 5213 Module

LeafLabs Garden » Support » Maple Hardware Support

Maple-Mini vs. Netburner 5213 Module

(98 posts) (15 voices)

Started 5 years ago by liam.ll
Latest reply from liam.ll

« Previous 123…7 Next »

TcLeaf
Member

Thanks gbulmer. One other question. I have not programmed a usb port before. The docs you pointed me to (sorry I didn't see those earlier) seem say the library supports terminal mode at standard serial speeds, which I'm guessing is something like 115,200, or programming mode.

Are there any usb device code examples and libs that enable the the higher speeds? I would like to get above 5Mbps. I have copied some of the text from the link below for reference. Does the current lib only support the slower terminal mode?

"The Maple stm32 microprocessor includes a dedicated USB peripheral which can be configured to act as a general USB slave device with transfer rates up to 12mbps (it unfortunately can not be configured as a host or on-the-go device). By default the peripheral is configured for two uses: first to receive sketch/program uploads from the IDE, and second to emulate a regular serial port for use as a terminal (text read/write).

The emulated terminal is relatively slow and inefficient; it is best for transferring data at regular serial speeds (kilobaud). Users requiring faster data transfer should consider implementing a different communications protocol;"

Posted 5 years ago #
gbulmer
Moderator

I am not a 'Leafblower', and I haven't benchmarked the USB speed myself. Maybe someone else can comment.

I do know that the current USB serial is not optimised for speed.
5mbps of data on a 12mbps USB is tough.

I think I read that FTDI (who make dedicated USB to serial chips) were getting 600kBytes/second on one of the 3 platforms (Windows/Linux/Mac), and that was pretty reasonable.
As you may know, USB is polled by the host PC, so some of the performance is the host PC driver.
What does your application need to do apart from move data at 5mbps?

Posted 5 years ago #
TcLeaf
Member

I wanted to see if a data logging app would be suitable for this part. Read A/D in digital I/O, do some floating point on the A/D readings, and stream it to a PC. Its still in the spec stages, I just thought that going for less than half the rated speed would be a good high end target. It seems a lot of parts with onboard usb don't have the actual software support to do much better than the FTDI usb to serial, which I have used before. A speed of 115,200 is way too slow, and there isn't enough onboard memory on this part to buffer the data.

Posted 5 years ago #
gbulmer
Moderator

I have had a bit more google, but still failed to find the USB benchmarks :-(

I don't think I have seen anything over 799kByte/second over a Full Speed USB for any OS using things like the Full Speed FTDI. But I haven't been digging very hard.

If you dig around this forum, you'll find some work on data logging using an FTDI ft2232h, which is a USB High-Speed 480Mbits device. A Maple can talk to it over 8 parallel pins. FTDI claim 8MBytes transfer, with 25MBytes possible.

If you have access to PCB making, then youcould do pretty much anything, but if not, and if it's within budget, there are modules with that chip that you could, e.g. http://uk.farnell.com/ftdi/ft2232hq-mini-module/module-usb-2-port-ft2232h-based/dp/1697465

Posted 5 years ago #
liam.ll
Member

I made up my mind between Netburner's MOD5213 and the Maple Mini. Among other reasons, I'm going with the Maple Mini because the price is right and it does not require a developer's kit. Also, I'm very impressed with the community forum. So I'm willing to wait a bit for the Maple Mini. Is it still on schedule for mid January, 2011?

Posted 5 years ago #
mbolivar
LeafLabs

gbulmer:

Someone was working on printf, but I don't know where that is.

some more details on printf(): it's available, with a little work.

if you want it, you'll first need to edit the libmaple/libmaple/syscalls.c implementation of putch():
https://github.com/leaflabs/libmaple/blob/master/libmaple/syscalls.c#L94

for example, if you #include usart.h and replace putch() with

void putch(unsigned char c) {
usart_putc(2, c);
}

then printf() will do its job over Serial2 (you will of course need to set up the USART using Serial2.begin()). similarly, usart_putc(1, c) would work over Serial1, etc. this approach is not likely to solve the problems related to high-speed transmission over USB. we haven't put a lot of effort into getting high speed USB communications yet, so you're stuck with the ST manuals for now.

liam.ll: yes, we are on schedule for a mid-January 2011 release. we have ported libmaple to work with our existing prototypes, and the hardware has passed our tests. still to do before release:

- make some straightforward hardware changes (to correct some design errors we discovered)
- spin another prototype to make sure everything works.

glad to hear you're going with the mini! we're excited about releasing it.

Posted 5 years ago #
liam.ll
Member

Regarding the mid-January 2001 release, by now the hardware tweaks have been corrected hopefully. Has another prototype been created? Do you have a photo? Thanks.

Posted 5 years ago #
liam.ll
Member

Thanks for the 12/16 Mini update in the blog. Glad you finally released the documentation too.

Posted 5 years ago #
mbolivar
LeafLabs

hi liam.ll,

thanks for the bump. as i mentioned in another post, we've been slowed down over the past few days, due in part to staff illness (we're a small shop). as a result, our target schedule for having manufacturer spin what we expect to be the final prototype before we began the main run, which was the end of last week, has slipped. while we're doing our best, it's starting to look like the "mid" part of our january release is less likely to happen.

hopefully later in january is still in time for your purposes. however, if that's too late, and you decide to go with another platform, then we're sorry to see you leave, and hope you'll keep us in mind in the future.

Posted 5 years ago #
mabl
Member

Hi mbolivar,

thx for the info. I hope the stuff gets well soon! Does this also mean the Native will be pushed back? I could really use the additional memory..

Posted 5 years ago #
liam.ll
Member

mbolivar -- thanks for the update. Setbacks are a normal part of business and its how you respond to them that counts. Notifying your customers ahead of time is good practice, and its even better practice laying out how you plan to remedy the setback. I don't plan on moving on to Netburner yet - the development kit requirement is a turnoff. Now, I have a technical question regarding the Maple Mini. Did you have issues with noise and crosstalk between inputs & outputs? Thanks.

Posted 5 years ago #
okie
Administrator

liam.ll,

The noise and crosstalk between the ADC pins and the non-ADC pins is very low (as is the case with Maple for the 6 isolated analog pins). Maple Mini has separate voltage regulators and all 10 ADC pins are together and isolated from other signals. I measured the standard deviation of thousands of readings on the ADC pins with USB running and providing the power, with the pins connected to a buffered voltage divider referenced to and powered by VDDA, to be right under 1-bit using the default libmaple analogRead, which is slightly lower than Maple-r3 and r5. Of course, you will see crosstalk on a neighboring ADC pin if you crank up SPI1, USART2, or one of the PWMs that are also analog pins; that's inevitable. For the best analog performance, don't use analog pins as fast-changing digital outputs.

Posted 5 years ago #
ianmga
Member

What's the timeline for the mini?

Posted 5 years ago #
liam.ll
Member

While browsing the wiki documentation (much better) for the Mini, I was dismayed that I could not find anything on the C++ "struct" type. Does the Maple language support it? Also, will the Mini documentation be ready before or after releasing the module? Thanks.

Posted 5 years ago #
gbulmer
Moderator

liam.ll - The Maple 'language' (like the Arduino 'language') is:
- C++
- with very useful libraries, declarations and headers, and
- some text processing to build (forward) declarations of all the functions in the project, and remove the need to do #includes.

Have a look here:
http://leaflabs.com/docs/language.html#note-for-c-c-hackers

So the omission of 'struct' is caused by LeafLabs trying to write yet-another-manual-for C++, and forgetting some stuff.

You can download the Maple IDE (it is Open Source) and try it out.
Verifying that a program is legal code works without having a Maple.

(I am not LeafLabs employee)

Posted 5 years ago #

RSS feed for this topic

« Previous 123…7 Next »

Reply »

You must log in to post.