That looks great Luis And great to hear that Lucky is on board too! This could be THE go-to board for coloring your own DMD, with all the bells and whistles on board!

looks great, awesome but the soldering looks a bit to hot here and there.
But that is a hand made prototype i think.

Shouldn't have to do any soldering at all.

shut-up-and-take-my-money_(resized).jpg

Wouldn't it be a good idea to only offer the more powerful processors? Or THE most powerful one. I assume we're only talking a few Euros more, and who knows what future features you or others may want or need to add? Plus with one SKU support and problem solving and testing are easier, and there's no segmentation and complete feature and performance parity. Standardisation makes things much easier in getting this off the ground as a standard and offering a competitive product, and really removes potential confusion for buyers.

There's no reason for it not to become the go to solution full stop. It's far more elegant and featured than ColorDMD or SmartDMD ever were. The community just needs to get behind it, and it's great that despite initial enmity, lucky1 and Luis are now collaborating.

Not only would we then have a better solution than anything now available, but it'd also force the market leaders to move forward and hopefully innovate and lower prices rather than hiding behind assinine patents that will never cover most of the world and would be unlikely to retain validity if tested in the home market.

Im in for a pair of these at least.
maybe 6!!!!

Exactly what I have said. For the cost of say £10/$14/€14, all of the boards would be the same (faster processor, Wifi etc) making support MUCH easier, and also firmware more compatible, since there wouldn't be different releases for different builds etc.

And yes, a much more refined and completed version for everyone

Ok, thanks Jörg. Good if you have already wifi source code for this ESP-12.
Im testing boards now, as soon as all is ok, I will send three boards to Dominik, then he give two for you.

Yes, these are prototypes, but even so with an ultrasonic cleaner, board may look perfect. At any rate I do not solder it by hand, but in an oven reflow, and resins need to be cleaned with ultrasounds to offer a good look board though work perfect too.

Yes, probably all boards will install F427 microcontroller, because of I have get a very good price to order it in quantity. Price difference between F407 and F427 is only 1.3 USD, so I think will be better install only F427. At any rate I must test it yet to confirm its 100% compatible.

There is also a more powerful solution with F745 or F746 microcontrollers, but higher price difference. These are a very fast microcontrollers ARM 216 Mhz, 1Mb Flash Memory ,320K Ram. May be for future special features that require very fast and powerful processor. Current PCB may install also these microcontrollers without modify hardware.

How much more are the 745 or 746? Probably not that much more? I can't stress how much better it would be to have one standardised platform, and one which will be future proof for quite some time. If you can forsee a situation where the extra speed and resources could be useful, then it's best to go for it.

F745 not available in China, they order to USA and resell. So price is same that order in Europe, and its double price than F427.

Finally I think I may offer board with two processors, F427 or F745/F746. With F745/F746 may increase price in about 10 Euros, but I must test it yet to see if its full compatible with current microcontrollers.

.

Some more pictures, these about USB.

For this kind of Vertical USB connector, I suggest use this 90 degree cable. Mainly if you must keep it connected for long time, like happen in a virtual pinball, because of a standar USB cable may break the solder points of vertical USB connector.

dmd_st32_008_(resized).jpg
dmd_st32_004_(resized).jpg

€10 isn't going to make anyone less likely to buy it, and if the processor is more powerful and allows more ambitious features to be added to the firmware and supported software, then it's likely to help the project a lot. If availability isn't an issue, then going for the bigger, juicer, slightly more expensive one would be better (745/6); I'd certainly want it, even if there is no pressing need for it right now.

Good idea re: right-angle connector ... I really don't understand why they're not more widely used. Only cables that are commonly used with the 90 degree connector are SATA.

.

I have here one STM32F746 chip, I will assemble one Controller Board with it and will test.

Its a very powerful ST microcontroller, F7 compared with F4, double its performance, 462 DMIPS vs 225 DMIPS. This may be interesting to apply keyframes and colorize with real pinball.

ST32_007_(resized).jpg

stm32_sc1169_(resized).jpg

We like the F746 option

If it offers that option, it sounds like an easy decision to make

You plan on implementing frame colorization support in your firmware? That sounds exciting.

I suspected about that, if some software timings are calculated according with clock frequency, if modify clock will need modify source code or display will not work properly.

So for current firmware must keep same clock for F427 and F7 though microcontrollers may run faster until create new firmware version specially for F427 and F7 with properly timings and faster clock (180Mhz and 216Mhz).

Unless it's a huge amount of work to adapt the firmware(s), then I really do think it's unwise to start on one and later move to another, especially when the latter option is available now.

I'm following this at a glance and don't entirely understand the power options but would I be able to power this board and/or the LED panels using Coyote's power supply board? I take it his board is only used to power the panels, right? If I need a separate power supply for the PinballSP 5v that's not a problem for me.

https://pinside.com/pinball/forum/topic/rgb-led-panels-for-dmd-replacement/page/19#post-3056895

i can answer this part, at least - I use it to power both the LED panels and the STM board and shield. (It powers everything for the display.)

I'd be curious how much more amperage this board would take with the stronger processor, wifi, etc. That may push the whole system over 4a.

4A is enough to provide power supply to two leds panels ??. Currently I install a 5v 10A power supply, and seem everybody suggest minimal 6-7 Amperes. Chinese provider replied to me that each panel need about 3-4 Amperes, so two panels will need 6-8 Amperes.

Microcontroller need a few mA, probably about 60-80mA, and WIFI about 300-400mA but only when active, can dissabled it from microcontroller with an AT command. Therefore when the controller board works to get the animations on the display, only need about 60-80mA. When microcontroller run faster will need a little more mA, but not much more.

In the test I ran - full brightness, white color, on TZ over two 30-min games, the amperage for EVERYTHING (both panels plus controller boards), there was never more than 2.3A drawn. (2.28a, I believe, but I rounded up.) This included LITZ mode on one game, and powerball on both. (Heavy animation, bright lit, etc.) This did not have any frame-keying done, no palette switching, etc. And, in case it matters, I'm also using the PinLED display driver board, which Lucky can attest, uses a different timing mode than normal WPCs. I doubt that a different timing mode, however, can over double the amps drawn.

Well that's an interesting find. Is that peak power usage or average, I assume peak? I think we need more info, maybe do a display test and see how much the inverted row/column tests draw maximum at 100% brightness over the length of the tests. I'm not saying your findings aren't accurate, but more information/sample data to solidify this would be great. At least we have some real world measurements to go on now.

That's a pretty small number. Especially if your firmware is going to be just one of several used on this hardware.

Very interesting, I never thought that two led panels may require 10 Amperes, do not understand why all I have read suggest install a 10A power supply, when may work perfectly with a 5A power supply, that is cheaper.

I want test it, with ALL leds ON, probably this never happen with a real game, but in theory its possible all leds ON, and this will be the real maximum amperage that may require the led panels. I will modify firmware to run a test that put all LEDS ON to know about it.

At any rate, I have checked price difference between 5A and 10A commercial power supply and its only 2 Euros en Europe. And price from China for a 10A power supply is only 6 USD, so probably it is not important put 5A or 10A power supply because of price is very similar. I do not know how much is your cost, but if a full assembled power supply 5V 10A has a price of only 6USD, do not know if worth to manufacture and assemble by hand.

This does actually happen in real games. Off the top of my head, Sopranos in Satisfaction mode, flashes the display all on for gun shots.

I did this not for cost, but because I have a strong diversion to having an arcade power supply sitting in the bottom of my cabinet, and having to either plug it in manually or cut into existing wiring. So, the board I made takes the transformer input, which normally plugs into display driver board (for WPC/-S machines), so no wire cutting, no manual plugging in power supplies, and it's all mounted to the back of the display, out of the way.

WPC machines have a display test, and the two screens before the Paging Tests are All On, and All Off. I can run a test tonight and measure.

Edit: Here you go. I even included a picture. This is pin2dmd's "white", at full brightness on v2.07.

White:
IMG_20160331_204905_(resized).jpg
Blue: Interestingly, blue took the least amount of amperage. And sorry for blurriness:
IMG_20160331_204923_(resized).jpg
And my custom deep orange color, which very very closely matches plasma displays:
IMG_20160331_205028_(resized).jpg

I did the other colors, if anyone wants a reading on a color, lemme know.

Blue LEDs are most efficient. How about red and green? Red should consume the most of the 3 and green should be somewhere in the middle.

Okay, I can't find my Red picture. I'll need to re-test that tomorrow. Green:
IMG_20160331_204932_(resized).jpg

Nevermind.

Power consumption depends on the duty cycle and even if its theoretically possible to may drive to panels with 3 or more amps each, this will never happen with pin2dmd in pinballs. These panels are so bright if you increase the duty cycle, you can't play anymore.

Has anyone ordered some 64x64 panels? Luis is 192x64 support in your firmware coming in the near future? I have a dying plasma and would like to buy a kit from you if it's in the plans.

FPGA would be a LOT more expensive most probably ... also, how many people potentially interested in writing software or adding firmware features are familiar with what FPGAs would require as opposed to ARM these days?

Certainly its hard work with FPGA, I have some experience with VHDL and Spartan based boards, but not so easy like work with C/C++ language.

There is an easier option with Raspberry or Beaglebone Black. I have do many developments wih Beaglebone Black, easy to program with QT Creator + GCC and with enough power to work with keyframes to replace frame by frame with new colorized pixel by pixel.

It is even possible to reuse part of the STM32 source code for new development with Beaglebone Black or RPY, both have also SPI and I2C ports, more I/O ports in BBB. At any rate, cheaper with STM32, and may be with F7 will may do advanced colouring.

Yes, I have order 3 panels 64*64 1/32 Scan, hope receive it in about one week, and then will start with firmware tests.

I think that pinout of data connector is same than 160*80 mm led panels, so probably will may connect same controller board, though the frame to fix the board seem has not same holes to fix them with screws.

This weekend Im testing controller board with F405, F407 and F427 MCU.

@pinballsp - Which panels would you recommend?? If you can post a link i'll get a couple ordered (for standard display size 128x32).

FPGA are perfect for this sort of thing they would scan those leds at crazy speeds without breaking a sweat. But as others have mentioned they are far more expensive and complicated to work with. Not sure id want to try and get a usb stack working on one?

One of my pinDMD2 prototypes was a Altera cyclone fpga with a high speed FTDI for usb coms. The thing was amazing it could support frame rates upto 120fps no flickering or tearing. BUT the BOM cost per board was insane far cheaper just to use the stm micro controller which did a good job at all tasks for quater of the cost.

I would like to play with the new fpgas that have the arm processer built in. You could have all ur control code there and usb coms. And just hsve a led scan module in the fpga side which only job is to scan the leds. Still cost a bit but far easier to work with. You can also get arm softcores but think you have to pay licence if using comercially?

Most of led panels providers are chinese available in Aliexpress. Seem most of people order to Angela but I received several defectives and she is not so cheap like another providers.

And avoid the 64*64 1/16 scan because of though may split the led panels in 2 x 64*32, can NOT split the frame and if try to cut it the look is very poor.

Cheaper than FPGA are SOC like Raspberry or Beaglebone Black. And about MCU core for FPGA I think also its expensive, or may you try to develop your own MCU core.

I think the cheap and faster alternative to microcontrollers are Raspberry and Beaglebone Black SOC boards. Current Raspberry 3 seem very powerful, but even with before versions sure may run a very fast system to apply keyframes and advanced colouring.

May be I will try to do something with Beaglebone Black. Or try if STM32F746 has enough power to do it too, so keep current hardware, simply replace MCU and improve software.

Raspberry Pi 3 ought to be more than powerful enough to do anything you wanted now, or in future. Also, as it's RP, its availability is ensured for ages and is likely to fall in cost more quickly than other options. If it has all the features needed, maybe it's the best option?

IMO I would prefer Luis's clean single board solution that is running software specifically designed for it. I really don't like Linux-based solutions for basic applications like this and it takes time to boot, you have to be careful with the SD card, etc.

Yes, you are right, all SOC boards like RPY are slow to boot.

I hope that my controller board with an F746 MCU has enough power to work with keyframes and pixel by pixel colorized pictures. And to store new colorized pictures to replace original, may use an external parallel eprom to get fast access, so may provide a cheap and powerful system to work with full and detailed colorized animations.

Also I want to check the Chrome-ART Accelerator and DMA2D system that include F427 and higher ST references, to control directly a TFT display, or even if may use it to improve LED panels control for advanced colouring.

At any rate a solution based on microcontroller, always will be cheaper than FPGA and faster to boot than SOC boards. I think with Cortex M7 microcontrollers like F746, F745, F756, F767, will may do a cheap and very powerful system.

Does your firmware exhibit color palette ghosting/trailing between planes during scrolling animations? DMD Extender and SmartDMD have this issue and both people told me it was timing issues with using an LCD display and these dim planes are built in to the display signal. I have confirmed this in PinMAME. These artifacts are not visible with plasma displays. I am not sure if this is because they are single color or because of the image retention/persistence of the plasma dots and how they handle brightness modulation.

image_(resized).png
image_(resized).png

Looks like timing issues?

Doesn't Extender and SmartDMD both have to grab all the sub frames from the dmd signal then merge them back into the full colour frame? It looks like its done this merging wrong basically grabbed some sub frames from one full colour animation frame and another from the next full colour animation frame. In a perfect world it should gather (ie for williams) both 1bit sub frames merge together into the 2bit full colour frame for each animation frame. But if it loses track where it is or misses a frame or something it will get out of sync and merge 2 different animation frames.

Looks purely like a timing issue when gathering the frame data from the original dot matrix signals.

That's exactly what I was thinking. It doesn't explain why PinMAME does it though. I opened the Addams Family attract mode in the goDMD animation editor and there are darker/ghosted frames between single frame transitions. I see the same thing in WPCEdit. I'm thinking if the PinballSP firmware reproduces the exact plane clock timing as the original plasma displays it shouldn't be a problem. A lot of it is probably because you're taking something non-standard like 122Hz and downsampling it to 60Hz for an LCD.