42 replies to this topic

[lololol]

Good morning,

 

the kl25z are very expensive now in europe, and in china, the price is 59€ (60 dollars) and there are even shops that require a minimum purchase of 200 dollars for a single kl25z.

 

we no longer find stock, is there another alternative to the kl25z? which works very well without any headaches? which is not expensive .

 

thanks 

[Gravy]

I think that manufacturing may be starting up again as an Australian supplier is stating more stock arriving 8/8/2023.

https://au.element14...m-kl25z&gs=true

[mjr]

> is there another alternative to the kl25z? which works very well without any

> headaches? which is not expensive .

 

Assuming you want to run the Pinscape firmware, I'm afraid the answer is no.  There aren't any compatible boards.  If you just want any microcontroller to run your own custom software, on the other hand, there are many options.

 

> I think that manufacturing may be starting up again as an Australian supplier

> is stating more stock arriving 8/8/2023.

 

Right.  Some people have been expressing fears that the current low availability suggests that the KL25Z is going out of production, but I haven't seen any indication of that.  It's still showing as an active part at Mouser and DigiKey and Newark.  Mouser quotes "39 week lead time" with no definite date, but Newark is showing 8/2023.  I wouldn't consider it a sure thing that they'll actually deliver on that date - they've been slipping their predicted Raspberry Pi 4 delivery dates for over two years now with no end in sight.  Hopefully the data they're getting from NXP about the KL25Z dates is more reliable.

For what it's worth, the KL25Z has gone through drought cycles before.  I think that's what this is - just many months worse because of the ongoing pandemic supply chain problems.  I think NXP makes the KL25Z boards the way most semiconductor manufacturers make things - they do a gigantic run every couple of years, filling orders from the warehouse between runs.  If they sell out before the next run, you just have to wait a couple of months.  This time it's been more like twelve months.  Hopefully that won't end up being 24 months or 36 months.

[drForce]

Not to be negative but realistically speaking: , there are two factor at work here:

1) The KL25z is a development board. At 15$ NXP is almost selling them at cost price.(stretching a bit here) By now NXP surely know that the usual stock is sold out faster because the boards are used for consumer goods and not as development platform/prototyping by developers. Perhaps production is part of a larger production run for different development board and automatically flagged, perhaps not. And in this last case, it is not in their interest at all to quickly do a new production run.

2) Above combined with the whole production cycle of the kinetis processors themselves, not being top tech it is most likely being manufactured in a not state of the art Fab. These fabs are overwhelmed because of high demand of all kinds of 'low tech' silicon devices since Covid.  And no investment because that is not that interesting either on low tech fabs.. Only much money to make at the high end and extremely large production runs. And sure sellers like car chips which is a core market for NXP.The rest is simply filler when Fab's have time to spare . And time is not that available at the low tech fabs right now. Only for sure money makers. That simple. 

 

So It might be quite some time before KL25z supplies are back up to usual. If at all. Nxp might have other plans unfortunately. Hope for the best but be realistic. But as MJR states, the board has seen drought periods before . So we might be lucky.

Edited by drForce, 29 May 2023 - 11:23 PM.

[boriskarloff]

I was looking for an alternative.

For example, if we need only get the info when the player pushs the cabinet, what can we use?

Because about the software maybe it is not so difficult to write it.

Thanks.

Edited by boriskarloff, 23 June 2023 - 03:05 PM.

[rickh]

It's July 11th and regular sources showed no stock for the FRDM-KL25Z.   Appears to be a 11-15 week lead time.   Last year I planned to code a RP Pico to use some external DIO and a gyroscope for a Pinscape alternative.   However, the cost of hardware and PCBA development turned out to be more expensive than a $17 USD FRDM-KL25Z.   I might pick this up again, as this is not the first time I ran into this availability problem and units available are being sold by scalpers at a much steeper price.    

[mjr]

Last year I planned to code a RP Pico to use some external DIO and a gyroscope for a Pinscape alternative.   However, the cost of hardware and PCBA development turned out to be more expensive than a $17 USD FRDM-KL25Z.

 

Yeah, it's hard to beat the price on the dev boards with a custom build.

 

The Pico is a nice little board, and super cheap, but you might look at the STM32F4DISCOVERY series as a more integrated alternative.  Some of those have an accelerometer built in, and like the KL25Z, they expose most of the GPIOs.  The Pico is stingy with exposed GPIOs.

(Just to be clear, this would be in the context of your porting project.  I don't want to give anyone false hope about just substituting a STM32F4 for a KL25Z.  They're completely incompatible with the existing software since all of the on-board peripherals are different.)

[rickh]

 

Last year I planned to code a RP Pico to use some external DIO and a gyroscope for a Pinscape alternative.   However, the cost of hardware and PCBA development turned out to be more expensive than a $17 USD FRDM-KL25Z.

 

Yeah, it's hard to beat the price on the dev boards with a custom build.

 

The Pico is a nice little board, and super cheap, but you might look at the STM32F4DISCOVERY series as a more integrated alternative.  Some of those have an accelerometer built in, and like the KL25Z, they expose most of the GPIOs.  The Pico is stingy with exposed GPIOs.

(Just to be clear, this would be in the context of your porting project.  I don't want to give anyone false hope about just substituting a STM32F4 for a KL25Z.  They're completely incompatible with the existing software since all of the on-board peripherals are different.)

 

Mike,

 

Thank you for chiming in.  You are always a wealth of information.  Yes, the Pico is indeed very stingy with I/O making it a painful alternative.  However, I paired it up with a MPU-6050 3 axis gyro/ accelerometer and a couple of TLC5947 for 48 channels of 12 deep PWM LED drivers.  Of course it would not be complete without a bunch of old fashion shift registers giving me a ~32 bits of inputs.  Getting back to the Pico, it currently sells for $5 USD and unlike its bigger siblings, large supplies still exist.  I'll need to put it together on a breadboard again and start coding.   Obviously, your help would greatly be appreciated.

 

 

Rick   

   

[mjr]

> Getting back to the Pico, it currently sells for $5 USD and unlike

> its bigger siblings, large supplies still exist.

 

Availability is definitely a big advantage for the Pico for the moment.  I'm still holding out hope that the supply chain will eventually heal and we'll be able to get KL25Zs and many other boards/chips easily again some day, but who knows, zombie apocalypse seems at least equally likely at this point.  Anyway, if you're going to go to all that trouble to add on-board peripherals, what I'd probably do is also add the power peripherals (MOSFET output drivers) on the same board, so that it's an all-in-one solution.  That would easily make the economics favorable against the KL25Z + expansion boards, since the expansion boards are up against the same pricing disadvantages that you would be (small runs with all components sourced at retail prices).

Edited by mjr, 12 July 2023 - 03:00 AM.

[rickh]

> Getting back to the Pico, it currently sells for $5 USD and unlike

> its bigger siblings, large supplies still exist.

 

  Anyway, if you're going to go to all that trouble to add on-board peripherals, what I'd probably do is also add the power peripherals (MOSFET output drivers) on the same board, so that it's an all-in-one solution.  

Mike, I agree 100%.  I recently purchased a large qty of IRF540NS MOSFETs at a ridiculously low price from an electronics surplus that is going out of business.  So adding a bunch of MOSFETs to the board should be fairly painless.  I'll keep you apprised of my progress.        

[rickh]

I completed the first hack of a schematic for my Virtual Pinball Interface board using a RP Pico.   I chose the RP Pico for the following:

• Very new (released in 2020)
• Extremely popular 
• Very powerful, it has a PIO function that runs like a state machine
• The fastest game pad available uses a RP Pico with latency times < 1 ms.  

As Mike mentioned, the I/O is very limited on this this device and it lacks an accelerometer.  These shortcomings caused me to make some compromises.  Regardless, I have the following specs:

• 16 each Button inputs
• 24 each, 12 bit PWM outputs using the TLC5947.  16 each 250mA outputs and 8 each 10A fully isolated outputs.
• 10A outputs can be configured to use pairs of 24V or 12V
• X/Y accelerometer using the Invensense MPU6050 or NXP MMA8652FCR1 (same as the KL25Z)
• 12 bit ADC for Ball Plunger, alternately may include provisions for a I2C VCNL4010.
• Expansion card for additional 24 PWM outputs (16 ea 10A outputs, 8 ea 250mA outputs)

Questions I have:

Should I offer more I/O for inputs on expansion card?

Should I offer 24 high current outputs?  Perhaps I could offer 8 ea 250 mA outputs and 16 ea 10A outputs?

Override switch for night mode that turns all outputs off (BLANK)?

  

Details

I am developing this project using DesignSpark PCB.  I picked this package because I am using a subscription of their mechanical package and this enhanced schematic capture version is part of my subscription.  I can describe this software as 'painfully bad'.  I would have picked Autodesk Eagle, but the subscription is beyond my budget. 

 

The PCBA will be primarily be all surface mount.  The RP Pico has castlelated terminals that simply solder to the main board.  Only the connectors and MPU6050 gyroscope board are through hole.  I have been working with SMT since 1981 and feel comfortable designing and building with this technology.  Furthermore, through hole components are becoming scarce.

 

The MPU6050 is NRND. Rumors around the campfire claim that the MEMS based device is not ROHS compliant and destined to die.  These components are frequently found on a board module and sold to the education/hobby market.  I have read several reports that these boards have an extremely high DOA rate and I suspect that these parts may be factory seconds.  Perhaps this would explain why the modules cost less than the actual chip.  Regardless, I will eventually move to the NXP MMA8652FCR1 when I complete my debugging.     

 

As a way of cost reduction, I am using JST PH 2.0 series of connector headers for all button inputs.  The connectors are small, cheap and prefab cables are plentiful for a variety of switches and cable lengths.  The medium and high current outputs use a two pin, through-hole, non-connecterized terminal block.  A larger compliance 2-pin connector will be used for power.  The board has provisions for 3 sets of power inputs that compliment the power requirements with all outputs at 50% of full rating. 

 

I contemplated on using a simple 5V TTL output to drive my IRF540 MOSFETS, as this could save me a dollar or two, but I realized that using an opto-coupler allows a higher gate voltage, saturating the MOSFET guarantying it will not operate in the linear region, hence it will run cooler.  Last, the output of the TLC5947 had constant current outputs that was perfect for driving an opto-coupler.   I made the MOSFETs totally isolated from the logic power and signals.  I created a bank of four MOSFETS with jumpers for either 12V or 24V.  This should come in handy when activating a 12V knocker using 24V.  It gives it an extra kick.

  

I have most of the circuit on a SK10 breadboard and will start playing with the software shortly.  Please let me know what you think. 

 

  Regards,

 

Rick

Edited by rickh, 25 July 2023 - 07:54 PM.

[mjr]

That all looks pretty good to me.  One change I'd make is to add another shifter chip to get to 32 button inputs - 16 really isn't enough for a fully decked-out cab.  The NXP accelerometer is a solid choice, judging by its excellent performance on the FRDM-KLZ25Z boards.

 

It sounds like you're planning to provide common flyback diodes (your "banks" of MOSFETs at common voltages).  The Pinscape expansion boards didn't do that - they just have open-drain MOSFETs and leave it up to the builder to add a flyback diode per device.  I find that design a lot more flexible for a pin cab because it lets you use heterogenous supply voltages without having to carefully plan out blocks of devices at common voltages.  Ultimately I guess it doesn't matter in that you can always add your own flyback diode for a device at a higher voltage than the common, but a lot of people find that confusing - that was a frequent question/point of confusion back when LedWiz's were the top device here.  It's really easy to understand the perfectly uniform rules for per-device diodes.

 

Sorry to hear about your design software woes.  I've tried a couple of those vendor-supplied packages just to see if they could replace EAGLE and was left with a clear No.  EAGLE does still have their free version, I believe, although the board size limitation is probably too restrictive for this project.

 

You should look into PCBA services if you're going all surface-mount.  It would be especially good to look into that early because a lot of the PCBA vendors have their own parts lists that get you much better pricing than parts they'd have to source from Mouser/DigiKey.  Using as many of their generic parts as possible would greatly reduce the cost of the final product.  The downside is that you have to select a vendor before the design phase.  I've never bothered to do that with any of my projects because I didn't want to be locked into a vendor, but if you're planning to produce these at any kind of scale, it could make the economics work out a lot better.

As for "how many outputs", the more the better.  I'd go with at least 32 high-current outputs if possible.  At least provide a way to daisy-chain expansion cards, if you can't fit them on a single board.

Night mode switch: no, do it in software.

[rickh]

That all looks pretty good to me.  One change I'd make is to add another shifter chip to get to 32 button inputs - 16 really isn't enough for a fully decked-out cab.  The NXP accelerometer is a solid choice, judging by its excellent performance on the FRDM-KLZ25Z boards.

 

It sounds like you're planning to provide common flyback diodes (your "banks" of MOSFETs at common voltages).  The Pinscape expansion boards didn't do that - they just have open-drain MOSFETs and leave it up to the builder to add a flyback diode per device.  I find that design a lot more flexible for a pin cab because it lets you use heterogenous supply voltages without having to carefully plan out blocks of devices at common voltages.  Ultimately I guess it doesn't matter in that you can always add your own flyback diode for a device at a higher voltage than the common, but a lot of people find that confusing - that was a frequent question/point of confusion back when LedWiz's were the top device here.  It's really easy to understand the perfectly uniform rules for per-device diodes.

 

Sorry to hear about your design software woes.  I've tried a couple of those vendor-supplied packages just to see if they could replace EAGLE and was left with a clear No.  EAGLE does still have their free version, I believe, although the board size limitation is probably too restrictive for this project.

 

You should look into PCBA services if you're going all surface-mount.  It would be especially good to look into that early because a lot of the PCBA vendors have their own parts lists that get you much better pricing than parts they'd have to source from Mouser/DigiKey.  Using as many of their generic parts as possible would greatly reduce the cost of the final product.  The downside is that you have to select a vendor before the design phase.  I've never bothered to do that with any of my projects because I didn't want to be locked into a vendor, but if you're planning to produce these at any kind of scale, it could make the economics work out a lot better.

As for "how many outputs", the more the better.  I'd go with at least 32 high-current outputs if possible.  At least provide a way to daisy-chain expansion cards, if you can't fit them on a single board.

Night mode switch: no, do it in software.

Mike,

Thanks for chiming in.  As always, you are a wealth of information.  I haven't decided if I should use a SPI based microchip MCP23S17 bus expander (16 I/O) or a I2C TI TCA6424A bus expander (28 I/O).  I have both buses available with another I/O I can use as an INT for either choice.  I agree with the accelerometer, NXP are excellent products.  I'm curious, what is the sample rate you are sending out to the USB vs the sample rate used by VPX for the X/Y axis?  

 

As for the MOSFET drivers, I discovered that the opto-couplers was the easiest way of driving the MOSFET.  The completed design looked almost identical to yours with differences in driving current to the opto-couplers and the emitter current  that drives the gate of the MOSFET.  I initially placed flyback diodes at each output, but since removed them.  I think that a flyback diode should always be placed near the coil to reduce any noise (pops) and the MOSFET already has a flyback/zener diode.  

 

I agree, Eagle CAD is awesome, but now that they charge $70/month I can't justify it.  I'm just whining because I was comfy with OrCAD for several years and it is like driving a Yugo after owning a Mercedes. 

 

The information provided about turnkey PCBA manufacturing is a great thing to know.  I was wondering how Mike from Oak Micro could sells those boards below component cost. If all possible, I want to keep this project made in USA, I don't anticipate high quantities for this product.  

 

As for 32 bits of high current outputs, I think that is a capital idea.  

 

Again, thank you for spending the time offering your feedback and expert advice.

 

Regards,

 

Rick

Edited by rickh, 26 July 2023 - 10:04 PM.

[mjr]

> I'm curious, what is the sample rate you are sending out to the USB vs the sample

> rate used by VPX for the X/Y axis?  

 

The device needs to send at about 5 to 10 ms intervals for good results.  VP uses DirectInput to read the joystick interface, so VP's polling rate doesn't ultimately matter; what matters is the DirectInput polling rate. That's usually 10ms.  So the device needs to send updates at least that fast.  The device actually doesn't have any control over that, as USB HID uses a polling model driven on the host side, so all you can do on the device side is make updates available when they're read.  If you're going to write your own software, I'd recommend making the update frequency user-configurable, because the optimal rate probably varies across machines.  If you send updates too rapidly, DirectInput will just drop the extras, and if you send too slowly, DirectInput will stutter the inputs.  If you're too far off in either direction the DirectInput stream will start looking a lot like random numbers.  It's a bad design on the Windows side (which is probably part of why Microsoft obsoleted DirectInput a long time ago) but we're stuck with it for now.

[rickh]

> I'm curious, what is the sample rate you are sending out to the USB vs the sample

> rate used by VPX for the X/Y axis?  

 

The device needs to send at about 5 to 10 ms intervals for good results.  VP uses DirectInput to read the joystick interface, so VP's polling rate doesn't ultimately matter; what matters is the DirectInput polling rate. That's usually 10ms.  So the device needs to send updates at least that fast.  The device actually doesn't have any control over that, as USB HID uses a polling model driven on the host side, so all you can do on the device side is make updates available when they're read.  If you're going to write your own software, I'd recommend making the update frequency user-configurable, because the optimal rate probably varies across machines.  If you send updates too rapidly, DirectInput will just drop the extras, and if you send too slowly, DirectInput will stutter the inputs.  If you're too far off in either direction the DirectInput stream will start looking a lot like random numbers.  It's a bad design on the Windows side (which is probably part of why Microsoft obsoleted DirectInput a long time ago) but we're stuck with it for now.

Wow!!!! I sold a Virtual Pinball machine a few years ago to someone who complained that the flipper response time had a lag.  He claimed that it happened on all tables.  I suspected  it might be the game controller (I was using a KLZ25Z with pinscape). During my research I concluded that if the buttons had a response time under 8ms (120Hz video refresh) it would not be an issue.  However, if this response time is dependent on DirectInput this might explain the lag.  I assume this is true with both VP9.x and VPX?  

[Zuxif]

Surprised no one has posted a link to the great virtual pinball parts seller Phil at Cleveland Software Design. He actually just recently released his newest item and it sounds like exactly what you are looking for. It looks like it has great potential and will be much cheaper as well. Still is modable and can be the brains for your entire build with expansion boards if you like, just like the kl25z. It has built in outputs for plunger as well and built in gyroscope for nudge. It is using Arduino he says to power it. It's also fully compatible with DOF and such. I've ordered many things over the last year from him when finishing my cabinet build. He makes very good quality parts that are well priced and built to last. He has a wide selection of toys, lights, motors, solenoids, and tons of various boards and expansion boards. He is also constantly improving all the various products and adding new ones all the time. I highly recommend him. His website is www.clevelandsoftwaredesign.com

Here is the direct link as well to the board in question I was referring to:
https://www.clevelan...onnection-board

It looks very interesting and promising! If I hadn't already finished my build a few months back I would probably go this route. It looks very efficient and much cheaper overall, and utilizing all of his latest improvements and additions.

[rickh]

Surprised no one has posted a link to the great virtual pinball parts seller Phil at Cleveland Software Design. He actually just recently released his newest item and it sounds like exactly what you are looking for. It looks like it has great potential and will be much cheaper as well. Still is modable and can be the brains for your entire build with expansion boards if you like, just like the kl25z. It has built in outputs for plunger as well and built in gyroscope for nudge. It is using Arduino he says to power it. It's also fully compatible with DOF and such. I've ordered many things over the last year from him when finishing my cabinet build. He makes very good quality parts that are well priced and built to last. He has a wide selection of toys, lights, motors, solenoids, and tons of various boards and expansion boards. He is also constantly improving all the various products and adding new ones all the time. I highly recommend him. His website is www.clevelandsoftwaredesign.com

Here is the direct link as well to the board in question I was referring to:
https://www.clevelan...onnection-board

It looks very interesting and promising! If I hadn't already finished my build a few months back I would probably go this route. It looks very efficient and much cheaper overall, and utilizing all of his latest improvements and additions.

 

Indeed!  I have seen Phil's products on Ebay and have visited his website.  He has a $40 entry level product based on an Arduino atmega32u4 daughter module and also a KL25Z based Pinscape for $85.  His website claims that the configuration setup is easier than Pinscape, is this true?  I also found that Zeb Boards also has a TAO board based on an Arduino daughter module that has 24 LED outputs, 18 High current outputs that can be further expanded.  Prices start ~$165 USD. 

 

As I pointed out earlier, the Invensense MPU 6050 gyroscope/accelerometer is EOL.  This is an amazing device.  I worked at Invensense back in 2004 when they had only a rough prototype, so it is dear to me.  I also worked with it at Fitbit where it was used on one of our health tracker products.  However, the party is coming to an end.  This is why I am looking into the NXP accelerometers.

     

BTW- I used this thread as a marketing requirements opportunity.  My plan is to design and build my own pinball controller that will be used on my cabinets which I plan to sell.  

 

Regards,

 

Rick

[rickh]

After researching accelerometers, I may have discovered why the FRDM KL25Z is in short supply.  The onboard accelerometer, MMA8450Q is now EOL.  NXP may simply need a drop-in replacement (a BOM change), or it may require a re-spin of the board.   I noticed that some distributors claim that they will have limited quantities of the FDRM KL25Z later this month, but full quantities don't appear until the middle of next year. 

As for my new board design, I have completed the schematic and board layout, but I too will need to look for a replacement for the MMA8450Q, as I used this on my board.  I'll have more info on this later.

BTW-  I was investigating what it would take to reproduce a scaled down version of the KL25Z and the parts alone would cost twice as much as the standard $18 for the premade assembly.   

[rickh]

I have completed my hardware design for my pinball interface.  I have 16 ea. button inputs, 8 ea. 150mA LED drivers (PWM controlled),  16 ea. High current output drivers (also PWM controlled).  I changed the NXP accelerometer to a ST microelectronics (LIS2DHTR).  I am currently using a TI LED controller chip (TLC594) that features 24 each 12 bit PWM channels.  I am contemplating on switching to NXP version that is 1/2 the price of the TI, but only has 8 bit PWM.  Anyway, below is an image of the top and bottom copper and silkscreen layer of the PCB.  Let me know what you think.

Album:

images

comments

Edited by rickh, 16 August 2023 - 03:34 AM.

[htamas]

I think if you manage to get it to market with the described feature set for the mentioned price, you'll have a winner.

Looks like a great job so far.

Edited by htamas, 16 August 2023 - 02:22 AM.