Revive an EM pinball with OPP and MPF.

254 views
Skip to first unread message

Benoit PETIT-DEMOULIERE

unread,
Feb 18, 2020, 4:06:50 PM2/18/20
to MPF Users
Hello,
I've summarized my current stage in a pinside topic : https://pinside.com/pinball/forum/topic/trying-to-revive-an-old-em-dealer-s-choice-with-opp-and-mpf#post-5478294

I'd need help from all of you as I'm really noob on plenty of domains involved by this topic :)
Thanks
Ben

Benoit PETIT-DEMOULIERE

unread,
Apr 18, 2020, 3:24:27 AM4/18/20
to MPF Users
Here is the dealer's choice topic, following a long list of questions and answers in another thread : https://groups.google.com/forum/m/#!topic/mpf-users/MCgyK--Hfpk

Let's put things in a correct order.
The story continues here!

Benoit PETIT-DEMOULIERE

unread,
Apr 18, 2020, 5:07:53 AM4/18/20
to MPF Users
STM32 flashed with success. 
VB Trace is cut (do you have some points to test the continuity ?)

Now it's time to configure it.
My first card would allow testing as bare minimal config.

WING0 : Neopixel with data wire of the serial led on PB15
WING1 SOLENOID ( 4 flippers with EOS, dual wounds)
WING2 SOLENOID ( slinghsot left, lsingshot right, ball through, empty)
WING3 INPUTS

I'm not sure about the solenoid config for ball through ?
Are inputs no WING0 available already in MPF ? (seems it was added in 0.54.0-dev 15., right ? is it usable at the moment ?)
Not sure about the color config too, it's just copy paste at the moment.
I could have 30 for inserts, not sure how to config the RGBW 
here is my actual config enclosed

Ben
STM32-card1.py

OPP

unread,
Apr 18, 2020, 7:28:43 AM4/18/20
to MPF Users
Your config file looks correct.  I might have put both solenoid wings on one side because then the card is more "symetric" and doesn't look like there are wings jutting out of both sides of the card.  Programmatically it doesn't matter, but it might make it easier to mount or take up slightly less space.  That config should allow you to run in white wood mode.  Color config is right if you have 8 LEDs.  That is a hard configuration, so the number needs to be correct or you won't be able to control them.  (cobra18t just sets it to a number that he higher than he knows he is going to use).

I just look at the trace and do a source to the pin and make sure there is no continuity.  As mentioned to somebody else, I cut it in two places and actually remove the copper between those points.  (That makes absolutely sure it won't conduct.  I cut too many traces with an xacto knife only to find out, I had sketchy connectivity).

On the ball trough, if the kick isn't forceful enough to serve the ball, increase it for 0x10 (16 ms) to whatever makes it serve reliably.

First get things running in white wood mode.  Then start adding in MPF.  In my opinion, that makes problems easier to isolate.
-H

Benoit PETIT-DEMOULIERE

unread,
Apr 18, 2020, 9:06:46 AM4/18/20
to MPF Users
Great ! 
My elder son and I are currently soldering diodes and cleaning the coils to set that up. 
Might be running this weekend in whitewood mode !

but plenty of other things running, like 3d printing some faceshields ^^

Hugh R Spahr

unread,
Apr 18, 2020, 9:21:25 AM4/18/20
to mpf-...@googlegroups.com
Nice.  It shouldn't be that much wiring, so depending on how much cleanup you have on the solenoids, it sounds like you have a good shot at it even today.  Verify all high voltage wiring twice to make sure you have the connections like you think they are.  Make sure you have the -V of the two power supplies tied together.

Nice that you are printing face shields.  My neighbor has been making them non-stop for the last week and a half.

Benoit PETIT-DEMOULIERE

unread,
Apr 18, 2020, 3:26:51 PM4/18/20
to MPF Users
And...that won't be today ^^
I took my time to do the power supply wiring, and diode soldering.
card is flashed and config is done. wings are soldered
I need to do the wiring to all coils and switches, following your recommendations (Famous Hugh's rules of thumb !!)

Might reach that tomorrow, or later in the week.
I'll keep you posted here.
Have all a good day/night !

Ben

Hugh R Spahr

unread,
Apr 18, 2020, 5:49:40 PM4/18/20
to mpf-...@googlegroups.com
Sounds great.  Have a good night.  Nothing is better than the first day when the pinball machine shows signs of life.  When you can play your first white wood game and show the kids what all their hard work has produced.  Of course you have a bunch of other really nice games, but this one will be something that they helped create.  Hopefully something they will have a hand in designing the rules.  Something they can show off to their school buddies as look what I helped make.

Benoit PETIT-DEMOULIERE

unread,
Apr 20, 2020, 4:25:16 AM4/20/20
to MPF Users
Hello Hugh and folks,

Yesterday, it was wiring time ! 
5V, 24V and their neutral are wired for all coils below the playfield.
I also put wires on flippers buttons and 5V-neutral.
And connected all power units.

So basically, everything was ready and i plugged in ..

and It worked ! directly, without any glitch, i had the 2 slingshots, and 4 flippers working by pressing buttons.
I was quite stunned to have no glitch at all.
That was minutes before the first glitch happenned (as all tinkerers had forseen ^^)
Left lower flipper has began to stay activated, even without the switch input. so i think it's the IRL540 that died, but i don't see why it happened.
Anyway, still some debugging, as through coil is not activated by the drain switch, and i need to raise a bit the kick value.
But It works !

thanks again Hugh for your help, I hope to find some time to move forward to a stable situation and do a proper video of my progress.

ben

Hugh R Spahr

unread,
Apr 20, 2020, 8:24:24 AM4/20/20
to mpf-...@googlegroups.com
Well, that is good-ish.  Think of it as an opportunity to not only teach your kids how to solder, but how to desolder and solder in a new part.  Using the IRL540, those MOSFETs should be rock solid.  There is a possibility (especially looking at the pictures of the coils because they look just like my old camelot coils) that there is a short in a coil.  I also had a single flipper coil that was giving me trouble, and I ended up blowing out 2 MOSFETs before I realized that the coil had a short in it.  You might be able to see that by measuring the resistance of the kick and the hold coils.  Just disconnecting the connector from the solenoid wing, you should be able to measure resistance.  Since you have 4 coils, it should easy to compare their resistance.  My coil showed the correct resistance when it was cold, but only shorted out when it got hot.  I eventually (mostly as a science experiment) by putting in a .1 ohm wire wound resistor and measuring the current and finding the short that way.)  I should have just swapped the coil and not wasted the 2 days looking into it, but it was a learning experience for me.

Second thing that I have had happen is I got some bad IRLs from ebay.  (Yeah, I know, don't buy gray market parts from Ebay, but sometimes they are just so darn cheap!  Especially the shipping when you only want 10 or so parts.).  The MOSFETs could not take the amount of current that they were spec'd for and stopped working.  Buying directly from a reputable supplier (i.e. Mouser or whatever the equivalent in France is.) removed that issue.

So first thing I would do is measure all resistances for your flippers and show they they are approximately the same amount.  (Measure the two coils individually so the kick coil should be something like 4-5 ohms, and the hold coil should be something like 200 ohms.  Note:  I may have the resistance wrong because I'm thinking of a different era pinball machine, so these are just ball park numbers).  Hopefully that won't show an obvious problem because you might not have another dual wound coil sitting around.  Second thing, test your EOS switches are working properly.  This can be a simple visual inspection, and you have probably already done it, but doesn't hurt to do it again.  Third thing, turn down the power supply at least for the first few times.  There is a potentiometer on the power supply that should allow you to dial it back to 18 or so volts.  That will reduce the current and will lessen the chance that another FET gets blown.

It sounds like the wiring is correct.  I would again check that 5V-neutral is connected to 24V-neutral near the power supplies.  That also has the potential to "blow" things up.  This is another good reason to use a phone charger or PC power supply at the beginning to generate the 5V.  You have no chance of messing up a laptop that is supplying the USB voltage in the computer.

I'm sorry to hear about the FET.  Those are a pain to replace.  Well good progress...kinda.
-H

Benoit PETIT-DEMOULIERE

unread,
Apr 20, 2020, 9:04:25 AM4/20/20
to MPF Users
Positive thinking at its best ^^

The coil had worked few times during minutes so i expect more a faulty FET. Plus the coil was fully working when it was still few days ago as an EM, so I'm not suspecting the coil by itself.
I have a dual wound coil lying somewhere, so i think I'll just put it to give a try with replcaing the FET and see how it's going.
it was indeed a chinese cheap FET(#stupidme)

Will keep you post.

jabdoa

unread,
Apr 20, 2020, 1:17:16 PM4/20/20
to MPF Users
Spike has a neat thing to detect bad coils or shorts. They include a shunt on the 48V supply on every board. Then they measure the voltage drop (which is linear to the current). During start-up they measure the current with all coils off then they pulse all coils for a very short time and measure the current. If it is too high they sequentially pulse the coils and find the bad one. If anything goes wrong they will either disable the bad coil (if it is only one) or the complete power (via their power board) to protect the hardware. They also do that with lights (but they also use step down converters which does not apply to OPP).

@Hugh: Maybe we could implement something similar on the power distribution board and share it with all boards? What do you think?


Jan

Benoit Petit-Demoulière

unread,
Apr 20, 2020, 2:11:02 PM4/20/20
to MPF Users, jabdoa

Would be a neat addition to protect hardware.



---- jabdoa a écrit ----


Spike has a neat thing to detect bad coils or shorts. They include a shunt on the 48V supply on every board. Then they measure the voltage drop (which is linear to the current). During start-up they measure the current with all coils off then they pulse all coils for a very short time and measure the current. If it is too high they sequentially pulse the coils and find the bad one. If anything goes wrong they will either disable the bad coil (if it is only one) or the complete power (via their power board) to protect the hardware. They also do that with lights (but they also use step down converters which does not apply to OPP).



@Hugh: Maybe we could implement something similar on the power distribution board and share it with all boards? What do you think?





Jan



--

You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.

To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.

To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/208986f0-383e-40b0-aea4-ba1b33406b10%40googlegroups.com.

MOSFET

unread,
Apr 21, 2020, 9:50:37 AM4/21/20
to MPF Users
I've used the INA168 high-side current sense amplifier in the past, which would work will for this application. I'd be happy to whip up a breakout PCB design with a shunt resistor and INA168 if there's interest! This current-sense board could live by the 48V (or whatever high voltage supply you have, up to 60V DC) power supply and measure all the load current sourced by that supply.

I'm not sure how to get the analog output from the INA168 all the way to the OPP microcontroller without having a really noisy signal at the ADC input from the long cable run. I suppose twisted-pair wire would help somewhat. Any suggestions?


Would be a neat addition to protect hardware.


jabdoa

unread,
Apr 21, 2020, 11:26:44 AM4/21/20
to MPF Users
That would be cool. The analog input does not necessarily have to be on the OPP board itself. Any usb or serial would work for MPF. I would add generic support for current sense input which then can be tied to the PSU in MPF. Then we can orchestrate fault detection and coil protection for all components connected to it.

Jan

MOSFET

unread,
Apr 21, 2020, 11:42:29 AM4/21/20
to MPF Users
Cool. We could take it a step further and have a small microcontroller on the current-sense PCB itself with a USB connection to the control PC. Say, an ATtiny85 or STM32F103 (like OPP uses)?

If we go that route (self-contained, USB-connected current sense), would anyone be interested in collaborating on the firmware side? I don't have firmware experience for USB interfacing, so I'd appreciate help there. :-)

Benoit Petit-Demoulière

unread,
Apr 21, 2020, 12:06:11 PM4/21/20
to MPF Users, MOSFET

Would be a nice wing addition to OPP that could interact with the power filter board, no?




---- MOSFET a écrit ----


Cool. We could take it a step further and have a small microcontroller on the current-sense PCB itself with a USB connection to the control PC. Say, an ATtiny85 or STM32F103 (like OPP uses)?

If we go that route (self-contained, USB-connected current sense), would anyone be interested in collaborating on the firmware side? I don't have firmware experience for USB interfacing, so I'd appreciate help there. :-)

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/9cb0ae75-392f-4fa1-a6a4-1bb815f1ae3b%40googlegroups.com.

jabdoa

unread,
Apr 21, 2020, 1:01:53 PM4/21/20
to MPF Users
That is what I thought also. Just a new wing type. Then we can also just define one output on the same board to control the power enable relay. Since the board would probably sit in the backbox it could then control backbox lights, knocker and such.

Jan

Hugh R Spahr

unread,
Apr 21, 2020, 1:19:04 PM4/21/20
to mpf-...@googlegroups.com
Took too long to hit send and then most of my message has now been discussed by others.

"Maybe we could implement something similar on the power distribution board and share it with all boards? What do you think?"

I don't know how we would do this with the current OPP hardware.  I agree that it would be a great idea, but it would involve adding a processor to the power filter board (and some sort of power sensing resistor).  Right now the power filter board is as dumb as it gets.  It is mostly three big caps, and NFC to keep the power supply from shorting on the inrush current to the caps, and sometimes a PChan FET to enable/disable the high power rail.   It can be controlled by a processor, but there is actually no processor on it.  Even if it existed, it would need a separate op-amp to isolate the ADC from a potential 48V.  (If you have a shorted solenoid and the board is measuring across a known resistor, the voltage has the possibility of reading the whole potential.  Normally you would throw an op-amp to guarantee that the voltage seen by the processor doesn't kill the processor pin.)

@MOSFET brought up all the same points.

That being said, it would be tough at all to add another wing that could support this functionality.  As Jan and others have suggested, it would be a fine addition to the backbox of the pinball machine.  I just have to make sure it doesn't use wing 0 because that is locked in for serial LEDs.  I have no problems doing the code for it.  The STM32 has plenty of space leftover to support this type of addition.  It would really simply be the support for ADC reads which are really easy.  Checking the pinout for the STM32, all the analog pins are on wing 2 or wing 3 anyway, so that shouldn't be a problem.  The analog input would have to be scaled to <3.3V for the STM32.

MOSFET

unread,
Apr 21, 2020, 3:53:56 PM4/21/20
to MPF Users
I split out the current measurement topic into a separate thread https://groups.google.com/forum/#!topic/mpf-users/YRIc-ZR8wS0 to discuss there!

Benoit PETIT-DEMOULIERE

unread,
Apr 22, 2020, 1:30:07 PM4/22/20
to MPF Users
Hello again,

IRL540 changed, one solder corrected for the drain coil and ... everything is working now !!


Played a bunch of (slow) games as I've defined a really 10ms kick, but that works !!
MPF is not running, only OPP as whitewood mode.

I just wonder why my left upper flipper is really slow compared to the 3 other ones.
same coils, mechanisms is perfectly working.
It reacts as is the 'kick' coil was not reacting, only the hold one, but EOS seems ok.
I've put a closer picture of it, if you see any potential defect on it ?

Anyway, kids are happy, so am I.

I've used some spare time to wire additional inputs that will be used forthe MPF part.

Time to move forward

Hugh R Spahr

unread,
Apr 22, 2020, 2:06:26 PM4/22/20
to mpf-...@googlegroups.com
That coil looks like it is a center tap.  (both kick and hold coils common is in the center).  If that is the case, the 24V should go to the center and the to power both coils.  That coil almost seems like it is miswired to me.  I can't see where all the wires go, so I can't see where the EOS is attached to it.  I would fix that before playing it.  Right now the kick coil is attached in series with the hold coil.  Depending on how the EOS is done, current may be doing a strange loop.  Look at the other flippers and you should see what I mean.

OPP

unread,
Apr 22, 2020, 2:13:52 PM4/22/20
to MPF Users
Should have looked closer at that picture.  Your flippers have a normally closed EOS.  Sorry, I was thinking about old Bally machines.  I think you want the smaller wire in the center moved to the right side in your third picture.  That means you use the kick solenoid and the hold solenoid initially, then just the hold solenoid after the kick solenoid is disconnected.

Benoit Petit-Demoulière

unread,
Apr 22, 2020, 2:22:26 PM4/22/20
to MPF Users, OPP

Yes, it's normally closed EOS. They open when kick is done to put the two wound in serial.

All coils are wired in the same way..



---- OPP a écrit ----


Should have looked closer at that picture.  Your flippers have a normally closed EOS.  Sorry, I was thinking about old Bally machines.  I think you want the smaller wire in the center moved to the right side in your third picture.  That means you use the kick solenoid and the hold solenoid initially, then just the hold solenoid after the kick solenoid is disconnected.

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/208880a0-5a82-4217-91fb-6674e07eb3f5%40googlegroups.com.

Benoit PETIT-DEMOULIERE

unread,
Apr 23, 2020, 8:52:49 AM4/23/20
to MPF Users
Hello,

I'm not really sure how to put correct wirings on the flipper coils.
As a summary, these are dual wounds coils, ref FL-20-28 Williams.
I made a basic summary of the config. They are NC (Normally Closed) EOS.

The actual OPP config on the wing is "rs232Intf.CFG_SOL_ON_OFF_USE_SW, '\x00', '\x00'" for them.
I've initially wired the 24V input on C and the neutral was on A going to the solenoid wing.
Diodes were tied between A & B (band on the B side) and B & C (band on the C side).
But that did not give a correct reaction, as shown on the previously uploaded video.
Based on my understanding, the thicker wire (20 Gauge) is the kick coil, so when EOS is closed, 24V goes from through the kick coil (C to B) then through the EOS, giving the maximal amount of current to the kick coil.
When the EOS opens, current would go from C to B through the kick coil and B to A through the hold coil (smaller wire 28Gauge), leading to two coils in serial and a lower global amount of current and heat.

flipper-coil.png


It seems that something is wrong in my understanding, or in the way I'm setting it up.

I referred to this picture http://www.pinrepair.com/em/flipal1.jpg


EOS are giving a zero Ohm when closed as expected.


Can someone give me some clues ?





Hugh R Spahr

unread,
Apr 23, 2020, 10:34:33 AM4/23/20
to mpf-...@googlegroups.com
Your write up is correct.  Disconnect the solenoid connector from the solenoid wing.  Measure the resistance from the 48V to the Molex pin that attaches to the drain of the MOSFET.  It should read something like 5-10 Ohms.  With your hand, physically move the flipper to the up position.  Measure the resistance to the to the drain of the MOSFET.  It should read something like 200-300 Ohms.  Verify that the crimp for this flipper in the Molex looks good.  

Benoit Petit-Demoulière

unread,
Apr 23, 2020, 3:36:49 PM4/23/20
to mpf-...@googlegroups.com, Hugh R Spahr

I've measured coil resistance without any wire or diode.



EOS closed,  B-C = 1 Ohm ; A-B = 0 and A-C = 1

EOS open B-C = 1 A-B= 6 and A-C = 7


My Ohm meter is dead or..?



---- Hugh R Spahr a écrit ----


Your write up is correct.  Disconnect the solenoid connector from the solenoid wing.  Measure the resistance from the 48V to the Molex pin that attaches to the drain of the MOSFET.  It should read something like 5-10 Ohms.  With your hand, physically move the flipper to the up position.  Measure the resistance to the to the drain of the MOSFET.  It should read something like 200-300 Ohms.  Verify that the crimp for this flipper in the Molex looks good.  

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/CAD9bh02vr_5qO0kt762UGugtV6mHYqJBWaRWtg-%3DyoYxj81zow%40mail.gmail.com.

Benoit PETIT-DEMOULIERE

unread,
Apr 23, 2020, 5:14:56 PM4/23/20
to MPF Users
The numbers seems accurate when comparing to the ref table : FL-21-375/28-400

https://www.flippers.com/coil-resistance.html

OPP

unread,
Apr 23, 2020, 5:39:07 PM4/23/20
to MPF Users
Ok, good.  I just measured my SharpeShooter3, and it is in line with that.  (That is my only 24V machine)  Shows that I should not just rely on my memory.  Sorry for the confusion.  That does all seem correct.  Make sure that the molex crimp is good and has a good contact with the wire where it should.  A bad crimp could show those issues.  It is nearly impossible to inspect those crimps once they are put in the housing (and removing them from the housing is nearly impossible).  Also, make sure all the soldering joints are good, but since the wires were removed, it is all going to be new solder anyway.  At this point, you might want to just put it back together and see if it works.  If not it is pointing more towards the crimp, or maybe inspect the solder joints for that MOSFET.  Since the other MOSFETs are working properly, you know the ground connection to the source plain is working properly.

Benoit Petit-Demoulière

unread,
Apr 24, 2020, 2:37:32 AM4/24/20
to MPF Users, OPP

I don't suspect a bad crimp or solder as the four flippers are reacting the same way. They are both on the same wing and I've tested all solders of the wing

I will try to power only the kick coil to see how it reacts, without holding. So wiring the neutral on the B plot and I'll see.



---- OPP a écrit ----

Ok, good.  I just measured my SharpeShooter3, and it is in line with that.  (That is my only 24V machine)  Shows that I should not just rely on my memory.  Sorry for the confusion.  That does all seem correct.  Make sure that the molex crimp is good and has a good contact with the wire where it should.  A bad crimp could show those issues.  It is nearly impossible to inspect those crimps once they are put in the housing (and removing them from the housing is nearly impossible).  Also, make sure all the soldering joints are good, but since the wires were removed, it is all going to be new solder anyway.  At this point, you might want to just put it back together and see if it works.  If not it is pointing more towards the crimp, or maybe inspect the solder joints for that MOSFET.  Since the other MOSFETs are working properly, you know the ground connection to the source plain is working properly.

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/e384e9a1-0a3c-4bb7-ba18-85e9f22645a0%40googlegroups.com.

Benoit PETIT-DEMOULIERE

unread,
Apr 24, 2020, 1:46:55 PM4/24/20
to MPF Users
Putting more than 26V did not lead to a significant, and also killed an IRL540 (twice)

OPP

unread,
Apr 24, 2020, 3:29:30 PM4/24/20
to MPF Users
So an IRL540 is supposed to be able to be able to handle 110A for a pulse, and 28A continuously.  With the kick coil resistance of 1ohm, (so maybe you are getting 30 Amps for a brief moment while the ferrous core pulls in the solenoid), and then 30V/7 ohms or 4.5A continuously, that shouldn't stress that MOSFET at all.  That is also a configuration that I've run on my machine and it is rock solid.  (With running at 28 - 30V).  Since you have the power filter board, that contains an NTC thermistor which minimally adds about .1 ohm of resistance during steady state.  I'm struggling to figure out why the FETs are blowing.

Benoit Petit-Demoulière

unread,
Apr 24, 2020, 3:42:04 PM4/24/20
to MPF Users, OPP

Might be my set of IRL540 that is bad


I gave some try few minutes ago by changing the config to have a kick x10 and pwn 04. I've used the config from one of the python examples.

Then I soldered the neutral on the middle point, to skip completely the hold coil.

Worked a bit then fuse blow and 1 FET died


Changed back to the initial state and changed the FET.

Now, the flippers are activated few times then stop. If I press the reset button on the Stm32, they work again.. I'm lost and a bit fed up for today.


I think I'll set a new stm32 with solenoids.



---- OPP a écrit ----

So an IRL540 is supposed to be able to be able to handle 110A for a pulse, and 28A continuously.  With the kick coil resistance of 1ohm, (so maybe you are getting 30 Amps for a brief moment while the ferrous core pulls in the solenoid), and then 30V/7 ohms or 4.5A continuously, that shouldn't stress that MOSFET at all.  That is also a configuration that I've run on my machine and it is rock solid.  (With running at 28 - 30V).  Since you have the power filter board, that contains an NTC thermistor which minimally adds about .1 ohm of resistance during steady state.  I'm struggling to figure out why the FETs are blowing.

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/a7c7486b-ffa4-4175-ad03-c978c2bc10a0%40googlegroups.com.

Benoit PETIT-DEMOULIERE

unread,
Apr 29, 2020, 7:43:27 AM4/29/20
to MPF Users
Interesting finding..
I have tried to invert my wings config and did put the flipper coils on the second solenoid wing, and bam, two irl died on this wing.
So it's the flipper coils that are hurting the irl540. But, on my first wing, I have 4 of them that are not dying. Maybe I've found the stronger ones..
I've purchased the 1 ohms NTCs and I will put additional wires for the longer wires running through the playfield.

Joe Slatz

unread,
Apr 29, 2020, 8:56:30 AM4/29/20
to MPF Users
Hey,
Just a heads up re cheap aliexpress irl540.
I have gone down the same road with the 10 for $2.50 irls from 2 different vendors.
A waste of time with destruction in the path.
They may work for low power coils but in my experience will fry and subsequently lock coils on.
Solution: Purchase brand name irl540 with specs you require....they do come in different flavours. Expect to pay approx 3$ each. I am satisfied with items from Digikey or Mouser.
Good luck and don't pull too much of your hair out.

Joe

Benoit Petit-Demoulière

unread,
Apr 29, 2020, 9:31:07 AM4/29/20
to MPF Users, Joe Slatz

Amen to this..

Good lesson learnt.


Doubling all long power wires + jumping the NTCs in parallel together was a real improvement. It's nearly good so I expect that the lower R NTCs will solve the issue



---- Joe Slatz a écrit ----
--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/a70af3f6-361f-49b6-92db-aa64e0c21138%40googlegroups.com.

Hugh R Spahr

unread,
Apr 29, 2020, 10:27:06 AM4/29/20
to mpf-...@googlegroups.com
I 100% agree.  I'm a sucker for a good deal, and I went the route of the ebay/ali-express MOSFETs.  Wow, what a pain in the butt to dig myself out of that hole.  The x10 cost was fantastic, but the amount of MOSFETs that blew up, had to be desoldered, etc, was not worth it.  I was so wrong.  I don't even use lower current rated MOSFETs for my pop bumpers and such anymore.  I always use the high current ones because then I don't have to worry.  Just grab one from the tube, and I'm good to go.  On the price of the MOSFETs, on Mouser they are $1.10 each when you buy 10 or more.  IRL540NPBF.  If you are going to be making any pinball machine, you will probably need at least 10, and why not have some spares on hand.  Also, there is almost no time in my life where I can't come up with a stupid idea that takes at least 1 MOSFET.

Benoit PETIT-DEMOULIERE

unread,
May 1, 2020, 4:14:26 PM5/1/20
to MPF Users
Order placed, I should be good for this pinball and the next one!
I'll give a try with the neopixel stuff now, as I have targets and coils working on the first card.
Once these work, I'll put MPF and use a barebone config. Then, once everything works, I'll add a second card..

Benoit PETIT-DEMOULIERE

unread,
Mar 22, 2021, 6:43:40 AM3/22/21
to MPF Users
Nearly 1 year later, project is back on tracks !

I've added a ruleset, created with my older son and additional ideas from some guys here..

I will need any available help for the coding, as we have now a working coils + switches + score reels config.


Happy Ben


Hugh R Spahr

unread,
Mar 22, 2021, 7:20:22 AM3/22/21
to mpf-...@googlegroups.com
Congrats on getting back to it.  You've made the cardinal sin of trying to get back into a pinball project in spring.  Winter...when you are stuck indoors...is when all pinball projects should be conceived and created.  Good luck!

--
You received this message because you are subscribed to a topic in the Google Groups "MPF Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/mpf-users/C6n6ZnNYYAk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to mpf-users+...@googlegroups.com.

Benoit

unread,
Mar 22, 2021, 7:52:20 AM3/22/21
to mpf-...@googlegroups.com, Hugh R Spahr

Hopefully, lockdown will help me!!

In fact, I had other winter projects!



---- Hugh R Spahr a écrit ----

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/CAD9bh02gLH233LWUA-QH%2B7BQ7%2B_BGYkWSAww6dX1YC279hq9nA%40mail.gmail.com.

Benoit PETIT-DEMOULIERE

unread,
Apr 14, 2021, 10:30:31 AM4/14/21
to MPF Users
Hello, I need some help to configure the incandescent lights that are attached to one STM that has also a neopixel.
STM32 serial is 6
wingCfg = [ [ rs232Intf.WING_NEO, rs232Intf.WING_INCAND, rs232Intf.WING_INP, rs232Intf.WING_INCAND ] ]

Wing0 is NEOPIXEL
Wing1 is INCAND

There are 10 GRBW leds on neopixel. 
My current config :
lights:
 l_Bonus_10000:
  channels:
   green:
    - number: 6-0-0
   red:
    - number: 6-0-1
   blue:
    - number: 6-0-2
   white:
    - number: 6-0-3
 l_Bonus_9000:
  channels:
   green:
    - number: 6-0-4
   red:
    - number: 6-0-5
   blue:
    - number: 6-0-6
   white:
    - number: 6-0-7
 l_Bonus_8000:
  channels:
   green:
    - number: 6-0-8
   red:
    - number: 6-0-9
   blue:
    - number: 6-0-10
   white:
    - number: 6-0-11
 l_Bonus_7000:
  channels:
   green:
    - number: 6-0-12
   red:
    - number: 6-0-13
   blue:
    - number: 6-0-14
   white:
    - number: 6-0-15
 l_Bonus_6000:
  channels:
   green:
    - number: 6-0-16
   red:
    - number: 6-0-17
   blue:
    - number: 6-0-18
   white:
    - number: 6-0-19
 l_Bonus_5000:
  channels:
   green:
    - number: 6-0-20
   red:
    - number: 6-0-21
   blue:
    - number: 6-0-22
   white:
    - number: 6-0-23
 l_Bonus_4000:
  channels:
   green:
    - number: 6-0-24
   red:
    - number: 6-0-25
   blue:
    - number: 6-0-26
   white:
    - number: 6-0-27
 l_Bonus_3000:
  channels:
   green:
    - number: 6-0-28
   red:
    - number: 6-0-29
   blue:
    - number: 6-0-30
   white:
    - number: 6-0-31
 l_Bonus_2000:
  channels:
   green:
    - number: 6-0-32
   red:
    - number: 6-0-33
   blue:
    - number: 6-0-34
   white:
    - number: 6-0-35
 l_Bonus_1000:
  channels:
   green:
    - number: 6-0-36
   red:
    - number: 6-0-37
   blue:
    - number: 6-0-38
   white:
    - number: 6-0-39

That works well, but I'm now wondering how i can set the incandescent lights ?
What would be the number of the first and the last incandescent lights that are on WING1 and WING 3 ??


Thanks for your help
best
benoit

Benoit PETIT-DEMOULIERE

unread,
Apr 14, 2021, 11:09:56 AM4/14/21
to MPF Users
ok, I found !

 light_EB_Lane:
    number: 6-0-15
    subtype: matrix

is the latest light on wing1 incand

Benoit PETIT-DEMOULIERE

unread,
Apr 14, 2021, 11:58:03 AM4/14/21
to MPF Users
sadly, it's not enough.
some of the lights stays on whatever the off command i send in MPF.
Hugh, shall I use a INCAND or HI_SIDE INCAND ? not sure abouth this last one.
One bulb is blinking when MPF is not running, is it the standard behaviour ? 
Why some other lights might be not blinking ? defective wing? 

Thanks

Hugh R Spahr

unread,
Apr 14, 2021, 4:20:12 PM4/14/21
to mpf-...@googlegroups.com
rs232Intf.WING_INCAND should be correct for the wing.  (the high side incandescent wing is surface mount and was created to be backwards compatible with old Bally machines that I didn't want to rewire).

When the STM32 starts up (without connecting it to MPF) all bulbs should blink 1s on, 1s off, repeat.  That helps prove out that the wiring is correct.  MPF, I believe, immediately over writes this initial state to stop the blinking.  If all of your lamps aren't blinking there is probably a wiring issue.

I don't know the incandescent numbering scheme in MPF, but Jan probably knows off the top of his head.

Benoit

unread,
Apr 15, 2021, 2:21:36 AM4/15/21
to mpf-...@googlegroups.com, Hugh R Spahr

Thanks Hugh,

So mostly an issue on the wing as some led are not blinking, and stay on.


What's the best way to test the component? Shall I unsolder the wing?

Ben

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/CAD9bh02%3Dg2kRGzjv3verHDhJ6tn_qqS0X0p2DkqNF65Kau29nA%40mail.gmail.com.

cobra18t

unread,
Apr 15, 2021, 3:08:41 AM4/15/21
to MPF Users
Which version of MPF are you running? Are all the lights that are not blinking on one wing? Which one, if so? Which bulb is working?

Can you measure the resistance between drain and source pins on the FETs to confirm if they are good or not?

Wing position 0 contains bulbs 0 to 7. Wing position 1 contains bulbs 8 to 15. Wing position 2 contains bulbs 16 to 23. Wing position 3 contains bulbs 24 to 31.

cobra18t

unread,
Apr 15, 2021, 3:17:06 AM4/15/21
to MPF Users
For the resistance test, red probe on Drain and black probe on Source. If the resistance is high (maybe >100k) or the meter is overloaded, then it is good. If the resistance is low, then the transistor is bad. 

Benoit PETIT-DEMOULIERE

unread,
Apr 15, 2021, 3:41:52 AM4/15/21
to MPF Users
I'm running the 0.54 version.
For my tests, I've only used WING1 with 7 lamps on it.
Among them, 5 are blinking as expected and 2 are staying on and I can't control them via MPF.
I will unplug and test my FETs to determine if it's a component issue.
And about the numbering, the main thing to understand is that that are two lights labelled the same !
Led : 6-0-14 is the blue channel of my 4th led on the neopixel
light : 6-0-14 is the 7th incandescent light on the WING1
the light has to be called via subtype: matrix
This might be interesting to mention in the doc.

Hugh R Spahr

unread,
Apr 15, 2021, 8:36:30 AM4/15/21
to mpf-...@googlegroups.com
For the lamps not blinking, I would first make sure that your crimps are good.  I personally have seen more issues with crimps than with MOSFETs blowing on lamp boards.  Just to be sure, make sure that the MOSFETs are also put in the right orientation.  If you have that wrong it should stand out like a sore thumb because you have six of the lights working.  If it is easy to do, pop out a connection from one of the bulbs that is working, move it to a non-working bulb.  That way you can also easily test the wires.  Those connectors are pretty easy to pop the crimps out of if you have something like a cake tester.  (Yes, after using the cake tester for pinball for many years, I was forced to buy a new cake tester for my wife so we have one cake tester for baking, and one cake tester for pinball depinning.)

If the lights don't blink in white wood mode, they certainly won't blink in MPF.  White wood mode, it is your debugging/wiring issue finding friend.

Benoit PETIT-DEMOULIERE

unread,
Apr 15, 2021, 9:07:28 AM4/15/21
to MPF Users
Thanks Hugh.
With bad crimps, I'd suspect lamps to stay off, no ? How a bad crimp would lead to a non blinking light ?
Anyway, that will be my next debugging steps.
Will keep you posted :)

Hugh R Spahr

unread,
Apr 15, 2021, 9:59:29 AM4/15/21
to mpf-...@googlegroups.com
You are right.  The lamp would just never work if it was a bad crimp.  If it is staying on, could be wiring or could be the connection between the wing and the STM32.  Swapping the pin could help isolate a wiring issue because the lamp that stays on would track with a single lamp.  (i.e. move it from pos 1 to pos 3, now pos 3 stays on all the time).  If pos 1 still stays on even after swapping, that points to the MOSFET or connection between wing and STM32.

cobra18t

unread,
Apr 15, 2021, 4:25:01 PM4/15/21
to MPF Users
As of MPF 0.55 dev.22, the "matrix" subtype was stolen for use with an actual light matrix. An "incand" subtype was added to handle incandescent wings. This should not affect you since you are using 0.54, but I wanted to give you a heads up.

It is a good note about overloading the hardware numbers given the different subtypes. We should add that information to the Docs when we add the info about the light matrix and new subtype.

The transistors being installed backward would definitely explain this behavior since current would just flow through the body diode.

Benoit

unread,
Apr 15, 2021, 4:27:21 PM4/15/21
to MPF Users, cobra18t

Light matrix might interest me as well.

Has the doc been updated about wiring it?

Feasible with the actual incand opp design??



---- cobra18t a écrit ----
To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/b5f15b30-e531-4a3c-b1d2-39f6131148fbn%40googlegroups.com.

cobra18t

unread,
Apr 15, 2021, 4:49:13 PM4/15/21
to MPF Users
The light matrix requires a plank that has not been released yet, but it is sitting here next to my desk. It gives you an 8x8 matrix. Like any matrix, it requires diodes at each bulb. No documentation available for it yet, either. The bare board should eventually be available through Mezel Mods or you will be able to have your own fabbed with the source files.

 IMG_20210415_133313158.jpg

Benoit

unread,
Apr 16, 2021, 2:25:06 AM4/16/21
to MPF Users, cobra18t

Benoit PETIT-DEMOULIERE

unread,
Apr 17, 2021, 12:39:53 PM4/17/21
to MPF Users
ok, I've unmounted the STM32 and I've double checked the orientation of the MOSFET.
That's BS170 so I have to follow the orientation on the plank.
I've seen pictures with resistors on the incandescent wing (inBrdIntf file), but they are mentioned as non mandatory, so just BS170, ground, input and STM32 connectore are fine, right ?

I have some pins in one wing that are showing no resistance between light ground and STM32 pin, so I guess the MOSFET is bad in this case.
But the other ones seem ok. I'll continue my debugging !

Benoit PETIT-DEMOULIERE

unread,
Apr 17, 2021, 1:44:19 PM4/17/21
to MPF Users
More progression !

I've replaced some mosfets with success but i've seen two profiles among my BS170.

one type throw higher voltage, but is not completely off when it should. with a LED, that's not good as the led stays weakly lit when it should be off.
If I put a classical GE44 light bulb, that's working well, as the bulb goes off when voltage decreases. Let's call it Type 1

the other type (called Type 2 now) as lower voltage all around, so the GE44 is weaker when on. but LEDs work perfectly well.

So I would need to have only type 2, but i can't identify them before having them soldered..

Would an additional 10k resistance for the type 1 help ?

or shall I buy a new set of BS170, or 2N7000 but will they behave as type 1 or 2 ..?

thanks for your help

Hugh R Spahr

unread,
Apr 17, 2021, 1:52:58 PM4/17/21
to mpf-...@googlegroups.com
I would only use BS170s.  Instead of using GE44, I would consider GE47 bulbs.  GE44s take about 250mA of current. GE47 take about 150mA of current which stress the MOSFETs that much less because of heat from Rds(on).  I think GE44s are pretty old school vs the GE47s, and many people switch out their 44s to protect the plastics and backglass from the heat.  Furthermore, if you have the capability, run LEDs.  Those will take even less current and should produce the same amount of light as the 44s.  Hopefully that is helpful.
-H 

Benoit

unread,
Apr 17, 2021, 2:35:17 PM4/17/21
to mpf-...@googlegroups.com, Hugh R Spahr

In fact, I dont' know any bulb, led only.

But that helped me troubleshooting..

Is there a way to change the behavior of mosfet? Why do I have two different profile.. Strange

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/CAD9bh01z%2BYKDR_OFjbgH_ohDBVEBfR1mS1O1G4wwoYBNe_uWSA%40mail.gmail.com.

cobra18t

unread,
Apr 18, 2021, 1:16:27 AM4/18/21
to MPF Users
It is good you are using LEDs, not 44's. The BS-170 is inadequate for driving the 44's from the STM32 since it is a 3.3V device. Your "type 2" behavior is what I would expect where the 44's are dimmer. The STM32 just isn't driving the gate of the BS-170 enough so the resistance is too high and the bulb is dimmer.

For the "type 1" behavior, did those transistors ever drive 44's before you tried the LEDs? Running 44's on BS-170 driven by 3.3V might be enough stress to damage the transistor. If you damage the transistor it might not turn off all the way. I think this is unlikely, but had to through it out there. Even in the worse case Rdson of a poorly driven FET, it would only need to dissipate ~400mW, which is well below the quoted maximum of 830mW.

Which wing/channels are not turning off all the way? It would be interesting to trace those back to pins on the STM32 to see if there is something unique about them.

Hugh, do you drive the gates of the FETs low or do you use internal pull-downs?

I am going to take it as an action item to find an alternate FET that is more bulb-agnostic given the 3.3V drive voltage.

cobra18t

unread,
Apr 18, 2021, 1:52:48 AM4/18/21
to MPF Users
It is somewhere around 2x the price, but a TN0606 would be better for driving incandescent bulbs with the 3.3V STM32. A VN0104 is in between the BS-170 and TN0606 as far as price and would also likely work, albeit with higher resistance. The pinout of both of these is the same as a 2N7000.

It is possible that you are overheating the transistors when you solder them? The two different behaviors seems like damage of some kind.

Benoit

unread,
Apr 18, 2021, 3:27:48 AM4/18/21
to MPF Users, cobra18t

Thanks for your input.

The wing were soldered just for this purpose and never used for anything before the led.

I've just used the GE44 bulbs to debug the system.

I don't want to use bulbs, as I have some neopixels and the timing of enlightenment is too different between bulbs an leds, so full led.


I'll maybe just buy a new set of bs170 and go through trials to get a proper set.

I basically need only.. 40 leds driven, so 5 correct wings.

And the matrix plank won't work as there is a neopixel on the STM32 that would use it.

From what I've read, it's not compatible due to hardware limitations.

Best

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/9b3f104e-ff8e-4b22-a85a-2fbd685d341fn%40googlegroups.com.

Hugh R Spahr

unread,
Apr 18, 2021, 9:14:43 AM4/18/21
to mpf-...@googlegroups.com
Sorry, I started this response last night, and then went to bed without finishing it.

I should really looked at this more closely when switching from the PSOC to the STM32.  The STM32 is a 3.3V part, and the BS170 is not really that great of a MOSFET for that low of a threshold voltage.  Yeah, it will workish, but it is at the hairy edge of working.  When I was doing the swap, I was paying much more attention to the high current (IRL540).  I didn't bother going back through the lower power MOSFETs.

Doing a couple of quick searches, it seems that these are good choices: TN0702, TN0606, TN0610.  The TN0606 is the cheapest of the lot at $.69 if you buy at least 25.  Be careful because they are "reverse pinned" which means they would be soldered in opposite of the silkscreen on the boards.  At 3V, the TN0606 allows over 1A from drain to source.

Cobra also mentioned the VN0104 which can be used, but passes less current.  Those are $.50 if you buy at least 25.  (Slightly cheaper).

To answer cobra's question, the pins are actively driven in both directions, not just attached to an internal pull up/pull down.  The external pull downs on the card were only to prevent a "blink" before the processor boots and starts actively driving the signal.  During that time, most processors float their pins, and I wanted to drag them to a known state.  It really doesn't matter.  It could matter more on the solenoid wings.  (Old bally machines fire all their solenoids when booting up (at least mine did), and that really annoyed me for some reason.)

Benoit

unread,
Apr 18, 2021, 9:28:55 AM4/18/21
to mpf-...@googlegroups.com, Hugh R Spahr

OK, I've redone my wings with bs170 and all wings works well.

As these wings belong to the kids soldering party, they may have been damaged by higher temp..

I'm using a 5v from my atx, and there will be around 40 leds used, hopefully that will work well!

To view this discussion on the web visit https://groups.google.com/d/msgid/mpf-users/CAD9bh001SVhZvVxSwvArvBfO0GET%2BfW5OPhvwur58C23qoMQkw%40mail.gmail.com.

cobra18t

unread,
Apr 18, 2021, 3:25:46 PM4/18/21
to MPF Users
Glad you got it working!
Reply all
Reply to author
Forward
0 new messages