285 replies to this topic

[NobodyYouKnow]

I feel a bit like an armchair quarterback here. :-)

That is an excellent question. I am not completely sure this is even an issue. My v-pin table is still implemented mostly on paper, and is wonderfully easy to change. The point in my email is that with 13 amp MOSFETS (de-rated to about 7 without a heat sink) and PCB lands and traces sized for 5 amps, that the current limit is about 3 amps per channel due to the rating of the connector. This would be what I fuse for. For non-PWM devices I am happy running them through a relay.

I know this is not the best solution, but if I find 3 amps to be an unacceptable limit I will slightly alter your excellent design and replace one of the 16 pin SIP headers with a 32 pin DIP header, load sharing between 2 adjacent pins. This will get me comfortably to 5 amps for those channels.

This is all highly speculative. I expect to find all this is a non-issue once you complete testing with your prototype.

Regards, Pat

[NobodyYouKnow]

Here is another idea that makes full use of the MOSFET board high current interface channels as designed. Provide a "daughter board" that can be attached to some of the existing "High Power Out" SIP pins (JP-5 and JP-6). The daughter board would connect closely to the output pins on the MOSFET board, and would fan out some number of channels (maybe 3-4) to higher current connections. Key in this approach is to minimize the path length from the mother to daughter board, even to the point of soldering the daughter board directly to the motherboard SIP pins, instead of using a SIP socket. As a bonus, because of your thoughtful board design, I believe it would be possible to design the daughter board to fit any group of 3-4 pins by rotating the daughter board (i.e. switch from fitting JP5 1-3 to JP6 14-16), or flipping the daughter board (i.e switch from fitting JP5 14-6 to JP6 1-3). This would be a very versatile approach, and is what I will do first if the 3 amp limit becomes restrictive. Understanding, of course, that the 5 amp per circuit and 12 amp per board limits must also be observed.

 

(To Mike - I really enjoy these thought based design challenges, and am so very pleased with the design of your expansion boards.)

[mjr]

I feel a bit like an armchair quarterback here. :-)

That is an excellent question. I am not completely sure this is even an issue. My v-pin table is still implemented mostly on paper, and is wonderfully easy to change. The point in my email is that with 13 amp MOSFETS (de-rated to about 7 without a heat sink) and PCB lands and traces sized for 5 amps, that the current limit is about 3 amps per channel due to the rating of the connector. This would be what I fuse for. For non-PWM devices I am happy running them through a relay.

I know this is not the best solution, but if I find 3 amps to be an unacceptable limit I will slightly alter your excellent design and replace one of the 16 pin SIP headers with a 32 pin DIP header, load sharing between 2 adjacent pins. This will get me comfortably to 5 amps for those channels.

 

That's a great idea - I think a double row should fit and would work to double the current capacity of the pins.

 

 

Here is another idea that makes full use of the MOSFET board high current interface channels as designed. Provide a "daughter board" that can be attached to some of the existing "High Power Out" SIP pins (JP-5 and JP-6). The daughter board would connect closely to the output pins on the MOSFET board, and would fan out some number of channels (maybe 3-4) to higher current connections. Key in this approach is to minimize the path length from the mother to daughter board, even to the point of soldering the daughter board directly to the motherboard SIP pins, instead of using a SIP socket. As a bonus, because of your thoughtful board design, I believe it would be possible to design the daughter board to fit any group of 3-4 pins by rotating the daughter board (i.e. switch from fitting JP5 1-3 to JP6 14-16), or flipping the daughter board (i.e switch from fitting JP5 14-6 to JP6 1-3). This would be a very versatile approach, and is what I will do first if the 3 amp limit becomes restrictive. Understanding, of course, that the 5 amp per circuit and 12 amp per board limits must also be observed.

 

That should work nicely too.

[JiePieWie]

I'm currently struggling with the setup of my pinscape-HW so this board is looking pretty handy. I've sent an email to mk47 to see if a groupbuy in Europe is possible. Excellent addition to pinscape!

[roar]

I believe I'm going to do a group buy for anyone interested n Canada once we get the all clear that things went well with the testing. PM if your interested. 10 board minimum order from the supplier so I'll have 9 up for grabs. (If I do all 3 boards I will be looking to do all 3 boards in the group buy too, not looking to keep an inventory )

[mjr]

I've been doing some testing, and things are starting to work.  I've discovered one serious bug that affects the main board and MOSFET board - one of the control pins to the PWM chips was wired wrong.  It's possible to work around by running jumper wires to the affected pins, but only the truly hardcore electronics nerds will want to deal with that, so it'll necessitate making a second batch of boards to make this sufficiently plug-(and-solder)-and-play.  

With that resolved, the data signal to the PWM chips looks very clean so far.  That was an area of some concern based on things I've read from Arduino people working with these chips, but it looks really good right now.  I was hoping that would be the case with a decent PCB layout, but it's good to have confirmation.  It augurs well for the daisy-chaining setup for adding boards for more outputs.

[mk47]

I've been doing some testing, and things are starting to work.  I've discovered one serious bug that affects the main board and MOSFET board - one of the control pins to the PWM chips was wired wrong.  It's possible to work around by running jumper wires to the affected pins, but only the truly hardcore electronics nerds will want to deal with that, so it'll necessitate making a second batch of boards to make this sufficiently plug-(and-solder)-and-play.  

With that resolved, the data signal to the PWM chips looks very clean so far.  That was an area of some concern based on things I've read from Arduino people working with these chips, but it looks really good right now.  I was hoping that would be the case with a decent PCB layout, but it's good to have confirmation.  It augurs well for the daisy-chaining setup for adding boards for more outputs.

 

For sure my prototype didn't completely work for the same reason. mjr already shared the details with me. I hope to finish a similar fix here tonight. If it works as expected we can be confident that the design works. Especially because my boards were produced by a different vendor in Germany and I applied small changes like replacing the transistors by common European models.

[NobodyYouKnow]

http://www.mouser.co...ontact/1725711/

 

OMG - I am like a kid in a candy store here.

 

I think this will work well to address my 3 amp current limitation consideration. I did not know these terminal blocks come in a 2.54mm pitch, but they do, and they are rated for 5 amps to boot. In theory, I will attach a 2-3 wide group of these to the bottom of the MOSFET board, to assure clearance with the SIP header and housing sitting just 2.5mm to the side on the top of the board. I will then install a SIP header on the top of the board with a pin count reduced by the terminal block width. This way I get the whole 16 outputs, and have access to a couple of 5 amp circuits if I need them. The next question becomes what the most judicious placement of the terminal block will be. Looking at the trace widths and length, I think pins 14-16 on both JP5 and JP6 would be the lowest loss path based on the length and width of the lands for MOSFET source and drain. It also helps to have the terminal block completely to one side as to not break JP5/6 into two separate headers. I am doing some "dead-reckoning" physics here. Thermodynamics never was my favorite subject.

 

I will also use an 8-wide block on the Chime board high power output, and on JP9 on the main board. I would have used it on JP-4, but the form factor does not allow.

 

Anyway, I'd love to hear some commentary.

 

Happy New Year, folks!

Edited by NobodyYouKnow, 31 December 2015 - 05:01 AM.

[mjr]

http://www.mouser.co...ontact/1725711/

 

I think this will work well to address my 3 amp current limitation consideration. I did not know these terminal blocks come in a 2.54mm pitch, but they do, and they are rated for 5 amps to boot. In theory, I will attach a 2-3 wide group of these to the bottom of the MOSFET board, to assure clearance with the SIP header and housing sitting just 2.5mm to the side on the top of the board. I will then install a SIP header on the top of the board with a pin count reduced by the terminal block width. This way I get the whole 16 outputs, and have access to a couple of 5 amp circuits if I need them. The next question becomes what the most judicious placement of the terminal block will be. Looking at the trace widths and length, I think pins 14-16 on both JP5 and JP6 would be the lowest loss path based on the length and width of the lands for MOSFET source and drain. It also helps to have the terminal block completely to one side as to not break JP5/6 into two separate headers. I am doing some "dead-reckoning" physics here. Thermodynamics never was my favorite subject.

 

Those connectors definitely look like an interesting thing to try.  The one possible snag is that their pins might be a tight fit - the data sheet wants you to drill at 1.1mm, whereas the regular pin header pads are drilled at 1.016mm.  It's a very small difference, probably smaller than the manufacturing tolerances on the boards, so it might not actually matter.

 

I don't think it should matter too much which outputs you use for the higher-power connectors.  All of the MOSFET sources connect directly to the big ground trace back to the power plug ground.  The shortest path length there is the low-numbered (leftmost in the EAGLE view) connectors on JP5.  But I'm hoping we're not living that close to the edge in the first place that this is even a concern.

 

I was thinking that on the next rev of the board, I could either double up the rows on both sides, for 2 pins for each output, as you suggested earlier, or set aside a few (say 4) pins on JP6 for one of the big Molex connectors.  There's room for either (or both, really).  The double row plan is kind of nice because it flexible - you can use single-pin connectors for outputs that are comfortably within the 3A limit, and only double up the outputs with the power-hungry devices.

[mk47]

 

http://www.mouser.co...ontact/1725711/

 

I think this will work well to address my 3 amp current limitation consideration. I did not know these terminal blocks come in a 2.54mm pitch, but they do, and they are rated for 5 amps to boot. In theory, I will attach a 2-3 wide group of these to the bottom of the MOSFET board, to assure clearance with the SIP header and housing sitting just 2.5mm to the side on the top of the board. I will then install a SIP header on the top of the board with a pin count reduced by the terminal block width. This way I get the whole 16 outputs, and have access to a couple of 5 amp circuits if I need them. The next question becomes what the most judicious placement of the terminal block will be. Looking at the trace widths and length, I think pins 14-16 on both JP5 and JP6 would be the lowest loss path based on the length and width of the lands for MOSFET source and drain. It also helps to have the terminal block completely to one side as to not break JP5/6 into two separate headers. I am doing some "dead-reckoning" physics here. Thermodynamics never was my favorite subject.

 

Those connectors definitely look like an interesting thing to try.  The one possible snag is that their pins might be a tight fit - the data sheet wants you to drill at 1.1mm, whereas the regular pin header pads are drilled at 1.016mm.  It's a very small difference, probably smaller than the manufacturing tolerances on the boards, so it might not actually matter.

 

I don't think it should matter too much which outputs you use for the higher-power connectors.  All of the MOSFET sources connect directly to the big ground trace back to the power plug ground.  The shortest path length there is the low-numbered (leftmost in the EAGLE view) connectors on JP5.  But I'm hoping we're not living that close to the edge in the first place that this is even a concern.

 

I was thinking that on the next rev of the board, I could either double up the rows on both sides, for 2 pins for each output, as you suggested earlier, or set aside a few (say 4) pins on JP6 for one of the big Molex connectors.  There's room for either (or both, really).  The double row plan is kind of nice because it flexible - you can use single-pin connectors for outputs that are comfortably within the 3A limit, and only double up the outputs with the power-hungry devices.

 

 

I used these connectors on my first self-made Pinscape extension board. I still have some of them here and just tried to mount them on the MOSFET prototype board. They perfectly fit, so the 1.1mm drill doesn't seem to be an issue.

 

But watch out, the screws are really small.

 

BTW if you're concerned about the connector, why don't you just solder the wires directly to the board?

Edited by mk47, 02 January 2016 - 12:59 AM.

[NobodyYouKnow]

I am delighted to hear these little terminal blocks fit.No surprise the screws are really small. From the drawings, it looks like the throat is just barely big enough to handle 18 AWG wire. I am anxious to get a set of boards in my hands to start working with the real deal, instead of playing around with these mental machinations.

 

And yes - soldering directly is always an option, but by personal preference not my first inclination.

[Sam66]

This is my PCB based on the original Pinscape hardware.  It has room for up to 16 high power outputs (only 10 MOSFETS are fitted in the pictures because that's all I had on hand). 

 

 

This is NOT the expansion board this thread is about but I thought if may be of interest.  To be honest if I had known Mike was working on a PCB I would have saved myself a lot of effort and waited for his

 

I'm still testing it, but so far everything works ok.  It's the first PCB I have done so I'm quite happy with the single fault of two pins being connected together; this is possible to fix in the software or by cutting the PCB track.  

 

Based on this one order only, I highly recommend  http://www.shenzhen2u.com   Low prices, lots of shipping options, good comms, reasonable turn around and accurately produced PCBs.

 

I have a few spare boards and would be happy to send them out for the cost of postage to anyone who doesn't need the extra expansion board capabilities and wants to give these a try. 

[vampirolatino2]

Are you located in USA?

Edited by vampirolatino2, 02 January 2016 - 08:03 PM.

[Sam66]

Sorry, should have said.  I'm UK based but happy to send anywhere as long as you cover the postage.

[vampirolatino2]

It's ok, but thanks!!

[mjr]

This is my PCB based on the original Pinscape hardware.  It has room for up to 16 high power outputs (only 10 MOSFETS are fitted in the pictures because that's all I had on hand). 

 

Congratulations on getting that implemented!  Glad to see more options out there.  If you plan to publish the plans, I can put a link and/or host copies on my Pinscape Resources page, if you'd like.

[NobodyYouKnow]

I have another "What if . . ." question. As I said earlier, my vPin cabinet is still 95% on paper. I have been musing over whether it was better to power the feedback devices with 12 volts DC, or, where possible with 24 volts DC or even higher. I may need to open another discussion on this topic, but lets put that aside for a minute.

 

Here is my question. If have both +12Vdc and +24Vdc busses in my cabinet, can I intermingle multiple accessories using both voltages on the same MOSFET board? For example, JP5-1 has an accessory at +12Vdc at the same time JP5-2 has an accessory at +24Vdc. Are there any combinations of duty cycle 0-100% on either channel that could pose an issue? We need to assume the two busses have a common ground, properly rated MOSFETs, etc.

 

My first instinct looking at the MOSFET board schematic is that the two should co-exist harmoniously since the board is wired to switch on to ground. The only risk I see is some failure mode that caused the MOSFET and opto isolator to short everything to the 12Vdc side, or maybe the connector at JP4 failed causing the output side of the board to float.

 

That said, if I am wrong the magic blue smoke is going to get out of a lot of pricey things.

[mjr]

Here is my question. If have both +12Vdc and +24Vdc busses in my cabinet, can I intermingle multiple accessories using both voltages on the same MOSFET board? For example, JP5-1 has an accessory at +12Vdc at the same time JP5-2 has an accessory at +24Vdc. Are there any combinations of duty cycle 0-100% on either channel that could pose an issue? We need to assume the two busses have a common ground, properly rated MOSFETs, etc.

 

I run exactly this way with my current setup, and lots of other people have similar setups with LedWizes and the like.  It seems to be completely fine.  I'm not an electrical engineer, but my understanding is that low-side switching (transistor between load and ground) is readily compatible with a mix of positive voltages like this.  In fact, Arduino people commonly use this property to construct voltage level changer circuits when they have to interface (say) 3.3V and 5V logic.  That's not power circuitry but I think the principle is the same.

 

 

 The only risk I see is some failure mode that caused the MOSFET and opto isolator to short everything to the 12Vdc side, or maybe the connector at JP4 failed causing the output side of the board to float.

 

If the outputs get disconnected and float, the device doesn't have a path to ground, so it just stays off.  A completely benign failure mode.  In fact, whenever the MOSFET is switched off, that's effectively what it's doing anyway - the electrical path between drain and source is extremely high impedance when the MOSFET is switched off, which is equivalent to the wire being disconnected.

 

To create a path that endangers the logic circuitry, it seems to me that the MOSFET has to fail in such a way that it creates a short between drain and gate, and at the same time the opto fails in such a way that it shorts between its phototransistor collector and its LED anode.  I think either one of those failure modes is pretty improbable, especially the latter.  The optos have a gigantic isolation voltage - 5000V (!) according to the data sheet.  Lots of Arduino people control power circuitry with MOSFET gates connected directly their GPIO pins, so I think even this MOSFET failure mode is pretty darn improbable.  I think the only way MOSFETs typically fail is that they develop a short (or near short) between drain and source, which makes them switch on permanently.  That would be immediately obvious and wouldn't endanger any logic circuitry. (Although it could damage a knocker coil or the like - even the Chime Board time limiter wouldn't help there, since the timer is on the control side of the MOSFET; if the MOSFET breaks down and gets stuck on, switching off its gate won't help.)

 

And anyway, I don't think any of these risks are affected one way or the other by the mix of supply voltages (or lack thereof).  The main risk that I see arising directly from a mix of voltage sources is that you create a short between the +12V supply and the +24V supply, say, due to a wiring error or dropping a screwdriver in the wrong place at the wrong time.  A short between +12V and +24V is exactly like a short between +12 and ground, and it'll fry something along the short circuit path - the power supply, wires, coil, etc.  I have all of my outputs individually fused, so hopefully the component that goes up in smoke if this happens in my cab is the fuse.  

[NobodyYouKnow]

Thank you for the confirmation. I am not an EE either, which I think leads to some of my questions. One of my quirks (and the reason I have the job I have), is that a big chunk of my mind considers what happens when things go bad. Your discussion above is almost exactly the same path my mind took when this 12v / 24v mix question came to mind.

 

A couple of things I really like about your design of the secondary power source is 1) that you have doubled up on the ground pins and 2) you use the +12 from the secondary power source (which is on the same connector as the ground) to power the gates on the MOSFETs. These design features are further insurance against a ground fault causing the MOSFET outputs to float.

 

Fuses? We don't need no stinkin' fuses! In truth, this project has led me to try my hand at designing PCBs for production by a board foundry. My first product will be a fuse board to help protect all those outputs.

[Swisslizard]

Regarding the 12v/24v question, there is also another route you can go if you want really strong and load feedback from your contactors.

It is possible to operate with a high voltage for a short period when power is turned on and reduce power to a much lower level after that (just enough to keep the contactor active.

In my cab the contactors are triggerd with 31V for 25ms and after that power is reduced to approx. 3-4V. Since I use some exotic contactors (they are made for AC mainly, but would accept maybe 10v DC) the optimal values might be different for other types.

The solution speeds up the contactors, generates more noise and feedback, but at the same time the contactors dont suffer from overheating. At least my contactors do also react faster when the power turned off. probably because there is less of a magnetic field after power has been reduced.

I think PWM should also work for a similar scenario, but I havent been really sucessfull on that route.