PI CHIP

[Pete the Builder]

Sorry, sido, still haven't got up to do anything. (I'm a crappy designer, huh?) We've been house-hunting today, and we found one. So that's gonna mess up the next few weeks. But meanwhile, I should still be able to do design and layout stuff. I've installed Protel on my laptop, which runs about as fast as a 486 doing rendering, but it *does* run. So I'm going to get started tonight. I've got the footprints and schematic components imported, and from the photos it's clear that your rotary switch is a simple 3-pin 0.1" pin pitch around a .500" (500 mil) square. I've added additional pads and padded out the footprint, it will work well. That's it for now. I'll post an update later tonight.
Cheers,
Pete

Oh, yeah, I was so excited to see Paul's go-ahead, I forgot to say Thank you, moodz, you're a champion of the poor, downtrodden tinkerers! You made my day with that thumbs up, and I know sido had a smile too. Thanks, mate. Much appreciated. Now I have to really do something instead of tout my skillz!! (You utter, utter, utter, utter... )

[sido]

Sorry, sido, still haven't got up to do anything. (I'm a crappy designer, huh?) We've been house-hunting today, and we found one. So that's gonna mess up the next few weeks. But meanwhile, I should still be able to do design and layout stuff. I've installed Protel on my laptop, which runs about as fast as a 486 doing rendering, but it *does* run. So I'm going to get started tonight. I've got the footprints and schematic components imported, and from the photos it's clear that your rotary switch is a simple 3-pin 0.1" pin pitch around a .500" (500 mil) square. I've added additional pads and padded out the footprint, it will work well. That's it for now. I'll post an update later tonight.
Cheers,
Pete

Oh, yeah, I was so excited to see Paul's go-ahead, I forgot to say Thank you, moodz, you're a champion of the poor, downtrodden tinkerers! You made my day with that thumbs up, and I know sido had a smile too. Thanks, mate. Much appreciated. Now I have to really do something instead of tout my skillz!! (You utter, utter, utter, utter... )

Thanks Pete, that will be great, much appreciated

Please do not include Mosfet switching transistors on this main MCU board as this could introduce RF interference due to no shielding and close proximity to the LCD Module, Just add pins to the MCU board for the Mosfet switching which will be located on the next future board to be designed and added with Hex Spacers........which will have the mosfet switching and TX combined so as to be shielded with PCB groundplane which is located away from the main MCU and LCD Module. I will post some more info as to where the I/O pins of the Mosfet switching devices should be located later tonight. I say on the top of the LCD Module protuding behind MCU board. Design it with 2.54mm spacing for the required pin count to use straight male header connectors.

Use only single line male headers where needed for sub board connection. You can actually buy the the female headers from Element14 (stock number 488-1724) to suit the LCD Module interface to MCU PCB. They are 2.54mm spaced also. So the only connectors used for interfacing PCB's are male/female single line PCB headers or PCB jumpers (stock number 334-561). The male header connectors stocked at Element14 are (stock number 251-8351).

Not sure about Power Supply. Should we have a on board 5V PS on MCU?? for ease of removing MCU so as to be dependant on its own for other projects.??

Dont know if you read my post on PCB design software, but i have decided to learn PCB design using Diptrace. I still have a long way to be able to produce quality boards for now, so this PI controller PCB, the heart of PI will assist me later to design further sub modules as to this layout, that you are greatfully assisting which is much appreciated.

Yea, Kudos to Moodz approval

Cheers Sid

[moodz]

awww shucks.

[sido]

awww shucks.

Moodz, i hope you are not regretting it now

We are 1/4 of the way there, hehehehe, only jooookiiiinggggg.

Cheers and thanks
Sid

[sido]

Put this on-board power supply Regulator circuit also with 2 male pin headers for input voltage. Dont include the 8 Mosfet switching drivers. So its MCU, crystal, caps, power supply, Toggle and pin header connectors etc.

Also include the ICSP connector somewhere near the edge of the board. I believe you are able to get 90 Degree Connectors to suit.

See below circuit.

Cheers Sid

[ODM]

You suggest no mosfets on the main board, but still a non synchronous switching regulator without two-stage filtering? A linear reg option would be nice to leave in, just in case the switcher/layout/inductor proves noisy. I imagine the universal board's layout would have plenty of free space around parts and elsewhere for potential modding - always nice for prototyping.

Synchronous noise is far easier to live with when sampling, so a synchronizable switcher option is worth considering if the switcher reg stays - the cpu can provide this. National makes sync capable switch mode regs in their "simple switcher" product family as well.

The coil driver can be rather noisy with fast high current switches, sure, but it's synchronous and thus doesn't usually prove an issue for PI frontends unless it's excessive, usually from poor bypassing or crazy layout.

And yes, mosfet drivers should be located as close to the switched transistor as possible. Those make for pretty high current transient loops.

[sido]

You suggest no mosfets on the main board, but still a non synchronous switching regulator without two-stage filtering? A linear reg option would be nice to leave in, just in case the switcher/layout/inductor proves noisy. I imagine the universal board's layout would have plenty of free space around parts and elsewhere for potential modding - always nice for prototyping.

This is for a basic UNIPI timing device that can be adapted to say Goldscan IV, SurfPI, Tinkerers great design suggestions, SD2000 frontend/backend audio with modified updated circuitry or a new latest low noise frontend device that may become available in future new designs when developed, which can be experimented with such MCU setup. Power supply circuit was for reference to power MCU only and did not consider a synchronous type. Any suggestions? Might as well include it for convenience.

Synchronous noise is far easier to live with when sampling, so a synchronizable switcher option is worth considering if the switcher reg stays - the cpu can provide this. National makes sync capable switch mode regs in their "simple switcher" product family as well.

Thanks for the suggestion. Will look into it. Plenty of free space to include it on MCU PCB.

Yes there will be plenty of free space to experiment with and will be isolated from TX and frontend amps.

That is my idea of going down this path as to minimize unnecessary internal noise, but most important to make it Universal in all aspects as outlined in previous posted pictures.

The coil driver can be rather noisy with fast high current switches, sure, but it's synchronous and thus doesn't usually prove an issue for PI frontends unless it's excessive, usually from poor bypassing or crazy layout.

And yes, mosfet drivers should be located as close to the switched transistor as possible. Those make for pretty high current transient loops.

Isolating the Mosfet drivers and TX frontend stage this way allow further shielding like full shielding in metal or copper PCB eclosures between each sub-modules and maybe the use of feedthrough caps at critical points for minimal internal noise if there is any benefit.

Another reason why i would prefer to isolate the Tx/sampling Mosfet drivers is that some circuit designs in the TX stage may not require them ( PCB component footprints) as other driver designs may be incorporated in different experimental circuits.

Moodz great work offered is universal in schematic/software design but there is no practical layout to assist for experimental use, which may be implemented into a workable in the field detector with additional sub-modules

Thanks for your input. Any more suggestions are welcomed.

Cheers Sid

[ODM]

Suggestions? Leave space for a popcorn 3 terminal linear regulator, those are usually pretty quiet in emc contribution. One thing to consider is having low value series resistors (series termination, too) for digital io lines, if driving long jump wires or pin connections.

The analog connector port should have the analog ground and ADC supply voltage connections, to continue those voltages - they're the "analog" ground and supply, and they usually connect right next to ADC to the digital world, otherwise separate to keep the analog vdd/gnd from being a conductor for digital current loops. AVDD should be filtered and separate from the digital VDD, but often needs to be present at the same time as the digital VDD. This is sometimes a make/break case for separate ADC even if the microcontroller's ADC would be "good enough" otherwise.

It's also important to consider the current route into a transient inducing component, the input of a switch reg or transistor can be a route of radiated/conducted noise as well as its output.

And finally all signal connector pins in/out the board should have an according gnd pin close to them, hopefully all "unused" pins, to make the smallest possible current loop area. Especially for digital signals which usually emit interference, and analog signals which can be sensitive to receive interference. We usually want only the intentional search coils in our device

Thanks for the effort for everyone involved - there are often lots of people with suggestions but few people actually spending the sweat with the pcb program and workbench! If these tips saved a few more minutes of debugging than it took to read them, then it was successful...

[sido]

Suggestions? Leave space for a popcorn 3 terminal linear regulator, those are usually pretty quiet in emc contribution. One thing to consider is having low value series resistors (series termination, too) for digital io lines, if driving long jump wires or pin connections.

Great consideration, What value resistor's should be considered? Is this for all I/O connection lines between sub-modules (analogue/digital)?

The analog connector port should have the analog ground and ADC supply voltage connections, to continue those voltages - they're the "analog" ground and supply, and they usually connect right next to ADC to the digital world, otherwise separate to keep the analog vdd/gnd from being a conductor for digital current loops. AVDD should be filtered and separate from the digital VDD, but often needs to be present at the same time as the digital VDD. This is sometimes a make/break case for separate ADC even if the microcontroller's ADC would be "good enough" otherwise.

It's also important to consider the current route into a transient inducing component, the input of a switch reg or transistor can be a route of radiated/conducted noise as well as its output.

How about input power supplied for MCU on-board regulator is isolated totally from further sub modules via its own 2wire shielded power cable and connected to the main power supply module further inline with additional RF bypass capacitors, therefore bypassing in-between sub modules.

We usually want only the intentional search coils in our device

Not sure what you mean here, but i will take a guess that your referring to unintentional component leg(s)/connector's Induction loops created due to the inevitable longer sub module Interface Header connections with such proposed outlined design??

Thanks for the effort for everyone involved - there are often lots of people with suggestions but few people actually spending the sweat with the pcb program and workbench! If these tips saved a few more minutes of debugging than it took to read them, then it was successful...

I am no circuit designer by trade. Just a regular common Tech to be understood, but willing to learn though and your suggestions are a real benefit and most welcomed. Keep suggestions forthcoming as the main PCB design evolves.

BTW, i just started to learn a PCB layout program ( Diptace-----thanks Mechanic), which with upon further reading in low frequency circuit designs, will aid me in employing new experimental designs with critical PCB layout principles as such.

Cheers Sid

[ODM]

Ideally you would treat the input side of a switching regulator with the same care as its output, even if buck regulators are less noisy in this regard than boost regulators. The current comes from somewhere too, emc design is basically about thinking of the current loops, and parasite capacitances. Long wires can make for large loop areas unless they're twisted. This is why it's important to have as solid "ground plane" as possible, if using one. Not cut underneath high transient current loops.

I don't think you need to consider the power supply wiring, just have a capacitor-choke-capacitor input filter. The choke can be just a lossy ferrite bead. Often just a simple ceramic capacitor at the power input pins will cut radiated disturbance by ten dB range. It would be preferable to use surface mount capacitors to have as short leads as possible - capacitor leg leads have inductance, and a ceramic capacitor has very low inductance compared to an aluminum electrolytic capacitor.

Series resistors, it's hard to give values - for wires it's usually higher resistances like 50..100 and for pcb traces lower, 10..50. Most of the time they're not necessary at all if the load (and its wiring!) is well behaved, more resistive than reactive, and they add unnecessary parts to a layout. But for long cables, and sensitive fast signals like clocks, it's a good idea to have them in place.

An oscilloscope is a good rough guide for showing where the issues are, if possible try to make in board measurements with as short ground wire as possible from the probe. The usual probe's long ground wire and gator clip can make well behaved things look dirty, and hide some higher frequency issues. One thing to try out is to just wave the scope probe tip at high gain above the circuit without touching leads, to see if there is radiated e field at some location.

[sido]

Ideally you would treat the input side of a switching regulator with the same care as its output, even if buck regulators are less noisy in this regard than boost regulators. The current comes from somewhere too, emc design is basically about thinking of the current loops, and parasite capacitances. Long wires can make for large loop areas unless they're twisted. This is why it's important to have as solid "ground plane" as possible, if using one. Not cut underneath high transient current loops.

I don't think you need to consider the power supply wiring, just have a capacitor-choke-capacitor input filter. The choke can be just a lossy ferrite bead. Often just a simple ceramic capacitor at the power input pins will cut radiated disturbance by ten dB range. It would be preferable to use surface mount capacitors to have as short leads as possible - capacitor leg leads have inductance, and a ceramic capacitor has very low inductance compared to an aluminum electrolytic capacitor.

Series resistors, it's hard to give values - for wires it's usually higher resistances like 50..100 and for pcb traces lower, 10..50. Most of the time they're not necessary at all if the load (and its wiring!) is well behaved, more resistive than reactive, and they add unnecessary parts to a layout. But for long cables, and sensitive fast signals like clocks, it's a good idea to have them in place.

An oscilloscope is a good rough guide for showing where the issues are, if possible try to make in board measurements with as short ground wire as possible from the probe. The usual probe's long ground wire and gator clip can make well behaved things look dirty, and hide some higher frequency issues. One thing to try out is to just wave the scope probe tip at high gain above the circuit without touching leads, to see if there is radiated e field at some location.

When we look at frequencies in the Mhz range, we should consider the possibilities of isolating the source, (like the 10Mhz Cpu oscillator in this case).....and LCD module sync and vice versa when interconnecting with low frequency dependant circuitry.

In respect to the resistors mentioned, with I/O and (especially for low frequency circuitry) they can implement a inductive pickup and mess things up than not including them overall.

Experiment is the key with the lowest wire lead possible.

Surface mount concepts can achieve this, but beyond this simple PI controller in LF.

Cheers sid

[ODM]

Most microcontroller application notes have the same instructions for "hygiene", and for good reason. Crystal/cap traces should be as short as possible, and the route for vdd/gnd traces should go past capacitor terminals, individual caps for all vdd/gnd pairs. Pads for surface mounted filter caps are not that big to leave on the board as a precaution.

Series resistors serve a dual purpose. They are series termination for reflections which may grow to be issues with longer traces/wires, and they slow down the usually fast rise and fall times which contribute to radiated interference. The resistance value depends on what needs more attention - termination or rise times. They provide an added measure of protection against "workbench accidents" as well

Often series resistors can be unnecessary, and the need for them might only show up for longer traces on an otherwise clean design. But it's a consideration.

[sido]

Understood, thankyou.

Its all a new learning curve for me and no better way to learn more from this Forum with detector design and peoples past experiences in common PCB design principles like yourself.

The more information posted, the more we can all learn.

Cheers Sid

[Pete the Builder]

This is all really interesting info, and it's great to have it in one place, so other folks can learn from this discussion.

I'd like to just say, in the interests of not causing any confusion, all of the above is well and truly understood. I've attached an example of what I generally produce for clients. Hopefully, you can verify that in this case at least, I've followed the rules!

Just so I can add something useful, the rule of thumb is to add one monolithic 0.1uF cap every 3 digital packages. Also, power and ground traces should run in different directions, to minimise impulse coupling.

I'll have a think about the transistors, it's a complex issue with this design, and I need to understand exactly what signals and currents are going where. I do agree, it would be nice to screen stuff off, but this will only work if the power is properly routed and so on. I'll re-read the info above and if I need more details, I'll definitely ask!

It's good to know we've got good knowledge available from people here! Hopefully newcomers will benefit from all that's been mentioned so far here!

Cheers,
Pete
P.S. I'm off to see the specialist to try and get an 'all clear'... Fingers crossed... (I hate these visits, they waste so much design time - showering, driving, waiting, driving back, collapsing for a few hours to recover, then back to designing again! Wheee! )

[sido]

Hi Pete, i am in no hurry, so take your time. Priorites first and goodluck with your results. Any good news is better than bad.

Interesting board and great looking one at that.

Btw, fire away any questions you may have.

Cheers Sid