Shop OBEX P1 Docs P2 Docs Learn Events
How much breadboard is too much breadboard? — Parallax Forums

How much breadboard is too much breadboard?

escherescher Posts: 138
edited 2018-06-07 01:42 in Propeller 1
l7ot7Ys.jpg

After implementing a high bandwidth signal between two Props, I started getting some nondeterministic behavior with my setup consisting of 3 separate interconnected breadboards - video dropping out, props resetting, etc.

At this point I have to conclude that the parasitic inductance caused by the dozens of large signal loops (combined with the breadboard's notoriously poor noise and high frequency handling) is the culprit.

But what other options are there for clean, rapid, plug and play prototyping? Vector and perf boards all require time consuming and messy soldering or wire wrapping etc.

Thoughts?

Fun fact: your smartphone's high speed video mode is an excellent tool for debugging video issues which only occur during periods of movement. But now I'm out of storage space...
«1

Comments

  • Wire wrap is better than garden variety breadboard and some breadboards are better than others. I don’t think there’s any good answer othèr than to keep leads short, use twisted pairs
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2018-06-07 02:48
    Oh my, surely you're not serious, wiring up Props like that. I can't even see where the single decoupling cap is connected directly across the VDD and VSS pins with "short" leads, and preferably both sides. At least use much smaller MLCC 0.1uF caps but don't rely on connecting to the bus strips and back again, that is way too much inductance. The poor crystal always seems to be stuck in the air but just remember that the crystal input pin is a very sensitive "receiver" and will pick up all kinds of noise that is never going to be filtered out, but only glitches the whole chip. At the very least trim the leads down to a minimum.

    However I would at least solder the Prop, the caps, and the crystal to a small matrix board and then plug it in or failing that, just solder the caps and crystal directly onto the top of the Prop package and be done with it quickly and surely.

    BTW, despite being plugin breadboard, it looks neat and impressive. It brings back memories of Humanoido and his "Big Brain".
    Another thing, don't daisy chain your grounds but run them separately back to a common point where the power feeds in.
  • Oh my, surely you're not serious, wiring up Props like that. I can't even see where the single decoupling cap is connected directly across the VDD and VSS pins with "short" leads, and preferably both sides. At least use much smaller MLCC 0.1uF caps but don't rely on connecting to the bus strips and back again, that is way too much inductance. The poor crystal always seems to be stuck in the air but just remember that the crystal input pin is a very sensitive "receiver" and will pick up all kinds of noise that is never going to be filtered out, but only glitches the whole chip. At the very least trim the leads down to a minimum.

    However I would at least solder the Prop, the caps, and the crystal to a small matrix board and then plug it in or failing that, just solder the caps and crystal directly onto the top of the Prop package and be done with it quickly and surely.

    BTW, despite being plugin breadboard, it looks neat and impressive. It brings back memories of Humanoido and his "Big Brain".
    Another thing, don't daisy chain your grounds but run them separately back to a common point where the power feeds in.

    Whoa lot of good info in here, slow down...

    I followed the recommended circuit from the prop documentation, don't remember any decoupling caps. Where and why would I include those?

    What are you referring to when you talk about the "bus strip" and the crystal?

    And where am I daisy chaining ground paths?

    Thanks!
  • RaymanRayman Posts: 13,800
    Ok, so my first circuits looked a lot like that... Yes, you need lots of bypass caps everywere.
    Especially on Prop and also one all other chips...
  • I thought I was giving you "just enough" information :)

    Filter/decoupling/bypass capacitors are ALWAYS required for any digital circuitry otherwise it will fall victim to its own switching transients which occurs when signals go high/low. So you must have them and you can google up some more information or check this one. If you work with digital circuitry, do not skip these fundamentals, they are very important but you can be forgiven because I have had "qualified" electronic engineers work for me that never knew any of this, to my absolute shock, horror, and wonder.

    When you have those horrible plugin breadboards they are made up of strips of metal forming fingers at each hole. There is a strip of 4 or so for each "pin" that you plug in plus there are the long strips on the outside that are used for power and (a horrible) ground. These long strips are your "commons" or as I just called them "bus strips".

    If you can at the very least get some of these bypass caps onto the board in the way that I mentioned you may have some hope of it working more reliably.
  • The Propeller package has 4 pairs of Vdd/Vss pins, one on each side of the package. These are connected on the chip itself, but cross the chip to do so. The demands of current from different areas on the chip vary a lot and there's some resistance in the connections across the chip. Result is very brief voltage variations across the chip which cause noise problems and, in the worst case, can cause damage to some sensitive structures. The most well known of this is the loss of clock signal from the crystal clock's PLL/multiplexor.

    Normally, you should have 0.1uF decoupling capacitors across each pair of Vdd/Vss pins, one for each side of the chip, preferably adjacent to the package or underneath the package. On PCBs, these pairs of pins are paralleled under the chip. Minimally they should be connected via very short leads.

    System clock time is normally 50ns and some signals internal to the chip can switch faster. Rise and fall time can be on the order of nanoseconds and light (and power) doesn't move very far that fast. The decoupling capacitors help supply power to different parts of the chip over this sort of timeframe ... if the leads are short enough.
  • Mike Green wrote: »
    The Propeller package has 4 pairs of Vdd/Vss pins, one on each side of the package. These are connected on the chip itself, but cross the chip to do so. The demands of current from different areas on the chip vary a lot and there's some resistance in the connections across the chip. Result is very brief voltage variations across the chip which cause noise problems and, in the worst case, can cause damage to some sensitive structures. The most well known of this is the loss of clock signal from the crystal clock's PLL/multiplexor.

    Normally, you should have 0.1uF decoupling capacitors across each pair of Vdd/Vss pins, one for each side of the chip, preferably adjacent to the package or underneath the package. On PCBs, these pairs of pins are paralleled under the chip. Minimally they should be connected via very short leads.

    System clock time is normally 50ns and some signals internal to the chip can switch faster. Rise and fall time can be on the order of nanoseconds and light (and power) doesn't move very far that fast. The decoupling capacitors help supply power to different parts of the chip over this sort of timeframe ... if the leads are short enough.

    Thanks to you and @"Peter Jakacki" for the info!

    If I'm only powering a Prop via one set of VDD/VSS pins, why would I need to decouple the other set as well?
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2018-06-07 03:56
    escher wrote: »
    Thanks to you and @"Peter Jakacki" for the info!

    If I'm only powering a Prop via one set of VDD/VSS pins, why would I need to decouple the other set as well?

    There is an acronym "RTM" that you may have come across. It stands for "Read The Manual" although you may find the ruder and more exasperated equivalent "RTFM".

    This is really important if you want stuff to work, and to work reliably and well. ALWAYS connect all power and ground pins of a chip otherwise suffer the consequences which can lead to a fried chip. ALWAYS connect up bypass caps as described or suffer the consequences. These requirements are no different to say "supply voltage of 3.3V" because you may know that if you decide to connect 5V or 12V even to the chip that it will not accommodate your foolishness for very long or even for a second.

  • An explosion in a wire (antena) factory! As the Grinch would say, noise Noise NOISE!!!!! There are cut wire kits that work very well for on board connections, and use very short spans to other breadboards. If you don't want to spend for those, you could get a spool of the same 22 gage wire pretty cheap and cut to length each time. Use the plastic insulated easy to strip wire for these. AS to bypass caps, I've always used them (well maybe over used) at one 0.1uF cap per TTL package back when. You can always use leaded caps, but I like the look of the dip caps, just sit it at the end of a package and wire away. Just looks neater and less chance of picking up stray signal pr shorting to something too close to the leads. Also the neatness of using dip caps and cut wire makes the circuit far easier to compare with the drawings, chasing faults or making changes in the design. Also easier to get a probe in without shorting or ripping something loose.
  • First of all ... you really should run short jumpers connecting the four Vdd pins together and the four Vss pins together. The jumpers would have a much lower resistance than the cross chip metalization connections. You also need the decoupling capacitors at (or very close to) the Vdd/Vss pin pairs because the chip needs a power reserve for surges on the order of a nanosecond or a few nanoseconds. Look at how existing Propeller boards are laid out in the area right around the processor package.

    I believe the Propeller Education Kit (using the DIP package and a breadboard) was designed before complaints started coming in of chips that would work when running from either RCFAST or RCSLOW, but wouldn't work using a crystal as a clock source. You could download programs to them (uses RCFAST), but the chip didn't seem to work otherwise. This occurred when power was supplied to only one or possibly two Vdd/Vss pairs and only one decoupling capacitor ... noise in the power source and voltage drops across the chip caused some structures to be stressed to the point of failure. Adding the decoupling capacitors and parallelling the Vdd/Vss connections fixed the problem.
  • 1UUHbHv.jpg

    Thanks for the info everyone! Added in all the decoupling caps and the behavior disappeared as I'm sure you all expected it would!

    This is just a test setup, and I'll have my next test bench caps be minimum pathed to the pins instead of across the commons, but even in this rudimentary setup the system is now stable!
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2018-06-07 04:32
    Great! Thanks for the feedback too.

    BTW, that lower set of caps look like they are totally ineffective. Do make sure you connect all the power and grounds.
  • Great! Thanks for the feedback too.

    BTW, that lower set of caps look like they are totally ineffective. Do make sure you connect all the power and grounds.

    Yep that realization smacked me in the face while I stared at the picture haha.
  • Use _ceramic_ 100nF capacitors for decoupling every supply pin on every logic chip.

    Make the leads as short as possible and the capacitor as close to the pin as possible.
    This usually means the cap bridges over the chip, which is a little annoying, but means
    the decoupling will be sound, not flaky.

    For large assemblages of fast logic on breadboard like this its probably wise to use a ground-plane under them all,
    and keep all the jumpers short and neat - this means cutting each one to size from a spool of single
    strand hookup wire - this keeps all the wires low to the ground plane and reduces crosstalk
    and noise pickup a lot.

    If you are doing low frequency analog stuff you do not need so many precautions, but fast
    logic is damn fast and its easy to get erroneous clocking in a rats-nest setup. dV/dt is ~ 1GV/s
    for CMOS logic signals, that means its easy to get cross talk through small amounts of stray
    capacitance. Keep wires as short as possible, avoid large loops. Plenty of ground wires to
    form a mesh across the breadboard will also help provide low inductance return current paths.
  • @escher, I am the only non-professional in this thread so I'm taking note of everything they said to you but I did want to add something you might find useful:
    Sometimes I use the probe of a wire tracer to listen for electrical noise. An oscilloscope would be ideal but since I don't yet have one the probe has helped me troubleshoot circuits on occasion.
  • Also, consider using the FLiP module. It pretty much takes care of its own power conditioning, plus it include the programming interface, the EEPROM, a couple of LEDs and that ain't countin' a reset button! Oh, and keep your wires short.
  • RickBRickB Posts: 395
    edited 2018-06-07 17:04
    There is no such thing as to much breadboard.
    If your name is AtomicZombie, you can do this on a breadboard at 25 Mhz with 74HC and a AVR.

    https://avrfreaks.net/forum/avr-helps-out-vulcan-74-7400-logic-vga-mega-game-system



  • For Parallax's recommended Wiring of the breadboarded dip propeller see:

    Figure 3-6 page 2 of "32305-PEKit-40-PinDIP-Enhancements-v1.0.pdf" or

    Figure 3-9 page 38 of "122-32305-PE-Kit-Labs-Fundamentals-Text-v1.2.pdf".

    Source - Parallax Shop - "Propeller Education Kit - 40-Pin DIP Version"
    Product ID: 32305
    >Downloads & Documentation
  • Cluso99Cluso99 Posts: 18,066
    Few things...
    The DIP prop chip only has 2 power connections, the QFP & QFN have 4 sets.

    DO NOT ONLY POWER ONE SET OF POWER PINS !!!!

    Power and ground pins are there for a reason. The sure way to blow the PLL inside the prop chip is to not connect all power and ground pins and not decouple each pair.
  • Wirewrap can actually make some impressive stuff. Back in the day there were products that went into production with wirewrap fabrication.

  • There is an acronym "RTM" that you may have come across. It stands for "Read The Manual" although you may find the ruder and more exasperated equivalent "RTFM".

    " Read the 'fine' manual"
  • Cluso99Cluso99 Posts: 18,066
    Wirewrap can actually make some impressive stuff. Back in the day there were products that went into production with wirewrap fabrication.
    The mini-computers I worked on in the 70s had wire-wrapped motherboard cages.
  • kwinnkwinn Posts: 8,697
    Wirewrap can actually make some impressive stuff. Back in the day there were products that went into production with wirewrap fabrication.

    Back in the day quite a few products were produced using wire wrap. Computers and keypunches are two I got all too familiar with. There were even articles in the IEEE pubs that talked about "the thickness of the nap" limiting the size and power of computers.
    Of course that was in the days of using discrete transistors and very early IC's.
  • Heater.Heater. Posts: 21,230
    edited 2018-06-09 17:09
    We were using wire wrap to create prototype boards with Motorola 6809 and such new fangled processors at Marconi Radar back in the day. Great for hooking up RAMS, ROMS, peripheral chips and sprinkling of 74xxx TTL.

    I have seen wire wrap used in many production runs. Especially of backplanes on racks full of boards.

    All those nice wire wrap sockets were always too expensive for me to use in hobby projects.

  • Long time ago, I worked at a place that put some wire-wrapped stuff into production. Had a room with maybe 8 or 10 ladies, all with hand held wire-wrap guns. Sounded pretty weird in there.
  • I have here a book on programming a fully experimental I8080 system. It has photo in it of the full system, and a very large breadboard assembled for that purpose.

    And in fact I have one for the I8080, and the author strongly advocated for wire-wrapped sockets.

    In fact my current project will be around a RasPi Zero W, rig, and its supported modules will all be wire-wrapped.

    Since we're beginning to drift seriously off topic, what say we correct that drift and return to discussing why the original idea has merit, but also some problems, eh?
  • ErNaErNa Posts: 1,738
    When it comes to communication between a source and a sink, what you say it not what is heard. It's a question of impedance matching. Wire wrapping point to point connection shortest path give minimum impedance variations, as all wires are equal . Only at the crossings interaction may occur. So creating meek and mild signals during a conversation meetings is part of the hardware/software protocol and only is valid information using a key, e.g. to give a news conference after contemplating. saying that, “us signals were kind of insulting” and he “will not be pushed around.” very honestly shows, that you used 5V output to an 1.8 input and the fuses were blown and had to be replaced. So a breadboard needs a proper ground plane. If you know a zero level, all information passed is related to this level. If you enter a casino, don't expect a game of chess. If you take part in a perpetual chess, you should be prepared. And breadboard should not be confused with breitbart, otherwise you will remain confused in solitude.
  • escherescher Posts: 138
    edited 2018-06-14 20:29
    ErNa wrote: »
    e.g. to give a news conference after contemplating. saying that, “us signals were kind of insulting” and he “will not be pushed around.”... If you enter a casino, don't expect a game of chess. If you take part in a perpetual chess, you should be prepared. And breadboard should not be confused with breitbart, otherwise you will remain confused in solitude.

    4335715-really.gif

    For Parallax's recommended Wiring of the breadboarded dip propeller see:

    Figure 3-6 page 2 of "32305-PEKit-40-PinDIP-Enhancements-v1.0.pdf" or

    Figure 3-9 page 38 of "122-32305-PE-Kit-Labs-Fundamentals-Text-v1.2.pdf".

    Source - Parallax Shop - "Propeller Education Kit - 40-Pin DIP Version"
    Product ID: 32305
    >Downloads & Documentation

    Awesome resources, thanks! Why aren't these diagrams in the P8X32A base datasheet?
  • I agree with the fellow wearing an odd hat.
  • I definitely know what you mean - I have not been able to come up with a good solution to this. I sort of compromise by putting arrays of things such as resistors, buttons and LEDs on strip board with male headers that plug into a breadboard and flying leads for power. It can save a lot of space. I also use Adafruit protoboards so I can take a stable breadboard design and just solder it up. These are by no means solutions!

    I did come across this, but I haven’t tried it yet:

    http://www.roadrunnerelectronics.com/epages/BT3782.sf/en_GB/?ObjectPath=/Shops/BT3782/Categories/Wiring_Pencils
Sign In or Register to comment.