Shop OBEX P1 Docs P2 Docs Learn Events
Single sided SX28 PCB — Parallax Forums

Single sided SX28 PCB

LemonLemon Posts: 34
edited 2005-01-18 05:28 in General Discussion
I have spend·all night trying to figure out how to place the components for a sx28 circuit, on a single sided pcb, with both crystal (jumper select) and SX-Key connectors, but I just can't get it all to add up without the use of one or a couple of jumperwires.

Beeing a bit fanatic about the use of wires on a PCB, I hoped you guys could help me out.

Btw as I said, I haven't had much sleep, so I may or may not make any sence.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Why do·they call it a spacestation? There's hardly "space" for anything in there.

Comments

  • Guenther DaubachGuenther Daubach Posts: 1,321
    edited 2005-01-11 10:43
    Lemon,

    instead of wire jumpers, you could use zero-ohm resistors - at least, this makes the pcb look like not having any wire jumpers smile.gif .

    Seriously, I made various pcb designs for the SX in the past but never tried to design a single sided board, and I doubt that it is possible without any jumper wires, and if yes, traces would have to be very thin, and might get long.

    In the meantime, I prefer a four-layer design with one middle layer connected to Vdd, and the other one to Vss. This dramatically reduces the EMI generated by an SX clocked at 50 MHz. Even a double-sided board with as many ground planes as possible on both sides helps reducing EMI, but a single sided pcb will end up in a nice transmitter with strong harmonics up to 200 MHz and even above.

    The filter cap (typically 100 nF) connected between the SX supply pins does not really filter out those signals because even when you use a low inductance ceramic capactior, its leads are long enough to build an inductance, reducing the capacitor's filter effect at such high frequencies.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Greetings from Germany,

    G
  • ForrestForrest Posts: 1,341
    edited 2005-01-11 11:59
    I only see a single jumper on this single-sided SX28 design http://www.rickard.gunee.com/projects/video/sx/gamesys.php
  • Guenther DaubachGuenther Daubach Posts: 1,321
    edited 2005-01-11 16:21
    Forrest,

    yes, you are right, I also could only find one jumper but a lot of resistors that can act as "jumpers" as well, i.e. act as a bridge over say 5 to 10 pcb traces.
    In addition, the eight RB pins go "straight" through to the D-SUB connectors which makes the layout easier, and the four RA pins are not connected at all.

    I better correct my statement that I doubt it is possible to design a one sided PCB for the SX 28 without jumpers, and better say that it might be impossible, depending on the complexity of the design, and what external components are to be connected to the SX 28.

    Nevertheless, my major concern when preferring double sided, or even four-layer PCBs is the reduction of EMI produced by the SX. For a hobby application, this may not be too important but when you make designs for commercial applications this becomes an important issue. Without a carefully designed PCB, the system would hardly pass any EMI tests.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Greetings from Germany,

    G
  • LemonLemon Posts: 34
    edited 2005-01-11 17:29
    Gunther

    Hopefully, I will have to people to do the EMI tests, and find a solution to any problem. But thats not relevant at the moment, since I can't afford to even think "commercial application". Never the less, thank you for pointing that out - since it (and electrostatics) might actually explain why my SX responds to a button, even before i press it.
    About the 0ohm resistor, I can only say Clever! Now why didn't I think of that - oh yes, cause I'm not as pro as I would like to be.

    Forest

    Thank you, I'll take a look at that, as soon as I've had something to eat. [noparse]:)[/noparse]

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • Paul BakerPaul Baker Posts: 6,351
    edited 2005-01-11 22:49
    Speaking of EMI, does anyone know of literature discussing EMI and PCB design? I know the basics of trying not to have traces which are 1/4 or 1/8 wave and that sharp corners should be avoided but thats about it. I will have 2 moderate length and one shorter length 50MHz traces and 1 moderate length 100MHz trace, I'd like to not inadvertantly design something that will cause a stereo in the next room (or any other nearby electronic device) to behave strangely. Will using an aluminum project enclosure significantly reduce EMI as opposed to a plastic enclosure?
  • LemonLemon Posts: 34
    edited 2005-01-11 22:59
    Just curious, what is "moderate length" ? [noparse]:)[/noparse] And what does 1/4 / 1/8 wave mean?
    (I'm starting to get VERY interested in the EMI subject)

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • Paul BakerPaul Baker Posts: 6,351
    edited 2005-01-11 23:42
    My definition of moderate is in the ballpark of 2 inches, shorter is around 1 in. Wavelength = c/f, where c is the speed of light and f is the frequency. 1/4 and 1/8 wave is the wavelength divided by 4 and 8. If you hear the term "(some number) meter band" they are talking about the wavelength corresponding to a certain frequency. If you want to get really good reception of a particular band you make your antenna 1/2 or 1/4 the wavelength, depending on how much space you have to put an antenna, the same is true for transmitting antenna. The traces on a PCB act like little antenna, and if your not carefull you can end up creating a board which emits a considerable amount of EMI. The higher your operating frequency, the more of an issue this becomes.

    I plugged the frequencies into here: http://www.csgnetwork.com/freqwavelengthcalc.html·and the 1/4 wave for 50MHz is just shy of 5 feet, not a problem. But there are harmonics 2*f, 3*f etc which are also produced and a wrong trace length can act like a resonator for that harmonic wreaking havoc on your EMI numbers and can even cause problems within your circuit. I don't feel like going into an explaination of this because microwave transistor design was the one course in grad school I really struggled with, and it brings up really bad memories of weeks of little to no sleep and still bombing the exam cry.gif .

    Paul
  • LemonLemon Posts: 34
    edited 2005-01-12 01:14
    Just as I suspected. Even thou the sentences makes sence, and even thou the math makes sence, I have abso*******lutely no clue what I am supposed to do with that information.
    I don't mean to sound disrespectfull. The subject has cought my attention, and it seems to be a source of both bugs and hardwarefailure, but at the same time the main reason why I can control a model plane, while stading on the ground, with nothing but a pack of batteries and neatly wired transistors in my hand.

    Is there any simple way to secure hobby projects from EMI? Would a wider trace reduce EMI? Does it even matter when I'm operating my SX at 50mHz but all other hardware on (up to) around 20kHz?

    If it's too huge a subject, I'll just go ask google - I'll probably do that anyway, atleast sometime soon.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • LemonLemon Posts: 34
    edited 2005-01-12 01:28
    Guenther Daubach said...
    Lemon,

    Seriously, I made various pcb designs for the SX in the past but never tried to design a single sided board, and I doubt that it is possible without any jumper wires, and if yes, traces would have to be very thin, and might get long.

    In the meantime, I prefer a four-layer design with one middle layer connected to Vdd, and the other one to Vss. This dramatically reduces the EMI generated by an SX clocked at 50 MHz. Even a double-sided board with as many ground planes as possible on both sides helps reducing EMI, but a single sided pcb will end up in a nice transmitter with strong harmonics up to 200 MHz and even above.
    I probably underestimated the importance of EMI? I guess I've always had an if-it-works-thats-fine approach to techy things - which, at times, actually brings up some neat solutions to some heary problems.

    Please currect me if I'm wrong, but it seems you imply that long thin traces emit more EMI than... ? Short bold traces? I don't know if thats what you ment, I'm just trying to figure out what measures I have to take, to avoid building an EMI emitter, instead of a (sometimes) usefull application [noparse];)[/noparse]

    Also, would it actually helt to have ground traces all around the PCB and when posible connect the unused PCB space for large grounded areas?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • danieldaniel Posts: 231
    edited 2005-01-12 01:44
    Paul,

    Several posts ago you asked about EMI references.· Have you seen Dave Van Horn's EMI info (http://dvanhorn.org/Emi/Index.php)?· His strong opinions seem to have been molded in the crucible of experience.

    He would probably also answer a direct question about good EMI references.

    HTH,

    Daniel
  • LemonLemon Posts: 34
    edited 2005-01-12 02:02
    That's good reading daniel.

    I guess the answer to all my questions summs up to somethink like; avoid looping traces at all costs. Minimize the distance current has to flow and test the application next to an FM radio to get an rough indication of the emitted EMI.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • LemonLemon Posts: 34
    edited 2005-01-12 18:32
    I vas just thinking, is there a simple way to construct an application for mesureing EMI? Like a wire loop, a darlington transistor and an A/D converter, hooked up to an SX which displays the readout on a LCD display.
    With the right components and the right calibration, would it work for hobby-scale projects?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • Paul BakerPaul Baker Posts: 6,351
    edited 2005-01-13 06:22
    not really, maybe, but its way outside my expertise. Problem is EMI is harmonics of your operating frequency (if your running at 50MHz, then 100, 150, 200, ...). Its pretty hard to measure a 200MHz signal with a 50 MHz part (Google Nyquist Theorem) It states to capture an analog signal your digital system has to sample the analog signal at twice the highest frequency of the analog signal to be gaurenteed of capturing that highest frequency. Otherwise aliasing errors occurs (this is something every DSP course covers). Have you ever seen a video of a spinning wheel where the spokes seem to either stand still or move backwards? This is an example of an aliasing error of any·videotaping process and is akin to the problem with this idea.

    Post Edited (Paul Baker) : 1/13/2005 6:27:51 AM GMT
  • LemonLemon Posts: 34
    edited 2005-01-13 06:46
    I see. Well, I'll just let the EMI experts worry about EMI, and (besides applying the basic rules described) focus on making stuff work, and taering apart stuff that works, to se how and why, it used to work.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Why do·they call it a spacestation? There's hardly "space" for anything in there.
  • StephenStephen Posts: 53
    edited 2005-01-18 05:28
    If it's any help, here's an app note that discusses EMI with regard to the SX...

    Cheers!
    pdf
    pdf
    AN3_EMI.pdf
    174K
Sign In or Register to comment.