Shop OBEX P1 Docs P2 Docs Learn Events
Design rules for stable propeller operation and overclocking — Parallax Forums

Design rules for stable propeller operation and overclocking

SapiehaSapieha Posts: 2,964
edited 2009-06-03 12:17 in Propeller 1
Hi All

As I stated it is possible to run Propeller with more that 80-96MHz.
But it is not posible on badly constructed PCB boards.

In nature Propeler chip is variable power consumption and that will have consequences with omit (BIG
4.7-10 uF if possibly Tantalum) capacitors Near Propeller chip parallel with one smal ceramic capacitor.
Standart Voltage Regulators can't handle that fast power variations withouth them.
Even with 80MHz it is recomended to add them. As many people have reported it have problem with PLL
without that capacitors.
It is normal behavior on PLL that it will have stable Voltage threshold to function corectly and with
omiting that capacitors it is not posible to have stable Voltage in Propeller chip.
With every COG start else exit from WAIT instructions Propeler chip dramaticaly change power consumption
and that have impact on PLL.
It even consume more power with unstable Voltage (Not all transistors on chip can work stable in ON and
OFF state) and that impact on Propellers function.


Best Regards.

Sapieha - Christoffer J

Comments

  • LeonLeon Posts: 7,620
    edited 2009-06-02 12:54
    I always use a 100 nF capacitor on each Vdd-Vss pair, close to the pins, with careful routing of the power tracks, with any MCU.

    Why "Sapieha", BTW?

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle
  • SapiehaSapieha Posts: 2,964
    edited 2009-06-02 13:00
    Hi Leon.

    Only 100 nF is to smal to handle fast power variations on Propeller chip.
    That can function on 80MHz with programs that not have many active WAIT instructions.
    But if You run extensive program that have all COGs on and many WAIT instructions that put temporary Propeller chip in LOW then HIGH power consumption it can have impact on its functionality.

    Ps. If You Overclock it that can have very big impact

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Nothing is impossible, there are only different degrees of difficulty.
    For every stupid question there is at least one intelligent answer.
    Don't guess - ask instead.
    If you don't ask you won't know.
    If your gonna construct something, make it·as simple as·possible yet as versatile as posible.


    Sapieha
  • Cluso99Cluso99 Posts: 18,069
    edited 2009-06-02 13:15
    Sapieha is saying you require a 100nF and a 4.7uF Tantalum capacitor close to the power pins of the propeller chip, especially near the crystal power supply pins. This becomes even more critical if you overclock or are running PLLx16.

    It seems that it is preferable to run a crystal at say 12.5MHz and PLLx8 rather than 6.25MHz and PLLx16. From what Sapieha has said, the crystal will always be multiplied up PLLx16 and then if necessary it will be divided back down. Now with a 6.25MHz crystal, no division would take place and the clock would "JITTER" causing more power usage and possible PLL failure. However, with a 12.5MHz crystal, a division by 2 would take place and the clock would be "Jitter" free.

    Thanks Sapieha smile.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Links to other interesting threads:

    · Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
    · Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
    · Prop Tools under Development or Completed (Index)
    · Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
    · Search the Propeller forums (via Google)
    My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
  • LeonLeon Posts: 7,620
    edited 2009-06-02 13:15
    100 nF is the usual value on much faster chips than the Propeller, often with a smaller value in parallel. Parallax only uses one capacitor on the Proto board, for some reason! I always use a 10 uF as well, of course, but it doesn't have to be close to the chip. A lot depends on how much current the outputs are switching, and how fast.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle

    Post Edited (Leon) : 6/2/2009 1:20:53 PM GMT
  • Cluso99Cluso99 Posts: 18,069
    edited 2009-06-02 13:45
    Leon, this is an issue with the prop requiring the bulk capacitor close to the power supply pins. It is much more important if overclocking, but is a recommendation to avoid pll failure due to the switching currents when coming out of wait instructions and starting cogs.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Links to other interesting threads:

    · Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
    · Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
    · Prop Tools under Development or Completed (Index)
    · Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
    · Search the Propeller forums (via Google)
    My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
  • SapiehaSapieha Posts: 2,964
    edited 2009-06-02 13:54
    Hi Leon.

    It is not only·Output pins that is problem.
    It is momentary power varies in Propeller core with Start of new COG else Exit from WAIT instructions.
    Standard chips have constant power consumption that give them more reliable Voltage to its Core.
    Propeller chip have Variable Core power consumption that give more requirements on power/Voltage stablity

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Nothing is impossible, there are only different degrees of difficulty.
    For every stupid question there is at least one intelligent answer.
    Don't guess - ask instead.
    If you don't ask you won't know.
    If your gonna construct something, make it·as simple as·possible yet as versatile as posible.


    Sapieha

    Post Edited (Sapieha) : 6/2/2009 2:06:34 PM GMT
  • Bill HenningBill Henning Posts: 6,445
    edited 2009-06-02 13:58
    Thank you for the information, I will update my PCB design today smile.gif
    Sapieha said...
    Hi Leon.

    It is not Output pins that is problem.
    It is momentary power varies in Propeller core with Start of new COG else Exit from WAIT instructions.
    Standard chips have constant power consumption that give them more reliable Voltage to its Core.
    Propeller chip have Variable Core power consumption that give more requirements on power/Voltage stablity
    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Largos - a nano operating system for the Propeller
    www.mikronauts.com - a new blog about microcontrollers
  • LeonLeon Posts: 7,620
    edited 2009-06-02 14:19
    Cluso99 said...
    Leon, this is an issue with the prop requiring the bulk capacitor close to the power supply pins. It is much more important if overclocking, but is a recommendation to avoid pll failure due to the switching currents when coming out of wait instructions and starting cogs.

    Parallax doesn't seem to bother about it, though. They only have one decoupling capacitor for the Propeller and a bulk capacitor by the regulator, a couple of inches away, on the Proto board.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle

    Post Edited (Leon) : 6/2/2009 2:25:51 PM GMT
  • BradCBradC Posts: 2,601
    edited 2009-06-02 14:22
    Cluso99 said...
    Leon, this is an issue with the prop requiring the bulk capacitor close to the power supply pins. It is much more important if overclocking, but is a recommendation to avoid pll failure due to the switching currents when coming out of wait instructions and starting cogs.

    I thought the deal with the smoked PLL's was more related to not having a sufficiently low impedance connection between all the propeller various power pins, not decoupling as such. I've been wrong before however.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    "VOOM"?!? Mate, this bird wouldn't "voom" if you put four million volts through it! 'E's bleedin' demised!
  • LeonLeon Posts: 7,620
    edited 2009-06-02 14:24
    Yes, that's what I thought. Some people seem to think that having two supply pins means that only one need be connected.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle
  • SapiehaSapieha Posts: 2,964
    edited 2009-06-02 14:29
    Hi BradC.

    You are corect on smoked PLL's.
    But JITTER on PLL's and bad behavior is direct related to decoupling capacitors (Rather momentary curent/Voltage pumps).
    10uF capacitors in that case acts as momentary bateries for Propellers CORE's/PLL's to give them stable Voltage.

    Ps. But it has same problem. To bad voltage to ON/OFF of PLL's transistors that can give them unreliable state's. And that can open 2 transistors in opposite state.



    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Nothing is impossible, there are only different degrees of difficulty.
    For every stupid question there is at least one intelligent answer.
    Don't guess - ask instead.
    If you don't ask you won't know.
    If your gonna construct something, make it·as simple as·possible yet as versatile as posible.


    Sapieha

    Post Edited (Sapieha) : 6/2/2009 2:37:40 PM GMT
  • LeonLeon Posts: 7,620
    edited 2009-06-02 14:35
    Why doesn't Parallax do it, then?

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle
  • SapiehaSapieha Posts: 2,964
    edited 2009-06-02 14:39
    Hi Leon.

    If You look on USB ProtoBoard them added 2 Tantalum capasitors for stablity on it.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Nothing is impossible, there are only different degrees of difficulty.
    For every stupid question there is at least one intelligent answer.
    Don't guess - ask instead.
    If you don't ask you won't know.
    If your gonna construct something, make it·as simple as·possible yet as versatile as posible.


    Sapieha
  • Cluso99Cluso99 Posts: 18,069
    edited 2009-06-02 14:59
    Leon, some pcbs obviously get away with it. This is newish information and is advice for new pcbs, especially for those who want to run higher clocks. With the extra tantalum cap(s) 100MHz using a 12.5MHz xtal seems to be a given.

    Sapieha has done a lot of work in overclocking and has discussed with Chip his findings.

    So 100nF, 4.7uF Tant, 12.5MHz, PLLx8 and 100MHz here we come... smile.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Links to other interesting threads:

    · Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
    · Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
    · Prop Tools under Development or Completed (Index)
    · Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
    · Search the Propeller forums (via Google)
    My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
  • LeonLeon Posts: 7,620
    edited 2009-06-02 15:03
    I can't see them on the schematic - just a single 1 uF and the 10 uF by the regulator. Perhaps it hasn't been updated.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2009-06-02 15:23
    One thing you have to watch out for with tantalums (or any cap for that matter) is the capacitance tolerance. I recently did a Prop design with a 3.3V regulator that requires at least a 10uF tantalum for stability. But, over temp, the tantalums I looked at in the required package size were all +/- 20%, so it was necessary to specify a 15uF cap.

    -Phil
  • SapiehaSapieha Posts: 2,964
    edited 2009-06-02 15:31
    Hi Phil Pilgrim (PhiPi).

    On my designs in that case I use 33uF Tantalums.
    4.7 - 10uF is smallest posible that give Propeller stablity and I never use that smal capacitors to stabilize Regulators Voltage if I have more that only one IC

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Nothing is impossible, there are only different degrees of difficulty.
    For every stupid question there is at least one intelligent answer.
    Don't guess - ask instead.
    If you don't ask you won't know.
    If your gonna construct something, make it·as simple as·possible yet as versatile as posible.


    Sapieha
  • ColeyColey Posts: 1,110
    edited 2009-06-02 19:35
    Great information, Thanks Sapieha cool.gif

    I too will take this into consideration for future designs....

    Regards,

    Coley

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    PropGFX - The home of the Hybrid Development System and PropGFX Lite
  • Bill HenningBill Henning Posts: 6,445
    edited 2009-06-02 19:46
    Personally, I'll stick with 6.25MHz x 16... I don't know the long term effects of running the main PLL at 200MHz, I'll feel much better about running the PLL at 100MHz.

    And I am NOT saying this because I am going to be selling 6.25MHz crystals, but because of the high PLL clock.

    I am however taking Saieah's excellent advice, and adding extra capacitors to my layouts as per his advice. 100MHz here I come!
    Cluso99 said...
    Leon, some pcbs obviously get away with it. This is newish information and is advice for new pcbs, especially for those who want to run higher clocks. With the extra tantalum cap(s) 100MHz using a 12.5MHz xtal seems to be a given.


    Sapieha has done a lot of work in overclocking and has discussed with Chip his findings.



    So 100nF, 4.7uF Tant, 12.5MHz, PLLx8 and 100MHz here we come... smile.gif
    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Largos - a nano operating system for the Propeller
    www.mikronauts.com - a new blog about microcontrollers
  • SRLMSRLM Posts: 5,045
    edited 2009-06-02 20:11
    So, is it safe to say that if you're overclocking you can add a whole bunch of capacitors from the 10nF range up the the 100uF range in parallel?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2009-06-02 20:38
    SRLM,

    Not all caps are created equal. Adding a "bunch of caps" is not the same as selecting those that do the job intended. You basically need two kinds: a filter cap, in the tens-of-uF range, that conforms to the requirements of the regulator, and multiple, localized ceramic bypass caps in the 0.1uF to 2.2uF range. The latter are used for transient control to eliminate power dips and ground bounce at a chip's supply pins. They need to have a very low ESR to be effective, which ceramic caps provide.

    -Phil
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2009-06-02 20:47
    To get the best decoupling capacitors you must add one that has infiite storage yet infintesimal responce time, these do not exist. Staggering the scale and type of the capacitors will help ie a small value one for extreme responce, one with better storage (but lesser responce) etc, etc ...

    Even then the inductance of any piece of the cct will affect their effectiveness, right down to the internal parts of the substrate itself. You could end up worrying about everything and building nothing. ESR is usually measured around 100KHz which is ok for ccts that don't pulse things at 100MHz. The slew rate of those 100MHz+++ clock edges are the cause of power drag and therefore heat. For stability that power must be available in an impedance suitable for those rates, this also means the chip will have more power to suck into itself, regardles if any of it is routed out again.

    This why there is a stated max to all those figures, after that "your on your own ..."

    Even my tormented Prop goes at 14.3MHz /8
  • BradCBradC Posts: 2,601
    edited 2009-06-03 07:35
    Sapieha said...
    Hi Leon.

    If You look on USB ProtoBoard them added 2 Tantalum capasitors for stablity on it.

    I'm fairly sure those Tantalums are there for a different reason, however I can't find the thread at the moment.
    Going from my rusty memory, the USB proto-board used a slightly different regulator configuration and there was a problem with some of the regulators squealing as they were unstable. This was worked around by adding those Tantalum caps. There was a long thread about it with lots of help and interaction from Paul Baker at the time.

    <edit> found it
    http://forums.parallax.com/showthread.php?p=735731

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    "VOOM"?!? Mate, this bird wouldn't "voom" if you put four million volts through it! 'E's bleedin' demised!
  • LeonLeon Posts: 7,620
    edited 2009-06-03 08:44
    It only has a single 1 uF capacitor for the Propeller, as I mentioned earlier, according to the schematic.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2009-06-03 12:17
    One thing that has always puzzled me is, why do 40 pin DIP chips like the prop and AVRs cross over the supply pins on their two sides?

    This must discourage the connection being put onto each without extra layout work. Not much I know, but threads seem to tell stories of this happening.
Sign In or Register to comment.