Shop OBEX P1 Docs P2 Docs Learn Events
PLL Reliability — Parallax Forums

PLL Reliability

BatangBatang Posts: 234
edited 2011-06-28 20:44 in Propeller 1
I noticed from browsing the Prop forums that PLL failure is mentioned many times.

As we are considering using the prop in a new design I would like to get some statistics from parallax about this,

Thanks.

Comments

  • LeonLeon Posts: 7,620
    edited 2011-06-10 02:23
    Problems with the PLL are caused by poor PCB design. If you decouple both supply pins it should be OK.
  • Heater.Heater. Posts: 21,230
    edited 2011-06-10 02:30
    As far as I know there is not much to be said about this. If you run the chip within spec, voltage wise and frequency wise. If you connect up all the power and grounds correctly. If you use adequate decoupling caps. In other words do what you would with any other device. Then it is very reliable and not an issue.

    Now, as with any device, if you go out of spec or abuse it something will fail. It's just that with the Prop the PLL seems to be the thing that fails first so it gets all the attention. Really it is as rugged as any other device. Perhaps more so.

    Of course if you are really serious about high reliability systems you will demand mean time between failure (MTBF) specifications from Parallax. As yet I have not seen such specs in any data sheet. I would suggest pursuing that with Parallax Semiconductor in that case.

    Anecdotally, myself and others have lashed up Props to bread boards willy-nilly, paying little attention to layout decoupling power supply etc, many times, with out any issues. Most failures seem to be attributable to abuse of I/O voltages and such.
  • BatangBatang Posts: 234
    edited 2011-06-10 06:17
    Thanks for the response guys.

    Decoupling caps and device used within spec are a given.

    I would still like to get the "official" response to this question though.
  • LeonLeon Posts: 7,620
    edited 2011-06-10 06:19
    Try Parallax Semiconductor support.
  • BatangBatang Posts: 234
    edited 2011-06-10 06:47
    It seemed logical to post here as there are suppose to be FAE's monitoring here.
  • LeonLeon Posts: 7,620
    edited 2011-06-10 07:25
    They seem to be conspicuous by their absence. :)
  • SapiehaSapieha Posts: 2,964
    edited 2011-06-10 07:33
    Hi Batang.

    I have worked/Experimented with Propeller and never had problem's with PLL up to 15MHz PLL8.
    (Over-clocking with standard Voltage with ambient temperature 0 to 40C)
    Only problems that I have read on was People that use 40 pin dip package in Bredboard with insufficient wire else omitted some Voltage wires. And even omitted Decoupling/Bulk capacitors.
    As Propeller can have very big differences IN curent consumption it needs for Voltage stability AT last one Bulk capacitor as near voltage pins on XTal side as possible.

    Batang wrote: »
    It seemed logical to post here as there are suppose to be FAE's monitoring here.
  • RaymanRayman Posts: 14,978
    edited 2011-06-10 07:33
    PLL failure is mentioned a lot because it's usually the first thing to break when you do something very wrong with your Prop.

    I wouldn't worry about it if you stay within the Prop's specs.

    But if you do want to treat your Prop badly, just use an external oscillator instead of the PLL. Then, it could stand up to more abuse.

    BTW: Parallax is on the West Coast, so you'll have to wait a while until they get out of bed...
  • MacTuxLinMacTuxLin Posts: 821
    edited 2011-06-10 11:46
    Sapieha wrote: »
    As Propeller can have very big differences IN curent consumption it needs for Voltage stability AT last one Bulk capacitor as near voltage pins on XTal side as possible.

    I'm sorry to intrude but when you said "Bulk cap", you mean 1uF?, 10uF?
  • Mike GreenMike Green Posts: 23,101
    edited 2011-06-10 12:35
    Usually you want something like 100nF (0.1uF) between Vdd and Vss on each side of the package, then something on the order of 10uF nearby (within a couple of inches as the PCB traces go).
  • SapiehaSapieha Posts: 2,964
    edited 2011-06-10 14:50
    Hi MacTuxLin


    You NOT intrude at all.

    As Mike said --> 10uF. Usually 10 to 33uF.

    Small ones Mice describe are to suppress AC bursts on top to DC to/from IC' Power traces.
    Bulk (10-33uF) to hold reserve current for IC with very fast swings in power consumption (Propeller are that one).


    MacTuxLin wrote: »
    I'm sorry to intrude but when you said "Bulk cap", you mean 1uF?, 10uF?
  • Daniel HarrisDaniel Harris Posts: 207
    edited 2011-06-10 20:12
    Hello Batang (and all),

    I think Heater said it pretty spot on:
    As far as I know there is not much to be said about this. If you run the chip within spec, voltage wise and frequency wise. If you connect up all the power and grounds correctly. If you use adequate decoupling caps. In other words do what you would with any other device. Then it is very reliable and not an issue.

    Rayman also said something that I think is very true as well:
    PLL failure is mentioned a lot because it's usually the first thing to break when you do something very wrong with your Prop.

    Recently, I asked Beau Schwabe about the PLL failing. He said that the most common cause is due to not powering all four (for QFP and QFN packages) or two (for the DIP package) power rails on the chip. Slight variation in voltage on the chip's internal power planes causes internal heating, which is a PLL's enemy.

    Bottom line: stick within the Propeller's absolute maximum parameters and you will have a very reliable processor.

    I suppose this is a good segue into MTBF/MTTF. We recently had a customer contact us for some information about the Propeller so that he could estimate MTBF. The estimation depends greatly upon operating conditions and the model for estimation being used. This customer was using the MIL-HDBK-217F-2 specification. I contacted Chip about the required metrics and he verified the estimation as per the model's given table values and the Propeller's design characteristics. The following is that info, though I would like to reiterate that this is just one particular model's estimate for MTBF.

    Handbook which contains specification (look at section 5-3):
    http://snebulos.mit.edu/projects/reference/MIL-STD/MIL-HDBK-217F-Notice2.pdf

    Using those formulas, values in tables, and some Propeller design metrics, the formula estimates approximately 109 failures per million hours.

    Again, I would like to say that this estimation is based off of a formula that uses some canned values and a few design metrics. I think a more accurate value could be obtained from the fab facility. They have invested serious time and money to properly estimate reliability for the particular process they used.

    Thank you for reading all of this. If I happened to mis-state something, feel free to call me on it :). I'll do my best to get you guys the most accurate information possible.

    -Daniel
  • BatangBatang Posts: 234
    edited 2011-06-10 21:49
    Daniel,

    Thanks for the info.

    What Is the recommended PCB trace layout for the QFN part, your documents to not appear to have this information.
  • MacTuxLinMacTuxLin Posts: 821
    edited 2011-06-11 04:27
    Thank you, Mike & Sapieha. Yes, have been placing the 0.1uF bypass cap but was not aware about another larger one for the prop. OK, engraved in my memory now :)
  • Kevin McCulloughKevin McCullough Posts: 62
    edited 2011-06-24 17:12
    Batang wrote: »
    What Is the recommended PCB trace layout for the QFN part, your documents to not appear to have this information.

    Hi Batang,

    We show a recommended land pattern and stencil for the P8X32A-M44 (44-pin QFN) on page 35 of the Datasheet.

    Is this what you were looking for? If you are looking for more of a reference design layout with additional components, then I would suggest possibly taking a look at the PropStick USB (although it may be a tighter form-factor than you may be looking for). http://www.parallax.com/Store/Microcontrollers/PropellerChips/tabid/142/ProductID/411/List/0/Default.aspx?SortField=ProductName,ProductName
  • Cluso99Cluso99 Posts: 18,069
    edited 2011-06-28 05:25
    Just to add my comments as I probably push the prop (yes you can shoot me for not using propeller) more than most. I run them at 6.5MHz PLLx16 = 104MHz with lots of good decoupling (4x 100nF, 1x 10nF, 1x 10uF all under the prop QFP, with a proper ground and power plane). Never killed a prop yet (now where is a lump of wood?) :)
  • LeonLeon Posts: 7,620
    edited 2011-06-28 05:33
    Don't you mean copper pour? Or are you using a four-layer board with power and ground planes?
  • Cluso99Cluso99 Posts: 18,069
    edited 2011-06-28 20:44
    Basically no & no.
    I only use 2 layer pcbs. I pour separate planes under the prop chip. The ground is under the chip (component/top side) and power is on the underside. The bypass and bulk capacitors are under the prop chip on the power plane. Then I connect those planes to the standard pours where ground is on the underside and power is on the topside. By hand routing, I ensure the bulk capacitor is directly on the power feed to the prop alone (sometimes with the eeprom) to maximise its effect for the prop.
Sign In or Register to comment.