Yesterday 23:29
Forum: Proton Plus Compiler v3
Starter: Tim
Views: 0
Replies: 3
Yesterday 21:44
Forum: Proton Plus Compiler v3
Starter: IstvanK
Views: 0
Replies: 3
Yesterday 18:17
Forum: LCD module
Starter: towlerg
Views: 0
Replies: 2
Yesterday 18:06
Forum: Proton Users File Server
Starter: Dompie
Views: 0
Replies: 0
Yesterday 16:56
Forum: Proton Plus Compiler v3
Starter: Henk57
Views: 0
Replies: 10
Go to last post By: Lester
Yesterday 14:50
Forum: Website / Forum Issues
Starter: Lester
Views: 0
Replies: 16
+ Reply to Thread
Results 1 to 9 of 9
  1. #1
    Member matty's Avatar
    Join Date
    Dec 2003
    Posts
    115
    Thumbs Up
    Received: 6
    Given: 3
    Total Downloaded
    228.03 MB

    0 Not allowed!

    Default DSPIC33EP512MU810 - LTC2983 speed question

    Hello everyone,

    Project Update + speed question about an LTC2983 20 channel SPI temperature sensing chip

    Firstly thanks to Les for Porting a hardware SPI routine for the 33EP PIC family,

    This allowed me to communicate very quickly using the 33EP512MU810 with an LDC1000 Inductive sensor which I am using for vibration monitoring of a turbine shaft.

    The next part of the project is to have many lots of temperature inputs, Controlled by the same 33EP512MU810

    My spec is as follows

    24 * Type K Thermocouple amplifiers with cold junction compensation.
    12 * PT100 RTD Amplifiers

    So rather than having to design op amps to do all this my PCB has a very limited space.
    I found this great chip called a LTC2983 this is a 20 Channel SPI chip for reading RTD's and Thermocouples,

    The LTC2983 Is ideal, It does all the cold junction compensation for type K Thermocouples and all linearization for PT100 and thermocouples.

    I ordered the LTC2983 development Kit,
    It comes with a breakout board and a strange Demo MCU Called a Lindiono any way using the Linduino LT's version of an arduino UNO with there example C code I got the LTC2983 working and it looks a great chip and has amazing 24 bit resoloution for temperature readings.

    Then To My Horror I realised how slow the LTC2983 chip is,

    It explains after initiating a conversion cycle it takes 2 or 3 conversion cycles to measure each temperature channel. This equates to 167ms or 251ms.

    I cant get all my channels on one LTC2983 chip so I Planned to use three LTC2983 Chips split as follows

    LTC2983-1: 12 * Type K thermocouples
    LTC2983-2: 12 * Type K thermocouples
    LTC2983-3: 12 * RTD PT100 Sensors

    So to read all the channels it would take 251ms * 36 Channles = 9036ms = 9.036 Seconds to read all channels.

    My spec which I only realised after ordering the LTC2983 Development kit silly me !! is 100mS to read / scan all 36 temperature channels. Its an awfully harsh spec but 9.036 Seconds is like 90 times to slow!

    It says that you can start multiple conversions on the LTC2983 but the readings are still done consecutively as in one after the other.

    To conform this I asked a nice man from Arrow an applications engineer he answered saying the part only has 3 ADC's so the conversions cant be done in parallel.
    Even if I Initiate multiple conversions by writing bits B[4:0] then it will still take n x 167ms or n x 251ms per channel to get the conversion result. If you times that by my 36 bleeding channels it takes 9 Seconds to get all the results I need all the reults in 100ms thanks to my bosses harsh specification.....

    Im waiting to have this confirmed by Linear Technology Support but I am having trouble getting an answer from them.

    Can anyone confirm it if I Post a link to the LTC2983 data sheet I cant attach it its too big a file size.
    www.linear.com/docs/45742

    Q Does Initiating Multiple Conversions speed up the conversion time I don't think it does

    I will probably have to use some custom AD8494 IC's which are 8 pin type K Thermocouple amplifiers and have 24 of these chips going to 24 analogue inputs of the 33EP512MU810 then do all the dam linearization tables for each channel manually

    Thanks

    Matty

  2. #2
    Member tumbleweed's Avatar
    Join Date
    May 2011
    Posts
    265
    Thumbs Up
    Received: 32
    Given: 0
    Total Downloaded
    76.31 MB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    Does Initiating Multiple Conversions speed up the conversion time
    Maybe slightly, but not enough to make any real difference.

    Since you have three LTC2983's the biggest speed improvement gain would be to trigger all three and then go back and get the data. That'll give you a threefold increase in speed, but even at the 167ms speed you're still at 12x167=2secs.

    Getting 36 channels in 100msecs is improbably unless you use a boatload of external hardware. The internal ADC of the 33EP512MU810 isn't going to get you anywhere near the accuracy of the LTC part.

  3. #3
    Prolific Poster joesaliba's Avatar
    Join Date
    Sep 2004
    Posts
    2,632
    Thumbs Up
    Received: 61
    Given: 18
    Total Downloaded
    2.27 GB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    Matty,

    In past I have used the MAX6675 for K-type thermocouples. There is no need to do the linearization tables. However, I never bothered how fast they actually work, as I only use one channel.
    Regards

    Joseph

  4. #4
    Prolific Poster joesaliba's Avatar
    Join Date
    Sep 2004
    Posts
    2,632
    Thumbs Up
    Received: 61
    Given: 18
    Total Downloaded
    2.27 GB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    A small correction for the above. The MAX6675 was superseded by the MAX31855. Just make sure to order the K-type.
    Regards

    Joseph

  5. #5
    Prolific Poster joesaliba's Avatar
    Join Date
    Sep 2004
    Posts
    2,632
    Thumbs Up
    Received: 61
    Given: 18
    Total Downloaded
    2.27 GB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    And the MAX31865 for PT100. See this post here.

    And here is the code for the MAX31855. Thanks to Ken.
    Regards

    Joseph

  6. #6
    Prolific Poster joesaliba's Avatar
    Join Date
    Sep 2004
    Posts
    2,632
    Thumbs Up
    Received: 61
    Given: 18
    Total Downloaded
    2.27 GB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    What I am interested in is the LDC1000 and how you use it. I have a model turbine and was wondering if it can be used for balancing.
    Regards

    Joseph

  7. #7
    Member matty's Avatar
    Join Date
    Dec 2003
    Posts
    115
    Thumbs Up
    Received: 6
    Given: 3
    Total Downloaded
    228.03 MB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    Dear Tumble Weed,
    Thanks allot for taking the time to have a look at the LTC2983 for me.

    I have just heard back today from official applications engineer at Linear Technology and he has confirmed Yes it does take 167ms x n or 251ms x n per channel even if multiple conversion are initiated.

    It is such a shame because the LTC2983 saves so much board space Is ideal but it just isn't fast enough.

    The I/O Project connects to a PLC running a Gas turbine, the PLC Governor program that controls the turbine has a recursion rate or program cycle of < 100ms. My boss says to protect the turbine from over-temperature this I/O Module must scan all the temperatures within 100mS to quickly protect and shut down the turbine if a high temperature quickly develops.

    The man from Linear technology suggested another part to me LTC2249 this is a 24 bit A/D convertor 8 Channel's, There is an application note called DN400F.pdf from linear showing how to use the LTC2249 with a RTD and a Thermocouple.

    I don't like the application note as it shows it needs 4 Channels to measure 1 RTD,
    Also in the application note there is no Cold Junction Compensation for the Type K
    It Looks to complex and the SPI Interface looks hard and at 24 bits would need very special PCB layout etc.

    My other idea was for the 24 Type K thermocuples use an AD8497 amplifier which gives 0-6v for 0-1200 deg c
    It has all the type K cold junction and some linearization built in.

    I agree the resoloution would not be any where near as good
    A/D 33EP512MU810 = 12 bits 0-3.3v
    Temp span = -50degc to 1200 deg c conditioned to 0-3.3v
    Resoloution would be poor
    1250degc range / 4095 bits = 0.305 degc / per bit
    My boss said might just get away with this but is not as good as the LTC2983 by a long way.

    The only other solution get a fast 14 bit or 16 bit adc which operates over SPI then have all the AD8497 thermocouple amplifiers feed into that to get a better resolution.

    I would need several 8 channel ADC's or use one and have a 24 way MUX switch with CMOS feeding each channel at a known time into an external single channles SPI high resolution 14 or 16 bit A/D Convertor.

    Either way one thing is clear Its going to Need A boat Load of Hardware to do this a big boat load. It he accepted 2 seconds I could use 3 * LTC2983 Chips on 3 Spi Ports trigger them all at one to do the 12 channles each then come back 3 seconds later and read all the temperatures.

    But 3 seconds is to much 100mS is such a harsh spec that's why I wanted to use this 33EP512MU810 as it is much faster than the 18Fxx Parts I am used to designing with and I knew this I/O Expansion board has to be lightening fast.

  8. #8
    Member matty's Avatar
    Join Date
    Dec 2003
    Posts
    115
    Thumbs Up
    Received: 6
    Given: 3
    Total Downloaded
    228.03 MB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    Hi joseph,
    Thanks for the helpful reply.

    In my specification, I must read 24 * Type K Thermocouples and 12 * RTD PT100 sensors all within 100ms

    The reason for the stupidly fast spec this is an I/O Expansion board for a PLC governing a GAS Turbine. According to my manager the PLC governor has a cycle time of <100mS . For protection of the turbine the this I/O Card must scan all the temperatures within 100mS to protect the turbine quickly and tell it to shutdown if a high temperature is sensed.

    The MAX 31865 for PT100 according to the data sheet takes 21mS To do a conversion cycle.

    If I had 12 * RTD so 12 * MAX31865 Chips with 12 chip selects then maybe I could initiate a conversion on many chips then come back more than 21ms later and read all the results.

    Similarly, MAX 31855 K type Thermocouple amplifiers, the data sheet says it takes 70ms to do a conversion.

    Maybee I could have 24 * Max31855 Chips, with 24 different Chip select lines,
    I could initiate a conversion on multiple chips,
    Come back around 70 + ms later and clock all the data in and read all the channles.

    I remember doing something similar with a DS1820B + dallas one wire temperature sensor.

    The DS1820B+ wire can take about 600ms-750mS to do a temperature conversion.
    In my program the program was very busy doing other things so I wasn't going to hang - around for 650mS-750mS waiting for the conversions to complete,

    If I remember I initiated a conversion DQ $44 then came back much later and read the results after the main program ran doing many other things.


    I wonder If I can ask Ken about this?
    I saw his example code and he used 50 of these MAX31855 K type chips on one board.
    If the data sheet says it takes 70mS to do a conversion on the MAX 31855, surely you can initiate a conversion using all the chip selects on several MAX 31855 chips, then go off and do something else the come back over 70mS Later and read all the data ??

  9. #9
    Member matty's Avatar
    Join Date
    Dec 2003
    Posts
    115
    Thumbs Up
    Received: 6
    Given: 3
    Total Downloaded
    228.03 MB

    0 Not allowed!

    Default Re: DSPIC33EP512MU810 - LTC2983 speed question

    Joseph Regarding the LDC 1000 it is an Inductance to digital convertor.

    I use it to measure vibration of a turbine shaft.

    Basically you have inductive accelerometer probe it faces the turbines rotating shaft.
    You have a tiny gap 0-3mm = 0-32,000 digital counts.
    As the turbine shaft rotates if it becomes of balance the 0-32,000 proximity register changes,
    The raw proximity data then needs to be differentiated and low pass filtered eventually to give vibration.

    Pictures and all details of hard ware software LDC100 in this post # 27 by me a few months ago
    http://www.protonbasic.co.uk/showthr...r-33EP512MU810

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

     

Similar Threads

  1. Replies: 29
    Last Post: 19th May 2017, 18:51
  2. New Analog Devices highest precision highest speed motor speed sensor.
    By normnet in forum New devices and techniques
    Replies: 0
    Last Post: 2nd November 2014, 11:22
  3. Replies: 0
    Last Post: 2nd November 2014, 11:22
  4. SPI master vs. speed clock differential question
    By PicNoob in forum Proton Plus Compiler v3
    Replies: 4
    Last Post: 25th January 2011, 22:07
  5. AC speed controler Drive question
    By Tim in forum The Lounge
    Replies: 12
    Last Post: 28th April 2006, 10:30

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts