DS2438 support
- Thread starter CAI_Support
- Start date
CAI_Support
Senior Member
How do you want to control sampling rate?
I don't think keep sampling the chip would change its thermal behavior or generating heat to affect its temperature or voltage measurement. That is a good question for Maxim, at what sampling rate it would cause their sensor chip introduce error. From Maxim datasheet, their sensors are all calibrated, that means accuracy not affected by environment temperature change.
I don't think keep sampling the chip would change its thermal behavior or generating heat to affect its temperature or voltage measurement. That is a good question for Maxim, at what sampling rate it would cause their sensor chip introduce error. From Maxim datasheet, their sensors are all calibrated, that means accuracy not affected by environment temperature change.
Efried
Active Member
You trigger the measurements via software all the time continuously and provide them for the user code in advance. Or do you start data acquisition only if a value is needed by the user code? Of course operation of the sensor heats it up and leads to higher temperatures if it is operated in an environment where no fast heat transfer takes place, but may be everybody knows that we have +- 0.5K accuracy
last Page http://www.micropik.com/PDF/ds1820.pdf
thanks
last Page http://www.micropik.com/PDF/ds1820.pdf
thanks
Efried said:
0.5 deg K ACCURACY but 0.06 deg K PRECISION.
Even if you don't need the accuracy, the precision is often nice to have - just to observe trends, to average etc. And with 0.06 degrees PRECISION, self-heating may be more evident.
That said, it is my understanding that the WC8 is constantly reading the attached sensors via background scheduler, and updating in-memory copies, so the code doesn't have to block/wait while it reads a sensor. I've never noticed a WC8 pause to read a sensor, but neither have I seen a problem with temperature readings that I can attribute to the frequency of reads.
Bassmanisme
Member
I have never had a problem with any 1 wire chips heating up on wc8 or any other host system. Why would they design a chip that has temperature measurement built in if it had a heating issue at recommended system timing? Doesn't make sense to me.
CAI_Support
Senior Member
Efried, What you refer to is DS1820 chip spec, that is 9 bit accuracy. I believe Maxim calibrated in its range error not exceeding specified, which in DS1820 is +/- 2C. If you really concern about accuracy, then we would recommend to use DS18B20, which is 12bit conversion, accuracy is +/- 0.5C. DS2438 will have even 13bit conversion, its accuracy is +/- 2lsb, so that accuracy is 2x 0.03125C degree or 0.0625C degree. Maxim guarantees its accuracy in the whole measurement range. If anyone doubts Maxim accuracy claim, it needs to have test result to proof. I think even I blow air use mouth to the sensor will introduce more error than sensor itself due to conversion.
We do sensor reading in background, so that PLC execution will not be slowed down. DS18B20 or DS1820 sensor takes 200ms to finish conversion. We checked with Maxim that DS2438 only takes 10ms to finish conversion. That would give us possibility to poll the chip more often. However, we do not plan to let user to change our background job scheduler, since we sliced time properly to make sure everything executed fine. If user changed the scheduler timing, that could cause something not working. If you have reason to concern certain thing, please tell us exactly the concern, so that we can tweak the scheduler for making that working better.
DS2438 has CCA and DCA register running internally in the background every 3ms to update its value all the time, that is not changeable by us or user. We do not plan to pull those value by WC scheduler, rather when user issue the PLC command to read them, it would slow down PLC execution. In that way, we can save RAM. Since they probably only read once a day, we think that slowness is okay to most people.
Once this is done, whoever having bootloader can request this firmware download to update his boards. We will then based on feedback tweaking it to work best we can.
We do sensor reading in background, so that PLC execution will not be slowed down. DS18B20 or DS1820 sensor takes 200ms to finish conversion. We checked with Maxim that DS2438 only takes 10ms to finish conversion. That would give us possibility to poll the chip more often. However, we do not plan to let user to change our background job scheduler, since we sliced time properly to make sure everything executed fine. If user changed the scheduler timing, that could cause something not working. If you have reason to concern certain thing, please tell us exactly the concern, so that we can tweak the scheduler for making that working better.
DS2438 has CCA and DCA register running internally in the background every 3ms to update its value all the time, that is not changeable by us or user. We do not plan to pull those value by WC scheduler, rather when user issue the PLC command to read them, it would slow down PLC execution. In that way, we can save RAM. Since they probably only read once a day, we think that slowness is okay to most people.
Once this is done, whoever having bootloader can request this firmware download to update his boards. We will then based on feedback tweaking it to work best we can.
Efried
Active Member
Many thanks for explaining. In most cases this regime is beneficial, in some it is not. I'm not talking about resolution or chip errors but influencing the sensor by applying frequent measurements dissipating heat in the sensor. I would appreciate if I could control the intervals for two reasons: 1. heat build up 2. power consumption and may be a third life time of the sensor.CAI_Support said:
CAI_Support
Senior Member
DS2438 is already doing CCA and DCA measurement every 3ms, as long as power is connected. What our measurement is way less frequent than that. If there is any heat built-up, the internal logic from the chip itself is way higher than anything we do. We measure the value in more than 100 times less frequent than chip itself doing. Any heat buildup, power consumption or life expectancy impact from WC logic will also be < 1/100 of chip itself doing. We are sorry that we don't see any user control would benefit. Please note, this chip during its conversion uses only 50 micro amp typical, max 100 micro amp, that is 0.1mA.
Because this chip is constantly doing CCA and DCA measurement, it requires power pin being connected all the time.
Because this chip is constantly doing CCA and DCA measurement, it requires power pin being connected all the time.
Efried
Active Member
CAI_Support said:
and the shut down current would be 25 µA
http://elcodis.com/parts/6055367/DS2438_p27.html#datasheet
May be this heat dissipation problem is not so interesting, and the frequent polling not the problem for relability and we should talk more about automatic identification of broken wires... Sorry this is the aftermath of some problems I had with sensors going off-line. May be the new boards allow rebooting so the sensors are initialized and manual rebooting not necessary any more...
CAI_Support
Senior Member
But we have not released any firmware with this sensor support yet. How could it goes off-line? Did you get any testing firmware from someone else?
CAI_Support
Senior Member
Different sensor chip uses different logic, due to command structure and data structure are different, processing speeds are also different. However, they do co-exist in the same 1-wire bus.
Efried
Active Member
CAI_Support said:
ok, but I wish you could introduce a mechanism able to recover an unbound sensor state, if my idea of taking stress from sensors polling them less often is not an option.
The need for a board reset was already discussed. If not feasible I have to implement a watchdog myself.
CAI_Support
Senior Member
When an 1-wire sensor running, it should always on the bus. However, external noise or bounce signal on the same bus could cause problem to the 1-wire sensor operation. To find the problem and fix the problem is most important than restart the board. It is not easy for most users to watch the 1-wire bus and determine the cause of the problem. However, make some adjustment on TSDO value could help in many cases when 1-wire sensor not reliable stay on bus. The reason could be the sensor has slightly different timing then others. I don't know your board firmware level. If that firmware is not the latest, we would recommend you to update to the latest one. The latest 3.02.17g firmware does have a change to 1-wire bus timing to tolerate wider response time. If you do need firmware update, please contact us directly through email or our web site. We want to thank you for your contribution of questions.
With all those said, I would also encourage you to start a new discussion subject about 1-wire bus communication, so that everyone can contribute and share on that topic. This one is for discussing DS2438 sensor specific questions, so that when later users need to report problem or to contribute or to search about DS2438, it is little easier for them to concentrate on this. If you have a specific problem, starting a new topics not only make it easier for others to contribute, but also easier to be searched later on by others.
With all those said, I would also encourage you to start a new discussion subject about 1-wire bus communication, so that everyone can contribute and share on that topic. This one is for discussing DS2438 sensor specific questions, so that when later users need to report problem or to contribute or to search about DS2438, it is little easier for them to concentrate on this. If you have a specific problem, starting a new topics not only make it easier for others to contribute, but also easier to be searched later on by others.
In a previous thread CAI support posted that change to network settings acts like a power up reset. You don't actually need to change anything just hit send. That works well if you need to manually reset the board.Efried said:
To do it automatically would need to dedicate an output pin and Vcc switch to power cycle the WC.
Just to be clear, WC does recover from temporary 1-wire bus problems but it does not rescan the bus, so no plug and play.
/tom
