Voltage regulator getting hot

I will update more later - The board has not failed - Its still working.
 
TTL outs - they drive simple ssds.
 
The reason I said "failing" because temp sensors start disappearing after few hours. After cooling the CAI- they come back!! (they are from Amazon, US supplier - So hopefully has some QC compared to Chinese suppliers)
 
The reason we need "industrial" boards is because they just need to work without thinking of cooling etc. The same box housed SSRs and PLCs - They work. Now I have moved it out in the sun(and they still work!)
The board is being used at roof(there are no furnaces here). The HVAC (Carrier) is on roof and in sun- And it houses a lot of electronics(their CCN network, which are 2 full sized ISA sized boards), even LED display.
 
I understand there are a bit of compromises - 40 odd USD cannot be compared to 200+ Siemens PLC, but still - This is no outside use board. It needs protection - in the same league that of pi boards etc.
 
PS - Direct sunshine is usually 5-7C higher as we measure on tiles on the roof. The Electrical box is open from below - so temp inside the box should be approximately the same as air temp outside- and today was 45 max.
 
More grief-
Looks like Ethernet has crapped out. However times I try to restart, ping, its not coming up. Hearbeat is on. Will try in evening when temperature is lower(for me to work!)
 
Point to note- This connectivity is provided by 39E homeplug, Tchibo - Made in China, which is subjected to same temperature as CAI. CAI is dead - Cheap Tchibo adapter still works(and its in sun)
 
I have wasted so much time on this so called childish hobby board which lables itself as "industrial board". My bad - I thought US companies have much better QC and less bragging  than Chinese companies. I was wrong.
 
Is THIS the same board you are using?
 
If so, it boasts an operating temperature of 85 degrees C (185 F)!  :eek:
 
webcntrlspec.jpg
 
 
I have only seen industrial components rated to a maximum of 60 degrees C, so I'm questioning those specifications.  Were they actually tested in an environmental chamber?
 
More grief-

Looks like Ethernet has crapped out. However times I try to restart, ping, its not coming up. Hearbeat is on. Will try in evening when temperature is lower(for me to work!)


Point to note- This connectivity is provided by 39E homeplug, Tchibo - Made in China, which is subjected to same temperature as CAI. CAI is dead - Cheap Tchibo adapter still works(and its in sun)

I have wasted so much time on this so called childish hobby board which lables itself as "industrial board". My bad - I thought US companies have much better QC and less bragging than Chinese companies. I was wrong


Definitely not "Industrial" quality. No industrial board would allow signals directly into CPU pins off a board without isolation and/or buffering.
 
The WC8 board has always had reliability problems with it's bit-banger 1wire interface. Some have had good luck with multiple sensors when they all have short cable lengths and  stuck to the basic temperature probes like DS1820 and DS18B20.  Once the Webset command and support for more complex 1wire chips were added the system craps out due to sharing memory causing the system processes to wait for each other to complete. The more complex and faster DS2438 has since been discontinued.  IOW you lose the multitasking ability of the WC8 system and when a 1wire device is hard to communicate with all other processes stop. My WC8 stops transmitting Ethernet information for 20-40 seconds at a time sometimes a few dozen times per day. Most others I have discussed this with have just thrown the board into their junk bin in frustration.
 
In short, get you PS voltage down, keep any voltage dropping parts outside the box, get the box protected from radiant heat, get a fan on it, keep your sensor leads very short, and don't expect 1wire devices to be running while the Ethernet processes are busy.
 
From all the users communicated with I can determine that none have problems with hardware 1wire interfaces on other systems using the same sensing probes ($1.95 from China) and very few have problems running multiple probes in WC8  versions of firmware before the Webset PLC command existed. Most are not aware as that communications stop at all and don't care in their applications. There are them that have a vested interest in promotion of the product.
 
The board is useful in certain limited applications but "industrial" is not one of them. I would have been fired years ago if I attempted to use one in my field of employment but as a hobby it could be fine. The RPi has so much more potential but the learning curve is much greater.
 
WebControl is a board build with industrial quality parts.  The CPU manufacture, Microchip, specified this particular CPU we are using only has Industrial grade, temperature range is -40C to +85C.  Other parts on board we also required to be able to operate in same temp range. We have tested the board operating in open at 45C, board working fine with free airflow from environment.  However, if you plan to use WC board to build a system in an enclosure, or operate at higher temperature, you will need to add cooling or active cooling into consideration.
 
Siemens PLC with its smallest system priced at $850 each includes cooling system design. In its spec stated if you allow designed air flow, it will tolerate up to 60C, and if you do not allow air flow, it can operate at 40C.  So that if you design a system using CPU to design your own CPU board, or buying WebControl as your system CPU board, you will need to put air flow and cooling as your own system design consideration.
 
We tested with good active cooling, most CPUs -- including WebControl CPU, will be able to work in 60C.  But in an enclosure without enough airflow, CPU temperature can go up quickly.
 
CAI_Support said:
WebControl is a board build with industrial quality parts.  The CPU manufacture, Microchip, specified this particular CPU we are using only has Industrial grade, temperature range is -40C to +85C.  Other parts on board we also required to be able to operate in same temp range.
Click to expand...
Sounds to me like you really need to change this specification. ;)
 
Thanks for the suggestion.  WebControl PLC board is based on our suppliers' spec.  Actually the whole heart of this problem is the temperature inside CPU.  Did the CPU over heated in the process?  I think it is possible.
 
From our measurement with 30C room air, CPU temperature can be higher than 50C. That means 20C temperature difference is required to open air cool the CPU at 30C.
But at 40C, its temperature difference to open air cool may require 30 degrees C, in another word, in open air, when room air is 40C, CPU can be as high as 70C.  That is still 15C below the CPU spec.  We tested at 45C in open air cooling condition, WebControl board is opening fine without problem.  We had customers told us that they put WebControl in 55C Arizona heat in open air passive cooling condition and did not have problem, but we did not test that ourselves.
 
In most applications, it will never need to worry about operating under 45C.  However, above 45C system designer should consider to turn on fan to active cool the board.  In Southern California, even in the Palm Spring, rarely it can reach to 45C or 113F.  With active cooling, a lot more molecules in the air bouncing off the chip heatsink and pins, carry the heat away from the CPU and other components, that will cool off the whole board much efficient than passive cooling.  Siemens PLC is a complete system with cooling parameters designed into its whole system, so that it can test and publish a more realistic number(40C blocked cooling and 60C with cooling).  WebControl board is a component in the final system, we should mention more about cooling, but it is hard for us to specify the actually temperature due to system designer may base on its environment to design cooling into his own system.
 
Siemens PLC also measures room air and shut off part of its system if temperature is above certain degree(like 45C).  If you are designing a critical system, those temperature warning and self protection logic could be programmed into PLC logic. At the minimum, to turn on a fan and send an email alert.  We hope this discussion will help most visitors of this forum, because  people always want to design a very reliable system, cooling is one area really need to consider also.  When NITIN100 first brought up this topics, we did not realize he was planning to use this in an closed environment with very high temperature, otherwise, we would suggest adding active cooling.
 
I see your points, but let me put it this way.
 
I also specify electronic equipment (PLC's, datalogger, communications, I/O, etc...) that have to operate in a Nevada desert, sometimes in an enclosure.  I look for temperature operation ranges and rarely choose anything under 60 degrees C and am willing to pay the extra fees usually required for this performance.
 
I see that 85 degree operating specification on your product and it immediately comes under scrutiny mainly because:
 
1. It's near impossible to accomplish with the components I suspect are installed
2. I doubt it has been truly tested in an environmental chamber to that specification.
 
Instead, you should quote a temperature spec that has been tested.  Not one that is based on the sum/guesses of the individual components the product is comprised of. 
 
If a user still wants to use your product, they will then provide cooling or some other means to get their environment to match your temperature spec.
 
Of course, (as all of my posts) this is JMO! ;)
 
Thanks for your opinion.  We should ask Microchip in what kind of condition their CPU is rated at 85C. So that we can test under same condition.  There is probably not many ways to mount SMT CPU to PCB, we do not think the way we put microcontroller on the PCB in anyway lowered chip's temperature rating.
 
Since this is a board product for system integrators, we have no control over how system integrator will cool the CPU and board.  Even we published our testing with method of cooling, the system integrators may not use same cooling method. Once we found out from Microchip how they tested the temperature range, we will recommend that to users, so that users in the extreme environment could reference Microchip method.
 
We appreciate all input and opinions.
 
Here is first answer from Microchip:

I expect the problem is that the device is in a box that the outside of the box is at 55C. Inside the box is probably much warmer, and possibly over 85C. There is no doubt that the CPU gets hot. The data sheet specifies well over 130 mA for an idling ethernet connection and up to 180 mA in operation (typ in both cases). If you multiply that by 3.3V you can see you are basically looking at ~400-600 mW which is quite a bit. In any case, I'd suggest checking the surface temperature of the device, the temperature of (what I assume is a) sealed box, etc. You may need to add some basic cooling such as a fan to your box.



 
 
We are still working to obtain more detailed temperature testing method they were using.
 
Another reply from Microchip Support engineer regarding this issue:

The device is designed to meet the stated specification of 85C. If it is failing at 55C, it most likely be due to overheating in the environment it is in.



 
 
We will not give up and keep asking the detailed temperature testing method.
 
Hello
I will chip in - (after the initial rage)
 
The board is in power db box, 45cx45cmx20cm. The box contains relays(no heat), power supply(little heat), SSRs(no heat as they drive relays). Cooling and air shifting is provided by 2 vented holes, approx. 10x20x20cm openings.Air shifting is required as humidity sensor is connected directly to the board.
 
I do think that mc is failing or hanging due to temperature, which is routinely 45C+ in the day. The board still works - It starts hanging in day(temp high) - where it may loose Ethernet and/or shows temp sensors unbound. Typical sign that temperature is doing its work on mc.
 
At the same time, a cheap powerline network adapter subjected to same stress works. The 12V power supply (dumbed down to 10.3v and further dropped by 0.7v by diode) works.
 
The siemens PLC I use are Logo kind - which are 80-120$ each(sometimes Chinese copies). I have never bothered to open even a small ventilation as they just work. A few are housed in IP65 boxes in sun.
When I read "industrial" word - I automatically assumed a few things.
 
Hi Nitin100,
 
Maybe you could do us a favor by measuring the temperature inside your box as well as on the CPU heatsink?  Microchip engineer asked for those temperature measurement.  If you have spare DS18B20, you can use the ground pin to tightly touch on the MC heatsink, since DS18B20 ground pin is the temperature probe.
 
We had tested before at 45C and not having problem. If your board quit working at 45C, maybe you could consider sending it back to us, so that we can check to see if any defective part on it.  Once you did temperature measurement, we will provide that to Microchip and working on this ticket.
 
As to Siemens PLC, its chassis has mostly top and bottom grid to allow air flow, also have active cooling inside. We don't know what do you mean logo kind.
 
Cai_support - Ofcourse the temps mentioned are from spare DS in the box (45C). Outside temp today (highest) was 39.7.
After 45C, the boards starts to hang(like network not working, DS sensors going unbound).
In evening, after reset, it works full night.
 
CPU = 30C+Room = 75C. So, my CPU is hanging at 75, which is very near to the operating limit.
The board has NOT failed yet.
 
Crux is - Please donot call this as industrial board - Its not. Usually all industrial stuff is supposed to work till atleast 65C (with or without active cooling)
The carrier CCN network I am mentioning doesnot have any active cooling. (Its a rooftop HVAC so I am guessing, they used industrial components)
 
Edit - Siemens Simatic logo is a plastic block, DIN mounted, no fan etc inside. Available on ebay for 20$ onwards used, 100$NIB. Tech sheet says -25 to +70C. It works.
Your tech sheet says +85...
 
Nitin100,
 
Do you consider take advice on Microchip engineer to enable a fan at 45C?  We would really appreciate if you could measuring the temperature both on MC and inside chassis, so that we can provide that last measurement before locking up to Microchip.
 
Our supplier Microchip in Arizona stated clearly this microcontroller is industrial grade, and its operation range is -40C to +85C.
We are still seeking clarification from them on this temperature measurement, so that can advise our customers how to achieve the +85C our supplier told us.
From what your reporting, it seems that the +85C  could be CPU core temperature.  If that was the case, we should change the spec to +45C without active cooling and +60-70C with active cooling.  We want to get confirmation from Microchip first.
 
Even without fan, using a larger heatsink on MC can significantly reduce CPU core temperature, especially if that heatsink has large enough area. Because hot air is lighter, built-in passive air tunnel in chassis can also act like little cooling air flow.
 
Siemens is our customers using our network equipment. We have a lot respect to them. Siemens logo has no network and it programmed through RS232 port. Many WebControl PLC functions do not exist in that.  If that meets all your needs, it is a good deal.
 
You must log in or register to reply here.

Similar threads

beelzerob
My house wiring odyssey
5 6 7
Replies
92
Views
18K
beelzerob
beelzerob
Back
Top