EN 60950 and component heating
Hello All, In testing some product for excessive temperatures I have come up against the following problem. Consider a diode (part of a bridge rectifier circuit) and the PCB underneath the component. If one measures the temperature of the diode it does not come close to the specification for the part. However if we place a thermocouple on the pad where the diode is attached to the PCB and we consider that as a temperature measurement for the PCB material itself ,then the temperature obtained is above the board manufacturers spec of 110 C (when we take into account our maximum permissible ambient temperature of 49 C). By the by all this is NOT operator accessible if that makes any difference. Questions. 1. Is this a valid temperature measurement for the PCB? I'm of two minds on this. It could said that I'm really measuring the diodes temperature and not the PCB. On the other hand the diode pad does touch the PCB . 2.Would it be more reasonable to measure the temperature in the same neighborhood as the pad but make sure that the probe does not touch the PCB pad? Would a notified body be of the same opinion? 3.If you feel that the first method is a valid measurement technique then do you know of any ways to work around the problem? There are many power devices that can easily and safely exceed a PCB material spec of only 110 C. Thanks for your opinions! Best Regards, Kevin Harris email harr...@dscltd.com
RE: EN 60950 and component heating
Kevin, Good question. By measuring the pad there are several items to consider. 1st, is the thermocouple in electrical contact with the pad? If so, you may have an incorrect reading caused by currents from the pad through the probe. 2nd, In measuring the pad, you are measuring the junction temperature of the soldered connection. This may or may not be a valid measurement. It is if you want to find out if the solder is going to melt or get stress cracks from repeated heating and cooling. It does not necessarily represent what the pcb material itself is seeing for temperature. 3rd, my practice is to measure temperature of the pcb near the device, either next to or underneath it, depending. 4th, to decrease the temperature of the pad, try adding more copper aorund the pad. A larger surface area, especially if on both sides of the board, will spread the heat out more. You may need to provide multiple current paths to the pad to keep one of them from heating up more than it should. You can also add ripples or bumps to the copper to increase surface area even more. 5th, if you are more concerned with the pcb and not the pad, then you might try moving the component off the board using longer leads, standoffs, etc. I have placed up to 25 watt resistors on boards by using longer leads and leaving an air gap of 1/4 to 1/2. Last, there are always heatsinks and fans to use to keep component temperatures, and thus pads and boards, cool enough to eliminate problems. And since it is a non-user access area, that makes it even easier. Regards, Scott Douglas Principal Compliance Engineer ECRM Incorporated Telephone: 1-508-851-0207 Facsimilie: 1-508-851-7016 e-mail: sdoug...@ecrm.com ___ From: Kevin Harris on Tue, Sep 16, 1997 12:35 AM Subject: EN 60950 and component heating To: EMC-PSTC (E-mail) Hello All, In testing some product for excessive temperatures I have come up against the following problem. Consider a diode (part of a bridge rectifier circuit) and the PCB underneath the component. If one measures the temperature of the diode it does not come close to the specification for the part. However if we place a thermocouple on the pad where the diode is attached to the PCB and we consider that as a temperature measurement for the PCB material itself ,then the temperature obtained is above the board manufacturers spec of 110 C (when we take into account our maximum permissible ambient temperature of 49 C). By the by all this is NOT operator accessible if that makes any difference. Questions. 1. Is this a valid temperature measurement for the PCB? I'm of two minds on this. It could said that I'm really measuring the diodes temperature and not the PCB. On the other hand the diode pad does touch the PCB . 2.Would it be more reasonable to measure the temperature in the same neighborhood as the pad but make sure that the probe does not touch the PCB pad? Would a notified body be of the same opinion? 3.If you feel that the first method is a valid measurement technique then do you know of any ways to work around the problem? There are many power devices that can easily and safely exceed a PCB material spec of only 110 C. Thanks for your opinions! Best Regards, Kevin Harris email harr...@dscltd.com -- RFC822 Header Follows -- Received: by macgtwy.ecrm.com with SMTP;16 Sep 1997 00:34:53 -0400 Received: by highlight.ecrm.com (AA10846); Mon, 15 Sep 97 23:30:10 EDT Received: from ruebert.ieee.org by maildrop.ecrm.com (XAA13124); Mon, 15 Sep 1997 23:31:42 -0400 (EDT) Received: (from daemon@localhost) by ruebert.ieee.org (8.7.5/8.7.3) id SAA03229 for emc-pstc-list; Mon, 15 Sep 1997 18:11:39 -0400 (EDT) Message-Id: 21ED5A1AFBBFD011B07000805F49DF4309252E@NT_MAIL From: Kevin Harris harr...@dscltd.com To: EMC-PSTC (E-mail) emc-p...@ieee.org Subject: EN 60950 and component heating List-Post: emc-pstc@listserv.ieee.org Date: Mon, 15 Sep 1997 18:14:01 -0400 X-Priority: 3 Mime-Version: 1.0 X-Mailer: Internet Mail Service (5.0.1457.3) Content-Type: text/plain Sender: owner-emc-p...@majordomo.ieee.org Precedence: bulk Reply-To: Kevin Harris harr...@dscltd.com X-Resent-To: Multiple Recipients emc-p...@majordomo.ieee.org X-Listname: emc-pstc X-List-Description: Product Safety Tech. Committee, EMC Society X-Info: Help requests to emc-pstc-requ...@majordomo.ieee.org X-Info: [Un]Subscribe requests to majord...@majordomo.ieee.org X-Moderator-Address: emc-pstc-appro...@majordomo.ieee.org
re:Re[2]: modeling RFI sources randomness
Moshe: One possible source of the phenomenon you've observed could likely be beating caused by assynchronous clocks. It's even more defined when the frequencies are slightly skewed by an amount slightly less than your receiver bandwidth. Take for example, you have two processors running off their own 100 MHz clock (VCXOs). These clocks are not only assynchronous but due to tolerances, there'll be some slight delta in actual operating frequency. The beat frequency formed by the vector addition of these sources at your transducer appears as an erratic emission (amplitude instability). There are two ways to verify this: - scan up to higher harmonics until the delta in frequency is greater than resolution bandwidth of the receiver--you'll observe two humps instead. - continue to lower the receiver bandwidth until the BW is less than the delta between the frequencies--you'll see two humps again. Regards, Geoff Skanes EMC Engineer Nortel In message Re[2]: modeling RFI sources randomness, moshe_vald...@isr-rhv-p1.ccmail.compuserve.com writes: I understand what you mean, but I'm not sure this is the whole story. I often check a product which is in idle mode, i.e. the processor is running in a tight loop, repeating itself every X microseconds (which is probably less than the measurement equipment integration time). In this situation everything should be static, but still the Spectrum Analyzer /receiver give an unstable reading (+/- several dB). What other factors could be related to this? Maybe it is related to the measurement equipment limitations? thanks for your opinions moshe valdman snip
RE: PCB Temperatures
Of equal concern for minor excessive temperatures, is that the bond strength of the copper on the PWB will be the first thing affected. The copper may pull away from the laminate due to thermal stresses and eventually result in high impedance connections. Such poor electrical connections could lead to the laminate decomposition mentioned by Bob DB or reduced clearances in a tight space. Regards, Peter L. Tarver Nortel ptar...@nortel.com -- From: DenBleyker, R. (919-543-7251 TL 441)[SMTP:bo...@vnet.ibm.com] Sent: Tuesday, September 16, 1997 4:41 AM In my opinion, the designer is responsible for the PC board temperature as well as the semiconductor. The PC board is both a structural element of the product maintaining creepage/clearance distances, and an insulator. PC board failure can easily result in a safety problem. Overtemperature will lead to decomposition of the epoxy binder and board faiure. Bobdb at vnet.ibm.com
Aircraft power
Hello everyone, Could someone please confirm that all commercial aircraft, world-wide, have North American outlets on board and that the power available at these outlets is 115 Volts at 400Hz. Thanks, Paul Schilke Advance Machine Company pschi...@advmac.com
Re: EN 60950 and component heating
To add to Art Michael's ways used to work around the problem of PCBs getting too hot, if the heat is being radiated (rather than conducted through the leads) from the component to the PCB, put a shiny reflective surface on the PCB (self adhesive aluminium foil, or just an area of copper on the top of the board under the component). Every little helps. Regards, Mark (whose PCBs went black and crunchy when they got a little too hot...) -- Mark Hone Wellman CJB Limited Email: m...@cjbdev.demon.co.uk Airport Service RoadTel: +44 (0)1705 664911 Portsmouth, Hampshire Fax: +44 (0)1705 697864 PO3 5PG, ENGLAND
RE: EN 60950 and component heating
We run into this quite often, and yes, I do consider the PCB measurement to be a valid measurement of the PCB temperature whether you are on a pad, a trace, or laminate. My experience with agencies is that they agree, and will allow (for example) a power resistor to be as hot as it's manufacturer says it can be, as long as the PCB underneath is not exceeding its limit. You may already be aware of this, but be careful of making measurements right on the pad (or any bare live part) for 2 reasons. The first is that some (inferior) temperature measurement equipment gives bad readings when noise is introduced onto the thermocouple by placing it on a noisy bare live part. It is easy to check for this by turning the EUT on and off and seeing if the temperature reading instantly changes by a large amount. The second thing to watch out for are the common-mode and channel-to-channel voltage ratings of the temperature meter We use one of 2 methods to get around the problem you describe. The first, if overhead clearance will allow it, is to raise the part off the board (usually needs 1/8 or more) using a spacer or by lead-forming a kink into the leads. The second is to put a barrier of some sort between the part and the PCB. We've seen as much as a 10C improvement on PCB temperatures when mounting hot parts flush on a piece of 0.010 Nomex. Hope this helps. Regards, Jim Eichner Statpower Technologies Corp. Burnaby, B.C., Canada jeich...@statpower.com Any opinions expressed are those of my invisible friend, who really exists. Honest. -Original Message- From: HarrisK@anetMHS (Kevin Harris){MHS:harr...@dscltd.com} Sent: Monday, September 15, 1997 6:20 PM To: emc-pstc@anetMHS (EMC-PSTC E-mail){MHS:emc-p...@ieee.org}; JEichner; bceresne Subject: EN 60950 and component heating Hello All, In testing some product for excessive temperatures I have come up against the following problem. Consider a diode (part of a bridge rectifier circuit) and the PCB underneath the component. If one measures the temperature of the diode it does not come close to the specification for the part. However if we place a thermocouple on the pad where the diode is attached to the PCB and we consider that as a temperature measurement for the PCB material itself ,then the temperature obtained is above the board manufacturers spec of 110 C (when we take into account our maximum permissible ambient temperature of 49 C). By the by all this is NOT operator accessible if that makes any difference. Questions. 1. Is this a valid temperature measurement for the PCB? I'm of two minds on this. It could said that I'm really measuring the diodes temperature and not the PCB. On the other hand the diode pad does touch the PCB . 2.Would it be more reasonable to measure the temperature in the same neighborhood as the pad but make sure that the probe does not touch the PCB pad? Would a notified body be of the same opinion? 3.If you feel that the first method is a valid measurement technique then do you know of any ways to work around the problem? There are many power devices that can easily and safely exceed a PCB material spec of only 110 C. Thanks for your opinions! Best Regards, Kevin Harris email harr...@dscltd.com
RE: EN 60950 and component heating
Good point. Local heating can damage a board locally. Although people often do not test the way you indicate, I've seen several failures due to long-term heat aging of the board under a hot component. Some things I've done: 1. Space the component body off the board and/or use longer leads to reduce heat flow to the board. Preformed leads or spacers help. 2. Use oversize pads and burried layers as a heat sink (heat spreader) to minimize local hot spots. 3. Use higher rated components which are physically larger (to get rid of heat better) or more efficient (to produce less heat). -- From: Kevin Harris To: EMC-PSTC (E-mail) Subject: EN 60950 and component heating List-Post: emc-pstc@listserv.ieee.org Date: Monday, September 15, 1997 3:14PM Hello All, In testing some product for excessive temperatures I have come up against the following problem. Consider a diode (part of a bridge rectifier circuit) and the PCB underneath the component. If one measures the temperature of the diode it does not come close to the specification for the part. However if we place a thermocouple on the pad where the diode is attached to the PCB and we consider that as a temperature measurement for the PCB material itself ,then the temperature obtained is above the board manufacturers spec of 110 C (when we take into account our maximum permissible ambient temperature of 49 C). By the by all this is NOT operator accessible if that makes any difference. Questions. 1. Is this a valid temperature measurement for the PCB? I'm of two minds on this. It could said that I'm really measuring the diodes temperature and not the PCB. On the other hand the diode pad does touch the PCB . 2.Would it be more reasonable to measure the temperature in the same neighborhood as the pad but make sure that the probe does not touch the PCB pad? Would a notified body be of the same opinion? 3.If you feel that the first method is a valid measurement technique then do you know of any ways to work around the problem? There are many power devices that can easily and safely exceed a PCB material spec of only 110 C. Thanks for your opinions! Best Regards, Kevin Harris email harr...@dscltd.com
Re[5]: modeling RFI sources randomness
How about another test case: A radio transmitter - carrier only. Turn it on. Then measure the RF output at my local test site. Does the spectrum analyzer report a constant/consistent amount? Seems to me we'd need to define fixed emission level and constant and consistent a little more carefully to nail this question down. In one case a clock is unintentionally radiating. It's total energy radiated is tons less than the intentional transmitter. So I'd guess it's closer to the background noise and harder to measure. In the other case the signal is larger and easier to measure. Still, I hear noise in the quiet time on the radio so I know the signal isn't perfectly consistent. Some measurements are just hard to do. Jon Bertrand j...@cirris.com __ Reply Separator _ Subject: Re[4]: modeling RFI sources randomness Author: moshe_vald...@isr-rhv-p1.ccmail.compuserve.com at Address-InternetPO List-Post: emc-pstc@listserv.ieee.org Date:9/16/97 1:58 PM I'll try to make it as simple as possible: Assuming I disable everything in a typical computer product, and leave just one oscillator with one clock trace, should I expect to see a fixed emission level in the clock harmonics? Did anyone try to perform such an experiment? Could minor variations in stray capacitance, shielding enclosure vibration etc, cause emissions variations? thanks moshe valdman mvald...@netvision.net.il ++ Moshe, Is there more than one clock source and is it at the same frequency? This happens many times where you get the symptoms you describe. What may be happening is the two sources are beating against each other, thereby giving you variations. Easy enough to confirm by temporarily removing one clock source. Eric Petitpierre Pulsecom Herndon, VA eric.petitpie...@pulse.com
Re: EN 60950 and component heating
Hello Kevin, This is not an uncommon problem as you have probably guessed. Rather than get embroiled in discussing the details of what one should measure, I'd rather offer a couple of ways I've seen used to work around the problem. A) Assuming you are using leaded diodes; Raise the diode off of the board and pass the leads thru tubular standoffs (rivets) which are staked to the board prior to wave soldering or pass the leads thru glass or ceramic beads prior to insertion in the board. Another standoff scheme is to bend a V into the leads prior to insertion into the board. I've seen diodes raised as much as 2 cm off the board. B) Use a higher temperature-rated circuit board. C) Use a combination of A and B, noted above. Let us know how this works out. Regards, Art Michael, Editor * * * * * * * * * * * * * * * * * * * * * * * International Product Safety News * *Check out our current offer on the * * Safety Link at http://www.safetylink.com * * * * * * * * * * * * * * * * * * * * * * * --- On Mon, 15 Sep 1997, Kevin Harris wrote: Hello All, In testing some product for excessive temperatures I have come up against the following problem. Consider a diode (part of a bridge rectifier circuit) and the PCB underneath the component. If one measures the temperature of the diode it does not come close to the specification for the part. However if we place a thermocouple on the pad where the diode is attached to the PCB and we consider that as a temperature measurement for the PCB material itself ,then the temperature obtained is above the board manufacturers spec of 110 C (when we take into account our maximum permissible ambient temperature of 49 C). By the by all this is NOT operator accessible if that makes any difference. Questions. 1. Is this a valid temperature measurement for the PCB? I'm of two minds on this. It could said that I'm really measuring the diodes temperature and not the PCB. On the other hand the diode pad does touch the PCB . 2.Would it be more reasonable to measure the temperature in the same neighborhood as the pad but make sure that the probe does not touch the PCB pad? Would a notified body be of the same opinion? 3.If you feel that the first method is a valid measurement technique then do you know of any ways to work around the problem? There are many power devices that can easily and safely exceed a PCB material spec of only 110 C. Thanks for your opinions! Best Regards, Kevin Harris email harr...@dscltd.com
Re[2]: modeling RFI sources randomness
I understand what you mean, but I'm not sure this is the whole story. I often check a product which is in idle mode, i.e. the processor is running in a tight loop, repeating itself every X microseconds (which is probably less than the measurement equipment integration time). In this situation everything should be static, but still the Spectrum Analyzer /receiver give an unstable reading (+/- several dB). What other factors could be related to this? Maybe it is related to the measurement equipment limitations? thanks for your opinions moshe valdman ^ It is not the random amplitude and phase changes that causes the peak, average or the quasi-peak readings to vary. Peak is the maximum reading obtained, period. Average and quasi-peak are filtering functions with respect to time and they will vary by the content of the varying signal. As an example, a CW sine wave, will have an identical reading at peak or quasi-peak or average!. The sources of interference are not fixed as you stated. The only fixed signal may be the clock. The bus lines, the address lines and other frequently operating signal lines will vary depending on the operating system and programs the operate. Then you have disk drives, monitors, DMA functions, et all, ad infinitum et nauseum. So, the bottom line is that electronics today are not as simple as they appear! even if they are small! Hans __ Reply Separator _ Subject: modeling RFI sources randomness Author: Non-HP-mvaldman (mvald...@netvision.net.il) at HP-ColSprings,mimegw5 List-Post: emc-pstc@listserv.ieee.org Date:9/13/96 10:02 PM Hello everyone, As we know, while measuring RFI generated from a product at a specific frequency , we get randomly changing amplitude (and I guess phase is also changing with time). This is why peak, average and quai-peak are different. On the other hand, the sources of interference (e.g. conductors running clock si gnals) are fixed and one would expect them to radiate a fixed signal. What is the explanation for this effect? (I.e. is there a model describing what influences the radiation generated?) Was there academic work in this area? thanks - Name: moshe valdman E-mail: mvald...@netvision.net.il Phone: 052-941200 fax: 03-5496369 List-Post: emc-pstc@listserv.ieee.org Date: 13/9/96 Time: 22:02:44 You are most welcome to visit my homepage at: http://www.geocities.com/CapeCanaveral/5233/ -
Re[2]: modeling RFI sources randomness
I understand what you mean, but I'm not sure this is the whole story. I often check a product which is in idle mode, i.e. the processor is running in a tight loop, repeating itself every X microseconds (which is probably less than the measurement equipment integration time). In this situation everything should be static, but still the Spectrum Analyzer /receiver give an unstable reading (+/- several dB). What other factors could be related to this? Maybe it is related to the measurement equipment limitations? thanks for your opinions moshe valdman ^ It is not the random amplitude and phase changes that causes the peak, average or the quasi-peak readings to vary. Peak is the maximum reading obtained, period. Average and quasi-peak are filtering functions with respect to time and they will vary by the content of the varying signal. As an example, a CW sine wave, will have an identical reading at peak or quasi-peak or average!. The sources of interference are not fixed as you stated. The only fixed signal may be the clock. The bus lines, the address lines and other frequently operating signal lines will vary depending on the operating system and programs the operate. Then you have disk drives, monitors, DMA functions, et all, ad infinitum et nauseum. So, the bottom line is that electronics today are not as simple as they appear! even if they are small! Hans __ Reply Separator _ Subject: modeling RFI sources randomness Author: Non-HP-mvaldman (mvald...@netvision.net.il) at HP-ColSprings,mimegw5 List-Post: emc-pstc@listserv.ieee.org Date:9/13/96 10:02 PM Hello everyone, As we know, while measuring RFI generated from a product at a specific frequency , we get randomly changing amplitude (and I guess phase is also changing with time). This is why peak, average and quai-peak are different. On the other hand, the sources of interference (e.g. conductors running clock si gnals) are fixed and one would expect them to radiate a fixed signal. What is the explanation for this effect? (I.e. is there a model describing what influences the radiation generated?) Was there academic work in this area? thanks - Name: moshe valdman E-mail: mvald...@netvision.net.il Phone: 052-941200 fax: 03-5496369 List-Post: emc-pstc@listserv.ieee.org Date: 13/9/96 Time: 22:02:44 You are most welcome to visit my homepage at: http://www.geocities.com/CapeCanaveral/5233/ -