Post by thread:[EE] Sanity check about hipot testing

First, a little background. I have designed industrial control trinkets for a number of industries over the past 20 years or so. These have all been 10-30 VDC input devices. Some in metal enclosures, some plastic and some potted epoxy. These devices, depending on the industries involved have been UL / CSA / FM approved with no problems. I just ran my first one through ETL, the allegedly cheaper UL knockoff lab. Obscurely hidden in the end of the final report is a requirement to hipot test this device with 840VAC or 1080VDC for 2 seconds. The product is nothing more than an electronic chart recorder with no control capabilites whatsoever. It just monitors thermocouples, control currents and voltages. My big concern is that I have never had any requirements to hipot test any low voltage equipment - EVER. I've done many mains power driven devices and they usually get a hipot test. Common sense engineering would lead me to believe that hipot testing low voltage DC instruments with these levels for 2 seconds would surely damage parts of the instrument, like the 36V TVS diodes across the power inputs. So, did ETL make a mistake in specifying a hipot test? Or am I missing something in the UL-61010 requirements? Thanks, Bill

> So, did ETL make a mistake in specifying a hipot test? Or am I missing > something in the UL-61010 requirements? Possibly. OR, possibly, the requirement is there to see if the device explodes/outgasses something toxic/bursts in to an unquenchable flame. If you're doing safety testing, which is usually the primary concern of UL testing, the tests are more often about making sure the device fails gracefully than making sure it doesn't fail. You should DEFINITELY ask them both whether the test is truly needed AND what the point of the test is. If the person you ask doesn't know, push the matter- someone has to, or you're just being asked to destroy one of your (expensive?) devices for no good reason. Mike H.

Couple of things: 1) Why is the hipot test needed for a low-voltage device? 2) I *assume* that the hipot test is done the standard way: connect both power input terminals together, then apply the test voltage between those input terminals and the case / enclosure. Output terminals (if any) are floating. Is that in fact what they propose to do? Regardless, I'd suggest finding out why they think that hipot testing of a low-voltage device is necessary. dwayne At 09:41 AM 6/18/2008, William Bross wrote: >I just ran my first one through ETL, the allegedly cheaper UL knockoff >lab. Obscurely hidden in the end of the final report is a requirement >to hipot test this device with 840VAC or 1080VDC for 2 seconds. -- Dwayne Reid <spam_OUTdwaynerTakeThisOuTplanet.eon.net> Trinity Electronics Systems Ltd Edmonton, AB, CANADA (780) 489-3199 voice (780) 487-6397 fax http://www.trinity-electronics.com Custom Electronics Design and Manufacturing

Mike Hord wrote: >>So, did ETL make a mistake in specifying a hipot test? Or am I missing >>something in the UL-61010 requirements? >> >>You should DEFINITELY ask them both whether the test is truly >>needed AND what the point of the test is. If the person you ask >>doesn't know, push the matter- someone has to, or you're just being >>asked to destroy one of your (expensive?) devices for no good reason. >> >>Mike H. >> >> Thanks for the response Mike. I have. And the exact response I've received so far--because we said so. And yes, it definitely will kill low voltage devices. And it's not just supposed to be a sample, it's 100% of production. This is the same project I asked the testing engineer how to calculate DC power -- and he didn't have a clue! That's why I'm getting very suspicious of this organization and I've had more than a few hackles raised over this project. Bill

At 11:41 AM 6/18/2008, you wrote: {Quote hidden}>First, a little background. I have designed industrial control trinkets >for a number of industries over the past 20 years or so. These have all >been 10-30 VDC input devices. Some in metal enclosures, some plastic >and some potted epoxy. These devices, depending on the industries >involved have been UL / CSA / FM approved with no problems. > >I just ran my first one through ETL, the allegedly cheaper UL knockoff >lab. Obscurely hidden in the end of the final report is a requirement >to hipot test this device with 840VAC or 1080VDC for 2 seconds. The >product is nothing more than an electronic chart recorder with no >control capabilites whatsoever. It just monitors thermocouples, control >currents and voltages. My big concern is that I have never had any >requirements to hipot test any low voltage equipment - EVER. I've done >many mains power driven devices and they usually get a hipot test. >Common sense engineering would lead me to believe that hipot testing low >voltage DC instruments with these levels for 2 seconds would surely >damage parts of the instrument, like the 36V TVS diodes across the power >inputs. > >So, did ETL make a mistake in specifying a hipot test? Or am I missing >something in the UL-61010 requirements? > >Thanks, > >Bill In industrial control, sensors such as thermocouples sometimes get shorted to mains voltage heaters. That could be a potential safety issue. Depending on the details, it might or might not be a silly requirement. Another reasonable possibility is that it's a cut-and-paste from a previous job that had mains power. Suggest you pick up the phone and ask! >Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" .....speffKILLspam@spam@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com

On Wed, Jun 18, 2008 at 11:41 PM, William Bross <wbrossKILLspamcinci.rr.com> wrote: > My big concern is that I have never had any > requirements to hipot test any low voltage equipment - EVER. This is certainly not true. I have run hi-pot tests for all kinds of low voltage device. This depends on the specification of the device. For example, for PLC I/Os, we need to have isolation between the backplane and field side, so we need to do HiPot tests between them even their power supply is normally 24V (10-30V). And if you have uninsulated connectors exposed, you may have to do hi-pot tests as well. > I've done > many mains power driven devices and they usually get a hipot test. > Common sense engineering would lead me to believe that hipot testing low > voltage DC instruments with these levels for 2 seconds would surely > damage parts of the instrument, like the 36V TVS diodes across the power > inputs. Do you have different circuits in your device which need isolation? What is lab say about the HiPot test? You know you can have hi-pot tests between different section of the circuits. You may not need to do Hi-pot tests between the DC voltage (say 24V and return). Xiaofan

William Bross wrote: > Thanks for the response Mike. > I have. And the exact response I've received so far--because we said > so. And yes, it definitely will kill low voltage devices. And it's not > just supposed to be a sample, it's 100% of production. This is the same > project I asked the testing engineer how to calculate DC power -- and he > didn't have a clue! That's why I'm getting very suspicious of this > organization and I've had more than a few hackles raised over this project. > > Bill > Talk to his manager or go elsewhere. Don't waste your time. - Martin

William, If the response is like "because we say" then they should be able to refer to the reference in the spec that requires it. In the tests we get done, every test is cross referenced to the relevant spec paragraph(s) so we can double check the requirement & limits etc. If they can't or won't do this then I'd look at moving somewhere else. Richard P 2008/6/19 Martin K <.....martinKILLspam.....nnytech.net>: {Quote hidden}> William Bross wrote: >> Thanks for the response Mike. >> I have. And the exact response I've received so far--because we said >> so. And yes, it definitely will kill low voltage devices. And it's not >> just supposed to be a sample, it's 100% of production. This is the same >> project I asked the testing engineer how to calculate DC power -- and he >> didn't have a clue! That's why I'm getting very suspicious of this >> organization and I've had more than a few hackles raised over this project. >> >> Bill >> > > > Talk to his manager or go elsewhere. Don't waste your time. > > - > Martin > -

I have not been overly impressed with my workings with ETL. More often than not, they would test to the wrong standard. When there was a change in UL508A, they kept telling us we had to retest to the new spec, but our equipment was listed to UL508, not UL508A. The inspection guides the field inspector had were generally the wrong ones (several years out of date), so he'd write up a bunch of noncompliance reports. I understand they recently charged $16k to safety test a heat sink with two solid state relays, a couple chokes, four circuit breakers, and a terminal block. Of course, it was six months work... I've worked with some of their RF test engineers in their anechoic chamber, and that seemed to go ok. Anyway, on hipot, you are generally shorting the power input leads together, then applying the high voltage to those shorted leads and measuring leakage to the grounded case. So, as long as your circuit is isolated from the case, it should not be damaged. Good luck! Harold -- FCC Rules Updated Daily at http://www.hallikainen.com - Advertising opportunities available!

According to EN 61010, which may not be exactly as UL 61010, a hipot test (dielectric strength) at 840VAC corresponds to 0.5mm clearance through basic insulation. BUT the correspondeing DC value should then be 1200VDC according to table 9 and not 1080VDC so the values may come elswere from. The clearance depends on a lot of factors - type of insulation, pullotion degree, category of measuring circuit, transient voltages, rated operating altitude, working voltage, combination of voltages for different circuits, mains voltage where the circuit is used and possibly more. So a hipot test may not be irrelevant. You should be able to get the paragraphs in 61010 where these requirements and values are stated (it should be in the test report). /Ruben {Quote hidden}> Mike Hord wrote: > > >>So, did ETL make a mistake in specifying a hipot test? Or am I missing > >>something in the UL-61010 requirements? > >> > >>You should DEFINITELY ask them both whether the test is truly > >>needed AND what the point of the test is. If the person you ask > >>doesn't know, push the matter- someone has to, or you're just being > >>asked to destroy one of your (expensive?) devices for no good reason. > >> > >>Mike H. > >> > >> > Thanks for the response Mike. > I have. And the exact response I've received so far--because we said > so. And yes, it definitely will kill low voltage devices. And it's not > just supposed to be a sample, it's 100% of production. This is the same > project I asked the testing engineer how to calculate DC power -- and he > didn't have a clue! That's why I'm getting very suspicious of this > organization and I've had more than a few hackles raised over this project. > > Bill > -

> On 6/19/08, Harold Hallikainen <haroldspam_OUThallikainen.org> wrote: > > Anyway, on hipot, you are generally shorting the power input leads > > together, then applying the high voltage to those shorted leads and > > measuring leakage to the grounded case. So, as long as your circuit is > > isolated from the case, it should not be damaged. > > > > That is what I think as well. > > Xiaofan > -- Yes, and 840VAC or 1080VDC is a relatively low voltage for these kind of tests. Remember - this is to test insulation breakdown, not to test what voltages the components or circuits can withstand (unless they are part of the insulation). /Ruben ============================== Ruben Jönsson AB Liros Electronic Box 9124, 200 39 Malmö, Sweden TEL INT +46 40142078 FAX INT +46 40947388 @spam@rubenKILLspampp.sbbs.se ==============================

On 6/19/08, Ruben Jönsson <KILLspamrubenKILLspampp.sbbs.se> wrote: > Yes, and 840VAC or 1080VDC is a relatively low voltage for these kind of tests. > Remember - this is to test insulation breakdown, not to test what voltages the > components or circuits can withstand (unless they are part of the insulation). Actually the OP says 840VAC and 1080VDC for 2 seconds. This kind of 2 seconds test is typically for production test only. I guess that 1080VDC is indeed a typo and should be 1200VDC as you mentioned. Normal UL or IEC Hi-Pot tests for insulation type tests are 1 minute. Xiaofan

Dwayne Reid wrote: {Quote hidden}>Couple of things: > >1) Why is the hipot test needed for a low-voltage device? > >2) I *assume* that the hipot test is done the standard way: connect >both power input terminals together, then apply the test voltage >between those input terminals and the case / enclosure. Output >terminals (if any) are floating. > >Is that in fact what they propose to do? > > >Regardless, I'd suggest finding out why they think that hipot testing >of a low-voltage device is necessary. > >dwayne > > >At 09:41 AM 6/18/2008, William Bross wrote: > > > >>I just ran my first one through ETL, the allegedly cheaper UL knockoff >>lab. Obscurely hidden in the end of the final report is a requirement >>to hipot test this device with 840VAC or 1080VDC for 2 seconds. >> >> > > > > Hi Dwayne, Yep, standard hipot test as you describe. They say that if the standard being tested against has the hipot requirement, they make production line testing mandatory. I've petitioned that this requirement be waived so we'll see what happens. I did have a fairly lengthy conversation with one of their engineers that seemed fairly knowledgable about the subject. He also seemed reasonable and was open to other opinions and interpretations of the spec. We'll see how it goes. Bill

Spehro Pefhany wrote: {Quote hidden}>At 11:41 AM 6/18/2008, you wrote: > > >>First, a little background. I have designed industrial control trinkets >> >> >In industrial control, sensors such as thermocouples sometimes get shorted to >mains voltage heaters. That could be a potential safety issue. > >Depending on the details, it might or might not be a silly requirement. > >Another reasonable possibility is that it's a cut-and-paste from a >previous job that had mains power. Suggest you pick up the phone and >ask! > > > >>Best regards, >> >> > >Spehro Pefhany --"it's the network..." "The Journey is the reward" > > Spehro, Good thought on the shorts to heaters. Been there, done that, have some charred remains back in the dungeon! But all that was taken into account in the initial design, which they can plainly see in the schematics and mechanical details. Everything is isolated from everything else. Yep, the rest is 'cut and paste' policy. Questions asked, statements made and now waiting for a decision. Thanks, Bill

Xiaofan Chen wrote: >On Wed, Jun 18, 2008 at 11:41 PM, William Bross <RemoveMEwbrossTakeThisOuTcinci.rr.com> wrote: > > >> My big concern is that I have never had any >>requirements to hipot test any low voltage equipment - EVER. >> >> > >This is certainly not true. I have run hi-pot tests for all kinds of >low voltage device. This depends on the specification of the >device. For example, for PLC I/Os, we need to have isolation >between the backplane and field side, so we need to do >HiPot tests between them even their power supply is normally >24V (10-30V). And if you have uninsulated >connectors exposed, you may have to do hi-pot tests as well. > > > Xiaofan, Yes, this equipment did get hipot tested in their lab and passed with flying colors. BTW this device is very similar to a PLC in function and form only more job specific, so it did get all those tests. My big concern is wasting production resources to test every blippin' unit off the line. {Quote hidden}>>I've done >>many mains power driven devices and they usually get a hipot test. >>Common sense engineering would lead me to believe that hipot testing low >>voltage DC instruments with these levels for 2 seconds would surely >>damage parts of the instrument, like the 36V TVS diodes across the power >>inputs. >> >> > >Do you have different circuits in your device which need isolation? >What is lab say about the HiPot test? > >You know you can have hi-pot tests between different section of the >circuits. You may not need to do Hi-pot tests between the DC voltage >(say 24V and return). > >Xiaofan > > Yes, there are multiple isolated sections and each was tested individually with the standard hipot tests. With up to 70 individually isolated I/O points, a testing jig might not be economically feasible and the probability of production messing up any one connection could render a section of hardware inoperable. My point to them is they proved the design passes the tests so why all of a sudden is production line testing mandatory? We've petitioned and we're waiting for a decision. Thanks, Bill

Timothy J. Weber wrote: >William Bross wrote: > > >>I have. And the exact response I've received so far--because we said >>so. And yes, it definitely will kill low voltage devices. And it's not >>just supposed to be a sample, it's 100% of production. >> >> > >Those two factors combined, if true, would seem to drive production >costs (and delivery dates) toward infinity. > > Agreed. Right now, it looks like the original engineer just 'boiler plated' the requirements without reading out product information packet before testing. It explained our point of view in detail in hopes of saving time. We're talking and waiting for a decision. Thanks, Bill

Martin K wrote: {Quote hidden}>William Bross wrote: > > >>Thanks for the response Mike. >>I have. And the exact response I've received so far--because we said >>so. And yes, it definitely will kill low voltage devices. And it's not >>just supposed to be a sample, it's 100% of production. This is the same >>project I asked the testing engineer how to calculate DC power -- and he >>didn't have a clue! That's why I'm getting very suspicious of this >>organization and I've had more than a few hackles raised over this project. >> >>Bill >> >> >> > > >Talk to his manager or go elsewhere. Don't waste your time. > >- >Martin > > These organizations are designed to be tough to contact anyone inside. Remember, their revenue stream comes 90% from beauraucratic babble and 10% from real work. Whenever I do talk to anybody that seems knowledgable, I go to great pains to get their direct contact info for future reference. We've just recently started working with ETL so I'm still learning their rules. Once I know their rules, I can use that info to beat them at their own game. ;-) Bill

Richard Prosser wrote: >William, > >If the response is like "because we say" then they should be able to >refer to the reference in the spec that requires it. > >In the tests we get done, every test is cross referenced to the >relevant spec paragraph(s) so we can double check the requirement & >limits etc. If they can't or won't do this then I'd look at moving >somewhere else. > >Richard P > > Thanks Richard. The final report didn't have the standard section referenced for the dielectric withstand test. I looked it up and cited the exact paragraphs I used as a reference for my reasoning. I got a total agreement from the new engineer I talked with and am now waiting for confirmation in writing that production line testing is not needed. All in all, I think I just got stuck with a bum engineer the first time around. We'll see on future reports. Bill

Harold Hallikainen wrote: {Quote hidden}>I have not been overly impressed with my workings with ETL. More often >than not, they would test to the wrong standard. When there was a change >in UL508A, they kept telling us we had to retest to the new spec, but our >equipment was listed to UL508, not UL508A. The inspection guides the field >inspector had were generally the wrong ones (several years out of date), >so he'd write up a bunch of noncompliance reports. I understand they >recently charged $16k to safety test a heat sink with two solid state >relays, a couple chokes, four circuit breakers, and a terminal block. Of >course, it was six months work... > >I've worked with some of their RF test engineers in their anechoic >chamber, and that seemed to go ok. > >Anyway, on hipot, you are generally shorting the power input leads >together, then applying the high voltage to those shorted leads and >measuring leakage to the grounded case. So, as long as your circuit is >isolated from the case, it should not be damaged. > >Good luck! > >Harold > > > > Hi Harold, I tend to agree with you that big, beauraucratic, governmental-wanna-be agencies are not very impressive as a whole these days. But, they've managed to make themselves seem necessary mostly due to the total lack of common sense in the general population. And yes, $16K seems about right to test a product, but I find UL more of a 6 - 8 month ordeal. ETL has generally been 4 - 6 weeks. I will put in a plug for their LEX noise testing facility though. They are fast, efficient and knowledgable. They will even call in the middle of a test if they find a problem with suggestions on how to remedy the problem. That saves bucks in lab time and shipping. And yes, the test unit passed all the hipot tests in the lab, it's just line testing every unit I've challenged. Bill

Xiaofan Chen wrote: >On 6/19/08, Ruben Jönsson <spamBeGonerubenspamBeGonepp.sbbs.se> wrote: > > >>Yes, and 840VAC or 1080VDC is a relatively low voltage for these kind of tests. >>Remember - this is to test insulation breakdown, not to test what voltages the >>components or circuits can withstand (unless they are part of the insulation). >> >> > >Actually the OP says 840VAC and 1080VDC for 2 seconds. This kind >of 2 seconds test is typically for production test only. I guess that 1080VDC >is indeed a typo and should be 1200VDC as you mentioned. > >Normal UL or IEC Hi-Pot tests for insulation type tests are 1 minute. > >Xiaofan > > > OK, you are more correct than I am. It's a typo. It was listed as 1180VDC. So, 1180,1200, whatever it takes. (Sorry, the movie Mr. Mom was on last night!) Bill

William Bross wrote: > And yes, the test unit passed all the > hipot tests in the lab, it's just line testing every unit I've > challenged. So apparently this unit provides either isolation or insulation from the input power to something else. That sounds like a important spec that could have bad consequences when others rely on it but it fails. It seems to me that a quick hi-pot test on every unit in production makes sense. Two minutes might be excessive, but you need some way to verify each unit is being produced with the isolation promised. How do you otherwise propose to find accidental dirt or solder or whatever accross the isolation barrier? ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000.

> Two minutes might be excessive, but you need some way > to verify each unit is being produced with the isolation promised. I just realized the OP said 2 seconds, not 2 minutes. 2 seconds sounds quite reasonable to me for the production test. What is the problem with this? ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000.

Olin Lathrop wrote: {Quote hidden}>William Bross wrote: > > >>And yes, the test unit passed all the >>hipot tests in the lab, it's just line testing every unit I've >>challenged. >> >> > >So apparently this unit provides either isolation or insulation from the >input power to something else. That sounds like a important spec that could >have bad consequences when others rely on it but it fails. It seems to me >that a quick hi-pot test on every unit in production makes sense. Two >minutes might be excessive, but you need some way to verify each unit is >being produced with the isolation promised. How do you otherwise propose to >find accidental dirt or solder or whatever accross the isolation barrier? > > >******************************************************************** >Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products >(978) 742-9014. Gold level PIC consultants since 2000. > > Good question Olin. All completed units get at least a 72 hr. extended temp. burn-in. Then the bad product gets weeded out in the final test / calibration / certification phase. The automated tester will kick out a unit even if it calibrates but the zero and span offsets end up outside our calculated normal range. Then all failures get a thorough visual inspection and then corrective action. That takes care of any isolated input side problems. The DC power input gets impedence checked and then input current is measured. Also, each isolated section has A/D channels monitoring VDD and regulator output current. Wouldn't any breach of any of the isolation barriers throw the measured current for each section off? But I can see the argument to still do the testing too. We'll wait for the ruling by the authorities. Bill

Olin Lathrop wrote: >>Two minutes might be excessive, but you need some way >>to verify each unit is being produced with the isolation promised. >> >> > >I just realized the OP said 2 seconds, not 2 minutes. 2 seconds sounds >quite reasonable to me for the production test. What is the problem with >this? > > >******************************************************************** >Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products >(978) 742-9014. Gold level PIC consultants since 2000. > > Just two small problems from my side. I've got half a dozen similar products doing similar tasks, both UL / ETL approved and none have ever had production line testing specified. All have passed the extended lab hipot testing though. The second is it's a PITA to integrate hipot testing into our automated low voltage test setup that they've been happy with for years. Bill

William Bross wrote: > Wouldn't any breach of any of the isolation > barriers throw the measured current for each section off? Not necessarily. Insulation breakdown is non-linear. If some dirt or a solder blob ended up partly bridging the isolation barrier, it would still work fine at low voltage, then suddenly conduct at high voltage. That's why you test at higher than expected voltage. > We'll wait for the ruling by the authorities. It seems to me you should want to do the hi-pot test for your own purposes anyway. If you claim a certain level of isolation, you really should test for it. I don't understand the objection. 2 seconds of additional test time seems cheap compared to just one field failure, and I don't see how else you can guarantee the spec is met. ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000.

William Bross wrote: > Just two small problems from my side. I've got half a dozen similar > products doing similar tasks, both UL / ETL approved and none have > ever had production line testing specified. That only means you got away with it, not that it was the right thing to do or a good idea. I guess this comes down to what exactly you specify to your customer. If you specify isolation, you need to be testing for it in production. Testing the lab sample only proves the design is correct. It does nothing to prove that each individual unit was manufactured correctly. Let's say your isolation clearance is 5mm. What if a solder blob is covering 4.5mm of that? You'd never know with any of your low voltage tests, but if that unit gets used in the field to the full isolation spec, it will likely arc over and some rather bad things could happen. Imagine what the dead guy's grieving widow's lawyer is going to make out of the fact you were too cheap to spend 2 seconds per unit to save her husband's life. You wouldn't want me on that jury. If on the other hand it appears you did the "reasonable and customary" thing and tested each unit, you've got a good argument to make. In that case I'd probably side with you unless they can prove negligence. ******************************************************************** Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products (978) 742-9014. Gold level PIC consultants since 2000.

On Fri, Jun 20, 2008 at 2:13 AM, William Bross <TakeThisOuTwbrossEraseMEspam_OUTcinci.rr.com> wrote: > Just two small problems from my side. I've got half a dozen similar > products doing similar tasks, both UL / ETL approved and none have ever > had production line testing specified. All have passed the extended lab > hipot testing though. The second is it's a PITA to integrate hipot > testing into our automated low voltage test setup that they've been > happy with for years. Last time I learned from in-house UL expert that no production testing are required from UL (we are using UL508). But there is a requirement for production testing form other standards like some from the European side. 2second is thus the best option to fulfill different standards. Last time I was doing sensors and we followed both EN50178 and UL508 and production testing was mandatory if isolation circuits exist in the device. Xiaofan

Olin Lathrop wrote: {Quote hidden}>William Bross wrote: > > >>Just two small problems from my side. I've got half a dozen similar >>products doing similar tasks, both UL / ETL approved and none have >>ever had production line testing specified. >> >> > >That only means you got away with it, not that it was the right thing to do >or a good idea. > >I guess this comes down to what exactly you specify to your customer. If >you specify isolation, you need to be testing for it in production. Testing >the lab sample only proves the design is correct. It does nothing to prove >that each individual unit was manufactured correctly. > >Let's say your isolation clearance is 5mm. What if a solder blob is >covering 4.5mm of that? You'd never know with any of your low voltage >tests, but if that unit gets used in the field to the full isolation spec, >it will likely arc over and some rather bad things could happen. > >Imagine what the dead guy's grieving widow's lawyer is going to make out of >the fact you were too cheap to spend 2 seconds per unit to save her >husband's life. You wouldn't want me on that jury. If on the other hand it >appears you did the "reasonable and customary" thing and tested each unit, >you've got a good argument to make. In that case I'd probably side with you >unless they can prove negligence. > > >******************************************************************** >Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products >(978) 742-9014. Gold level PIC consultants since 2000. > > Sorry for the belated reply. In the meantime, I have actually talked to a new engineer at ETL who was actually able to walk me through why they wanted this thing production line tested. And Olin, I'm pretty much in agreement with everything you listed here. It's more or less a rather common sense approach that when properly documented is pretty hard to refute. Thanks to everyone who had info for me or replied to this post. Smaller companies have their perks but one of the major downfalls is a lack of diversity in opinions. That's where you guys come in. I don't ask often but when I do I'm never disappointed with the array of answers. Thanks, Bill

> Sorry for the belated reply. In the meantime, I have actually talked to > a new engineer at ETL who was actually able to walk me through why they > wanted this thing production line tested. And Olin, I'm pretty much in Can you share the reason with us? /Ruben ============================== Ruben Jönsson AB Liros Electronic Box 9124, 200 39 Malmö, Sweden TEL INT +46 40142078 FAX INT +46 40947388 RemoveMErubenTakeThisOuTpp.sbbs.se ==============================

Ruben Jönsson wrote: >>Sorry for the belated reply. In the meantime, I have actually talked to >>a new engineer at ETL who was actually able to walk me through why they >>wanted this thing production line tested. And Olin, I'm pretty much in >> >> > >Can you share the reason with us? > >/Ruben >============================== >Ruben Jönsson >AB Liros Electronic >Box 9124, 200 39 Malmö, Sweden >TEL INT +46 40142078 >FAX INT +46 40947388 >rubenEraseME.....pp.sbbs.se >============================== > > In a nutshell, since this device is in a two part metal case with a ground bonding wire between the halves, it must be tested. Bill

sorta late on catching up on emails... I agree, its a quick test by the mfg line to test each unit. If you ever look at most power supplies, even commidty PC ones, they all have a sticker saying it passed the hipot test. Just a cost of doing business really. few jobs ago, I was designing and building high voltage power distribution and it was the same way, we supplied a hipot tester to the CM and they did the tests. --- On Thu, 6/19/08, Olin Lathrop <EraseMEolin_piclistembedinc.com> wrote: {Quote hidden}> From: Olin Lathrop <RemoveMEolin_piclistEraseMEEraseMEembedinc.com> > Subject: Re: [EE] Sanity check about hipot testing > To: "Microcontroller discussion list - Public." <RemoveMEpiclistspam_OUTKILLspammit.edu> > Date: Thursday, June 19, 2008, 9:00 AM > William Bross wrote: > > And yes, the test unit passed all the > > hipot tests in the lab, it's just line testing > every unit I've > > challenged. > > So apparently this unit provides either isolation or > insulation from the > input power to something else. That sounds like a > important spec that could > have bad consequences when others rely on it but it fails. > It seems to me > that a quick hi-pot test on every unit in production makes > sense. Two > minutes might be excessive, but you need some way to verify > each unit is > being produced with the isolation promised. How do you > otherwise propose to > find accidental dirt or solder or whatever accross the > isolation barrier? > > > ******************************************************************** > Embed Inc, Littleton Massachusetts, > http://www.embedinc.com/products > (978) 742-9014. Gold level PIC consultants since 2000. > --