EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Eliminateur on February 18, 2011, 03:08:08 pm
-
I just come from almost settings my eyebrows on fire due to shoddy design(or grave mistake on my part too)..., let's begin...
i'm trying to test for oscillations in the main filter of a SMPS, which is 311Vdc out the bridge rectifier, my analog CRO ha a 300vpp peak, but i have a X10 probe...
so i hook it up to 5V/div, X10 div and test the cap with mains unplugged(because it was charged) and surely, shows 292V~, nice and dandy...
i power up the smps and the moment i touch the ground croc to the negative of the cap, sparks and bits of melted copper/tin and solder flew everywhere, i jumped back half a meter(not very scared, i'm used to shit blowing in my face and working with high energy), the main breaker tripped, etc etc, much laughter ensued(not from me)...
So i have a blackened oscilloscope croc with a melted/missing front, a chunk of PCB trace missing, coworkers laughing and i'm totally baffled thinking i accidentally shorted two adjacent PCB traces, but no, it was the negative island all the way....
intrigued, i put the DMM in continuity test and to my horror i find that the input "outer rim" of the BNC connector is part of the chassis which is GROUNDED to mains ground!, WHAAAA, FAIL! which essentially grounded the neg output of the bridge rectifier(-150Vdc~) when i touched it, aren't CRO inputs supposed to be fully floating?.
or i need an esoteric differential probe?
so how do i go about testing this? if i can't ground it....
ofc not stopping there i changed main cap and since smps fuse didn't blew(incredible, a 1.5A fuse didn't blew but the 20A thermomag AC breaker did trip) i plugged it again, this time a huge flash on the component side took out the breaker >:(, turns out i took things for granted since fuse was ok and cap had high charge but the neg branch of the rectifier fused on the first accident... second one took out the fuse..., i hate this friday and this CRO...) >:(
-
intrigued, i put the DMM in continuity test and to my horror i find that the input "outer rim" of the BNC connector is part of the chassis which is GROUNDED to mains ground!, WHAAAA, FAIL! which essentially grounded the neg output of the bridge rectifier(-150Vdc~) when i touched it, aren't CRO inputs supposed to be fully floating?.
or i need an esoteric differential probe?
so how do i go about testing this? if i can't ground it....
Almost every scope I've used has had the neg side connected to mains earth. Working on an SMPS I tend to run the DUT through an isolation transformer which gets round this, I assumed this was absolutely common practise. You can buy scopes with isolated grounds, but I haven't come across them since back at uni in the power labs and I'm pretty sure they're a lot more expensive.... Commiserations on the scope though!
-
Ouch, sorry to hear and read, glad you are not hurt. More comments as I've time to reply.
Alas, all scopes are earth grounded, unless specifically stated they are not, like the portable Fluke 199 series.
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
-
Many people seem to learn the fact that scope input jacks are galvanically connected to the AC ground the hard way.
Just connect the DUT via isolation transformer and keep the scope grounded. Yes, you could use a differential probe but in this case isolating the DUT is better way, for general safety, as you are working near the DUT anyway. Isolating the scope is not recommended since you might accidentally connect the scope also to something else or someone else might touch the scope while you are measuring.
Regards,
Janne
-
I managed to do something similar with a Transmille precision calibrator.
If I am working on SMPS at work I use a Tektronics TPS2024 which is a DSO with isolated leads - both isolated from the mains (it is battery powered) but the leads are isolated from each other. There is a maximum voltage allowed between the ground leads for each but I can't remember what it is.
Neil
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
I like this idea, never crossed my mind, but i'm not sure if this CRO has X/Y substraction capability, i'm gonna check now.....
my other gross error was not using the lamp-series we have for this(TWICE!), that would have prevented the fireworks....
About isolation transformers(which i don't have...): is it a simple 1:1 transformer? Dave/Janne, how would it prevent the issue, even if it's isolated the neg input is still at -150V from ground, or because it's across the transformer it doesn't matters?
hehe, a simple visual inspection would have reveled the connectors are grounded, i simply never came across this situation before(always used a DMM to check main cap voltage), since this smps is behaving bad(that's why it's here) with low-freq 1.5V oscillations with triangular peaks in them(that don't show as variations on the DMMs i have) i was thinking about dry main cap..., and this happened...
Neilm, interesting device, floating AND isolated to eachother!
-
aren't CRO inputs supposed to be fully floating?.
Welcome to Oscilloscope 101. No, they aren't, unless explicitly mentioned on the oscilloscope and in the oscilloscope's manual.
-
About isolation transformers(which i don't have...): is it a simple 1:1 transformer? Dave/Janne, how would it prevent the issue, even if it's isolated the neg input is still at -150V from ground, or because it's across the transformer it doesn't matters?
Yes it is 1:1 (or slightly higher secondary voltage for compensating the droop under load). It prevents the issue because isolation transformer does not have any direct connection to AC mains ground (thus the term isolation transformer), thus preventing any current flow to AC ground. It is the current that kills, not voltage. Note that one should connect only one DUT behind one isolation transformer.
Regards,
Janne
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
I like this idea, never crossed my mind, but i'm not sure if this CRO has X/Y substraction capability, i'm gonna check now.....
my other gross error was not using the lamp-series we have for this(TWICE!), that would have prevented the fireworks....
About isolation transformers(which i don't have...): is it a simple 1:1 transformer? Dave/Janne, how would it prevent the issue, even if it's isolated the neg input is still at -150V from ground, or because it's across the transformer it doesn't matters?
hehe, a simple visual inspection would have reveled the connectors are grounded, i simply never came across this situation before(always used a DMM to check main cap voltage), since this smps is behaving bad(that's why it's here) with low-freq 1.5V oscillations with triangular peaks in them(that don't show as variations on the DMMs i have) i was thinking about dry main cap..., and this happened...
Neilm, interesting device, floating AND isolated to eachother!
I was not proposing subtracting the x and y just the x1 and x2 even the oldest analogue scopes can do this (think diff. oppamp)
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
I like this idea, never crossed my mind, but i'm not sure if this CRO has X/Y substraction capability, i'm gonna check now.....
my other gross error was not using the lamp-series we have for this(TWICE!), that would have prevented the fireworks....
About isolation transformers(which i don't have...): is it a simple 1:1 transformer? Dave/Janne, how would it prevent the issue, even if it's isolated the neg input is still at -150V from ground, or because it's across the transformer it doesn't matters?
hehe, a simple visual inspection would have reveled the connectors are grounded, i simply never came across this situation before(always used a DMM to check main cap voltage), since this smps is behaving bad(that's why it's here) with low-freq 1.5V oscillations with triangular peaks in them(that don't show as variations on the DMMs i have) i was thinking about dry main cap..., and this happened...
Neilm, interesting device, floating AND isolated to eachother!
A reasonable scope if you work with mains and they do a high voltage lead set if you are working on anything higher. The set I have at work are 300V CATII standard leads but we have a leadset that is 1000V CATII. I don't remember how much it was, I think it was around £2k mark from memory. Lacks the bells and whistles of the new Agilent scope, but that doesn't come with isolated inputs.
Neil
-
i don't have access to anything fancy like a 2K pund oscilloscope, not even a DSO, very tight budget, anyway, that's a different subject.
Simon, yeah my mistake!, when i wrote X and Y i was thinking about Ch1 and Ch2, later it came to me "oh, Y is vertical, doh!" :D
@boreadatwork: haha i passed that long ago, and seems like i forgot the basic stuff and simply took the negative as "ground"
-
well if it is a decent oscilloscope it will have a CH1-CH2 function, it is usually a button switch close to the channel selection buttons, often there is a CH1+CH2 function as well
-
it's an 11 yr old Goldstar one, quite decent but doesn't has that exact function,
anyway i think i figured out how to do it(but don't have the balls to test it :D)
ch1 vertical adjust has a "pull to add" which add to ch2
ch2 vert adj has a pull to inv
so i inv ch2 and set that as "ground", then use ch1 in add mode, that should effectively result in the difference ch1+(-ch2) (and displaying only ch1)
but i don't want to blow another bridge... besides i changed cap and the problem persists(opened a new thread about it) so i don't have to scope it anymore
-
well yes the scope does support it as you pointed out you set it up yourself, instead of putting 3 function buttons on they just put invert and add because if you invert one channel and then add them you acheive subtraction
-
intrigued, i put the DMM in continuity test and to my horror i find that the input "outer rim" of the BNC connector is part of the chassis which is GROUNDED to mains ground!, WHAAAA, FAIL! which essentially grounded the neg output of the bridge rectifier(-150Vdc~) when i touched it, aren't CRO inputs supposed to be fully floating?.
Congratulations, you learned that scopes are grounded in the true traditional time honored method! *pins badge of honor to your chest*
Dave.
-
There are a few scopes with fully floating inputs, isolated from each other - I think Tek have some, and also some of the higher-end handhelds are like this.
However for day-to-day use, I think I'd find these a bit of a pain, as it means you'd always have to connect the probe ground - for a lot of low-to-mid frequency 'is there any signal there' and 'Is there DC' type quick checks, especially on something connected to a PC, you can often get away with just relying on the mains ground.
On a similar subject - many years ago, my first scope was a big old Telequipment rackmount job. I had a TV to fix - think I'd pulled it out of a skip (dumpster), and as was typical in those pre-VCR days when the only input on a TV was a UHF antenna, the mains input was rectified, and the negative output of the rectifier connected to chassis (The aerial socket had an internal isolator).
I couldn't afford an isolating transformer, so I lifted the scope earth and connected it to the TV chassis, and sat the scope on a plastic sheet for insulation.
All went well until I decided to change the timebase setting, and discovered the hard way that although the knobs were plastic, they had a metal grubscrew which was _just_ close enough to the surface to not quite be finger-proof :-[.
-
I know connecting the chassis of old equipment to the mains was common, but why were they doing it in the first place? Simplify wiring?
-
I know connecting the chassis of old equipment to the mains was common, but why were they doing it in the first place? Simplify wiring?
Simplifies the PSU - no need for an isolation transformer. It was much cheaper to isolate the aerial socket. The PSU was essentially a line-voltage buck converter which supplied something around 100v to the line output stage, other voltages being derived from extra windings on the line output transformer - we're talking 1970s here, when switchmode PSUs were pretty new, and cheap power semiconductors in their infancy - thyristors were commonly used as the main switching elements, along with other oddball circuitry.
As soon as TVs started having composite inputs to support VCRs and other devices, they moved to a more conventional isolated SMPS topology.
-
Makes sense. How did they isolate the aerial socket? Some high frequency 1:1 tranformer?
-
Makes sense. How did they isolate the aerial socket? Some high frequency 1:1 tranformer?
Yes - either a transformer or a capacitive coupler - much easier & cheaper than an isolated mains transformer. This method was already being used from back in the valve (tube) days, where they connected all the filaments in series (plus a big hot dropper resistor) across the mains, and the Ht supply was just a halfwave rectifer off the mains.
-
Just a reminder to check the ground on your isolation transformer for continuity to mains ground before assuming it's floating. In the US it's illegal to float the ground of an isolation transformer...
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
Hi,
I am wondering how this type of advice get past the moderators ?
This is AC mains you guys are playing with, for goodness sake ! Go to a technical college and pay for professional training if you do not know what you are doing. It might be 'cool' to blow things up but being electrocuted or severely burnt is not. Not to mention the potential damage to expensive test equipment.
Beginners reading this type of advice may sadly think it is correct because it is published on a well-respected Engineering blog.
Regards.
-
you could run the scope off an inverter ? others will know more on this. Your other option is use both channels and not the earths making one channel "earth" and the other signal, then use the subtraction feature of the scope to give you the difference and voila'
Hi,
I am wondering how this type of advice get past the moderators ?
This is AC mains you guys are playing with, for goodness sake ! Go to a technical college and pay for professional training if you do not know what you are doing. It might be 'cool' to blow things up but being electrocuted or severely burnt is not. Not to mention the potential damage to expensive test equipment.
Beginners reading this type of advice may sadly think it is correct because it is published on a well-respected Engineering blog.
Regards.
Thank you, your are so full of observation you failed to spot that, the post was made by a moderator: me ! You also fail to point out which of the two options I provided you have a problem with
I did point out that others would hopefully have more advice on the matter and it has been suggested on here before. The moderators of this board and usually any board are not supposed to be experts, usually they are some of the more active members that "know the forum" and are around more to deal with spam and other trouble and of course are deemed to be gifted with sufficient common sense to keep things on an even keel. But then why do I explain all of this to an expert like yourself. I mean after 3 posts you must be an expert !
That aside welcome to the forum
-
The moderators of this board and usually any board are not supposed to be experts, usually they are some of the more active members that "know the forum" and are around more to deal with spam and other trouble and of course are deemed to be gifted with sufficient common sense to keep things on an even keel.
Agreed, but on the other hand beginners do tend to look up to people with high post counts, especially moderators.
I mean after 3 posts you must be an expert !
I fail to see any connection between the number of posts and the level of knowledge, most knowledge is not obtained through this forum, and spending lots of time on this forum doesn't make one an expert (not talking about anyone specifically). I don't see anything wrong with the technical side of what tjw said (don't float a scope).
That aside welcome to the forum
Seconded.
-
Yea i realize people tent to think user with higher post counts may have more knowledge. Big mistake (one I made often on another certain forum where mods often came out with outlandish claims to look clever but were actually providing fatally inacurate information like a heated discussion over RCD's and what the correct values for safety were. Due to the mods attitude I left as it was plain accuracy of knowledge was not a priority there only ego). Firstly some of us hang out here more, secondly someone who just joins may be very knowleageble but just be new. By 3 posts making one an expert i was refering to the use of this forum and knowlege of it's members not electronics in itself.
No offence intended of course, I'm happily corrected. One should never take any information for 100% accurate particularly where life is endangered and usually letting the discussion run for a bit allows for things to be thrashed out properly and a sane conclusion reached
-
If only someone could figure out an algorithm to calculate posters' signal-to-noise ratio instead of post count...
-
yes quite, unfortunately we live in the real world
-
In some boards, you can rate posts for their quality, but few forums use it consistently, you'd have to encourage it. It doesn't speak for accuracy of the content, but that folks found a post helpful. If peers read the post and gave it a bad mark, then the good and bad will balance each other. If the writer later deletes the post, votes disappear too, so percentage of votes can be used as an index of the poster, than raw thumbs up or down numbers.
If a member has only 3 posts, but all rated well, he'd have a 100% rating, but if another poster simply posted 'wow', 'nice', 'awesome' type posts and are not rated, even if has 1000 posts, he'd have no rating or even negative ratings.
Its not perfect, but its better than simple raw total posts.
If only someone could figure out an algorithm to calculate posters' signal-to-noise ratio instead of post count...
-
intrigued, i put the DMM in continuity test and to my horror i find that the input "outer rim" of the BNC connector is part of the chassis which is GROUNDED to mains ground!, WHAAAA, FAIL! which essentially grounded the neg output of the bridge rectifier(-150Vdc~) when i touched it, aren't CRO inputs supposed to be fully floating?.
Congratulations, you learned that scopes are grounded in the true traditional time honored method! *pins badge of honor to your chest*
Dave.
tehehe i'm gonna wear my badge of fail proudly, best way to learn stuff if by blowing it up :D
-
hm providing you use RCD's and any other precautions
-
If only someone could figure out an algorithm to calculate posters' signal-to-noise ratio instead of post count...
I am a huge critic of the whole concept of Artificial Intelligence. But the problem you define actually is one that is more or less solved, albeit by harnessing human opinions.
I would propose a system similar to that used to rank reviews on Amazon. Forum members would have an optional binary vote on each post they read. It could be "helpful"/"not helpful", or "interesting"/"not interesting", depending on what the goals of signal-to-noise consideration would be. Call it "good"/"bad".
Forum members could then be ranked by total goodness, instead of total posts. Or more extremely they could be ranked by absolute goodness, which is total goodness minus total badness.
I believe that this would work and would be relatively easy to implement (compared to AI).
However, I think that DLJ's EEVblog time is better spent researching and filming blog entries rather than hacking forum software or screening real life people to do so. Therefore we should expect members to make their own decisions about the credibility of individual posters, based on standards similar to those that have been used on Usenet for decades.
After several weeks' reading the forums, or less if delving back into historical posts, any sensible user should get an impression about each of the regular posters, and will then know if a poster is reliable, or is frequently corrected, or simply has to respond to every message, whether s/he can add anything useful or not.
-
User rating systems are often the a major source for problems and flamewars between users.
I'm not convinced that they are worth the hassle on this or any other board.
In some boards, you can rate posts for their quality, but few forums use it consistently, you'd have to encourage it.
[....]
-
Tektronix has a good app note on floating:
http://www2.tek.com/cmswpt/tidetails.lotr?ct=TI&cs=apn&ci=3214&lc=EN (http://www2.tek.com/cmswpt/tidetails.lotr?ct=TI&cs=apn&ci=3214&lc=EN)
-
I'm not a sysop, so I don't know how well it works. But in other technical boards, I'm been paid for my posts based on ratings, and its not much, but its a free 100+ US$ a year, its like getting free test gear every year. Some top posters 10,000+ a year based on the quality of their posts.
User rating systems are often the a major source for problems and flamewars between users.
I'm not convinced that they are worth the hassle on this or any other board.
In some boards, you can rate posts for their quality, but few forums use it consistently, you'd have to encourage it.
[....]
-
it is a tricky thing to get right as it is all based around peoples opinions and perceptions. Personally unless someone is offering a clear cut commercial deal I'd distrust any board that pays members although I admit I'm not falmiliar with how it works
-
In some boards, you can rate posts for their quality, but few forums use it consistently, you'd have to encourage it. It doesn't speak for accuracy of the content, but that folks found a post helpful.
It is a rather stupid mechanism. The result is that postings that make readers feel cozy and good and support their prejudice get voted up, not the ones that are correct.