EEVblog Electronics Community Forum
EEVblog => EEVblog Specific => Topic started by: IanB on January 07, 2012, 05:27:35 am
-
I think this might have been covered in the videos, I can't remember, but what happens if you should connect a battery to the output, for example attempting to charge it? Is there a danger of damaging anything if the power supply is switched off or has a lower set point voltage than the battery?
I'm wondering what might happen if the supply was powered down with the battery connected. I recall that voltage regulators may not like having the voltage on their output pin raised above the input voltage.
Wow. I asked and Dave followed up with a whole video, and even murdered some components in the cause of science. Kewl! :)
-
I had some time to kill in the lab before I headed home, so it seemed like something worthwhile!
Dave.
-
so when do we happen to see the case and pcb :),
stupid me just went and blew up the display on his own supply,
so you definatly have one kit locked in for purchase :)
-
Mmmmm, I bet that office smells like magic now. ;D
-
Mmmmm, I bet that office smells like magic now. ;D
An electronics lab without a faint smell of burnt components and soldering iron burns on the carpet is like a chemistry lab without stains on the ceiling - just not right.
-
I had some time to kill in the lab before I headed home, so it seemed like something worthwhile!
Think you should have been a bit more scientific about it. The LT3080 datasheet has a note saying the set pin has diode clamps to the output pin. Wouldn't be surprised to find a diode from the set pin to the input pin. So with 12v on the output and the input shorted was the 3A going through a broken down pass transistor emitter base or finding a couple of diodes?
The full power supply has the 5v regulator loading the input. Maybe the LT3080 will stand powering up the rest of the supply from a battery on it's output. Will be a bit weird having the whole power supply running from a battery on it's output. A blocking diode in series with the LT3080 input seems the best solution if you can afford the voltage drop.
Should try to get a comment from LTC, having to guess or experiment indicates the datasheet is lacking.
-
If more rigorious testing is required, I would put the LT3080 on a heatsink and do a long period test. Set up a test where the current is limited to like 1 or 2Amps (so it doesn't overheat), but keep it running for an hour. If it still survives , then it's good. The test I have seen is only overloading it for a limited period, which may not kill the clamping diodes *yet*.
My aim with that would be to see if it survives hooking up something like a lead-acid battery without being conscious that the PSU doesn't 'like' that. By adding a heatsink (as you would do normally) you eliminate burning it up by heat, but maybe the internal clamping diodes can't handle high overloads for long times. That would maybe still kill it in the end..
Personally, I would add a diode in the schematic just to be safe. It doesn't mean you have to mount it, most applications will be fine without.
I've had a buck converter running for 2 hours fine, but suddenly failed to start up >7V input. The shutdown/ON pin was connected to the input voltage (which is typically 20V), but the pin could only have a maximum of 5,5V (oops!). Glad I did long and extended testing on it and fixed the cause..
Also, I just checked RS and Farnell's on the LT3080 or LT3083. They have got the LT3080, but it's a US source, which is like 25 euro's extra to ship it (that's a lot if you just want to order 5 regulators for a total of ~32 euro's, combined price to a total of 11.4 euro each!). The ones that have EU (netherlands) stock are in useless small packages such as MSOP, SOT223-3 or something. So if I want to design my own version, I'm afraid I have to find another chip.. :-(
-
Ask yourself if you really need voltage under 1,25. If yes then you need to search more:P, if not there is the lm338. 5A capability but minimum voltage of 1.25.
-
Think you should have been a bit more scientific about it.
It's an "off the cuff" video blog :P
Dave.
-
'buck converter' ? Ins't that called currency exchange office ? ;)
Hans, Farnell has it pretty cheap - but I think you have to buy £20 of stuff to get free shipping (at least to the UK, not sure about the rest of europe).
-
Also, I just checked RS and Farnell's on the LT3080 or LT3083. They have got the LT3080, but it's a US source, which is like 25 euro's extra to ship it (that's a lot if you just want to order 5 regulators for a total of ~32 euro's, combined price to a total of 11.4 euro each!). The ones that have EU (netherlands) stock are in useless small packages such as MSOP, SOT223-3 or something. So if I want to design my own version, I'm afraid I have to find another chip.. :-(
I've said it dozens times now, if you don't like the LT3080 or can't get it, then just use an LM317 and be happy with 1.15V minimum. (or mod the design so you have a negative supply available)
Dave.
-
LT sells chips directly 'i think. I don't know the shipping cost though. Maybe cheaper.
Alexander.
-
so when do we happen to see the case and pcb :)
Very curious about your case choice, oh and the impending LCR meter showdown!
-
Very curious about your case choice, oh and the impending LCR meter showdown!
The case choice has been driving a lot of this project requirements.
Nothing fancy, just a standard case, but relatively small and the internal will be fairly tightly packed.
There won't really be a showdown as such, just separate reviews and teardowns, then the viewer can make their own mind...
The Agilent teardown was shot same day as the IET, but haven't edited yet.
Dave.
-
There won't really be a showdown as such, just separate reviews and teardowns, then the viewer can make their own mind...
The Agilent teardown was shot same day as the IET, but haven't edited yet.
Dave.
Thanks Dave - the Agilent was going to be my xmas present but I decided to wait and hear your thoughts on it first. Although Im sure both meters are great I'd still love to hear any thoughts on one vs the other - perhaps in a forum thread when the video is posted?
-
Thanks Dave - the Agilent was going to be my xmas present but I decided to wait and hear your thoughts on it first. Although Im sure both meters are great I'd still love to hear any thoughts on one vs the other - perhaps in a forum thread when the video is posted?
Yeah, I'm pretty sure both are going to meet their specs and do exactly as claimed, wouldn't surprise me if there is no clear winner, likely just personal preference. By all other reports, both are excellent units and perform well.
Dave.
-
Regarding the protection features of voltage regulators, Linear Technology offers some of the best protected devices on the market. For example, the LT1129 has current limiting, thermal overload protection, reverse input voltage protection and reverse output current protection (at least to a certain voltage at the output). Additionally, many of their voltage regulators have the tab of the package connected to ground, which is a very convenient feature. The downside is, of course, the price of such LT voltage regulators, which is several times higher than the price of the common MC78xx, LM317 or similar products.
@Dave
It may be interesting to consider an alternative approach to the laboratory power supply design, which would be based on a high power operational amplifier such as the OPA548. It can source and sink the output current and the overall design would require a small number of components. A disadvantage would be the need for the negative rail (in applications that require a low output voltage).
-
I think you hit the nail on the head, Frant. You're paying for the 'convenience' of a well-designed IC, so it can stand the abuse and other 'gotchas' which beginners might overlook in an initial power supply design. Of course, it's also keeping the chip count low, something that's hard to put a price on in many designs. I think Dave chose wisely, given the many factors & criteria of this supply.
This is (completely?) off-topic, but I've been designing my own supply over almost the exact same time period Dave's been showing us his design. It was straight up CREEPY, some of the similarities between our two designs; almost as if he were peering over my shoulder at some points, and me over his on a few occasions. However, the similarities end on the 'intelligent control' side of things. Mine is a three-plus channel, capable of 3.3, 5, and 12V output on two, and 5V or 0-20V on the third, each channel up to 10A. The adjustable voltage can go higher if you're willing to go 'out of spec' and none of the channels are under a 'high' load; then it'll go up to about 30V. None of the regulators have the pass element integrated, which is a key aspect in such a high-current design. I'm also adding low-current, super-accurate and adjustable outputs, only capable of a few hundred milliamps, to power microcontrollers, op-amps, etc. I made this supply for myself, because I'm doing some embedded automotive system designs, and I need the ability to chuck out large currents without a whole lot of noise (at least during initial testing; I'm still working out how to inject typical automotive/racing noise).
There are SO many 'gotchas' in this sucker, though. It's using I2C, SPI, USB, PWM, Gray code, intelligent programmable current/voltage limiting, data logging, some funky feedback setups, relays, perhaps wireless communication, etc. I wouldn't expect anyone short of an EE junior or senior to develop, or appreciate, such a monster. If anyone even suggested they wanted to build it and wanted my circuit schematics, I'd probably look at them like they were crazy, and ask why they couldn't make their own if they really needed it. Yes, I know, that sounds completely conceded. But understand that great harm can occur with such a setup, in the wrong hands. I'm not one of those chinese manufacturers that don't give a rat's ass about the general public. Do you really want a n00b handing a power supply capable of more than an amp of current, at more than about 10V? I know I wouldn't. At a minimum, it'd be irresponsible. At worst, I could cause a few houses to burn to the ground. As a conscientious human being, I cannot allow that to happen. Great power, great responsibility, yada yada...!
Please edit/delete this if it's not in the right place (or size!), Dave. Just had this on my mind when I read Frant's post. :)
-
@ SigEpBlue
You cant be serious?
You think electronic designs should be kept secret?
Great harm? Great power?
Now stop telling what other people can and cant do, and try to help them instead.
The FCC are made by people with your mindset :'(
-
I cannot work out if SigEpBlue's post about his power supply design is serious or a joke. It is exactly the same as Dave's design, and yet totally different.
-
...
But understand that great harm can occur with such a setup, in the wrong hands. I'm not one of those chinese manufacturers that don't give a rat's ass about the general public. Do you really want a n00b handing a power supply capable of more than an amp of current, at more than about 10V? I know I wouldn't. At a minimum, it'd be irresponsible. At worst, I could cause a few houses to burn to the ground. As a conscientious human being, I cannot allow that to happen. Great power, great responsibility, yada yada...!
Raither poinless concern as long as noob searches like "atx bench power conversion" produce 364000 hits on google...
Besides it's not your concern what somebody does with a (any) power supply.
All that you have to worry about as designer is that the product itself is safe. As long as it is not than its inresponsible indeed... for anyone
-
Uh...Jeezus..... Look, don't you guys think there's a bit of a difference between offering designs for (whom I assume to be) beginners to build or use, and offering designs to those same beginners that could easily harm them? That is the basic premise under which I was posting, so if you missed it, sorry for you.
@ SigEpBlue
You cant be serious?
You think electronic designs should be kept secret?
Who pissed in your Rice Krispies? When the hell did I say I'd keep things secret? Go get your Magic 8 ball recalibrated.
Now stop telling what other people can and cant do, and try to help them instead.
That is the most hypocritical sentence I've read in years.
The FCC are made by people with your mindset :'(
BWAHAHAHAHAHAHA No, those people are prudes that never have sex and think penis is a dirty word. ;D
Rather poinless concern as long as noob searches like "atx bench power conversion" produce 364000 hits on google...
Besides it's not your concern what somebody does with a (any) power supply. All that you have to worry about as designer is that the product itself is safe.
Good points. I just wanted (and apparently failed) to convey that I wasn't reaching for the same audience as Dave, that my design is nothing unique or special, but I still have concern for the end user. I don't want people to get hurt due to something I did, that's all; I don't think anyone should fault me for being a humanitarian. You made me realize that this may be an unrealistic goal in the end though. But yes, my product will be very safe in its final form.
I cannot work out if SigEpBlue's post about his power supply design is serious or a joke. It is exactly the same as Dave's design, and yet totally different.
:o Okay then...I'll try to make it more simple for you: it is a triple-channel DC source, but with more ampacity than Dave's design. You could almost arc weld with it if you found an adequate transformer (which again, is reason for my earlier concern: high amperage can cause problems, especially for those that do not know better). Mine uses completely different ICs, but shares the aspects of having a digital encoder, current sensing, I2C, etc. Nothing special there. It's just capable of more output, and we happened to be developing our projects at the same time. I find coincidence intriguing, especially when it's a personal project.
-
In light of Short Circuit's remark on modifying ATX power supplies, I'd have to agree that the question of safety is really beyond moot.
I feel that the primary hazard lies in end-user ignorance, and if the almost 4 hours (thus far) of edited video content enumerating just about every facet of the current prototype doesn't satisfy safety concerns, then Tim Tebow praising baby Jesus in a manger wouldn't be able to save that chap's life. To be sure, I have yet to encounter a design process comparable in documentation scope and made freely available outside academia.
-
Add spots for a reverse diode from output to input and two protection diodes between out and set.
-
nihao the set pin is internally diode clamped
-
I feel that the primary hazard lies in end-user ignorance, and if the almost 4 hours (thus far) of edited video content enumerating just about every facet of the current prototype doesn't satisfy safety concerns, then Tim Tebow praising baby Jesus in a manger wouldn't be able to save that chap's life. To be sure, I have yet to encounter a design process comparable in documentation scope and made freely available outside academia.
And I ain't done yet, just kinda video documenting as I go...
I really can't possibly cater for everyone's requirements here in terms input and output voltage and other protection. At some point you have to just say that this design is meant for an intended purpose, and if you want to use it outside of that scope then you'll likely have to modify it, and that entails knowing what you are doing. That is the same for almost every design, there is no escaping it.
Dave.
-
just watched psu#6. i'm concerned about the lt3080 burnt out. i had a situation of burnt npn power transistor that just simply short out the Vin to Vout, ie max voltage input=output. say we are powering 3.3V device and then poof the lt3080, if Vin something like 20V, then poof the device as well. so Dave, is there any protection/voltage sense on the output side to cut off supply in case the psu fails to regulate? or is it not necessary?
-
It's not hard to put an in-line fuse when connecting the external input voltage. It doesn't necessarily have to be a part of the PCB design.
Bourns make PTCs that go up to 3A holding current: http://www.bourns.com/data/global/pdfs/mfsm.pdf (http://www.bourns.com/data/global/pdfs/mfsm.pdf)
They're not hard at all to solder onto pads as they're pretty big.
Heck, there's heaps of through hole ones up to 13A! http://www.bourns.com/ProductLine.aspx?name=polymer_ptc_resettable_fuses_multifuse (http://www.bourns.com/ProductLine.aspx?name=polymer_ptc_resettable_fuses_multifuse)
-
just watched psu#6. i'm concerned about the lt3080 burnt out. i had a situation of burnt npn power transistor that just simply short out the Vin to Vout, ie max voltage input=output. say we are powering 3.3V device and then poof the lt3080, if Vin something like 20V, then poof the device as well. so Dave, is there any protection/voltage sense on the output side to cut off supply in case the psu fails to regulate? or is it not necessary?
You have to ask yourself in what way can the circuit fail?
In essence, you can't really fully protect the output of a variable supply from circuit failure, because you don't know what the required output voltage the user has set, nor what maximum voltage the external circuit can tolerate.
You simply have to rely upon the voltage regulator working correctly and not failing in order to protect your circuit, there is little choice in that. That's the same with every supply, unless you designing some elaborate super redundant power supply for a 20 year life span on a space craft or something.
If the LT3080 suddenly shorts out for some reason then there isn't much you can do about it, the circuit you are powering won't get the regulated voltage any more. Not that the LT3080 is going to suddently "short out", as there usually has to be a deliberate external factor in order for that to happen.
If the supply was fixed then it would be easier of course, you could have a big zener on the output as additional fail-safe protection for example.
It is simply unrealistic to expect anything like this level of redundancy or protection from a regular variable bench power supply.
Dave.
-
It is simply unrealistic to expect anything like this level of redundancy or protection from a regular variable bench power supply.
You are vastly more likely to blow up something with operator failure than circuit failure.
You could provide a secondary max voltage/current limit in firmware for free, giving the operator a second chance not to screw up. That or point users at a range of BIG zener diodes :)
-
yes i hope the lt3080 is not as simple as a single power transistor, which it should be (internal protection or cut out power to output during failure? etc). the same case with other regulator such as lm317 in case other ppl want to do minor modification. i was just asking my concern. i believe redundant protection will increase the cost and complexity. and the "already" current limiting feature is good enough to avoid the regulator from blowing up.
-
The LT3080 has built in over-current and thermal protection. It shouldn't be damaged under anything like normal operating conditions as long as the input voltage stays in range. Like most regulators, they are fairly bullet proof. The only thing that really kills these guys is reverse bias. If you short out the input and apply external voltage to the output (or have way oversized output capacitors) as dave did, you dump a lot of current through the substrate diode, which can destroy the chip. If that is an expected failure mode for your input, you just add a reverse diode from the output to the input that will take the fault current and protect the regulator.
A lot of the HP/Agilent programmable supplies do have programmable over-voltage and over-current protection that disconnect the output if they threshold is exceeded. This is nice for powering very sensitive devices, but is an expensive feature, and rarely used for things as mundane as micro controllers.
-
Many Agilent power supplies (E3630-series at least) feature an adjustable thyristor crowbar circuit at the output (called OVP), maybe that is what you are looking for? Of course, it requires that one sets the crowbar limit accordingly.
Having that is not always a good thing, we had one supply going into smoke at work because of it, when it was used for charging a battery and then somebody accidentally turned it off, the crowbar triggered and tried to short out the battery (without any success :P).
Regards,
Janne
-
ok, we learnt from agilent mistake :D
-
Having that is not always a good thing, we had one supply going into smoke at work because of it, when it was used for charging a battery and then somebody accidentally turned it off, the crowbar triggered and tried to short out the battery (without any success :P).
That's obviously a dumb thing to do. Few power supplies are happy when you force current into them, especially when turned off. Which is you put a diode in series when using one for charging.
OVP can help preventing a shorted pass transistor from killing your circuit. Depending on the costs and lead time of your circuit, a dead power supply may be preferable to a dead DUT.
-
Having that is not always a good thing, we had one supply going into smoke at work because of it, when it was used for charging a battery and then somebody accidentally turned it off, the crowbar triggered and tried to short out the battery (without any success :P).
That's obviously a dumb thing to do. Few power supplies are happy when you force current into them, especially when turned off. Which is you put a diode in series when using one for charging.
OVP can help preventing a shorted pass transistor from killing your circuit. Depending on the costs and lead time of your circuit, a dead power supply may be preferable to a dead DUT.
That is of course true, but one must always assume that people don't always read manuals. So I think that things should be as robust as possible against most common errors. I think the OVP should not spuriously trip on power up/down situations.
That thing is indeed mentioned in the power supply manual, use a series diode when charging a battery. Power supply was ok after replacing the thyristor and cleaning up the burned PCB. That wonderful "radio" smell.
(https://www.eevblog.com/forum/index.php?action=dlattach;topic=6244.0;attach=19475;image)
Regards,
Janne
-
Dave,
I was wondering if you looked at any other high side current sensing solutions other than the max4080. While it is convenient and does reduce your parts count not having gain resistors, it is also surface mount. There are many different rail to rail opamps with similar or better input offset voltages out there, and you could improve upon the full scale accuracy by using the 0.1% resistors you're already using elsewhere.
All spec reasons aside, I'd think the main thing would be being able to get it in a dip package.
edit - http://search.digikey.com/us/en/products/LT1636CN8%23PBF/LT1636CN8%23PBF-ND/891916 (http://search.digikey.com/us/en/products/LT1636CN8%23PBF/LT1636CN8%23PBF-ND/891916)
Something like that. It seems to outmatch the 4080 on Vos.
-
Dave,
I was wondering if you looked at any other high side current sensing solutions other than the max4080. While it is convenient and does reduce your parts count not having gain resistors, it is also surface mount. There are many different rail to rail opamps with similar or better input offset voltages out there, and you could improve upon the full scale accuracy by using the 0.1% resistors you're already using elsewhere.
All spec reasons aside, I'd think the main thing would be being able to get it in a dip package.
edit - http://search.digikey.com/us/en/products/LT1636CN8%23PBF/LT1636CN8%23PBF-ND/891916 (http://search.digikey.com/us/en/products/LT1636CN8%23PBF/LT1636CN8%23PBF-ND/891916)
Something like that. It seems to outmatch the 4080 on Vos.
That could possibly be a better solution.
I recall briefly considering a precision opamp, but decided a proper high side current shunt was neater.
With hindsight, the extra resistors and a DIP8 package would have been a real squeeze on the given PCB.
There would still be the So8 package on the uCurrent side though.
Too late now, the board is sent to manufacture.
Not out of bounds that I could eventually find an all though hole solution for another revision.
Dave.