linear power supply turn on delay

Thread Starter

amidpl

Joined Oct 5, 2011
6
we recently bought a set (4 of them) of 15VDC, 3A linear power supplies from a well known manufacturer. This is not a new thing, we have been buying these supplies for years, all the way back to the 1990's. Every time we buy them, we test them before installing them. We would apply a resistive load and test it to 100% of its rated output. Only this time, on all 4 of the power supplies, the output voltage would not come up to 15V. So we figured we got a bad batch, sent them back, and got a new set of 4. These would not work either. The outputs would only come up to about 0.3V. Eventually we found that if we waited long enough, like between 45 seconds to 2 minutes, the outputs would come up to 15Vdc. This wasn't a slow transition, over 30 seconds the output would rise from 0.3V to about 0.6 volts, and then suddenly jump to 15V, and would behave normally. Also, we found that if we only loaded it to 50 or 60% of its rated load, it would turn on immediately. we don't observe the delay unless the load is between 70 to 100%.

We called the manufacturer and explained this to them, and after a few days they claimed that there is nothing wrong with them. they said that the turn on time is not defined in the datasheet, so we should always assume it might take several minutes for them to turn on.

this seems fairly weird to us, is this just how things work and we can't do anything about it?

Also, we have at least 6 of the power supplies of the same type, just older from 10 years ago, and we tested them the same way right next to eachother, and the older ones turn on immediately under full load. Note that all of these power supplies use the lm723.

we use these power supplies in fairly sensitive equipment, and normally they're not under 70% load, but occasionally they are, and if they don't turn on it could cause some damage. Does anyone have any thoughts on this situation? is there an implicit understanding that linear power supplies might not turn on immediately, for as long as 2 minutes? shouldn't this be reflected in the datasheets?
 

#12

Joined Nov 30, 2010
18,224
As a professional designer of linear power supplies using the 723 chip in the 1970's, I can safely say, "BS".

Probably some recent college graduate got in there and set the foldback overcurrent limit too low. I see that the manufacturer has told you that you can either take it upon your self to redesign that part or get a different supplier. I say, "believe them".
 

wmodavis

Joined Oct 23, 2010
739
And how do you define "turn on immediately"? If you have a true need you should specify it. How immediate is immediate? If you have been buying them since the 90s without a requirement for some hypothetical turn on time why now? Obviously because something changed to make a previous non-issue which you had not even defined into an issue because some manufacturing/design change brought the undefined into view.

Did you investigate the components used on the new PS?

Vote with your feet if they don't meet your unstated/undefined requirements. Find a mfg who makes one that turns on with your defined turn-on time. But specify it!
 

#12

Joined Nov 30, 2010
18,224
I think your requirements to turn on in less that 2 minutes are entirely valid. Every other electronic device is up and running in a couple of seconds at the most. A linear power supply just isn't that complicated. It doesn't need to load several gigabytes of operating system before it starts. Whining about, "You didn't say it should start when you push the button" is disingenuous at best.
 

Adjuster

Joined Dec 26, 2010
2,148
Defining the reasonable is always difficult, but the legal mind sometimes comes to conclusions an ordinary man mght find odd. The argument runs that unless we specify all our requirements explicitly, we can expect nothing. In principle then, if the startup time is unspecified, could it be 1s, 10s, 100s, or many years? In fact, is it stated explicitly that the system will start at all?

When I last bought a car, I did not ask the dealer whether it would require more than a few seconds to start up, even though it uses a diesel engine which can require somewhat longer to start than petrol (gasoline). I simply assumed that the makers would not be foolish enough to market a vehicle with the starting characteristics of a coal-fired locomotive. Was this unwise?
 

Thread Starter

amidpl

Joined Oct 5, 2011
6
We investigated the pcb and components as thoroughly as we could, as messing with it too much would've voided warranty. there is a difference between component manufacturers, the old ones had motorola 723's and the new ones use ST 723's. the old ones have a TIP31C transistor, and the new ones use a TIP29C transistor. the tip29 is lower power. There is a second large transistor, both power supplies use 3055's.
 

oldtech33709

Joined Sep 24, 2011
26
Sounds like the new supply is having an in rush at start up under full load. Is there a varistor in line before the linear regulator? The fact that there is a long delay before the voltage suddenly jumps to a regulated output implies that the regulator input voltage is below dropout at startup.
 

#12

Joined Nov 30, 2010
18,224
Gee whiz! When did this become a legal advice site?
The prevailing "industry standards" are that a linear power supply is only a few dozen milliseconds slower than a straight piece of wire.

@amidpl
If you did your investigation well, you might have found the problem in that the different brand chip (ST 723) is not happy with the resistor network used to set the foldback current limiting circuit, and then, you might not have found the problem. I do not expect that the transistors are the problem. ST Microelectronics presents the same internal diagram that National Semiconductor presented 40 years ago. That gives me nothing to work with. I'm sure I could fix this in a New York minute if I had the power supplies on my bench, but that still wouldn't fix your problem. Your problem is that the power supplies do not comply with industry standards and the manufacture told you to go pound sand.

I am concerned that the problem is temperature sensitive because the power supplies start after what might be a warm up period. On a cold day, they might never start. You could try putting them in a refrigerator to cool them down before attempting a startup and seeing of that makes it worse. Or set them in bright sunlight and see if they start quicker when warm.

Still, if you aren't willing to void the warranty, you can only look for a different supplier.
 

Attachments

crutschow

Joined Mar 14, 2008
34,281
As a professional designer of linear power supplies using the 723 chip in the 1970's, I can safely say, "BS".

Probably some recent college graduate got in there and set the foldback overcurrent limit too low. I see that the manufacturer has told you that you can either take it upon your self to redesign that part or get a different supplier. I say, "believe them".
I agree. To state that there is nothing wrong with a power supply that takes several minute to turn-on is ridiculous. I would never go back to that vendor for any equipment.

And I would publish the name of the manufacturer so no one else buys their crappy designs.
 

Adjuster

Joined Dec 26, 2010
2,148
I don't support the exaggerated legalistic attitude, in fact I suspect that a lot of this supposedly "logical" argument is advanced by people who hope to profit from the situation. These days though, some suppliers of goods and services do seem to think themselves entitled to offer some absurdly bad conditions, which customers seem to accept.

One of my pet hates is the so-called "unlimited" broadband ADSL service sold as "up to" ##Meg, where the supplier knows full well that line conditions to a typical client's home permit only a very small fraction of this rate, and the download allocations are in fact capped by so-called fair use limits. Legislation is being considered to address these issues, but I will be pleasantly surprised if it makes much difference.

Coming back to the topic, I think that the supplies may well be in danger of not starting at all. The 45s to 2min variability on a small sample is ominous, as you cannot be certain of the maximum variation between units, let alone temperature and ageing effects.
 

Thread Starter

amidpl

Joined Oct 5, 2011
6
well i guess it's not a big deal now. we did actually switch to another manufacturer, their supplies are a lot nicer actually but more expensive too, but i guess you get what you pay for. i guess i can name the manufacturer(it's PowerOne, the supplies are HC15-3-AG), although i'd like to emphasize that we used their power supplies for many years and this is the only time that something has gone wrong, so you have to give them credit for a few things. I suppose they are probably dealing with the same stuff that everyone else is in the world, ie stalling economies, they are continually forced to cheapen their designs. i dont want this to turn into a flaming session for them. we aren't a very large company either, so from their perspective, we were probably just an annoyance.

i have a theory that the switch to the lower power tip29 transistor probably caused them to set the current limit more conservatively, so during the initial startup, which coincides with inrush currents and control voltages going all over the place, the supply gets into the foldback areas. what do you think? does anyone by chance have any of these power supplies (recently bought, of course) that somebody could test? see if anyone can corroborate these observations?

for anyone that buys power supplies and uses them in existing or new designs, do you test them always right off the bat to find stuff like this? if these power supplies take 2 minutes to turn on if they're loaded 70% - 100%, does anyone have designs that a power supply such as these could cause damage? is using a power supply at 80% load too low of a safety factor(should you use a larger supply?)? we use these power supplies in electron probe microanalyzers. they are similar to SEM's, and the ones we deal with are from the 1970's.
 

#12

Joined Nov 30, 2010
18,224
It's really impossible to guess as to whether someone, while redesigning (taking the quality out of) a proven machine, tried to protect a smaller (cheaper) transistor by changing the foldback circuit, without seeing it, especially after you said there were no changes in anything but the larger transistors and the chip manufacturer. Your statement implies that the foldback reistors were not changed in value.

Your experience over several years has proven that using a power supply at 80% load worked. Period. Getting suckered into buying a larger power supply to compensate for design errors is...getting suckered.
 

Thread Starter

amidpl

Joined Oct 5, 2011
6
i guess i should note that the foldback is set by a potentiometer on these models, so it's a bit hard to see if they are in a different position, especially since they're covered in glue to prevent people from messing around with them.
 

#12

Joined Nov 30, 2010
18,224
Glad to know it's a simple adjustment that could have been set a little too tight by a new-hire. Still NOT glad to know they told you to get lost and they claim their machines aren't designed to start within a few seconds.
 

ErnieM

Joined Apr 24, 2011
8,377
I've done my share of power supply design and testing. Two minutes is absurd. "Redesigning" a part qualified and used for 20 years without informing customer is asking for trouble.

Perhaps "legally" the manufactures here is not liable, but I would also be liable to find a different source.
 
Top