Topics

Switching power supplies

Brendan
 

I have a few questions about power supplies and would like some opinions. Are switching powers supplies inherently harder on components than linear power supplies? If you picked up a 70's-80's vintage scope with a SMPS would you replace power supply components before using it as a daily driver? Or do you treat all power supplies the same and visually inspect, check for ripple and call it good? From my reading it seems that when a SMPS melts down the chain reaction damage has the possibility of being bad.


Brendan

Chuck Harris
 

Tough question.

Yes, switching supplies are inherently harder on
components than linear, but they also use much better
parts. And they are more likely to be designed using
math, rather than rules of thumb. And they are more
likely to have extensive protection devices to prevent
catastrophic failures from occurring.

Linear supplies are heavy, are more likely to break your
toes, and make copious amounts of heat. That heat soaks
into everything around them, and causes component failure.

Tektronix used some really uber parts in their 70's vintage
supplies. In some cases, I cannot find modern parts that
were as good as what they used.

I would take it through calibration, testing the ripple
and voltages, and just use it. If it fails, fix the failure.

Odds are very, very, good that in prophylactically replacing
parts you are going to introduce failures, now and in the
future. Replacing a 10,000 hour rated part with a 1000 hour
part, isn't going to improve reliability.

-Chuck Harris

Brendan via Groups.Io wrote:

I have a few questions about power supplies and would like some opinions. Are switching powers supplies inherently harder on components than linear power supplies? If you picked up a 70's-80's vintage scope with a SMPS would you replace power supply components before using it as a daily driver? Or do you treat all power supplies the same and visually inspect, check for ripple and call it good? From my reading it seems that when a SMPS melts down the chain reaction damage has the possibility of being bad.


Brendan

Tom Gardner
 

On 26/03/19 20:37, Brendan via Groups.Io wrote:
I have a few questions about power supplies and would like some opinions. Are switching powers supplies inherently harder on components than linear power supplies? If you picked up a 70's-80's vintage scope with a SMPS would you replace power supply components before using it as a daily driver? Or do you treat all power supplies the same and visually inspect, check for ripple and call it good? From my reading it seems that when a SMPS melts down the chain reaction damage has the possibility of being bad.
Replace Rifa mains interference suppression delayed action smoke generators on sight; if there is any sign of cracking in the transparent case, don't even turn it on.

Where there are tantalum beads that are operating near their rated voltage (e.g. a 15V tant on a 13V rail), replace those with a higher voltage.

Visually inspect, measure, replace if there's a problem.

Otherwise, if it ain't broke, don't fix it.

Glenn Little
 

My take on this after replacing thousands of surface mount electrolytic capacitors is replace them all when you get a new to you device that is more than ten or so years old.

Switchers are harder on electrolytics because they run at a higher frequency than a linear supply and the bean counters usually required the least expensive part to be installed to get the instrument out of warranty.

My take on surface mounted electrolytics is that the fail for two reasons. The seal where the positive lead exits the case was violated during the reflow process. Or the capacitors was not properly selected for the task at hand.

When I replace power supply capacitors in switchers, I select the highest temperature capacitors available with the highest ripple current rating that will fit into the available space. I do not use 85 degree C capacitors unless they are all that is made in that value and voltage.

When doing repair work on cameras and tape decks in TV stations, I never saw a failure of a capacitor that I replaced with the above guidelines. I may have not waited long enough or may have been lucky.

Large computer grade electrolytics are the exception. Do not replace unless needed. These do not usually cause board damage that cannot be repaired.

See: https://www.eetimes.com/author.asp?section_id=183&doc_id=1279791#

Hope this helps.

Glenn

On 3/26/2019 4:37 PM, Brendan via Groups.Io wrote:
I have a few questions about power supplies and would like some opinions. Are switching powers supplies inherently harder on components than linear power supplies? If you picked up a 70's-80's vintage scope with a SMPS would you replace power supply components before using it as a daily driver? Or do you treat all power supplies the same and visually inspect, check for ripple and call it good? From my reading it seems that when a SMPS melts down the chain reaction damage has the possibility of being bad.


Brendan


--
-----------------------------------------------------------------------
Glenn Little ARRL Technical Specialist QCWA LM 28417
Amateur Callsign: WB4UIV wb4uiv@... AMSAT LM 2178
QTH: Goose Creek, SC USA (EM92xx) USSVI LM NRA LM SBE ARRL TAPR
"It is not the class of license that the Amateur holds but the class
of the Amateur that holds the license"

Kevin Oconnor
 

I have to agree with Chuck. Most test equipment with imbedded/integrated switchers are going to be very difficult to qualify parts if working. Their functionality can be extremely subtle. Blanket replacement of components will more likely cause additional reliability issues.
My 1978 Tek 485 has had spot repairs, but it’s still working fine. I’d never replace caps in it without good reason.
I have an HP 500MHz digital scope with one of this 3rd party modular switchers. (I forget the HP model but 5digits). Switcher went bad. No docs, no schematics. Really hopeless to diagnose. Best option was an eBay replacement switcher. $50 and never looked back.
Now if you are talkin about bad caps in an LCD monitor or TV, that’s where “replace em all” will serve u well! Lots of crap used in those devices.

K

Sent from kjo iPhone

Chuck Harris
 

SMD electrolytic capacitors are a tricky problem in several ways:

First, they are easily damaged by the heat of the reflow oven,
and solvents that may be used in cleaning flux,... though cleaning
flux isn't done much if at all, on consumer grade equipment.

Second, it is almost impossible to find out what was originally
installed. There are no standardized markings to identify the
manufacturer, or what grade capacitor was installed.

Third, it is hard for a re-worker to find a complete spectrum of
ESR values from which to select replacement capacitors.

When you grab an assortment of SMD electrolytic capacitors off
of ebay, you are unlikely to be able to match more than just the
capacitance and voltage of the original. Never the ESR or lifetime
rating.

When you buy from Mouser, or DigiKey, you will find that cute
little 20uf, 16V cap you need to replace, (because its ESR is 3 ohms
while its cohorts are all 0.5 ohm), can only be had with ESR's of
0.7 ohm, 3 ohms, and 12 ohms! Replace the 0.5 ohm ESR caps with a
0.7 ohm cap, and you have taken the expected life of the cap and cut
it in half (or more). Put in a 3 or 12 ohm version, and you will have
improved nothing.

Tricky indeed!

-Chuck Harris

Kevin Oconnor wrote:

I have to agree with Chuck. Most test equipment with imbedded/integrated switchers are going to be very difficult to qualify parts if working. Their functionality can be extremely subtle. Blanket replacement of components will more likely cause additional reliability issues.
My 1978 Tek 485 has had spot repairs, but it’s still working fine. I’d never replace caps in it without good reason.
I have an HP 500MHz digital scope with one of this 3rd party modular switchers. (I forget the HP model but 5digits). Switcher went bad. No docs, no schematics. Really hopeless to diagnose. Best option was an eBay replacement switcher. $50 and never looked back.
Now if you are talkin about bad caps in an LCD monitor or TV, that’s where “replace em all” will serve u well! Lots of crap used in those devices.

K

Sent from kjo iPhone



David Kuhn
 

Switching supplies can be a problem when they fail. I recently had a
Agilent VXI E4808A chassis that the main power supply failed, at least
its 12 volts out did. That power supply is huge with a logic board mounted
about it ( I think it's a power supply monitor board). It's a lot of
physical work just to get it apart to check caps. Anyway, I doubt
HP/Agilent made that power supply themselves. I really don't know, as to
get information out of them is like pulling hen's teeth. They no longer
sell, or support, those 4 slot VXI chassis's, so it would be nice of them
to release the schematics, or service information. I think you can only
hope to buy a used chassis somewhere. I would love to have the schematics
for them.

In other switches that fail to start, I often find what I call the
"start-up" capacitor in the primary is open or leaking or has a high ESR.
I call it the "start-up cap" as I do not fully understand switching
supplies, and often if there are schematics, there is not a theory of
operation with them, but there is often a capacitor in the primary circuit
that looks to be a short to ground and then charge-up to create the initial
switch swing to get it going and then afterwards, the power supply self
sustains.

So a lot of times with a dead switcher, I have been able to fix them by
replacing the small electrolytic in the primary side of the circuit. It is
usually a 4.7uf or 10uf. Other than that, if the rest are not physically
leaking or swelled and the supply is working I don't touch it.

One instrument that I work on has an on-board +5volt switching regulator
circuit. It is very reliable. Two times (since 1999) I have seen the
switch regulator fail where the output drive to the FET shorts to ground
allowing the supply voltage (~+12volts) go through to the output with no
over-voltage protection, not even a +5.6volt zener to short. It blows
every TTL chip on the +5volt rail, and some some regulators that follow it.
Stupid, stupid German design. You have to be very urber careful probing
that +5volt switching regulator circuit. One slip of the scope probe and
you can simulate the switching regulator IC output shorting to ground
turning on the pass FET full time. So, I've learned "If it aint broke,
don't fix it"!

Sorry it that was slightly off topic.

Dave

On Wed, Mar 27, 2019 at 8:29 AM Chuck Harris <cfharris@...> wrote:

SMD electrolytic capacitors are a tricky problem in several ways:

First, they are easily damaged by the heat of the reflow oven,
and solvents that may be used in cleaning flux,... though cleaning
flux isn't done much if at all, on consumer grade equipment.

Second, it is almost impossible to find out what was originally
installed. There are no standardized markings to identify the
manufacturer, or what grade capacitor was installed.

Third, it is hard for a re-worker to find a complete spectrum of
ESR values from which to select replacement capacitors.

When you grab an assortment of SMD electrolytic capacitors off
of ebay, you are unlikely to be able to match more than just the
capacitance and voltage of the original. Never the ESR or lifetime
rating.

When you buy from Mouser, or DigiKey, you will find that cute
little 20uf, 16V cap you need to replace, (because its ESR is 3 ohms
while its cohorts are all 0.5 ohm), can only be had with ESR's of
0.7 ohm, 3 ohms, and 12 ohms! Replace the 0.5 ohm ESR caps with a
0.7 ohm cap, and you have taken the expected life of the cap and cut
it in half (or more). Put in a 3 or 12 ohm version, and you will have
improved nothing.

Tricky indeed!

-Chuck Harris

Kevin Oconnor wrote:
I have to agree with Chuck. Most test equipment with imbedded/integrated
switchers are going to be very difficult to qualify parts if working. Their
functionality can be extremely subtle. Blanket replacement of components
will more likely cause additional reliability issues.
My 1978 Tek 485 has had spot repairs, but it’s still working fine. I’d
never replace caps in it without good reason.
I have an HP 500MHz digital scope with one of this 3rd party modular
switchers. (I forget the HP model but 5digits). Switcher went bad. No docs,
no schematics. Really hopeless to diagnose. Best option was an eBay
replacement switcher. $50 and never looked back.
Now if you are talkin about bad caps in an LCD monitor or TV, that’s
where “replace em all” will serve u well! Lots of crap used in those
devices.

K

Sent from kjo iPhone





Chuck Harris
 

In switcher parlence, the "start up" capacitor is called the
bootstrap capacitor because the supply picks itself up by its
bootstraps. It is often charged by using a high wattage resistor
connected to the power line, with a diode, capacitor, and zener
to limit its DC voltage. It supplies power to the control PWM
circuitry. When the supply starts, the bootstrap will be taken
over by voltage from a winding on the switcher transformer.

Half wave rectifying creates a lot of ripple current in these
capacitors, and they very often... and I mean VERY often go open
circuit, preventing the supply from starting.

I have fixed many tens of thousands of dollars of equipment ranging
from sewing machines to TV sets by replacing the bootstrap capacitor
and sometimes the bootstrap resistor.... Even HP supplies ;-)

Competently designed switching power supplies are a delight to
behold. They seem to protect themselves from everything, including
stupid technician mistakes.

-Chuck Harris

David Kuhn wrote:

Switching supplies can be a problem when they fail. I recently had a
Agilent VXI E4808A chassis that the main power supply failed, at least
its 12 volts out did. That power supply is huge with a logic board mounted
about it ( I think it's a power supply monitor board). It's a lot of
physical work just to get it apart to check caps. Anyway, I doubt
HP/Agilent made that power supply themselves. I really don't know, as to
get information out of them is like pulling hen's teeth. They no longer
sell, or support, those 4 slot VXI chassis's, so it would be nice of them
to release the schematics, or service information. I think you can only
hope to buy a used chassis somewhere. I would love to have the schematics
for them.

In other switches that fail to start, I often find what I call the
"start-up" capacitor in the primary is open or leaking or has a high ESR.
I call it the "start-up cap" as I do not fully understand switching
supplies, and often if there are schematics, there is not a theory of
operation with them, but there is often a capacitor in the primary circuit
that looks to be a short to ground and then charge-up to create the initial
switch swing to get it going and then afterwards, the power supply self
sustains.

So a lot of times with a dead switcher, I have been able to fix them by
replacing the small electrolytic in the primary side of the circuit. It is
usually a 4.7uf or 10uf. Other than that, if the rest are not physically
leaking or swelled and the supply is working I don't touch it.

One instrument that I work on has an on-board +5volt switching regulator
circuit. It is very reliable. Two times (since 1999) I have seen the
switch regulator fail where the output drive to the FET shorts to ground
allowing the supply voltage (~+12volts) go through to the output with no
over-voltage protection, not even a +5.6volt zener to short. It blows
every TTL chip on the +5volt rail, and some some regulators that follow it.
Stupid, stupid German design. You have to be very urber careful probing
that +5volt switching regulator circuit. One slip of the scope probe and
you can simulate the switching regulator IC output shorting to ground
turning on the pass FET full time. So, I've learned "If it aint broke,
don't fix it"!

Sorry it that was slightly off topic.

Dave

Tom Gardner
 

On 27/03/19 06:05, Kevin Oconnor wrote:
My 1978 Tek 485 has had spot repairs, but it’s still working fine. I’d never replace caps in it without good reason.
Personally I'd replace C911: 22uF 15V tant bead on a 13V PSU rail, diagram 17 (not the PSU diagram!), and IIRC there are a couple of others on that rail, hidden on the other diagrams 7, 9, 11, 16.

When C911 shorted, it toasted R965.

Kevin Oconnor
 

“Personally I'd replace C911: 22uF 15V tant bead on a 13V PSU rail, diagram 17 (not the PSU diagram!)”
Interesting.....
It’s been a while, but I think this is the first failure of my 485! Did this repair 15 years ago.
Sent from kjo iPhone