Topics

Tek 465 Negative 8 volt rail issues


Ed Pavlovic
 

I recently picked up a used Tek 465 scope for a price I liked and could pick
it up without having is shipped, and I'm getting started on checking it
over. Everything works, although both traces even when grounded look like
there's an uncompensated probe attached. But I figured I would look at the
low voltage power rails before I try to track down that issue. Early 465,
with a B010XXX serial number, needs some cleaning and TLC.



All the test points for the rails except the -8 volt are within spec which
was low, so I hooked up a probe to the test point. I'm getting a 1.75 volt
ripple at a frequency of 120 Hz, which makes me suspect the filter cap is
bad. I have the caps on hand to replace all of those power supply caps on
all the rails, but before I go and do that, I just figured I would ask if I
should look at anything else on this power rail before I change the filter
cap so I don't kill it right away. There's also a 33uf tantalum cap on that
rail to ground that I will replace as well as the service manual calls it
out as a 10 volt rated cap, which isn't much to spare on 8 volts.

I uploaded a photo to a folder I created of a shot of the ripple on the -8
volt rail here:



https://groups.io/g/TekScopes/photo/258534/0?p=Created,,,20,2,0,0



If the link doesn't work the name of the folder is "Tek 465 Negative 8 volt
rail" and there's only one picture in there.



Ed Pavlovic

KC9MMM


Dave Peterson
 

Ed, how serendipitous.

I too am looking at a power supply cap issue. Haven't nailed it down yet, but similar line frequency "ripple". In my case the "ripple" is more like unregulated/unfiltered full-wave rectifier output on the 120v and/or 55v rectifier output - there's no significant capacitance. Question is, which cap.

What caught my eye here was "I have the caps on hand". Do you mean you have a stock of original 465 PS caps? Or do you mean you have a supply of modern caps?

I've looked at the "465 Power Supply Capacitor Replacement Guide", and am feeling a little more confident about attacking these caps. But I'm still on the fence about what route I would want to go. I'm very much the "ain't broke, don't fix" kind of guy, and I'll say I'm 51% biased towards keeping things as original as possible.

Curious to know what you meant by on hand, and if you have suggestions for sources.

Dave

On Thursday, December 24, 2020, 09:11:31 AM PST, Ed Pavlovic <kc9mmm@comcast.net> wrote:

I recently picked up a used Tek 465 scope for a price I liked and could pick
it up without having is shipped, and I'm getting started on checking it
over.  Everything works, although both traces even when grounded look like
there's an uncompensated probe attached.  But I figured I would look at the
low voltage power rails before I try to track down that issue.  Early 465,
with a B010XXX serial number, needs some cleaning and TLC.



All the test points for the rails except the -8 volt are within spec which
was low, so I hooked up a probe to the test point.  I'm getting a 1.75 volt
ripple at a frequency of 120 Hz, which makes me suspect the filter cap is
bad.  I have the caps on hand to replace all of those power supply caps on
all the rails, but before I go and do that, I just figured I would ask if I
should look at anything else on this power rail before I change the filter
cap so I don't kill it right away.  There's also a 33uf tantalum cap on that
rail to ground that I will replace as well as the service manual calls it
out as a 10 volt rated cap, which isn't much to spare on 8 volts.

I uploaded a photo to a folder I created of a shot of the ripple on the -8
volt rail here:



https://groups.io/g/TekScopes/photo/258534/0?p=Created,,,20,2,0,0



If the link doesn't work the name of the folder is "Tek 465 Negative 8 volt
rail" and there's only one picture in there.



Ed Pavlovic

KC9MMM


Paul Amaranth
 

These caps dry out all the time, it's a very common problem.

The orginal cans are no longer available and, if you find some NOS
ones, they will probably dry out shortly if not already.

Get 105C snap caps and use one of the adapter board floating around to
match the footprint to the 4 terminal footprint of the original. Use
high quality caps by Nichicon, Panasonic, etc. Avoid cheap noname or
offbrand Chinese caps.

All 3 ground leads need to be jumpered since Tek used the can as a
jumper and the adapter boards take care of that for you. The adapter
boards make for a very professional repair. I hate seeing caps hanging
off the board hot glued to something.

If I have to replace one, I generally replace all of them. Then you
won't have to worry about it for 20 years.

By using very long leads on the adapter boards I have replaced the
caps in situ without pulling the main board by threading them down from
the top.

Unsoldering the main caps is fun, not. It's easy to overheat the board
and have trace separation. Be careful and use good desoldering tools
and technique for that. Some chipquick wouldn't hurt.

Paul

On Thu, Dec 24, 2020 at 05:59:34PM +0000, Dave Peterson via groups.io wrote:
Ed, how serendipitous.

I too am looking at a power supply cap issue. Haven't nailed it down yet, but similar line frequency "ripple". In my case the "ripple" is more like unregulated/unfiltered full-wave rectifier output on the 120v and/or 55v rectifier output - there's no significant capacitance. Question is, which cap.

What caught my eye here was "I have the caps on hand". Do you mean you have a stock of original 465 PS caps? Or do you mean you have a supply of modern caps?

I've looked at the "465 Power Supply Capacitor Replacement Guide", and am feeling a little more confident about attacking these caps. But I'm still on the fence about what route I would want to go. I'm very much the "ain't broke, don't fix" kind of guy, and I'll say I'm 51% biased towards keeping things as original as possible.

Curious to know what you meant by on hand, and if you have suggestions for sources.

Dave


On Thursday, December 24, 2020, 09:11:31 AM PST, Ed Pavlovic <kc9mmm@comcast.net> wrote:

I recently picked up a used Tek 465 scope for a price I liked and could pick
it up without having is shipped, and I'm getting started on checking it
over.  Everything works, although both traces even when grounded look like
there's an uncompensated probe attached.  But I figured I would look at the
low voltage power rails before I try to track down that issue.  Early 465,
with a B010XXX serial number, needs some cleaning and TLC.



All the test points for the rails except the -8 volt are within spec which
was low, so I hooked up a probe to the test point.  I'm getting a 1.75 volt
ripple at a frequency of 120 Hz, which makes me suspect the filter cap is
bad.  I have the caps on hand to replace all of those power supply caps on
all the rails, but before I go and do that, I just figured I would ask if I
should look at anything else on this power rail before I change the filter
cap so I don't kill it right away.  There's also a 33uf tantalum cap on that
rail to ground that I will replace as well as the service manual calls it
out as a 10 volt rated cap, which isn't much to spare on 8 volts.

I uploaded a photo to a folder I created of a shot of the ripple on the -8
volt rail here:



https://groups.io/g/TekScopes/photo/258534/0?p=Created,,,20,2,0,0



If the link doesn't work the name of the folder is "Tek 465 Negative 8 volt
rail" and there's only one picture in there.



Ed Pavlovic

KC9MMM













!DSPAM:5fe4d72d33731209017941!
--
Paul Amaranth, GCIH | Manchester MI, USA
Aurora Group of Michigan, LLC | Security, Systems & Software
paul@AuroraGrp.Com | Unix/Linux - We don't do windows


Dave Peterson
 

Thanks Paul!

I see I've scored. I'm going to have to pick your brain on this subject. That is, take advantage of the accumulated wealth of knowledge of the community. You guys are great.

I want to keep this conversation on the forum for the time being as the first thing I've been doing so far is searching and review threads with regards to 465 power caps. It's paying big dividends so far. But it is challenging to peruse the vast results. I'll do a search for adapter boards as this is the first I've heard of it, and am kind of liking the idea. The 465 PS Cap Replacement Guide is good, but I'd like to implement it a little neater. That's no dig against the original writer. We're all indebted for the shared experience. So hopefully my experiences and this conversation can contribute to the knowledge base.

I'm a little on the fence about the wholesale replacement of the original caps, but then I'm an ignorant newb. I hear your point about prophylactic replacement. I'll have to dwell on it for my own comfort. Also know that I replaced C1419/C1418 with new Nichicon ULD 60v caps, so I'm not a original preservationist. I guess I want to strike a balance between need and the risks and efforts of wholesale replacement. I suspect the adapter boards are constructed for replacement of all 5 big caps. I do like the assurance that new over-spec'ed caps provide.

As I say, I'll do some research, but if you have particular advice for adapter boards I'd love to hear from you.

Thanks again!
Dave

On Thursday, December 24, 2020, 10:41:39 AM PST, Paul Amaranth <paul@auroragrp.com> wrote:

These caps dry out all the time, it's a very common problem.

The orginal cans are no longer available and, if you find some NOS
ones, they will probably dry out shortly if not already.

Get 105C snap caps and use one of the adapter board floating around to
match the footprint to the 4 terminal footprint of the original.  Use
high quality caps by Nichicon, Panasonic, etc.  Avoid cheap noname or
offbrand Chinese caps.

All 3 ground leads need to be jumpered since Tek used the can as a
jumper and the adapter boards take care of that for you. The adapter
boards make for a very professional repair.  I hate seeing caps hanging
off the board hot glued to something.

If I have to replace one, I generally replace all of them.  Then you
won't have to worry about it for 20 years.

By using very long leads on the adapter boards I have replaced the
caps in situ without pulling the main board by threading them down from
the top.

Unsoldering the main caps is fun, not.  It's easy to overheat the board
and have trace separation.  Be careful and use good desoldering tools
and technique for that.  Some chipquick wouldn't hurt.

  Paul

On Thu, Dec 24, 2020 at 05:59:34PM +0000, Dave Peterson via groups.io wrote:
  Ed, how serendipitous.

I too am looking at a power supply cap issue. Haven't nailed it down yet, but similar line frequency "ripple". In my case the "ripple" is more like unregulated/unfiltered full-wave rectifier output on the 120v and/or 55v rectifier output - there's no significant capacitance. Question is, which cap.

What caught my eye here was "I have the caps on hand". Do you mean you have a stock of original 465 PS caps? Or do you mean you have a supply of modern caps?

I've looked at the "465 Power Supply Capacitor Replacement Guide", and am feeling a little more confident about attacking these caps. But I'm still on the fence about what route I would want to go. I'm very much the "ain't broke, don't fix" kind of guy, and I'll say I'm 51% biased towards keeping things as original as possible.

Curious to know what you meant by on hand, and if you have suggestions for sources.

Dave


  On Thursday, December 24, 2020, 09:11:31 AM PST, Ed Pavlovic <kc9mmm@comcast.net> wrote:

  I recently picked up a used Tek 465 scope for a price I liked and could pick
it up without having is shipped, and I'm getting started on checking it
over.  Everything works, although both traces even when grounded look like
there's an uncompensated probe attached.  But I figured I would look at the
low voltage power rails before I try to track down that issue.  Early 465,
with a B010XXX serial number, needs some cleaning and TLC.



All the test points for the rails except the -8 volt are within spec which
was low, so I hooked up a probe to the test point.  I'm getting a 1.75 volt
ripple at a frequency of 120 Hz, which makes me suspect the filter cap is
bad.  I have the caps on hand to replace all of those power supply caps on
all the rails, but before I go and do that, I just figured I would ask if I
should look at anything else on this power rail before I change the filter
cap so I don't kill it right away.  There's also a 33uf tantalum cap on that
rail to ground that I will replace as well as the service manual calls it
out as a 10 volt rated cap, which isn't much to spare on 8 volts.

I uploaded a photo to a folder I created of a shot of the ripple on the -8
volt rail here:



https://groups.io/g/TekScopes/photo/258534/0?p=Created,,,20,2,0,0



If the link doesn't work the name of the folder is "Tek 465 Negative 8 volt
rail" and there's only one picture in there.



Ed Pavlovic

KC9MMM













!DSPAM:5fe4d72d33731209017941!
--
Paul Amaranth, GCIH            | Manchester MI, USA
Aurora Group of Michigan, LLC  |  Security, Systems & Software
paul@AuroraGrp.Com              |  Unix/Linux - We don't do windows


Michael W. Lynch
 

Dave,

I have replaced several sets of these filter caps on the 465/68/475 series. Seller has multiple versions for different brands and types of equipment.

I used these:

Capacitor Adapter 15.5mm triangle recapping vintage equipment Tek 465 kit (x5)

eBay item number:
273254508468

Seller information:
cuog (749 )
100% Positive feedback

I believe that these are the ones that Paul is recommending in his earlier reply.

Good Luck!


--
Michael Lynch
Dardanelle, AR


Paul Amaranth
 

If you search on ebay for "Tektronix capacitor adapter" you'll get a few hits.

Someone posted gerbers on eevblog if you want to send them out yourself. I
could probably find the link to that. Actually, here's one:

https://www.eevblog.com/forum/repair/making-dave_k_s-capacitor-adapters-at-osh-park/msg1218875/#msg1218875

I made up a pattern in a drawing program and etched some using
pnp-blue. That worked fine, but I just buy them now. A number of
people have independantly come up with this solution, a sure sign
it's a good idea :-)

You could likely use the gerbers if you had access to a pcb milling setup.

"Wholesale replacement of caps" in this instance only refers to the
metal can filter caps. Other than the odd shorted tant I've never had
to replace any other cap and, in general, if it ain't broke don't make
it so :-)

It's possible the other cans could be good for years. To accurately
check them though you'd need to pull them, so ...

An ESR meter might give you some good hints without pulling them. The
ESR will go up as the cap dries out so if you're seeing 100 ohms on a
470uf cap, it's a good candidate for replacement.

You could just replace the bad one and you'd be good for a while. When
I've tried that I'm back in there in a year or so anyway. At least
they don't leak all over and destroy the PCB.

Paul

On Thu, Dec 24, 2020 at 07:08:47PM +0000, Dave Peterson via groups.io wrote:
Thanks Paul!

I see I've scored. I'm going to have to pick your brain on this subject. That is, take advantage of the accumulated wealth of knowledge of the community. You guys are great.

I want to keep this conversation on the forum for the time being as the first thing I've been doing so far is searching and review threads with regards to 465 power caps. It's paying big dividends so far. But it is challenging to peruse the vast results. I'll do a search for adapter boards as this is the first I've heard of it, and am kind of liking the idea. The 465 PS Cap Replacement Guide is good, but I'd like to implement it a little neater. That's no dig against the original writer. We're all indebted for the shared experience. So hopefully my experiences and this conversation can contribute to the knowledge base.

I'm a little on the fence about the wholesale replacement of the original caps, but then I'm an ignorant newb. I hear your point about prophylactic replacement. I'll have to dwell on it for my own comfort. Also know that I replaced C1419/C1418 with new Nichicon ULD 60v caps, so I'm not a original preservationist. I guess I want to strike a balance between need and the risks and efforts of wholesale replacement. I suspect the adapter boards are constructed for replacement of all 5 big caps. I do like the assurance that new over-spec'ed caps provide.

As I say, I'll do some research, but if you have particular advice for adapter boards I'd love to hear from you.

Thanks again!
Dave
--
Paul Amaranth, GCIH | Manchester MI, USA
Aurora Group of Michigan, LLC | Security, Systems & Software
paul@AuroraGrp.Com | Unix/Linux - We don't do windows


Ed Pavlovic
 

Dave,
I have current production caps that I just bought from Mouser. I specifically looked for smaller diameter caps so I wouldn’t have clearance issues between the caps, but a few I just couldn’t get as small as I wanted but they should work.

I did buy the adapters on eBay to get the same footprint as the original ones, I may use a small washer on the pins to make soldering then to the board easier. Not really looking forward to unsoldering the old cans.

I can post a list of what I bought, but I want to make sure everything fits before I do.

Ed


Dave Peterson
 

Thank you everyone for the pointers. I will follow those. I think I need to invest in an ESR meter. If this is going to be a problem I'm going to regularly encounter, and I think it's likely if I continue to buy and fix 465s (and I will), then it's worth the expenditure.

For now, it's time to go bake Christmas cookies.

Happy Holidays!
Dave

On Thursday, December 24, 2020, 11:36:12 AM PST, Ed Pavlovic <kc9mmm@comcast.net> wrote:

Dave,
I have current production caps that I just bought from Mouser.  I specifically looked for smaller diameter caps so I wouldn’t have clearance issues between the caps, but a few I just couldn’t get as small as I wanted but they should work.

I did buy the adapters on eBay to get the same footprint as the original ones, I may use a small washer on the pins to make soldering then to the board easier.  Not really looking forward to unsoldering the old cans.

I can post a list of what I bought, but I want to make sure everything fits before I do.

Ed


Dave Peterson
 

Paul,

I'm putting together a Mouser project for 465 PS caps. I pulled all 5 as they appear to have been leaking, it was going to be necessary to get at the suspected C1513, and once I got going it wasn't that bad, so might as well do all of them. I also have the adapter boards ordered and waiting for USPS to get them across country.

I was searching for radial Nichicon caps as I'm pretty happy with the order I made for C1419 replacements. I didn't catch your earlier reference to "snap" caps, and I see that these come in higher capacitance and voltage values than radial. Not that radial's aren't available, it just appears snap caps are inherently larger?

I've never heard of snap caps before. Aside from the obvious lead difference, what is the purpose/difference between radial lead caps and snap caps?

Dave


toby@...
 

On 2020-12-27 11:10 p.m., Dave Peterson via groups.io wrote:
Paul,

I'm putting together a Mouser project for 465 PS caps. I pulled all 5 as they appear to have been leaking, it was going to be necessary to get at the suspected C1513, and once I got going it wasn't that bad, so might as well do all of them. I also have the adapter boards ordered and waiting for USPS to get them across country.

I was searching for radial Nichicon caps as I'm pretty happy with the order I made for C1419 replacements. I didn't catch your earlier reference to "snap" caps, and I see that these come in higher capacitance and voltage values than radial. Not that radial's aren't available, it just appears snap caps are inherently larger?

I've never heard of snap caps before. Aside from the obvious lead difference, what is the purpose/difference between radial lead caps and snap caps?
Nichicon publishes app notes and press releases like these:

https://www.nichicon.co.jp/english/products/pdfs/e-al_gui.pdf

https://www.nichicon.co.jp/english/lib/new133.html
https://www.nichicon.co.jp/english/lib/new135.html
https://www.nichicon.co.jp/english/lib/new141.html

via https://www.nichicon.co.jp/english/lib/index.html

Hope some of that is helpful.

--Toby

Dave





n4buq
 

Regarding snap caps, the leads are bent into a crook so as to allow inserting processes to push the leads into the PCB holes and grip the cap in place until soldered. Otherwise, they're like their ordinary radial-leaded cousins.

Thanks,
Barry - N4BUQ

----- Original Message -----
From: toby@telegraphics.com.au
To: TekScopes@groups.io
Sent: Monday, December 28, 2020 9:25:25 AM
Subject: Re: [TekScopes] Tek 465 Negative 8 volt rail issues

On 2020-12-27 11:10 p.m., Dave Peterson via groups.io wrote:
Paul,

I'm putting together a Mouser project for 465 PS caps. I pulled all 5 as
they appear to have been leaking, it was going to be necessary to get at
the suspected C1513, and once I got going it wasn't that bad, so might as
well do all of them. I also have the adapter boards ordered and waiting
for USPS to get them across country.

I was searching for radial Nichicon caps as I'm pretty happy with the
order I made for C1419 replacements. I didn't catch your earlier reference
to "snap" caps, and I see that these come in higher capacitance and
voltage values than radial. Not that radial's aren't available, it just
appears snap caps are inherently larger?

I've never heard of snap caps before. Aside from the obvious lead
difference, what is the purpose/difference between radial lead caps and
snap caps?
Nichicon publishes app notes and press releases like these:

https://www.nichicon.co.jp/english/products/pdfs/e-al_gui.pdf

https://www.nichicon.co.jp/english/lib/new133.html
https://www.nichicon.co.jp/english/lib/new135.html
https://www.nichicon.co.jp/english/lib/new141.html

via https://www.nichicon.co.jp/english/lib/index.html

Hope some of that is helpful.

--Toby

Dave










Dave Peterson
 

Thanks Toby and Barry,

I'm becoming partial to Nichicon, not because I think their documentation is better, but I am becoming more familiar with it. And their capacitor series. I've found their Series charts very useful: https://www.nichicon.co.jp/english/products/alm_larg/index.html But I enjoyed reading and learning more about the large power supply cap types. So thanks for the app notes.

I am a bit concerned that the lifetime specs are so much less than the U** series caps, but I see that the "Large Can" types are intended for higher voltage. It's difficult to find a 1200uF, 100V U-series cap, but the "Large Can" L-series can handle that easily. But at a cost of lifetime/ripple current?

LGN2C122MELA40: 1200 uF, 160V, 25/40mm, 3000 hour lifetime @105C, 2300mA ripple current.
UHW1K122MHD: 1200uF, 80V, 16/40mm, 10000 hour lifetime @105C, 3500mA ripple current.

The application is a 465 C1513 which is filtering the unregulated 55v supply. Schematics, and measurement, shows this output is really in the range of 70+ volts. So very near the limits of U-series cap. The existing cap is a 100v cap, so we'll say that's the spec? But the U-series cap above has significantly better lifetime and ripple current. Which is the more important selection criteria here?

And then for the lower voltage, but higher value +15v, +5v, and -8v: the U-series again has significantly higher lifetime and ripple currents at the expense of a lower voltage rating, but still greater than the application. For example: UHE1E562MHD (5600uF, 25v, 10000h, 4200mA) vs. LGY1H562MELA40 (5600uF, 50v, 5000h, 2300mA)

The 1200uF cap seems a little on edge at 80v for U-series. But, within voltage spec, doesn't lifetime and ripple trump the ratio: spec voltage/operating voltage?

Dave


Dave Peterson
 

Spent a good amount of time yesterday studying available caps and Nichicon documentation. Came across this:

https://www.nichicon.co.jp/english/products/pdf/aluminum.pdf Section 2-9 "Life and Reliability".

"if a capacitor is used ... at the maximum operating temperature or below ... the life doubles for each 10deg C drop in temperature." (Ft)

"The degree that applied voltage effects the life of the capacitor when used below the rated voltage is small, compared to the degree that ambient temperature and ripple current effects life. Therefore, when estimating the life of a capacitor, the voltage coefficient to the applied voltage (Fu) is calculated as 1."

2-9-5 "Ripple Current and Life" is a complex section, but can be summarized as "ripple current causes temperature rise", and this temperature rise reduces lifetime. (Fi)

Section 2-9-6 "Estimated Life" gives lifetime as:

L = Lo x Ft x Fi x Fu

Where Lo = Datasheet lifetime (about 3000h to 10000h for snap-in and radial aluminum electrolytic capacitors). Or:

L = Lo x 2^((To-Tn)/10) x 2^(1-((In/Im)^2)/K) x 1

Fi = 2^(1-((In/Im)^2)/K) is complicated as heck, but note that its value ranges (roughly) from 1 to 2. Suffice it to say, run your ripple current at max at max temperature and Fi approaches 1, but is generally closer to 1.5 to 1.9.

So lifetime, grossly, at worst, is:

L = 3000h x 2^((105-40)/10) x 1 x 1 = 3000h x 2^6.5 = 3000h x 90 = 270,000h = 270,000h / (24h x 365d/y) = 30y

Finally, noted from later in section 2-9-6 "Please note that calculated life time is for reference only and not guaranteed. Typically, fifteen years is generaIIy considered to be the maximum for the estimated life obtained by the above formula." I take this to mean Nichicon does not assure a lifetime of greater than 15 years, in general.

So: given the worst snap cap lifetime of 3000h, run at or below rated voltage, at or below rated ripple current, at approximately 40C (104F) the lifetime of the cap is estimated to be double the guaranteed lifetime of Nichicon caps. You halve the lifetime for every degree above 40C (roughly), and all bets are off if you exceed voltage or ripple current. Run ripple current below rated and you tend to double the lifetime. The "Miniature Type" (U-series) caps do tend to have about double the lifetime and ripple current rating of "Large Can Type" (L-series). But for large cap and voltage values the L-series have more "headroom" with regards to voltage than the U-series.

For the 1200uF, 100v C1513 (the big cap that brought this scope down) I'm going with a snap-cap at 160v. A U-series cap would have been limited to 80v - there's nothing offered in this line above that. The output at this cap is just under 80v, but it's near enough that it gives me pause. And clearly the technology used in U-series is at its limits with this cap, where the L-series is offered in significantly higher voltages. So it doesn't seem as stressed. The big question remains, what is the actual ripple current?

For the 560uF, 100v C1512, the smaller value L-series caps have ripple current specs less than the 2000mA of C1513 above. So I went with a U-series cap for this one. At the lower value this series has a 160v cap available. So why not!? It's ripple current spec is the same, 2000mA, as C1513. The ripple currents for equivalent L-series caps are less than 2000mA.

The low voltage caps are a bit more limited in available capacitance values. The U-series has 5600uF but only at 35v. The L-series has the same at 50v. But then the L-series does not have a 5000uF. It gets complicated going through all the combinations. I found it interesting that in the "465 PS Cap Replacement Guide" used a 5600uF for all three of C1542, C1552, and C1562. I was inclined to do the same as again, the U-series seems to be more against their voltage limits than the L-series. I do wonder about the significance of using a 5600uF cap in place of a 3000uF cap. But these caps are all about biggest C in the space available (at the time), right?

The big unknown that remains is what IS the ripple current to these caps in these supplies? I wonder if I can put together a reasonably accurate simulation. Just to get some sense of order of magnitude.

I didn't come across (yet?) the effect of voltage on the capacitor, besides little to no effect if kept below spec. But common sense suggests a cap operated near its spec is more likely to fail than one that has some head room. The question is what is "some" head room? A good value seems 2x, but it's not well defined. Lifetime suggests no real gain as long as spec is not exceeded and it's just a matter of providing enough head room to ensure spec is not exceeded under all operating conditions.

Here's my final list ordered from Mouser:

C1512, 560uF: UCY2C561MHD, 560uF, 160v, 2130mA, 18mmx40mm, 7.5mm lead spacing.
C1513, 1200uF: LGN2C122MELA40, 1200uF, 160v, 2300mA, 25mmx40mm, 10mm lead spacing.
C1542, C1552, C1562, 5600uF: LGY1H562MELA40, 5600uF, 50v, 2300mA, 25mmx40mm, 10mm lead spacing.

These all fit the adapter boards from the seller on eBay noted above. I bought two sets of both adapters and caps. One as a back-up for my first attempt at installation, and should I succeed without damage I have a second set for my other 465.

This was not as bad to do as I'd feared, and the caps that came out of my "parts" 465 are so bad that I'm kind of a convert now. Leaking is quite evident, though not so much that the A9 board is affected. While I don't have an ESR type meter, I'm able to use the equipment I have to get some kind of capacitance measurement off working caps. It's difficult to impossible to measure the capacitance on the old PS caps as the internal resistance is so high that shorting the leads does not discharge the cap appreciably (C1513 specifically). But the measured cap is less than 1/10th the rated cap. There's a big fat R between the lead and the internal capacitance. I only find it odd that this scope worked at all. I really wish I'd gotten some ripple measurements off the caps of the supplies (before regulation) before it died. I suspect the regulators were compensating until the 55v regulator couldn't take it any more. I'll have to take a look when I get it back working. I wonder if any of the regulator components are affected by the stress.

Sorry for the long write up, but I hope the info is useful to someone down the road.
Dave


Bob Albert
 

It shouldn't be a big deal to measure ripple current.  The tough part is access to the component without changing the ripple current.  My HP 428B could probably measure it if the probe could be made to fit.
A sensing resistor would work if it's a whole lot lower resistance than the ESR.
The ripple current would be much higher during turn on of the scope; does that count?
Bob

On Tuesday, December 29, 2020, 10:56:48 AM PST, Dave Peterson via groups.io <davidpinsf=yahoo.com@groups.io> wrote:

Spent a good amount of time yesterday studying available caps and Nichicon documentation. Came across this:

https://www.nichicon.co.jp/english/products/pdf/aluminum.pdf Section 2-9 "Life and Reliability".

"if a capacitor is used ... at the maximum operating temperature or below ...  the life doubles for each 10deg C drop in temperature." (Ft)

"The degree that applied voltage effects the life of the capacitor when used below the rated voltage is small, compared to the degree that ambient temperature and ripple current effects life. Therefore, when estimating the life of a capacitor, the voltage coefficient to the applied voltage (Fu) is calculated as 1."

2-9-5 "Ripple Current and Life" is a complex section, but can be summarized as "ripple current causes temperature rise", and this temperature rise reduces lifetime. (Fi)

Section 2-9-6 "Estimated Life" gives lifetime as:

L = Lo x Ft x Fi x Fu

Where Lo = Datasheet lifetime (about 3000h to 10000h for snap-in and radial aluminum electrolytic capacitors). Or:

L = Lo x 2^((To-Tn)/10) x 2^(1-((In/Im)^2)/K) x 1

Fi = 2^(1-((In/Im)^2)/K) is complicated as heck, but note that its value ranges (roughly) from 1 to 2. Suffice it to say, run your ripple current at max at max temperature and Fi approaches 1, but is generally closer to 1.5 to 1.9.

So lifetime, grossly, at worst, is:

L = 3000h x 2^((105-40)/10) x 1 x 1 = 3000h x 2^6.5 = 3000h x 90 = 270,000h = 270,000h / (24h x 365d/y) = 30y

Finally, noted from later in section 2-9-6 "Please note that calculated life time is for reference only and not guaranteed. Typically, fifteen years is generaIIy considered to be the maximum for the estimated life obtained by the above formula." I take this to mean Nichicon does not assure a lifetime of greater than 15 years, in general.

So: given the worst snap cap lifetime of 3000h, run at or below rated voltage, at or below rated ripple current, at approximately 40C (104F) the lifetime of the cap is estimated to be double the guaranteed lifetime of Nichicon caps. You halve the lifetime for every degree above 40C (roughly), and all bets are off if you exceed voltage or ripple current. Run ripple current below rated and you tend to double the lifetime. The "Miniature Type" (U-series) caps do tend to have about double the lifetime and ripple current rating of "Large Can Type" (L-series). But for large cap and voltage values the L-series have more "headroom" with regards to voltage than the U-series.

For the 1200uF, 100v C1513 (the big cap that brought this scope down) I'm going with a snap-cap at 160v. A U-series cap would have been limited to 80v - there's nothing offered in this line above that. The output at this cap is just under 80v, but it's near enough that it gives me pause. And clearly the technology used in U-series is at its limits with this cap, where the L-series is offered in significantly higher voltages. So it doesn't seem as stressed. The big question remains, what is the actual ripple current?

For the 560uF, 100v C1512, the smaller value L-series caps have ripple current specs less than the 2000mA of C1513 above. So I went with a U-series cap for this one. At the lower value this series has a 160v cap available. So why not!? It's ripple current spec is the same, 2000mA, as C1513. The ripple currents for equivalent L-series caps are less than 2000mA.

The low voltage caps are a bit more limited in available capacitance values. The U-series has 5600uF but only at 35v. The L-series has the same at 50v. But then the L-series does not have a 5000uF. It gets complicated going through all the combinations. I found it interesting that in the "465 PS Cap Replacement Guide" used a 5600uF for all three of C1542, C1552, and C1562. I was inclined to do the same as again, the U-series seems to be more against their voltage limits than the L-series. I do wonder about the significance of using a 5600uF cap in place of a 3000uF cap. But these caps are all about biggest C in the space available (at the time), right?

The big unknown that remains is what IS the ripple current to these caps in these supplies? I wonder if I can put together a reasonably accurate simulation. Just to get some sense of order of magnitude.

I didn't come across (yet?) the effect of voltage on the capacitor, besides little to no effect if kept below spec. But common sense suggests a cap operated near its spec is more likely to fail than one that has some head room. The question is what is "some" head room? A good value seems 2x, but it's not well defined. Lifetime suggests no real gain as long as spec is not exceeded and it's just a matter of providing enough head room to ensure spec is not exceeded under all operating conditions.

Here's my final list ordered from Mouser:

C1512, 560uF: UCY2C561MHD, 560uF, 160v, 2130mA, 18mmx40mm, 7.5mm lead spacing.
C1513, 1200uF: LGN2C122MELA40, 1200uF, 160v, 2300mA, 25mmx40mm, 10mm lead spacing.
C1542, C1552, C1562, 5600uF: LGY1H562MELA40, 5600uF, 50v, 2300mA, 25mmx40mm, 10mm lead spacing.

These all fit the adapter boards from the seller on eBay noted above. I bought two sets of both adapters and caps. One as a back-up for my first attempt at installation, and should I succeed without damage I have a second set for my other 465.

This was not as bad to do as I'd feared, and the caps that came out of my "parts" 465 are so bad that I'm kind of a convert now. Leaking is quite evident, though not so much that the A9 board is affected. While I don't have an ESR type meter, I'm able to use the equipment I have to get some kind of capacitance measurement off working caps. It's difficult to impossible to measure the capacitance on the old PS caps as the internal resistance is so high that shorting the leads does not discharge the cap appreciably (C1513 specifically). But the measured cap is less than 1/10th the rated cap. There's a big fat R between the lead and the internal capacitance. I only find it odd that this scope worked at all. I really wish I'd gotten some ripple measurements off the caps of the supplies (before regulation) before it died. I suspect the regulators were compensating until the 55v regulator couldn't take it any more. I'll have to take a look when I get it back working. I wonder if any of the regulator components are affected by the stress.

Sorry for the long write up, but I hope the info is useful to someone down the road.
Dave


Dave Peterson
 

I've added photos (and schematic) of waveforms for C1512 and C1513: https://groups.io/g/TekScopes/album?id=258720

What I realized looking at these is that the unregulated 55v top of C1513 is basically the transformer/rectifier output with no filtering at all. This "noise" is propagated throughout the power supply. The 120v unregulated node is being pulled down by this noise via the C1512 coupling (i.e. it has some remaining capacitance), and the other supplies are also affected because the 55v regulated output is going to the + input of all of the other supplies' comparator input. The 55v regulator cannot compensate for the full swing unfiltered noise of the unregulated node. Note that the 120v unregulated noise is not pulling to ground where the 55v unregulated node is pulling all the way to ground. No cap at all. Thus the conclusion that C1513 is the culprit. And FWIW, the 120v regulated output is reducing the noise pretty well. It's just still too much to remove entirely.

Pulling the caps verified that C1513 is suffering massive internal resistance, and all other caps are suffering large internal R and reduced effective capacitance. C1513 has capacitance, but it's behind the large internal cap. It'll charge and discharge _very_ slowly. It shows about 1uF at the lead, nominal 1200uF.

That aside, what these waveforms illustrate for me is the nature of "ripple current" a DC PS filter cap needs to contend with. Through out its lifetime a filter cap is going to be pulling AC current at source frequency to smooth the DC value. The cap must pull current during these rectifier valleys to maintain the DC value. There are a number of ways to measure this current in-situ, but I'm not going to put a measuring R in the circuit - as interesting as that'd be - and I'm not going to buy a current probe just for this. To me the AC supply current the cap is pulling is the true stress on these caps, and transient, while potentially relatively large, likely doesn't contribute to significant heating and the associated reduction in lifetime.

Thought I'd share the images as they give a nice visceral depiction of what these caps are filtering.

Dave


Dave Peterson
 

To wrap up my PS cap replacement:

I replaced all 5 power supply filter caps. See thread above for details on replacements. I spent a couple of afternoons carefully pulling old supply caps. With all the new parts in hand I got the new caps and adapter boards put together in another afternoon. The next day I spent the morning installing the new caps. See photos in:

https://groups.io/g/TekScopes/album?id=258720

For the large negative terminal holes in the A9 board I used a pieces of aluminum ducting tape. Not silver duct tape of Mythbusters fame, but actual aluminum foil tape used for sealing stove exhaust vents. I used a three-hole punch to make nice little circles that matched the trace pad around the large hole. I just needed it to stop the solder from running through and pooling on the cap. Worked pretty darn well.

By afternoon I had everything put back together, held my breath, and pulled the power switch on. Nothing. Not even a low voltage light. No channel indicator lights, no trace. Turned it off. It didn't take long to find the problem. I first checked the outputs of the rectifiers at the new caps. Good voltages and no more deep voltage dips. Output of 110v regulator good, output of 55v regulator 0v. Or just a couple of volts. After some investigation I found that the 57.3v at the anode of CR1532 was zero. Strangely the voltage at R1528 was zero, tho the + terminal of C1512 was fine.

Tracing the trace between R1528 and C1512 I discovered that the via through A9 at the + terminal of C1512 lost its metallization when pulling the old cap. With no metal in the via reflowing the solder on the cap pin didn't allow solder to connect to the other side of A9 and left no connection to the trace to R1528. It took several attempts today to try to replace the metal via. I tried aluminum foil, but the solder wouldn't flow on aluminum. I realized I needed to use copper, and had some copper foil. I was able to reinstall C1512 and confirm the continuity to R1528.

So finally all caps are installed and working now. I've started the calibration procedure and all PS levels are working perfectly. Setting the 55v supply to 55.0v results in all other supplies being within 20mV of perfect. I've a ways to go through the calibration, but I'm back where I was when C1513 failed.

Dave