Extending the life of a Tek 2465?

[DaveC]

The U400 hybrid IC in the Tek 24xx 'scope models is infamous for its short
life.

Has anyone attempted heat-sinking it (if it is not already) and blowing it
with a small fan (a la pentium and PowerPC)?

Just crossed my mind in between random other thoughts...

 

[Jim Yanik]

The U400 hybrid IC in the Tek 24xx 'scope models is infamous for its
short life.

Has anyone attempted heat-sinking it (if it is not already) and
blowing it with a small fan (a la pentium and PowerPC)?

Just crossed my mind in between random other thoughts...

I haven't heard of anyone doing it,but it would seem to lend itself to
being heatsinked,the metal substrate is exposed on the topside of the
IC,and there's threaded studs to secure a HS. I don't think there's enough
clearance for a fan(cabinet-HS clearance)

It certainly couldn't hurt.
Although,I suspect it's some problem in the die construction itself.

 

[DaveC]

I haven't heard of anyone doing it, but it would seem to lend itself to
being heatsinked, the metal substrate is exposed on the topside of the
IC, and there's threaded studs to secure a HS. I don't think there's enough
clearance for a fan (cabinet-HS clearance).

Is U400 accessible w/o removing any PCB's? It sounds like it's on the outward
side of the main board, just beneath the case skin. Is this true?

Thanks,

 

[Jim Yanik]

Is U400 accessible w/o removing any PCB's? It sounds like it's on the
outward side of the main board, just beneath the case skin. Is this
true?

Thanks,

Yes,the metal substrate is on the TOP of the IC,exposed,completely
accessable. (but soldered in,40 pin DIP)
All one has to do is coat with HS compound,and bolt the HS to the
studs,IIRC,they're 4-40 thread.I don't think the HS could be more than
perhaps .375" high,though.(SWAG)
You don't want it to contact the cabinet,as I'm not sure if the substrate
is electrically neutral or not.

 

[DaveC]

Yes,the metal substrate is on the TOP of the IC,exposed,completely
accessable. (but soldered in,40 pin DIP)
All one has to do is coat with HS compound,and bolt the HS to the
studs,IIRC,they're 4-40 thread.I don't think the HS could be more than
perhaps .375" high,though.(SWAG)
You don't want it to contact the cabinet,as I'm not sure if the substrate
is electrically neutral or not.

-=-=-=-

Hey, what gives? I wasn't expecting this:

http://home.covad.net/~peninsula/U400.jpg

Isn't this already heat-sink'd? I wasn't expecting all the metal. I don't see
how a heat sink could help this situation much. It seems pretty cool, as is.

This is on a 2465DMS, s/n B051xxx. (And yes, it works. FIne.)

Comments?

Thanks,

 

[mike]

-=-=-=-

Hey, what gives? I wasn't expecting this:

http://home.covad.net/~peninsula/U400.jpg

Isn't this already heat-sink'd? I wasn't expecting all the metal. I don't see
how a heat sink could help this situation much. It seems pretty cool, as is.

This is on a 2465DMS, s/n B051xxx. (And yes, it works. FIne.)

Comments?

Thanks,

Well, it's not a 40-pin dip.
The board sez channel2 preamp.
Methinks you have the wrong U400.

But when you find the right u400, please post that pix.
I don't really want to take mine apart.
mike

--
Bunch of stuff For Sale and Wanted at the link below.
laptops and parts Test Equipment
Honda CB-125S
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[DaveC]

Well, it's not a 40-pin dip.
The board sez channel2 preamp.
Methinks you have the wrong U400.

I have the Tek svc manual. It says there is only one U400. Maybe it's a
40-pin under that massive sink?

I think that "channel 2 preamp" is the description for U200. Look in the
lower left corner of the picture: next to "U400" is the description
(partially hidden by the cable) "channel switch" which is what U400's
function is.

Jim Y. described its location accurately, as did the manual's component
location page, and that's where I found it. Just not the size/form factor I
was expecting...

 

[Jim Yanik]

http://home.covad.net/~peninsula/U400.jpg

That's not the horizontal output IC being discussed,that's the CH2 preamp
hybrid IC.

Odd,but most of the hybrids with this type of HS don't even need a HS,they
don't get warm at all,but the H-output IC doesn't have a HS,yet does get
warm. I guess it's just a fancy package to clamp the hybrid substrate to
the PCB socket.

 

[Jim Yanik]

Well, it's not a 40-pin dip.
The board sez channel2 preamp.
Methinks you have the wrong U400.

But when you find the right u400, please post that pix.
I don't really want to take mine apart.
mike

IIRC,it was U800 that is the 40 pin DIP horiz output IC.
(and not a hybrid)

 

[John Miles]

Well, it's not a 40-pin dip.
The board sez channel2 preamp.
Methinks you have the wrong U400.

But when you find the right u400, please post that pix.
I don't really want to take mine apart.
mike

My understanding is that the troublesome 246x part is U800, the
horizontal driver. This is a (28-pin?) DIP with an integrated heat
spreader on the bottom that's bolted to the board.

Does U400 cause a lot of problems, too?

-- jm

 

[DaveC]

That's not the horizontal output IC being discussed,that's the CH2 preamp
hybrid IC.

My mistake :-(

I'll look at U800, then.

Thanks,

 

[DaveC]

OK, I've done it. I added a heat sink to U800 in my 2465DVS (w/DMM option):

http://home.covad.net/~peninsula/Before.jpg

http://home.covad.net/~peninsula/After.jpg

The IC is on the main board which is located on the under side of the scope,
easily accessible once the case is removed (by removing 6 screws on the rear
of the scope).

This heat sink fit just perfectly. It is a stepped design, probably for a
TO-220. This allows it to raise up above the neighboring components without
interference.

The IC is already bolted to the PCB with nuts on two #4-40 studs. I had to
notch the ends of the sink to sit under the nuts, but that was trivial. It
was important to keep below the height of neighboring real estate so the heat
sink won't contact the case when reinstalled. I used a *light* coat of white
heat sink compound under the sink.

Who knows if this will extend the life of this fragile IC, but I consider all
semiconductor failures heat-related. So to my mind, this will help in some
incalculable way.

Enjoy,

 

[mike]

OK, I've done it. I added a heat sink to U800 in my 2465DVS (w/DMM option):

http://home.covad.net/~peninsula/Before.jpg

http://home.covad.net/~peninsula/After.jpg

The IC is on the main board which is located on the under side of the scope,
easily accessible once the case is removed (by removing 6 screws on the rear
of the scope).

This heat sink fit just perfectly. It is a stepped design, probably for a
TO-220. This allows it to raise up above the neighboring components without
interference.

The IC is already bolted to the PCB with nuts on two #4-40 studs. I had to
notch the ends of the sink to sit under the nuts, but that was trivial. It
was important to keep below the height of neighboring real estate so the heat
sink won't contact the case when reinstalled. I used a *light* coat of white
heat sink compound under the sink.

Who knows if this will extend the life of this fragile IC, but I consider all
semiconductor failures heat-related. So to my mind, this will help in some
incalculable way.

Enjoy,

A couple of thoughts.

What's on the backside of the IC? Do the studs connect to some metal
structure that can act as a heat sink?

What was the temperature of the tab before and after the modification?
You can also stick a probe thru the hole and measure the top of the IC
before and after just for grins.

I don't know anything about this particular IC.
BUT
In general, when there's a tab, virtually ALL the heat flows out the tab.
Heat sinking the plastic probably won't do any harm and may do some good
IFF you don't hinder the heat flow out the tab.

Can't tell from the picture...did you leave the nut on the tab and add
another for the heat sink? If you just used one nut, you probably
REDUCED the heat flow out the tab into whatever is on the backside
by a LARGE amount.

A couple of temperature measurements should tell the story. I wouldn't
run the scope until you measure that tab temp.

More mechanical details please.
mike
--
Bunch of stuff For Sale and Wanted at the link below.
laptops and parts Test Equipment
Honda CB-125S
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[DaveC]

Can't tell from the picture...did you leave the nut on the tab and add
another for the heat sink? If you just used one nut, you probably
REDUCED the heat flow out the tab into whatever is on the backside
by a LARGE amount.

A couple of temperature measurements should tell the story. I wouldn't
run the scope until you measure that tab temp.

More mechanical details please.

The tab seems to run through the middle of the IC:

http://home.covad.net/~peninsula/Before.jpg

http://home.covad.net/~peninsula/After.jpg

On the right end (as viewed in the photo), the tab runs out like a TO-220 and
is held to the PCB with a stud and nut. I checked the level of the top of
this nut and found that it was above the level of the plastic IC case, which
would have not allowed a flush-fitting heat sink.

On the left end, the plastic is notched (you can see a bit of the notch just
to the right of the left nut) which exposes a bit of the integral tab (Jim Y.
calles this the "metal substrate" of the IC). This nut is not in contact with
the metal, but I presume this nut's purpose is to hold the metal in contact
with whatever heat sink is below it the chip.

I replaced the right nut with one that was a bit shallower in hight, so it
could continue to hold the tab down, but also so the heat sink would clear
the top of the nut. The original nut incorporated a star lock washer, and the
replacement didn't (it would have been too tall). I used a *very small*
amount of Loctite thread locker (blue) on the top of the plain nut (where it
meets the stud) after installing it, to keep it from loosening.

I notched the ends of the heat sink where they intersected the #4-40 studs.

I put a *very small* amount of white heat sink compound on the top of the IC
and placed the heat sink on top of it, moving it around a bit to get good
distribution.

Then added the original nuts and tightened carefully.

Enjoy,

 

[Jim Yanik]

A couple of thoughts.

What's on the backside of the IC? Do the studs connect to some metal
structure that can act as a heat sink?
edit

I don't know anything about this particular IC.
BUT
In general, when there's a tab, virtually ALL the heat flows out the
tab. Heat sinking the plastic probably won't do any harm and may do
some good IFF you don't hinder the heat flow out the tab.

On this particular IC,the metal substrate is on the TOP side of the IC.Just
right for adding a heat sink! The PCB does not dissipate any heat from the
IC except for what comes thru the leads.I suspect the studs were intended
for a HS that someone else thought unnecessary,or an extra expense.

I would not overtighten those nuts,it might cause IC failure
itself;mechanical strain.

 

[mike]

On this particular IC,the metal substrate is on the TOP side of the IC.Just
right for adding a heat sink! The PCB does not dissipate any heat from the
IC except for what comes thru the leads.I suspect the studs were intended
for a HS that someone else thought unnecessary,or an extra expense.

I would not overtighten those nuts, it might cause IC failure
itself;mechanical strain.

I thought I understood until I read this. According to the pictures and
the previous description, the tab looks like it sits on the board.
There's room to add a nut who's top sits at the level of the IC top???

If there's any room between the tab and the board, the mounting stresses
on the package would be enormous. This assumes the studs go thru the
board. Just occurred to me that the studs could be pressed into the tab
and not go thru the board at all???

If the chip sits on the tab, the optimum setup might be to put a
threaded spacer on the left end that fits inside the notch, without
binding the plastic, to get a solid conductive path to the heatsink
on that end. Contact with the top of the chip should be irrelevant
if all the heat is coming out the tab. Also might want to grind the
anodize off the heat sink (keeping it FLAT) for better thermal contact
with the stud material.

I'd still like to see the temperature numbers. The question would be,
"What mode should the system be in for max heat?"
High sweep rate would put the most stress during sweep, but holdoff
would reduce the duty factor. So, tests at different sweep speeds
while activelly triggered might be in order. Just letting it free-run
at high sweep won't maximize the duty factor. Which brings up another
issue.

Back when I was designing pulse generators, I read an article that
suggested that the primary failure mode of output transistors was
thermal cycling. Given the thermal time constants in the package,
the worst case for thermal cycling was in the 10-100Hz. rep rate range.

So, if you leave the scope sitting untriggered in auto-trigger mode,
you're running the sweep circuits near maximum thermal cycling stress.
From now on, I'm gonna turn off auto-triggering on my scopes when I'm
not using 'em.

So, DaveC, can you give us some temperature readings in various modes?

Thanks, mike

--
Bunch of stuff For Sale and Wanted at the link below.
laptops and parts Test Equipment
Honda CB-125S
TEK Sampling Sweep Plugin and RM564
Tek 2465 $800, ham radio, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[DaveC]

So, DaveC, can you give us some temperature readings in various modes?

No. I'm not opening it up any more. It was risky to open it to do the
heatsink, and I'm not going to risk ESD or any other tomfoolery I might do
accidentally to this important piece of equipment in my lab. I just wanted to
do *something* to settle in my mind that I've *possibly* extended the life of
this component. I've done what I can do. Maybe not in your mind, but I'm
satisfied. I'll let y'all know if anything comes of it.

 

[Jim Yanik]

I thought I understood until I read this. According to the pictures and
the previous description, the tab looks like it sits on the board.
There's room to add a nut who's top sits at the level of the IC top???

If there's any room between the tab and the board, the mounting stresses
on the package would be enormous. This assumes the studs go thru the
board. Just occurred to me that the studs could be pressed into the tab
and not go thru the board at all???

the studs are mounted to the PCB.I don't even know why the engineers felt
they needed them.You could leave the nuts off and it would make no
difference.The bottom of the IC plastic package sits right on the PCB,the
substrate IIRC,doesn't hang past the ends of the IC,there's notches in the
ends of the IC for the studs. If I remember right.

 

[DaveC]

The bottom of the IC plastic package sits right on the PCB,the
substrate IIRC,doesn't hang past the ends of the IC,there's notches in the
ends of the IC for the studs. If I remember right.

One end of the substrate hangs out like a TO-220 tab. On the other end, it
does not; instead the plastic is notched for the stud.

See the photo here:

http://home.covad.net/~peninsula/Before.jpg

 

[Vince]

Dave,

OK, I've done it. I added a heat sink to U800 in my 2465DVS (w/DMM option):

http://home.covad.net/~peninsula/Before.jpg

http://home.covad.net/~peninsula/After.jpg

Where did you get this heatsink? I'd like to get one to add to my U800.

I spoke with a Tek engineer who told me that excessive heat was the
failure mode for U800 and that heatsinking it is a good idea, so I think
you're on the right track.

I know someone else who heatsinked his U800 and hasn't had any problems
so far. The only caveat he said is the studs are at -5V so one has to be
sure that the heatsink doesn't touch the case...

-V-