EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: linux-works on March 05, 2014, 03:23:17 am
-
hopefully this won't be too controversial (I saw the comments in the 2464b teardown thread), but I would like to at least rationally discuss the pros and cons of doing a DIY heatsink for the infamous u800 on tektronix 2400 series scopes (at least the ones that use this chip).
here is a photo I just took of my 2465b, with the chip in question center of frame and with its 2 star-nuts removed to show the chip and its single exposed heatsink tab.
(http://farm4.staticflickr.com/3802/12941114564_57cd96d246_o.jpg)
from what I can tell, there is an isolated/slotted section of the pcb right below this chip and it has a solder coating on the metal. however, it seems that there is an air gap between the heatsink pcb section and the lower body of the chip. I believe this is so since there is a star washer wedged between the pcb and the chip body on both (screw) sides of the chip.
some questions come to mind. I'd like to discuss these.
- was tek thinking that the star washer 'bites' (points) would be enough to carry heat from the bottom of the chip to the pcb solder plated board area? is the air gap intentional? maybe for heat expansion to avoid the chip cracking during stress?
- those 2 screws look like 'towers' to me. tek would not use large screws like that for no reason; they seem to be 'poor mans heatsinks', essentially. are they counting on that amount of metal to be enough to cool the chip down and keep it in spec?
- if I decide to add a DIY heatsink to this, what are the things to watch out for? for example, I am hesitant to put much weight or stress on the top of the plastic chip. I've seen some heatsinks for this chip screwed down with those mounting nuts on the 2 'towers' (screws). that seems risky to me; I don't know how much room there should be if that plastic chip pkg wants to expand under heat.
- one idea I had was to have a single piece of heatsink metal that sits on the metal tab (on the right of the photo), rises up just a bit higher than the chip's top and then over like a 'bridge' to the other screw. I'd leave an air gap between the top of the chip and the bottom of this 'bridge' so that nothing stresses the chip and there would also be some airflow on top of the chip. that seems the least risky way to do this.
I'd like to hear your comments on this. please try to be constructive (that should go without saying, but I remember some heated replies in that other thread and I'd like this to be a design-based thread, not a 'tek never makes mistakes' thread. please. ok?)
thanks ;)
-
Originally, it was probably calculated that in the case, with the airflow by the fan, it would not overheat. However, when the fan fails, all hell breaks loose. I remember reading somewhere that connecting the 2 tall screws would result in a short between some voltage rail, so your best bet be to probably use small heatsinks like these eBay auction: #331143627474 and maybe put a small fan on top. That would work the best IMHO.
-
ah, good point; I did not yet check that those 2 screws were at different potentials. if they are, then that puts a damper on the idea of a 'bridge' made of metal going between them. thanks for giving me the heads-up on that aspect.
this brings up another related question: I guess there is nothing in that scope series that will shut down the scope if the internal temp gets too high? clearly the fan is not a 3 or 4 wire fan, so there is nothing to watch its tach or change its speed. these days, gear that can be damaged due to fan failures usually send out 'events' (in some way or another) and also do a safe shutdown when things get beyond the 'red' mark. serious equipment does (networking gear, rackmount computer servers, etc).
is there nothing inside this scope that will shut itself down if the temp gets too high? I guess that that was not really a thought they had, back then?
I do think its a good idea. I'd much rather have the gear shut itself down than fry itself. especially now that the gear is no longer fixable via official channels and all the specialty parts are only buyable via 2nd hand sources, if even at all.
fans do fail and it sure would be nice if there was a 'management module' that cared about such things and would shut down the system if anything was looking bad (including even the key supply rails, come to think of it).
-
You are correct, the scope has no way to know when the fan fails or the chips are on fire.
-
I just checked those 2 metal screws: they are at the same potential. they would have to be, too: think about it. there is a solid strip of tinned pcb under the chip and the screws anchor to this on both sides. that's a short.
and so, I might be back to thinking about the 'bridge' idea, again ;)
either with an air gap between the top of the chip or some heatsink 'foam' as a cushion to at least bond it thermally a little while not putting stress on top of the chip.
I may also consider throwing some arduino board in there, somewhere (sorry if that offends any purists) to monitor the fan and maybe even internal temp of some key points. if things get 'bad' a series relay would trigger (or release) that would shut down scope down. ideally, the fan would continue to run for a short bit, too, even after graceful operator-initiated shutdowns (some cars have their fans run for a while after the car stops; its the same idea, here.)
I'm just thinking aloud. don't have definite plans, but looking for feedback.
-
I remember reading somewhere that connecting the 2 tall screws would result in a short between some voltage rail
thinking about it more, what the person may have meant is that there's a diff of potential between the screws (ie, baseof chip) and GROUND. so I won't allow the heatsink to touch ground (or anything else!). I never planned on that happening, anyway ;)
-
ah, good point; I did not yet check that those 2 screws were at different potentials. if they are, then that puts a damper on the idea of a 'bridge' made of metal going between them. thanks for giving me the heads-up on that aspect.
IIRC the 2 posts are at the same negative potential (-15V ? checked it on my 2467 but i can't recall exactly) and the issue is attaching a heatsink to one of those and it touching the case and shorting there. i have a small heatsink that would fit nicely and some thermal epoxy to attach it (and avoid the post poential issue) but I haven't gotten around to doing it. After I monitored the temp of the chip for a while I was less concerned about a heatsink but I hadn't considered the effect an unnoticed fan failure could have....hmmmmm.
-
i don't recall what the failure mode for these chips are but if im not mistaken the chip doesn't even generate enough heat to raise its temp 40C
i spent a bunch of time building this, it worked as long as the TEC did, which wasn't long. water got into the junctions and destroyed it.. unfortunately, its epoxied to the bottom of the chip.
http://johansense.com/other/2465/2465u800_5.JPG (http://johansense.com/other/2465/2465u800_5.JPG)
-
thanks, mtdoc, for confirming that the 2 posts are at same potential. I'll just be sure not to have anything that touches them, touch anything else.
I will check to be sure, but my guess is that anything that is as high as those 2 posts (or lower) will fit safely inside and not be at risk of touching anything 'bad'. that would be my height limit that I would not exceed.
I'm not a fan (lol) of adding a fan TO this local heatsink. I do want to have only 1 fan in the box, but I'm thinking that it might be a nice mod to have a temp and fanflow monitor.
-
i spent a bunch of time building this...
Holy Frankenscope! Well done!
-
all that extra ribbon cable lead-length didn't do any harm?
-
Great. These threads about fixing, improving, and repairing the old Tek 2400 series has made me want to buy a broken one and restore it to its former glory. If only I could find one cheaply enough. *sigh*
-
That two poles are connected together seamlessly at the pcb and its at (negative) "-5" Volt, not ground. :-\
(https://www.eevblog.com/forum/testgear/tek-24xx-series-u800-heatsink-diy/?action=dlattach;attach=84028;image)
This picture is stolen from Kibi's -> (Tek 2445 Repair thread (https://www.eevblog.com/forum/testgear/tektronix-2445-repair/)), also it has a picture of the U800 shot at the bottom that revealed the metal tab, this bottom part is the ideal place to cool this chip imo.
Just an idea, the ideal cooling is to make a custom CNC'ed copper plate that extends beyond the IC for larger cooling area, placed and sandwiched with thermal pad or thermal grease at the "bottom" of the IC. Of course this needs a major job to de-solder the IC, and solder two single row of sockets for the two rows pins. These sockets will lift the chip higher and makes enough space at the bottom of the IC for the copper plate. Not easy, but this is my obsession that I never have time to do it myself. :'(
Just becareful when turned on without the enclosure, make sure you have big fan, not tiny dc fan, blowing the whole big A1 board.
...... but I remember some heated replies in that other thread and I'd like this to be a design-based thread, not a 'tek never makes mistakes' thread. please. ok?)
:palm: damn, you just bring out the long forgotten bad memory. :-[
-
I got a 2465A with lots of extras on eBay for $105 shipped. The extra stuff alone was worth that.
But sadly it wasn't broken so there wasn't anything to fix inside. I added a heatsink with very strong thermal tape to U800 and cased it back up. Super easy, no special parts of bolting required, and I would hope someone wouldn't be bashing this thing to make the heatsinks come loose (but the scope would probably be destroyed before the light heatsinks come off).
-
I got a 2465A with lots of extras on eBay for $105 shipped. The extra stuff alone was worth that.
But sadly it wasn't broken so there wasn't anything to fix inside. I added a heatsink with very strong thermal tape to U800 and cased it back up. Super easy, no special parts of bolting required, and I would hope someone wouldn't be bashing this thing to make the heatsinks come loose (but the scope would probably be destroyed before the light heatsinks come off).
Uh. ANd the jealousy builds. I would REALLY love a 2467B.
-
Great. These threads about fixing, improving, and repairing the old Tek 2400 series has made me want to buy a broken one and restore it to its former glory. If only I could find one cheaply enough. *sigh*
the broken ones are cheap enough, but they may be TOO broken!
I was going down the 'buy a new 4ch rigol' path but decided to get an old classic tek instead. I may get a 4ch rigol someday, but for now, I think a good old tek is going to be a more trustable bit of kit (once I get it recapped and cal'd).
-
I'm hearing suggestions of unsoldering this chip (I dread it, but I can do it if it must be) and helping it cool from the bottom.
I may consider that. I do trust my desoldering skills but even experienced people can break traces doing that. its not a perfect process and there is risk involved.
those star washers really bug the hell out of me, down there under the chip. even flat washers would have spread the heat out and transferred it better, still giving the needed (so I am told) spacing to de-stress the chip pins.
the sip socket strips certainly will destress the chip pins. that's a nice win. maybe enough to convince me to take the risk and unsolder mine for the upgrade. and if the chip ever does blow, having a socket already there will make the pain a bit easier to live with ;)
-
I was going down the 'buy a new 4ch rigol' path but decided to get an old classic tek instead. I may get a 4ch rigol someday, but for now, I think a good old tek is going to be a more trustable bit of kit (once I get it recapped and cal'd).
If you haven't done so already, definitely join the Yahoo Classic Tek Scope group (https://groups.yahoo.com/neo/groups/TekScopes/info). It is very active with many new posts on various topics daily. Several retired Tek engineers and a wealth of knowledge.
The whole "prophylactic recapping" topic is somewhat controversial. I tend to fall in the "if it ain't broken don't fix it" camp. There was a prolonged recent thread on that group about someone whose attempt to recap a 2400 series scope went terribly wrong.
-
yes, I am already on that group. a very good and helpful group.
I already removed the SMD caps that are known to be bad on this board:
(http://farm3.staticflickr.com/2894/12826733144_33a59a8343_o.jpg)
and those are a 'must do' mod. its already too late since I removed the old ones and the new ones are due here tomorrow.
on the lvps board, I'm told that pretty much all of them should be replaced. I plan to do that.
beyond that, I'm not sure. replace the dallas NVRAM module (after copying the data out to a new one) is a must-do, also.
I would not just seek-and-replace every single lytic, but on the power supply board, I'm reading lots of reports of it being a 'must do' mod for a scope this old (mine has a 1992 date on the boards).
-
i don't recall what the failure mode for these chips are but if im not mistaken the chip doesn't even generate enough heat to raise its temp 40C
Mine is scorching hot and almost unbearable to touch with my finger, definitely above 40 C.
-
The whole "prophylactic recapping" topic is somewhat controversial. I tend to fall in the "if it ain't broken don't fix it" camp.
the other side of the coin is: if and when they fail, you'll have much bigger problems than just re-capping.
that's what worries me.
cap replacement is a preventative thing, the way I see it. if you wait *until* they fail, then you waited too long.
-
The whole "prophylactic recapping" topic is somewhat controversial. I tend to fall in the "if it ain't broken don't fix it" camp.
the other side of the coin is: if and when they fail, you'll have much bigger problems than just re-capping.
that's what worries me.
cap replacement is a preventative thing, the way I see it. if you wait *until* they fail, then you waited too long.
I guess it depends on how they "fail". My 2467 is pre SMD so none of the leaking smd cap concerns. My LVPS ripple is withing specs and the scope is working perfectly.
I'm a relative newbie but I have seen and appreciated the arguments from the experts on both side of that debate and it seems a toss up to me. But once you factor in my general laziness (and the fact that I have 2 other tek scopes and a new Rigol) the "do nothing" argument has won -- for now. I did stock up on a set of replacement caps for when the time comes...
-
There are 3 generations of 2465 scopes: 2465, 2465A, and 2465B. There was never a need to add a heat sink in the design. You can add a heat sink but you might end up damaging a perfectly good U800 chip that is becoming unobtanium. One of the main problems of the chip is that it is very fragile, not only thermally but also mechanically. If you overtighten the nuts you might break the chip so make sure you don't exert any pressure when mounting or gluing a heatsink on the chip.
-
One of the main problems of the chip is that it is very fragile, not only thermally but also mechanically. If you overtighten the nuts you might break the chip so make sure you don't exert any pressure when mounting or gluing a heatsink on the chip.
Hence the idea of using a socket instead of soldered permanently to the board. Cause when the chip is soldered to board, while its tighten against that two poles, bad thing tends to happen, especially when that two nuts are not tighten at the right point which is not easy, and have even pressure at that two mounting points.
Once the chip is installed on two sip sockets, it opened up many choices, like securing the chip "AND" with the heatsink attached without mechanically stressing them both at that two poles, or many other different options that are available.
Just look at this shot of my U800 at this sharp angle, imo, this is a mechanically bad design, also with thermal expansion and shrinkage at the board, chip body, the poles, soldered pins and etc, this securing technique will put a lot of stress at the chip body it self through thermal cycling.
(https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/?action=dlattach;attach=16705;image)
-
all that extra ribbon cable lead-length didn't do any harm?
i don't know but it didn't work long enough for me to find out.
any effect would be confined to lack of bandwidth on the horizontal sweep, which would show up as non-linearity, not as lost information.
my U800 failed within 3 months of very limited operation in 100F conditions, long time ago. it works as long as its below about -20C
-
Cheapest eBay buy now price here in central FL is $356. There are 2 that are being bid on but they will definitely go over the $105 price. Skycraft only has a couple of 2245's for $250 each. Turning green down to my toenails. |O
-
Hiya!! My two'pennorth, (FWIW!! ) :D
These guys http://www.qservice.tv/vpasp/default.asp (http://www.qservice.tv/vpasp/default.asp) stock a fair range of spares for the Tektronix range of scopes, & they have a small heatsink for U800 which they suggest you just stick on top of the chip with 'Thermally Conductive Epoxy'.
It also seems that about 80% of the 24XX series scopes problems are caused by PSU caps leaking/drying out.
I have a 2445 I purchased from a military surplus depot near me 20 years ago, and after a recent psu teardown/recap, is working a treat again.
Cheers, Guys
Fred
-
no offense to qservice, but I don't think they put much thought into it. I immediately rejected the idea of simply gluing a heatsink to the top. plastic is a poor conductor and you could be accumulating heat in that sink and pumping it right back ONTO the chip since the chip sits with its 'top' plastic part facing downward. does not seem smart to me...
-
From the pictures above, it appears that the tabs are already heatsinked to the PCB. It may be tough to improve on that, unless you improve the surface area and thermal conductivity of the mechanical connection between the chip and the PCB. A potential drawback of that could be greater susceptibility to stress on the chip with vibration/expansion/contraction of the PCB. The biggest risk of a top-mounted heatsink is eddy currents which could cause air to detour around some components that currently are adequately cooled, changing the thermal profile of the PCB nearby. Without careful research, is it worth making either type of mod? Is the U800 that prone to failure from overheating?
I'm interested in this thread because I have a 2465B which has worked fine for years, but I do plan to replace the Dallas chip and some of the leak-prone caps in the near future. I had not planned to touch this chip, but am always interested to hear arguments one way or the other about such issues. So far, haven't seen any convincing evidence to support touching this chip.
Dave
-
I got a few emails from the yahoo tek scopes group and mostly they said 'leave it alone!'. to do a proper heatsink you have to unsolder the chip (a little risky), add a shim of metal between the chip and board, torque the chip down just the right amount and either solder the chip back OR socket it (socket helps relieve stress but you lose some of the heat sink nature from the pins being soldered in).
to unsolder this chip you have to remove the motherboard. that is a major hassle and also can be error prone (some people have broken things by removal and reinstallation of the main board).
all in all, it sounds like its not worth it. the chip should not fail on its own; it fails due to being a maxim chip (not a tek chip) or if the fan fails (change your fan regularly!) or if you run the scope without its case on and don't have a service fan blowing in the right places. it seems that the chip won't usually blow on its own in normal every day use.
if it ever blows, THEN you have to deal with it, but it seems that to fix things that are not yet broken could do more harm than good. replacing caps is a good preventative measure but this chip is not in dire need of a heatsink (it seems).