EEVblog #695 - Mailbag Tektronix 2465 Analog Oscilloscope

[(In)Sanity]

I've got a mint condition fully functional and calibrated 2465BDM that I just love.  I still use my DS2000A series more often,  but that's just because I don't want to age the jewel of a Tek.   The ASIC's overheat without the case and the proper air flow.  Even when servicing you should have air moving over them.   I've added the heat sink to the U808 or something like that chip on the one I've got.   Great scopes.

This is misinformation:

1. The 2465 series scopes will run fine without the case, with no overheating issue.  You are confusing them with the 2430/2440 DSOs.

2. There is no evidence that adding a heatsink to U800 reduces the failure rate (which is not that high any more, seems like most of the bad U800s have already failed).

Good to know,  I'm however leaving the heat sink in place

Jeff

[c4757p]

2. There is no evidence that adding a heatsink to U800 reduces the failure rate (which is not that high any more, seems like most of the bad U800s have already failed).

No, but there's not much sample size to discredit the procedure either, and for the minimal effort it's definitely worth the attempt if it does do anything. The chip does run hot, which tends to accelerate most types of aging.

Though from what I've seen in the 2445/2455 (never had a 65), you seem to be right about the bad ones having mostly failed by now.

For a while, I had entertained the idea of designing modern replacements for some of the hybrids in that thing - someone else has done that too, might have been for U800. A lot of them could pretty easily be replaced with modern high-speed analog ICs now.

[edavid]

2. There is no evidence that adding a heatsink to U800 reduces the failure rate (which is not that high any more, seems like most of the bad U800s have already failed).

No, but there's not much sample size to discredit the procedure either, and for the minimal effort it's definitely worth the attempt if it does do anything.

Unless you damage U800 while installing the heatsink 

The chip does run hot, which tends to accelerate most types of aging.

It's not a normal aging issue, it's a die attach failure.  For all we know, high temperatures help it stay attached.

If a heatsink helped, you would think Tek would have added them at some point... they were certainly replacing U800s under warranty.

[SeanB]

Possibly they figured the heatsink would reduce the failure rate, but that you would have to add it with a new chip as well, as the old one would be somewhere between good and fail. Changing the few percent that failed in warranty, and then repairing at cost was probably cheaper than replacing all the chips with new.

[edavid]

Possibly they figured the heatsink would reduce the failure rate, but that you would have to add it with a new chip as well, as the old one would be somewhere between good and fail. Changing the few percent that failed in warranty, and then repairing at cost was probably cheaper than replacing all the chips with new.

If Tek thought a heatsink would help, they would have added one when they replaced U800, or when they had the scope in for some other repairs.

[(In)Sanity]

Possibly they figured the heatsink would reduce the failure rate, but that you would have to add it with a new chip as well, as the old one would be somewhere between good and fail. Changing the few percent that failed in warranty, and then repairing at cost was probably cheaper than replacing all the chips with new.

If Tek thought a heatsink would help, they would have added one when they replaced U800, or when they had the scope in for some other repairs.

I'm not sure why we are arguing about the merits of removing excess heat from a chip that runs really hot.   If we were talking about removing a heat sink then I can see it.  Perhaps a Tek employee can answer the question.   I think we should focus our attention on what could be wrong with the B version Dave has instead of worrying about a non-issue.

Jeff

[edavid]

I'm not sure why we are arguing about the merits of removing excess heat from a chip that runs really hot.   If we were talking about removing a heat sink then I can see it.  Perhaps a Tek employee can answer the question.   I think we should focus our attention on what could be wrong with the B version Dave has instead of worrying about a non-issue.

Just for fun, like all the other arguments we have here.  Anyway, U800 runs about 60C without a heatsink... hot but not "really hot".

[(In)Sanity]

I'm not sure why we are arguing about the merits of removing excess heat from a chip that runs really hot.   If we were talking about removing a heat sink then I can see it.  Perhaps a Tek employee can answer the question.   I think we should focus our attention on what could be wrong with the B version Dave has instead of worrying about a non-issue.

Just for fun, like all the other arguments we have here.  Anyway, U800 runs about 60C without a heatsink... hot but not "really hot".

Ok,  sorry...carry on.

[trackman44]

Hi Dave. I have a 1985 vintage 2445 that I fixed using parts from a second unit ( I replaced the missing U800 chip from the scrap one). You know what you always say when a unit doesn't work, check for voltages!  I replaced all the electrolitics on the PSU and downloaded the service manual and adjusted the voltages to spec and it works like a champ. Do this first and cross your fingers your vertical deflection chips are still OK!

Will

[Rasz]

different outcome after unit warmed up is a sign of dried caps, warmed up regain some capacitance and allow digital side of the scope to stay on (albeit hung/stopped). It does look salvageable

[G4ZWI]

Yep, what Rasz says,...  I picked up a 2445 for a song which was coming up with allsorts of selftest fails, hit & miss powerup etc,... after a PSU electrolytic cap replace, and set up, has given sterling service for years,.... Some custom parts are available from QService  ( on the Greek island of Rhodes)
Not cheap, but seem to stock most of the hard to get bits... I did stick a small heatsink on my U800, it has never given any trouble, but as I had the unit apart......
If you decide you don't need them, Dave, I'll send you the shipping to 'The Old dart',  I'd love em!!!!

Cheers,  Merry Christmas & a Happy New Year to you & yours,...

Fred G4ZWI

[Howardlong]

...and for those of us too lazy to fix 'em, I can recommend ebay seller f1power01 in Europe who frequently has both 2465 and 2467 in stock at what I consider to be reasonable prices (for the EU).

http://stores.ebay.co.uk/Electronique-professionnel/_i.html?rt=nc&_nkw=Tektronix&_sid=454615688&_trksid=p4634.c0.m14.l1581&_pgn=3

Last week they had a 2465B (£286) and a 2467B (about £350) they seem to be gone now, but there's a good selection of 2465A there now.

I've had both a 2465B and a 2467B off this seller. Delivery is about a week to arrive from France to The Old Dart. (BTW maybe I've lived under a rock, never heard that before).

I have no connection with the seller other than as a satisfied customer.

One comment I should add for new players is that when you first switch them on, be patient. Nothing visible happens for a few seconds. The first 2465B I bought about 12 years ago I thought was DoA until I realised impetuosity isn't a good quality to have on these occasions. Also getting the illumination right and finding the trace can take a short while particularly if you're new to the scope. Make sure you have trigger Auto or Auto Lvl set to aid this so it's continually triggering. I find myself frequently changing the trace illumination of these scopes, a feature common to all CROs to one degree or another, but I seem to have to do it a bit more on the 2465/7s.

[kevinpt]

Just a note for those inspired to hunt one of these down. The Original and A (350MHz) models of the 2465 keep the cal. data in a regular SRAM with a separate backup battery instead of the Dallas chip used on the B models. It is easier to replace the battery on these while powered up without upsetting the data than dealing with the Dallas parts. That is a positive factor in their favor. Also, the custom ICs on the older models were all made at Tek's fab and have a lower failure rate than the ones used on the early B models when IC production was shifted to an outside vendor. Apparently the issues were sorted out by the end years of production for the B's. The A's have most of the new features that the B models have except for the auto-measurement capability.

The service manual is also a good read. A whole section is dedicated to describing how they implemented the algorithm to make the 360-degree pots on the front panel act like digital encoders without a jump when the wiper crosses over the non-conductive region. They did it by using ganged pots with the wiper of one rotated 180-degress with respect to the other and switching from one to the other when nearing the gap. You don't see that sort of attention to detail anymore in instrument documentation.

[guido]

The original 2465 does not have a battery. Cal data and front panel settings are stored in an early form of EEPROM (called EAROM, electrically alterable ROM). See the service manual 

[Howardlong]

In case you're still in the market for one of these, I did a video of the difference between a 2465B and 2467B.

http://youtu.be/ZGi1Uw8l-ac

Here I dispell the urban myth that electronic engineers have to sit in darkened rooms. Well they don't if they have a Micro-Channel Plate (MCP) oscilloscope anyway.

The Tektronix 2465B and 2467B are generally identical oscilloscopes, except for the screen (aka CRT, or cathode ray tube for the younger viewers).

The 2467B CRT uses MCP technology which allows the user to easily see very small repetitive but relatively infrequent pulses. The 2465B uses a conventional CRT.

The same applies when comparing the 2465A to the 2467, where the primary differentiator between the 2465A and 2467 models was the MCP CRT. (There was no 2467A).

The test shown uses a Jim Williams pulse generator which generates a 6ns pulse every 82us or so, a duty cycle of about 1:13,500.

[edavid]

The same applies when comparing the 2465 to the 2467, and the 2465A to the 2467A, where the only differentiator between the 2465x and 2467x models was the MCP CRT.

Minor correction, there is no such thing as a 2467A.  The 2467 is closer to the 2465A than the 2465.

[Howardlong]

Thank you, I will correct that, lucky it's just words and not in the video!

[barjammar]

It was a touching moment to see young Sagan learning how to do a two-thumbs-up from his dad.  I think there is a developmental milestone somewhere for when kids learn to copy hand shapes and gestures but I couldn't find a reference tonight.  I saw Sagan again in the recent soldering of the 8 bit computer, four years later.  He reads well and has a fair comprehension of decimals, at least for solder wire thickness.  Great effort Sagan!