After replacing Tek2440 Ram U664 and U350 by FRAM no scope display anymore

[Alectronic]

Good luck with installing both new 32k U664 and U350 memory chips Bob!
Those are the only ones that I measured that both were connected to the Backup Lithium 3V Keeper Battery in my 2440 scope.
And they are the ones we need to restore to get the Tektronix back alive! So whoever tells us that the scope is still calibratable when the battery is fully drained and the 32k Srams (so not the Dallas NVRAM ones) completely lost their memory and so their calibration/configuration data is most likely ill advised. 

If the calibration still causes any problem, know that for the last CTE calibration the 100 KHz frequency is probably not critical. But I did use a good square wave pulse from my very old from the 70's Systron Donner 100A Pulse Giver to successfully finish that final calibration check. After that I also tested my scope's bandwidth with my homemade improved Jim Williams Fast Rise Avalanche Pulse giver. And it was as it is to be expected of a 300 Mhz fast oscilloscope. 

[David Hess]

The fast rise calibration is for matching the delay between channels.  I used the upgraded sync output from my 2 MHz function generator which was like 4 nanoseconds and plenty fast.  The oscilloscope measures the delay between channels 1 and 2 and then uses that value to shift one channel to match the other.

[bob91343]

Thanks Alectronic!  It will be a couple of months until I receive the parts, most likely.

I measured my other, working, 2440 and its bandwidth runs around 280 MHz, close enough I guess.  I measured it by looking at the output of my HP 8756B and increasing its frequency until the display dropped 3 dB from what it was at low frequency.  The usual monotonic droop in response runs -3 dB at cutoff, and -1 dB at half cutoff.  So to be accurate, I think I used 1 MHz which should be low enough.  How could I raise the bandwidth to 300?  Is there a hardware adjustment?

[David Hess]

I measured my other, working, 2440 and its bandwidth runs around 280 MHz, close enough I guess.  I measured it by looking at the output of my HP 8756B and increasing its frequency until the display dropped 3 dB from what it was at low frequency.  The usual monotonic droop in response runs -3 dB at cutoff, and -1 dB at half cutoff.  So to be accurate, I think I used 1 MHz which should be low enough.  How could I raise the bandwidth to 300?  Is there a hardware adjustment?

More likely the real bandwidth of your the 2440 was higher than 300 MHz, but the output of your HP 8756B (?) fell enough by 300 MHz to give a false measurement.  This could be confirmed using a sampling oscilloscope or sampling RF voltmeter.

Even the effects of the patch cable between the signal source and oscilloscope needs to be accounted for when making this type of measurement.

[bob91343]

I verified the signal input with my other fast scope, the 500 MHz HP 54542A and it comes in at about 600 MHz.  Same cable.

[Alectronic]

About measuring the scope's bandwidth after complete calibration I measured a pulse rise time set about 500pS/Div on 2 Div which gives about 1 nS on my 350MHz Tektronix 2465A oscilloscope with my upgraded Jim Williams Pulse Generator. Which was around 1GHz rise time and when calculated in the formula this results in a bandwidth of 350MHz on my 2465A and about 300MHz on the 2440.
How to make one with todays 2N2369A fast avalanche HF transistors that are protected against avalanching at lower Collector emitter voltages which makes them harder to use as an Avalanche Pulse giver can be read in this link:
https://jestineyong.com/making-a-working-jim-williams-based-pulse-generator/

[bob91343]

Here is the nonsense display I get on my 2440.

[m3vuv]

So only $15 total to diagnose your oscilloscope? Can you even go out to lunch for that? Good test equipment doesn't come cheap, I don't like paying more for the postage than the cost of the items but even so it sounds like a bargain to me.

what a tight ass!!

[BradC]

Here is the nonsense display I get on my 2440.

Yowza. Checking the schematics and theory of operation, U664 is the likely culprit. Looks like the system uP isn't running properly.

[bob91343]

Thanks Brad!  I will start to pursue that avenue.  What part is U664?  Working on these multilayer boards is worse than a pain.

[james_s]

It's easy if you have the right tools. Get a vacuum desoldering station, even one of the Chinese knockoffs will do. Trying to do a job like this without the proper tools is a recipe for a ruined instrument. Also when you replace anything that has a high failure rate, such as static RAM, install a good quality IC socket if at all possible.

[BradC]

Thanks Brad!  I will start to pursue that avenue.  What part is U664?  Working on these multilayer boards is worse than a pain.

One of the RAMs you replaced. U664 is the main processor memory along with the calibration data.

[james_s]

A thought occurred to me, if that RAM was replaced without proper desoldering equipment, it's possible the board has been damaged, such as tearing out the through-hole plating. If that has happened then that would explain why the scope is not working, and it could be very difficult to repair.

[bob91343]

That is entirely possible.  When I get the new ICs I will check for board damage and decide whether repair is required or advisable.