Tek 495P SA Repair

[NHnewbie]

Tektronix 495P Spectrum Analyzer Repair

I recently acquired two Tek 495P spectrum analyzers. One is in mint condition and performs in accordance with the initial checkout procedure in the user manual. The second unit came along with the first as a freebie “parts” unit. In booting up this unit I find that it is working to a substantial extent but has a couple of issues. Hoping the pros here can provide some insight into next steps for repair.

Both 495s are late serial number units (sn B030xxx in both cases). When the parts unit boots up, it posts a message on the screen identifying the firmware version as “492AP v9.7 panel v1.1,” not 495. In reviewing the service manual and other info from various online resources, I see that the firmware in the 49x series is somewhat generic, with DIP switch settings on the A54 memory board determining which model version of the firmware actually runs following boot up. I have verified that these DIP switch settings are as specified in the service manual. I have also verified that the DIP switches themselves are working properly, and that the correct EPROMs (v9.7) are installed on the board. (There are also DIP switches on the Z-axis board, which are verified to be in the correct positions.)

The seller reports that the memory board may have been damaged when removing the memory backup battery for replacement, or that the board may have been damaged by a leaking battery. I see no evidence of damage, however, the battery leads show some evidence of solder work and it appears that soldering on the plated through hole on one of the battery leads is questionable. In looking at the circuit diagram of the memory board, it is not apparent why a poor battery connection would produce the symptoms I am seeing. Unless someone here has a better idea, my next move is to try to purchase another memory board.

The other issue is an error message stating “Tuning failure – 1st LO.”  The 492AP and 495 share a common controller architecture but differ significantly in the analog/RF circuitry, with the 492AP having a freq limit of 21 GHz, while the 495 is 1.8 GHz. It seems somewhat likely that the 492AP firmware in attempting to tune a 495 1st LO is having problems doing so. I would not be too surprised to find that this error goes away once the unit is running the correct firmware.

I have a long, strong electronics background, but am a newbie when it comes to SAs. Any insights much appreciated.

Rick
K8EZB

[edavid]

Have you read John Miles's service notes?

http://www.ke5fx.com/49x_notes.pdf

[drussell]

Wow, that is a very interesting set of notes...

I found this particularly interesting in light of the OP's problem:

Use caution when toggling DIP switches found in these analyzers, especially the GPIB switch on the back panel.  These switches are very likely to fail when actuated for the first time in 20+ years, leading to difficult-to-diagnose problems

I see that he has stated:

I have verified that these DIP switch settings are as specified in the service manual. I have also verified that the DIP switches themselves are working properly,

... but to the OP, are you absolutely certain that the DIP switches themselves are operating correctly in light of the fact that other people have apparently had problems with the switches themselves?

Does it actually change the reported model identifier if you change switch settings?  I would think that a broken or contaminated trace from the ID switch bank or something could also certainly cause this kind of "incorrect ID" behavior...

[NHnewbie]

I have read John Miles notes carefully and have used Deoxit on each switch and verified the operation of each switch. (Also much useful info here on Tek and many other manufacturers: http://www.ko4bb.com/getsimple/index.php) Absolutely certain? No,  but reasonably certain. I will do some additional testing tomorrow.

 

I would think that a broken or contaminated trace from the ID switch bank or something could also certainly cause this kind of "incorrect ID" behavior...

Agree. I have inspected the board with a magnifying glass but, so far, not seeing a problem. I am suspicious of the solder joint on the battery lead as this may make connections between layers which are not visible. Worst case, I can take the memory board out of my working 495 and try it in the non working one. Have been resisting this due to not wanting to risk messing up one that works. There is an awful lot going on under the covers in this SA! Service manual (2 vols) runs several hundred pages.

[Kosmic]

Adjusting the dip switch is probably just to make sure a certain module can work in different spectrum analyser model. I would be surprise if the eproms contain the firmware of all the 49X spectrum analysers.

Since it look like the firmware is from another model, you should first try to find the binaries of a compatible firmware and then figure out how to get that on the eproms.

[Kosmic]

Ha and if you find the right firmware, you can diff the binairies to confirm that they are really different (or the same  ).

Bin of the 492 can be found here: http://w140.com/tekwiki/wiki/ROM_images

[Kosmic]

I just realized since you have 2 x 495, you already have the firmware. 

So then you need to get the firmware from the working usnit and copy it on the non working unit.

Should be simple enough 

[drussell]

I would be surprise if the eproms contain the firmware of all the 49X spectrum analysers.

Since it look like the firmware is from another model, you should first try to find the binaries of a compatible firmware and then figure out how to get that on the eproms.

That's not how it works on these units...  On these, the firmware is the same across series but the physical hardware is different among models.

Each set of 49x-series firmware images is shared across several models in the product line, with the DIP switches determining the instrument’s actual ‘personality.’  For example, the v9.0 firmware stored at the “Tek-495p_9_0” link is applicable to the 494A/P, 492A/P, 495/P, 2753/P, 2754/P, 2755/P, and 2756/P models, while the v9.7 firmware stored under “Tek-494AP_497P_9_7” added support for the newer 492B/P, 2755A/P, and 497/P models.
...
There is also an earlier version, “Tek-492ap_6_0,” which is suitable for use with the original 495/P and 492A/P ROM boards that have only eight EPROM sockets rather than the ten that are present on the later instruments.  Search the various notes and spreadsheets in the ROM Images section carefully for the latest information applicable to your model.
...
Users of the original 492 should look for the "Tek-492_1_6" link, which contains the original 1.6 firmware for the 492 plus a consolidated file suitable for writing to a single EPROM, with notes by Luis Cupido.  I’m unsure if this file can be used to support a 496, 496P, or 492P.

[Kosmic]

hmm interesting. I'm working on a 2794 right now and was looking for a updated firmware. I need to check in the manual if it's working the same way for that series.

Thanks,

- Kosmic

[metrologist]

subscribed for shotgun targets. My 496 come up with 1st LO lock failure, I think it was...

[NHnewbie]

I think drussell has it right, at least according to the info I have dug up so far. I did more checking and am confident the DIP switches are working as they should. All show ground when closed and 11K ohms when open, which corresponds to what the circuit diagram suggests. The memory board (A54) is from a 495P and has v9.7 firmware EPROMS, so that seems right. I am now somewhat convinced there is a problem somewhere on the memory board. I have found a board for sale (in Italy) from a 494AP that the seller says has v9.0 firmware installed. This version is said to be compatible with the 495P. I'll know more once I receive the board. It was cheap enoug that if it works it will be very well worth the price. Depending on what I find out, I may move the v9.7 EPROMs from my current memory board to the new one. One possible gotcha with this board substituting approach is that the ROMS on the new board are unique to the 494AP, however, all of the documentation I have found so far indicates that the differences among the 49x series are dealt with in the (socket mounted) EPROMs.

[Kosmic]

hmm interesting. I'm working on a 2794 right now and was looking for a updated firmware. I need to check in the manual if it's working the same way for that series.

Thanks,

- Kosmic

So for the records, the 2794 share a lot of modules with the 49X family but the memory module is not one of them. On the 2794 the dipswitches on the memory module are used to activate selftests and signal that some options are installed.

[Kosmic]

hmm interesting. I'm working on a 2794 right now and was looking for a updated firmware. I need to check in the manual if it's working the same way for that series.

Thanks,

- Kosmic

So for the records, the 2794 share a lot of modules with the 49X family but the memory module is not one of them. On the 2794 the dipswitches on the memory module are used to activate selftests and signal that some options are installed.

Hmm I might be wrong. I just reviewed the 492AP service manual and it's pretty similar to the 2794. I'll have to experiment with the dipswitch to see if one of the switch "unused" in the manual actually do something.

[Kosmic]

hmm interesting. I'm working on a 2794 right now and was looking for a updated firmware. I need to check in the manual if it's working the same way for that series.

Thanks,

- Kosmic

So for the records, the 2794 share a lot of modules with the 49X family but the memory module is not one of them. On the 2794 the dipswitches on the memory module are used to activate selftests and signal that some options are installed.

Hmm I might be wrong. I just reviewed the 492AP service manual and it's pretty similar to the 2794. I'll have to experiment with the dipswitch to see if one of the switch "unused" in the manual actually do something.

OK, well I was able to switch my 2794 to a 2792 by flipping one of the dip switch. So it really work the same way as the 49x family.

[andy2000]

Since you have one that works, you can always do a little board swapping to narrow it down. 

[NHnewbie]

Success! Finally found a solution to the problem of the 495P booting up as a 492AP. As previously suggested, the problem seemed to be on the A54 memory board. Replacing several leaking electrolytics didn't make any difference, so I turned to a replacement board sourced from Italy via eBay. This board was priced right but had v9.0 firmware EPROMs which were not compatible with the v9.7 firmware in other EPROMs on other boards in the SA. These v9.0 EPROMs were soldered in place so it was necessary to remove them and install sockets so I could move the v9.7 EPROMs from the defective board to the "Italian" board. Once this was accomplished and the board installed, the SA booted up as a 495P (correct), displayed the correct initial settings on the screen, and the previous "Tuning Error - 1st LO" error message was no longer displayed. So, that problem is solved.

I may attempt to trouble shoot the bad memory board. I obtained a set of extender boards that will allow access to the board while the SA is running. It appears that the DIP switches which the microcomputer reads to determine which model of the 49x family it is supporting are not being read at boot up. These DIP switches are enabled to the main data bus through a tri-state buffer during the boot up process. It seems that there is possibly a problem with this IC or the signal that enables it to put the switch settings on the bus. We'll see.

Now that this problem is solved, I am moving on to some additional refinements. It turns out that the 1st mixer in this instrument was replaced at some point with an aftermarket Mini-Circuits mixer spec'd for 10MHz to 3GHz, likely affecting the lower freq performance. I am now looking for a replacement Tek mixer module to replace the Mini-Circuits module. Perhaps mission impossible, but then finding a replacement memory board seemed that way at first.