Wavetek 2520a 0.2-2,200MHz RF sig gen repair

[hAudi]

Dazz1, Great job and perseverance on repairing the SG.  You had quite a slog wading through all of the problems.  The good news is that you are totally familiar with the unit now so future service should be easier.
I got one likely close in age to yours as I believe my SG is running Ver 5.2 code (that displays at the end of the startup self test).  I found a bad 5v regulator in the back of the main board that was wandering between 4-6v.  That was a simple fix.

I did the same backlight strip replacement as you did (before I found your work) so my internet search brought me to the same choice.

The unit originally let me get to the calibrate menu by shorting across the keyswitch with a jumper before startup (you can short 2 pins on the CPU board also instead) but after pulling the NVRAM module and saving the code from it I loaded that code on one of the new modern substitute boards with the coin cell on top.  The unit since will not go into cal menu.  It runs through the normal self check then the level window displays fast changing display but it never gets to the cal menu.  All functions of the generator seem to work with fast lock changing frequencies and normal am/fm internal/external modulation with level control.  I can use it as is however I don't like the idea of not being able to get to the cal menu.  Putting the original Dallas NVRAM module back in does not correct the problem.  I even tried blanking the new NVRAM module to erase state however the SG won't even go into the self check at startup.  It would seem the NVRAM needs some code other than retained from normal front panel operations.  Could you add any more information about the NVRAM operation and/or post the contents of your Dallas module?  I wonder if my module was on the way out with corrupted bits?

BTW, the code you posted of the EPROMS verified identical to what I got from my machine.  Also, my boards are dated 12-88.

[dazz1]

Hi
0K  I was dealing with the after effects of a TBI at the time and I found fault finding was therapeutic.    I couldn't assemble a plastic model F1 race car, but I could fault find 80's vintage electronics. 

I built an NVRAM programmer to read/save/write the contents to the new NVRAM  module.   I think I wrote this up.

I don't think it helped because my NVRAM was so old, I wouldn't rely on the memory still being intact.

I'd suggest you do some hardware debugging around the cal mode switch signal.  If you want me to run a duplicate test to spot diffs, let me know.  If you don't have the manuals get them. 

I don't think the NVRAM holds code.  I suspect a blank NVRAM is filled with default values, then over-written in the cal process.  That would be a period correct approach because it would make manufacture easier.
I did suspect the NVRAM held code because the phase modulation feature on my SG didn't work, then it did.

I did a cal on the SG amplitude output and got it to within +/- 0.1dB.  By modern standards the harmonics are not great but good enough for most work.

Do you have the manual?  It shows how to run a bunch of built in test procedures that look like they were designed to be used on the production line for fault finding.

Did you use the same blue colour backlight?  It looks very cool.

[hAudi]

I got the manuals from Artek before starting work on the unit.  I have one of the Chicom Xgecu T48 units that seems to be able to work with most chips.  It even can work with the 25v programming voltage chips that some programmers can't supply enough voltage to program.  It read the code in the Dallas module and let me reprogram it as well as my new replacement one with the lithium coin cell on top.

Getting to the other tests if I remember reading correctly requires getting to the cal menu...which I can't get to currently.  Correct me if I'm wrong on that.  I do plan on studying the circuitry in from the switch.  I did notice it is 5v high in normal operation and goes low obviously when the switch is turned to cal.

I'm not sure what you mean by default values on the NVRAM other than I would suspect default fallback generic calibration data.  That would mean one would have a brick if one had a SG with a totally failed NVRAM module.

I agree the spectral purity is not as great as some of the other SGs out there however it is easy to work on...has self cal...and I like having a spin knob interface.  It works for me.

Here is how my backlight turned out.  I used the same inverter as it seemed to still work ok.  I'm not too impressed with the brightness however it works ok for me on the bench.

And Bob's you uncle.

[hAudi]

You-da-man as we say up here.  Problem solved.  I must have accidentally put one of the 2 required switches open (1 and 5) on SW301 when I was exercising switch contacts that had not been contact wiped in years  .
I ran the 19 routine freq cal with no problems.  I can put the cover back on it.

Are there any other routines you would suggest from the menu?  The only level measurement item I have is a Tiny SA Ultra which is only relatively accurate over the range I normally work.  I would have to get access to a much better power meter before I would consider trying the amplitude cal routine.  The output looks pretty close on the Tiny SA Ultra as is currently.

Thanks again for the pointer.

[dazz1]

Hi
Glad to hear it is working.  There are a LOT of parts that can go wrong on these things, but the advantage is that they are replaceable.

Failure of the NVRAM does not make the SG unrepairable/unrecoverable. A blank NVRAM and the SG will write default cal values to it by running Test 20. 


The frequency audo-cal procedure figures out the transition frequencies between each oscillator, so no test equipment required.  The level output cal requires a good RF power meter.

I was lucky I was able to get my SG calibrated at a calibration centre for $zero .  I don't know how long the calibration will last before it drifts.
I would suggest you try the electronics department of a University if one is nearby. 
If you have a spectrum analyser, you could use that to do a confidence check.

Mine past the power-on self test before I found any of the faults.

Test 18 is important because it tests the control loops.  If this test fails, the SG is seriously ill.

The internal inverter to drive the display back light is quite weak.  I think this was probably a design feature to extend the life of the back light. This is consistent with the backlight on/off switch on the rear panel.  I would describe the back light as adequate.  Not great, and not bad.  It works as designed, and that is good enough for me.  Most importantly, it looks cool       

[hAudi]

Thanks for the info on the NVRAM.  It is helpful to know about Test 20.  This blog series should be helpful for anyone else wanting to get their 2520A back in operation.

I had another failure yesterday that threw me for a loop.  The gen would not output below 137.5MHz which uses the 500MHz VCO which is in the aluminium enclosure to the right as viewed with the top lid removed.  Upon opening the module, the VCO is in the small section of the enclosure.  Probing the VCO stripline with my sniffer probe there was no 500MHz output with the SG set to 100MHz.  The screen also read unleveled and unlocked.  After studying the schematic I figured out the feed through pins that lead into the smaller 500MHz VCO section.  There is a 595 serial to parallel shift register on the main top side board that controls the band switching.  That voltage toggled from 0 to 2v as the frequency went back and forth through the 137.5MHz point so the micro was flipping the 595 as expected.  I was perplexed why the voltage high was not 5v as expected so I flipped the machine over and probed the 595 chip VCC pin only to find it measured 2v also.  It turned out the problem was the 5v regulator that I replaced earlier in the week (pulled from my old junk box stash) failed early.  I replaced that reg again and all was well again with the SG.  I usually use parts that I got from Digikey or Mouser however hamfest cheap finds also populate my junk box.

I'm going to let the machine run for a few days to be sure it is finally stable.  It is a PITA to remove all of the screws to pull the bottom covers.
The self test and self calibration was a big selling point for the generator.  That and the spin knob were important to me. Test 18 ran quickly and passed.  I plan on feeding all of my equipment with a Trimble cell site pulled GPS standard 10MHz reference when I get around to adapting and old Grass Valley video DA for multiple outputs.

[dazz1]

Hi
Glad it is all working.
I got good at removing the screws, pcbs etc.   These SGs have that hand made look and feel about them. 
For anyone else fault finding this make/model SG, get the manuals.  I would rate them as essential items.

There are two things I like about 80's/90's vintage test gear.  I can fix them and they are old enough to retain their value.  If I buy a new item of test equipment, it loses at least 50% value before I switch it on.  When it is superseded by the next model, the value depreciates even further.    The depreciation is offset by the capability provided by modern test gear, and the benefit to cost ratio of modern gear is astonishing.  So not knocking modern test gear at all.

I have considered GPSDO but it is not required for my use cases. The inbuilt oven oscillator is good enough for me.
These days I only buy test gear if it adds capability that I know I will use.
Buying an SG added significant capability that I didn't have.  A GPSDO does not add capability that I would use.  It is a personal thing and there is no right answer for everyone.