TDS3014 adventures

[james_s]

So I scored this dead TDS3014 for free, initial investigation was not promising, all the voltages looked good but there were no signs of life other than the backlight. Then upon further investigation I noticed that one of the two oscillator cans didn't seem to have any output. Could it really be that simple? I poked around and measured both and found identical conditions on the other pins but this one had no output. I thought perhaps something was shorting the output but I'd expect to see *something* on the output in that case but this was flatlined.

Ok so I start digging through my stash of scrap boards I found a similar looking 28MHz oscillator, nowhere near the 75.75MHz of the original but I thought it might at least invoke some small sign of life and tell me if I was on the right track. So I popped that in and imagine my surprise and elation when I hit the power switch and the scope burst to life with the splash screen immediately appearing on the screen. Woohoo!! Then it proceeded to boot up fully and display a trace, although not too surprisingly it feels very sluggish. Still, progress! Seeing formerly dead equipment spring back to life is the sort of thing that makes life worth living 

Sooo now I need a 75.75MHz oscillator, I have to assume that Tek had a good reason for using an odd value like that. I have not had much luck with the usual suppliers but maybe I've missed something. The original part is a Fox 401 series with a 5x7mm package, powered by 3.3V. I'm not opposed to adapting something in a different package if I need to. I'm open to buying a used part if anyone has one, maybe one of you has a scrap  TDS mainboard with bad ASICs or something?

I need some knobs too, oddly a bunch of them are crumbling to bits, those are readily available on ebay though. I'd also love to find the optional comm module but first things first.

[texaspyro]

There are a lot of programmable oscillator chips out there these days.   Most distributors of these devices offer programming services.

[DaJMasta]

If it'll run on a 28MHz oscillator, you may be able to get something in the same ballpark (70-80MHz) that will work fine.  It would be worth checking timekeeping functions and stuff, but if it's got an RTC, the sampling should be controlled by its own clock, so it may not actually be that critical to nail the frequency.  There's definitely the chance it's important, as you mention, but it's worth trying.

It's an odd number for a signal generator, but if you can generate the frequency (or maybe like a 25.25MHz squarewave and filter out the fundamental) you can probably test to see if it actually is the thing that needs the fix.

Another potential replacement option is a VCO or DDS generator chip capable of generating your frequency, but it's probably worth looking around for a crystal first, as it would be a bunch of extra stuff to do.


Is it possible that it's not actually a 75.75MHz oscillator and that's some other package marking?

[james_s]

I actually was just coming here to say I ordered a programmable oscillator from digikey that theoretically ought to be a drop-in replacement. It was only about 8 bucks so worth a shot at least.

It boots with the 28MHz oscillator but it doesn't really work properly. It will display the compensator waveform and the menu works for the most part but a lot of things feel weird. I'm going to wait until the new part comes before I bother to mess with it further.

In the meantime I've been doing a deep cleaning. The plastic on this thing is crazy brittle and I've had to glue a few clips and tabs. Almost every one of the knobs has cracked into pieces too, maybe the high temperature from being in Hawaii? I bought a set of new knobs that will fix that right up. The screen also has a bit of weirdness that turns out to be some delamination of the rear polarizer film. Since there is also one stuck pixel I may splurge and spend the 80 bucks or so on a replacement but that can also wait until it's fully working.

Is there a way to hack it into the 500MHz model without the comm module? Looks like people want a few hundred bucks for that option module and I have no real use for it beyond unlocking features.

[james_s]

So I was messing around at my bench today and the oscillator from the scope was sitting there still so I decided to solder some wires to it and play around. In the process I cleaned the rust off and noticed that it actually says 75.757 on it which I'm fairly sure is the frequency as there is another identical part in a different area that says 48.000 which is a frequency I've seen elsewhere.

Anyway I hooked it up to power and yep, still stone dead. I tried turning the voltage up as high as about 6V with no signs of life, then I turned the voltage down and was surprised to note that when I got down to around 2.2V it woke up, with an output of about 25MHz. I could turn the voltage up and the waveform would start to get ugly and then abruptly around 2.8V it would suddenly go dead, turn the voltage down and it would wake up again. I experimented, heating it up to the point where the wires desoldered from it and up until then the characteristics changed very little except it would run at a slightly higher voltage. I then used freeze spray to make it very cold and again surprisingly little difference.

So I'm really curious at this point what is going on with this thing. Is it likely that it is itself a programmable oscillator with a 25MHz source clocking an internal PLL that has failed? It's academic at this point since it's obviously defective but this is not a failure mode I've ever encountered or expected.

[james_s]

Oh hang on a sec, the frequency it's running at is almost precisely 1/3rd of the expected 75.757 MHz, is it possible that the crystal is oscillating in the wrong mode? I am not very knowledgeable about crystal oscillators but I do recall there are multiple frequencies they can oscillate at depending on how they are driven.

[edavid]

Oddly enough, it seems to be a standard frequency... this company sells them for $1.44 each:

https://www.knjn.com/ShopOscillators.html

Or maybe you could get a sample from Ecliptek:

http://www.ecliptek.com/stocksearch/stock.aspx?PartNumber=EHF1125TS-75.757M

I think it would be better to avoid using a programmable oscillator, since they tend to have high jitter, which is really not great for an oscilloscope sampling clock.

[james_s]

Oh hey that's great, that place hadn't come up in any of my earlier searches. Never heard of them before but I ordered one and will see what happens.

[james_s]

The new oscillator arrived today so I installed it and I'm pleased to report that the scope is now working perfectly! While I was waiting for that, I completely disassembled the scope and cleaned everything, bought a full set of replacement knobs for $60 since the originals were all cracked or broken, and I bought a NOS display for $65 because the original had the rear polarizer starting to delaminate, a stuck pixel and the backlight was getting tired. I reflashed one of the option modules with the code that unlocks all the features and now finally I put it all back together. All four channels work, all the buttons and knobs work, floppy drive works, everything great. I'm super excited to have this.

The only negative is that the plastic is all crazy brittle for some reason, I had to glue a number of clips and I'll need to be gentle with the rest of the thing, I have no idea why that's so bad or if there's anything that can rejuvenate brittle ABS. Also it doesn't have a comm module so I can't do the 500MHz bandwidth upgrade. Anyone got one of those for less than ebay prices, maybe even  one I can borrow briefly to do that? Serial or GPIB is fine.

[giosif]

Still holding a grudge for getting the scope before me  , but have to admit it's definitely in better hands than mine.
I would have never thought of checking something like the oscillator.
Nice job with the repair! 

[james_s]

Well if it makes you feel any better, I never win anything and am never the first to encounter and jump on a deal like this, so this is probably a one-off experience   

As far as the oscillator, once I'd checked all the voltages, it kind of seemed like the next logical place to look since there was no sign of any activity anywhere. Once I get sucked into a project like this I tend to obsess over it and power through it until I'm finished.

[giosif]

Ok, no grudges anymore. 

That is good, to get focused on a project until it's done.
Others... ok... Me, I tend to start a few projects in parallel and I have the feeling that leads to more of them ending up in limbo state.

[james_s]

Oh I absolutely have that problem as well, but then occasionally something like this gets me excited and I really focus on it and don't get stuck to the point of burning out.

I ended up taking this apart again last night after I noticed the RTC was stopping when it was powered off. Obviously the battery in the dreaded DS1742 was failing although it still was keeping the data in the SRAM. At one point in time I thought those Dallas chips with the integrated battery were a neat idea but after dealing with discontinued older ones I now in a handful of devices think they are absolutely stupid, I mean how hard would it have been to build a coin cell holder into the top? Lithium coin cells almost never leak but I digress. Anyway I desoldered the chip and installed a socket, then popped it in my TL866 and read it as an EEPROM saving the contents to a file.

Then on to the surgical procedure, first cut the plastic shell with a razor knife and carefully peel off a section to expose the epoxy. With the help of a little hot air I softened the epoxy and carefully dug out the old battery which I then measured out of curiosity and found it was only 0.6V, pretty remarkable that it held the NVRAM contents. In this scope there was sufficient space that I was able to graft a CR2032 holder right onto the top of the chip, it's not pretty but it does the job.

Pop it back in the TL866 and load the contents back on it, then install it back in the scope and power that up for a test. All looks good, reference waveforms still intact, date and time are waaaaay off so I set those and it appears to be working fine. Then I check the error log and wait a sec, ok that's not right, I'm reasonably confident that this scope has not been powered on continuously since the year 213 BC unless I've stumbled into something *really* strange. I suppose it would explain the brittle plastic though.

Ok so back to the drawing board. Pop the newly modified Dallas chip out and back into the TL866 it goes. A couple hours of messing around and dozens of back and forths between the scope and programmer and eventually I figured out that reading these Dallas NVRAMs in the TL866 is not reliable, it seems the first read is almost always corrupted to some degree but read it a couple more times and it gets a clean read. Unfortunately I had only that one dump that I did initially so now I'm *really* glad that the calibration values are not stored in this as on some older scopes, incidentally neither are the power-on cycles. I worked out experimentally that the power on hours are stored in a series of bytes starting at x7E0. I'm not absolutely positive of the format but I suspect the time is stored as seconds, before realizing this I experimentally found that x0210 resulted in a display of 2252 hours which is pretty close to what it had prior to this little adventure. At this point I decided to quit while I was ahead and call it good before I break a pin off the DS1742 or plug it in wrong and fry something while trying to get the hours exactly right.

Sooo back together, put it through its paces and now I'm fairly confident that it's 100% working now. I bit the bullet and bought a GPIB/RS232/VGA comm module for $240 which is more than I had invested in the whole project up until now but I figure I'm still ahead as this should allow me to turn it into what is effectively a TDS3054 which I can take advantage of when needed by borrowing probes from my TDS784C boat anchor.



Once the dust settles and I'm satisfied that this is going to stay working I think I might try to repay the generosity that got me this thing and offer the TDS320 I've been using as my "portable" scope until now to someone who needs a scope but I need to find a suitable box. It has displayed a calibration error ever since I tried to back up the stupid Dallas chip it uses and getting it calibrated would cost more than it's worth. I'll post that elsewhere when I decide what to do.

[TheSteve]

Will you trace out the serial portion of the plugin module, it does look like a DIY version of that should be possible.

[james_s]

Will you trace out the serial portion of the plugin module, it does look like a DIY version of that should be possible.

Yeah that's the plan, it's probably not worth replicating the whole thing but the serial part looks like nothing more than a level shifter and possibly a buffer.

[tautech]

Nice adventure, result and write up James ! 

[james_s]

In case anyone cares, I took a look at my notes today and the power-on hours are simply stored as the number of minutes in hex format in locations x7E0 through x7E4 in the DS1742W. I'm not sure why this wasn't immediately obvious to me last night, I guess I was tired.

[james_s]

The comm module arrived today and I was able to use the GPIB interface to easily set the model to TDS3054. After that it showed a DC offset so I ran the SPC routine which passed and took care of that, then I connected my pulser to verify the bandwidth and this confirms a substantial improvement relative to the reference waveform I saved prior to the upgrade. With this done I updated the firmware to 3.41 necessitating another SPC and now it's all good to go. I'm a bit surprised there are still so many lesser models that have not been upgraded to 500MHz given it's so trivially easy to do.

I do intend to reverse engineer the serial portion of the comm module, it might be a few weeks before I have time to dig into that though.

[james_s]

After working fine for a while my repaired DS1742W crapped out. It would still function fine with the scope on but every time I shut it off the memory and time get randomly corrupted. The battery is fine and still connected, after messing around with it for a while I suspect the power controller IC inside it has failed, perhaps it was damaged accidentally with 5V from the programmer.

After scouring datasheets I found the still current DS1744 which appears to be identical except for being a 32k part vs 2k. It is also available in a PowerCap package instead of the ridiculous potted battery. If I tie the top four address lines high this should get me something equivalent to the DS1742W. I whipped up a quick adapter board this afternoon, once I've built and tested one I'll post the files.

[ArcticGeek]

James,

I think your solution sounds pretty good.   There are a couple other options that I considered but haven't been motivated enough to try yet:

1.  A small interposer board that has male pins on one side (24 pins) and a female socket on the other for a 28 pin DS1744W part.  The DS1744W is still available, and tying the upper address pins would work fine.   The only problem with this approach is the DS1744W part is going to be sitting roughly ~3/8" higher above the board due to the height of the sockets/headers.  I don't know if this is a problem or not in a TDS3000 series scope, I haven't checked the Z height constraints.   The other issue is that in another 15 years or so you'll be stuck finding a replacement for DS1744W because its battery will be dead too, and that part might be obsolete by then.

2.  A small interposer board that has male pins on 1 side (24 pins) and a discrete version of a RTC clock/cal and NVRAM.  From what I can tell, a DS1558 would be a compatible RTC and NVRAM controller....and then you would add a 2K SRAM, a 32Khz crystal, and a coincell battery.   This would have the advantage that you could change the battery down the road should it ever expire and you would not have to worry about it becoming obsolete.  It also would not be as tall as #1 above.  The disadvantage is the design is more complex.

[james_s]

I briefly considered the first option, then decided I don't want to invest money in another stupid potted block, I'd just be dealing with it again 20 years from now when the parts are even more hopelessly obsolete niche items. Height would also be an issue as you suspect, the housing for the rear accessory port where the comm module, DSI interface or printer install only clears one side of the (socketed) DS1742 by about 1/8" so an offset adapter would be necessary. Given this constraint and my dislike for the potted battery, the surface mount PowerCap package seemed like the obvious choice, even if the replacement PowerCap becomes obsolete it's not potted so I can easily hack a new battery into it later.

I also like option 2 and considered that as well, I came across the DS1744p before locating a suitable RTC though. I may design something around the DS1558 just for giggles, I would not be the least bit surprised if that's what lies within the DS1742 and related parts. I know from x-raying similar Dallas bricks that they contain off the shelf ICs which are almost certainly Dallas' own offerings. The DS1386 for example has a large QFP that may well be a DS1588. The limiting factor in the imaging of these is in digitizing the film as I lack a proper film scanner.

[james_s]

DS1558 won't work, the register map is different, it has a lot more features and some things are moved around. Maybe there is another similar part that is the same as used in the DS174x parts.

[james_s]

In retrospect I should have xrayed the virgin DS1742W before I hacked out the dead battery. Anyone got a dead one that hasn't been hacked up? Doesn't *really* matter but it would be cool to add that to the collection of images of these things.

[2N3055]

How about M48T12 ?

[james_s]

It's 5V only so it wouldn't work in these scopes. If someone wants to send me one I'm happy to xray it though.