TDS 644B SPC Cal Failure

Questions in this forum area are community supported. Tektronix does not regularly monitor posts in this area.
Post Reply
Steveh
Posts: 4
Joined: August 30th, 2012, 11:00 am

TDS 644B SPC Cal Failure

Post by Steveh » August 30th, 2012, 12:45 pm

Hello,

Sorry for the long first-post.... I've searched through 40+ pages of this forum in search of info on this issue but so far, have found almost nothing regarding TDS 644B failures but a lot on the 420B, 5xxA/6xxA/7xxA models. I have a TDS 644B, SN B010474, bought last December from epay. Initially, it passed all self-tests with no problem. But after a month or so, I noticed it'd fail the SPC cal test initially but then pass it when run manually. Having other things to do at the time,the 644B was shelved for a few months and when recently pulled back out, failed SPC cal every time. However, it still seemed to work fine; but I was using it only for audio signals at the time. I did not think to check higher sweep speeds until recently, but then I noticed that at sweep speeds from 2 uSec/div to 20 nSec/div, in addition to the input signal, all 4 channels now display noise that fills the screen top to bottom. Above 20 nSec/div, the trace is noisy but much lower in amplitude. The input attenuators have no effect on the amplitude of the noise until sweep speeds above 10 nSec/div, and then only at sensitivies of 10 mV/div or lower. When sensitivity is decreased above 10 mV, the displayed noise is around 1/4 vertical division but increases at sensitivities higher than 10 mV/div. The scope still seems to work normally at sweep speeds of 20 uS/div and lower.

On the face of it, the symptom of noisy fast sweep speeds seems related to at least one post I saw on the TDS 540B for the infamous SMD electrolytic failures... except this is a 644B model which is not supposed to have those problematic electrolytics, not a 644A. I am thinking that a low battery in the DS1650Y corrupted some default signal path cal constants for the higher sweep speeds or sensitivities, and that the high noise level is due to incorrectly-increased gain which is corrupting actual SPC measurements during cal, causing the scope to fail SPC calibration. Could this be?

The processor board is PN G090-2043-00 with date code of 9633D. The DS1650Y NVSRAM also has a 1996 date code. The processor board holds about a dozen 10 uF/35V or 33 uF/10V electrolytics. There are no SMD aluminum electrolytics on the acq board, which is PN G9A-2027-00, and while I don't see a date code, all the chips are '95 or '96. I've checked across each cap on the processor board with a DMM and all have DC voltages within the voltage rating of the caps (unlike what I've read about the earlier scopes where it sounds like 10V electrolytics were installed on some 12VDC lines). I see no signs of electrolyte leakage or burnt traces around the caps. The caps on the processor board look exactly like the ones on the printer board which, I've read, apparently was still populated with the troublesome caps on the 644B.

There are NO entries in the error log despite consistently failing SPC cal.

Is it possible that a low battery in the DS1650Y could have corrupted some of its contents, and that replacement of that chip plus a Tek calibration would restore everything, or is it more likely to have some hardware failure? I've seen references to a Maxim replacement for the 1650Y but did not see a PN; what is it?

I've also seen posts where folks have proposed cutting off the top of the 1650Y and replacing the internal battery; but I haven't read of anyone actually doing that yet. Is this a possibility? Anybody know what that battery is? If I were to do that, would Tek still calibrate the scope?

Should the DS1486-120 timekeeping chip also be replaced (date code 9635); i.e., does it also contain a battery that must be getting low by now? Mine still keeps reasonable time but I think it was off by a few minutes from when I last set it some 6 months ago.

Thanks!

cap
Posts: 397
Joined: May 1st, 2009, 8:16 pm

Re: TDS 644B SPC Cal Failure

Post by cap » August 31st, 2012, 9:23 am

Hello,

Your scopes SMD electrolytics are better than TDS500A/TDS600A and earlier scopes, they rarely leak. If you measure the ESR’s you will find some that are high, usually 3 to 5 of them. Replace all of them, good or bad. Use 33uf/35V/105C caps on the acquisition and print option board. Think of this as maintenance.

Any info regarding a TDS684A/TDS6XXB or later has a very similar acquisition system to yours.

Make sure dip switch #3 is not toggled as that instructs the scope not to do a self-test at power on, which is slow on this scope. If your scope quickly fully powers on, it’s toggled to the skip testing position.

If your NVRAM battery is dead the scope will prompt you to run SPC after turned on, at least if it is turned off for a while. Probably get cal initiation errors or similar. Error log entries may be forgotten w/power cycling (depends on how dead the battery is). You can do the battery lobotomy, I have, but your first try will be crude. A DS1250Y will work fine, other brands too. You should be getting an error log entries if SPC fails...

I have wondered if operation does not impart a bit of charging action on a marginal battery, just enough to keep it going if the scope is used regularly.

Pursue the above before you worry about your calibration. I would do the above, run spc about 20 time to get some human readable result or confirm the problem has gone away. Then determine if there is something to repair or calibrate.

Keep us posted.

cap

Steveh
Posts: 4
Joined: August 30th, 2012, 11:00 am

Re: TDS 644B SPC Cal Failure

Post by Steveh » August 31st, 2012, 10:47 am

Hi, cap, thanks for the reply; I've read many of your posts with great interest.

The acq board doesn't have any visible aluminum electrolytics, just ceramic. But I'll have to replace the 4 on the printer board and the 12 or so on the processor board. I haven't looked at the front panel board yet but there probably are some there.

The only S1001 switch I've flipped is #8, while following the service manual troubleshooting diagnostics, which told me absolutely nothing; none of the tree diagrams go as far as this problem. Flipping #8 makes the LED display show some alphanumeric characters while stopping boot-up, which works as described in the service manual.

I haven't yet allowed SPC to loop but I'll set that up while doing other things. It seems strange that there is no error log entry after it fails SPC, doesn't it? I can understand it "forgetting" the error if I turn it off and the battery is nearly dead; but it seems like it should log an error when turned on and something fails... Perhaps the battery is too dead for that to happen. Is there also a battery in the 1486? I assume the 1486 must be for storing user settings and setups and such?

What about the fast sweep time and noise failure?? That's the most critical thing for me right now as we got this scope partially for the high speed but also so we can print plots, which I can't do with my old 7904 scope.

And what kind of battery is in the 1650Y?

cap
Posts: 397
Joined: May 1st, 2009, 8:16 pm

Re: TDS 644B SPC Cal Failure

Post by cap » August 31st, 2012, 12:35 pm

The front panel will have some caps to replace as well, I forgot to mention that, none on the display driver board or in the power supply (other caps will leak here so it is not a bad idea to take a peak and blow the box on w/canned air).

The #8 diagnostics is a waste of time.

The 1486 is the time keeper, and I think your correct w/storing setups too. As one would guess, if this battery goes your clock will be radically off at each power on.

Usually the error log will sustain itself so long as power is not cycled. So, cross your fingers and replace the caps and if you’re lucky that will make a difference. If not replace you’re your NVRAM. IF the scope is telling you that you need to run SPC at power on, this almost certainly means the battery is dead (DS1650y). Maybe the 3v lithium battery is acting as a short and preventing write operations, or in combination with a bad cap.

IF the NVRAM is non-functioning, you will need to get past that to do anything with the acquisition system. And quite possibly, the acquisition system will prove not to be the issue.

cap

Steveh
Posts: 4
Joined: August 30th, 2012, 11:00 am

Re: TDS 644B SPC Cal Failure

Post by Steveh » September 17th, 2012, 12:47 pm

Just replaced all 6 of the 10uF/35V and all 16 of the 33 uF/10V electrolytics. The vendor sent us 22uF/35V, I guess size 1206, at first in place of the 33 uF/10V. The plastic bag had the correct PN for the 33 uF/10V caps but they evidently binned the 22 uF guys in the 33 uF bin. Was easy to tell something was wrong since the 22 uF/35V guys were the next-up size.

Funny thing: Panasonic doesn't have an ESR spec for these caps (HA-series). Almost all the old caps measured the right capacitance at 120 Hz but around 7 ohms ESR. The new caps also measure about the right capacitance but around 4 - 5 ohms. Yes, OHMS, not milliohms. At 10 kHz, both new and old caps are around half the rated capacitance and maybe an ohm more ESR. So what's with the talk I've seen mentioning milliohms of ESR?? :?

Also, I found TWO of the original 33 uF/10V caps had NOT been soldered on one side to the board :shock: Unfortunately, I didn't notice this until I had them all replaced, so I'm not sure which board they came from; but I think they were on the processor board, as I remember several caps almost popped theirselves off while I was desoldering them.

Anyway.... I also ground into the 1650Y chip and have access to the battery top and most of the side, and it measured just 0.08 VDC. But I can't figure out how to connect to the battery's body without trying to grind under the battery to pop it out, and possibly causing devastating damage to the chip by grinding into something else. This potting compound is the pits :x How did you connect to the old battery's body, Cap??

Thanks!

cap
Posts: 397
Joined: May 1st, 2009, 8:16 pm

Re: TDS 644B SPC Cal Failure

Post by cap » September 17th, 2012, 1:53 pm

You can safely use 33uF/35V parts anywhere to replace the SMD electrolytics. One the 600B’s it is fairly common for the factory to intentionally not install some of these caps. Not a big deal, if you are certain the pad and stenciling is the same you can opt to add caps if you choose w/o harm.

On that scope, I simply left the old battery on the chip in place and soldered to one solder tab, and scratched up and soldered to the battery body. Very crude, worked for the short term, I used a larger battery to compensate for the internal battery losses of the old battery. You can do this and make the battery easily replaceable. Buy a new NVRAM battery once you have convinced yourself that you have repaired the scope.

Or, you can try to dig the battery’s innards out w/a dremel tool or??? Tough to do if it is still mounted to the motherboard. Maybe tape everything off like they do in surgeries. Use the bottom half of the battery as a shield to prevent over penetration. It might be possible to peal the battery out once you’ve ground a hole in it to grab onto, may end badly. Presumably, you have much more time than monie… A PITA.

You can use a DS1250Y chip or another brand that crosses over.

cap

Steveh
Posts: 4
Joined: August 30th, 2012, 11:00 am

Re: TDS 644B SPC Cal Failure

Post by Steveh » September 17th, 2012, 3:30 pm

I was able to make a brass clip that fit over the old battery and gripped the battery body, then soldered a wire to the old strap which, surprisingly, is evidently steel. Stripped the ends of the 2 wires and taped them to a new 3V battery (CR2032). I was unable to solder to the body of the old battery; I'm pretty sure the battery is stainless.

Anyway, it didn't work. CRT display flashes rapidly, and the LED displays 8 for a moment, then quickly 1, 3, 7, 8 then goes into a race condition, flashing each segment too rapidly to follow. Display continues to flicker and all front panel LEDs are continously illuminated. Instrument won't boot. I've got +12.06 and +5.08 on the processor board. Tried it again with the battery disconned but same thing; I suspect the 1650 has gone completely dumb :cry:

Guess I'll try to replace the 1650 with a 1250Y but where can I get it flashed?? I don't have access to a programmer and have no code with which to flash it, anyway. If I send it back to Tek, I suppose they can flash it but for how much more, and are they able to do it with the 1250Y installed? Either way, I'll install a socket on the board (unless someone tells me Tek won't touch it if it's modified like that).

andy
Posts: 73
Joined: April 17th, 2009, 9:48 am

Re: TDS 644B SPC Cal Failure

Post by andy » September 18th, 2012, 9:32 am

I believe the only important thing in the NVRAM of the 644B is the option list. It should work with a blank NVRAM, but you will lose any software options (which would have already happened due to the dead battery anyway). I'm not sure what it would do if the NVRAM was completely nonfunctional, but it might be worth replacing it if you're not sure.

cap
Posts: 397
Joined: May 1st, 2009, 8:16 pm

Re: TDS 644B SPC Cal Failure

Post by cap » September 18th, 2012, 1:06 pm

Let’s see. The scope will not start if the chip is missing, not fully re-soldered in, or if the D1 bus is installed upside down upon reassembly (all things I’ve, I even soldered on in backwards once on a 684C, all was well when I turned it around, whew…). NVRAM does not do well with negative voltage incursions.

The 600B memory architecture is different from earlier scopes, I’m not certain you will lose options. Do not turn the scope on with the memory protect switch in the unprotected position to protect them (also a possible reason the scope will not power on). Did you have your FFT option with the dead battery?

Most chips on eBay are too old and/or way overpriced. Find a crossover chart on the web for the DS1250y, and consider those parts as well. A used part might have a corrupt CRC and that might be a problem. New chip goes in, scope tells you to run spc, you do and all is well.

Socket may place the chip a bit high if you have option board 05. Get the chip 1st before choosing a socket, some nvram such as that on the tds3000’s have funky shaped pins that do not like machined sockets after a few insertions.

The PC board quality is high. Solder wick works but is slow and fairly crude. Low temperature de-solder alloy is also slow (be patient) and messy (tape the area off w/capstan tape) but virtually guarantees no board damage. A high quality de-soldering gun like the Hakko 808 will make short work of the job and is the best approach; the all in one economy solder/de-solder stations on eBay are all junk for this task. No matter what method you choose a couple of pins will be stubborn.

cap

robertd
Posts: 3
Joined: November 27th, 2017, 12:16 pm
Country: United States

Re: TDS 644B SPC Cal Failure

Post by robertd » November 27th, 2017, 2:11 pm

Can I resurrect this after 5 years?

I have a 644B with a very similar problem. All channels seem to work properly on all vertical and horizontal ranges except channel 2. Channel 2 seems to work properly on all vertical ranges and on slower horizontal rates, but for sweep rates 2uS and faster, channel 2 is messed up as described below:

1. The DC offset is wrong and different between most sweep rates. Selecting a different horizontal rate usually results in a different DC offset.
2. For most of the sweep rates, the signal appears inverted as compared to the other three channels.
3. The signal amplitude is wrong.
4. For two or three of the sweep rates, there is "noise" that resembles rail-to-rail clipping which follows the envelope of the original signal. (see attached -)
DSC_4344.JPG
DSC_4344.JPG (744.15 KiB) Viewed 3098 times
The scope had this problem when I got it from someone else, and I do not know if the previous owner had ever done anything to it, but when I got the scope, interestingly the very first entry in the error log was "FATAL: 250 nv storage ram copy corrupt...acq cal library errors" and after that, the error log is filled with "Calibration errors." (see attached -)
DSC_4346.JPG
DSC_4346.JPG (886.24 KiB) Viewed 3098 times
So, my guess is that something happened to the nvram that stores calibration data and this is throwing off some processing for channel 2.

I removed the two Dallas nvram modules, saved .bin images, and installed sockets on the board. The timekeeper is a DS1486, and the other module is a DS1250. (I think I saw some versions of the 644B use a DS1650 instead?) If I fill the DS1250 with 00 (zeros) and install it, I get some sort of calibration library error message that appears at boot up telling me to run SPC (which always fails). More importantly, I noticed that the other 3 channels now misbehave in a similar way. So it seems likely to me that the error occurred in the DS1250, since changing its data causes other channels to do similar things.

I've been searching and searching for someone with a backup image of their 644B nvram, but have not found any. I've a found 644A copy, but that did not work (as I surmised that the A's are too different of architecture from the B's). Installing the 644A image causes a similar response as writing all zeros.

A related question: what is the difference between the "calibration constants" found in the DS1250 and the "calibration constants" contained in the two serial eeproms on the ACQ board (U1052 and U1055, I think)? Is it possible those are messed up instead/in addition to the DS1250?

For now, I am asking anyone for a .bin image of a known, working 644B nvram DS1250 or DS1650.

Thank you for any help you can offer.

hexley ball
Posts: 16
Joined: October 26th, 2017, 9:44 am
Country: United States

Re: TDS 644B SPC Cal Failure

Post by hexley ball » November 27th, 2017, 5:07 pm

I also have a TDS644B, so some of the things you are seeing are a bit familiar.

First thing to know is that the scope changes modes in a fundamental way when moving to sweep rates of 2 usec/div and faster. At slower sweeps, the ADCs can digitize the signal directly -- the ADCs are rated for something like 20 Msamples/sec, and the fastest sampling needed by the scope at sweeps slower than 2usec/div is 10 Msamples/sec (corresponding to a sweep rate of 5 usec/div).

At 2 usec/div and faster, the scope involves the "sampler" chips, which apparently are analog storage devices that use CCD delay lines or equivalent technology. The TDS644B has four of these samplers, one for each channel. The chips are marked 156-6782-00. This is described as “IC, ASIC; CMOS, Custom; Sampler IC, 16K Memory” in the Tek literature.

The function of the samplers is to grab the incoming analog data at the real-time sampling rate (from 20 Msamp/sec to 2.6 Gsamp/sec, depending on the horizontal sweep rate), store those samples, then send that analog data out to the ADC at a more manageable speed -- which seems to be fixed at about 8 MHz. With this architecture, the scope can use inexpensive ADCs that only have to handle 8-10 MHz sampling rates.

These analog samplers are not ideal -- they have significant offset errors, and those errors vary from cell to cell. So one of the functions of the SPC seems to be to measure these offset values and store them to apply later on the fly to incoming data.

To get a feel for these offsets, here is a scope shot of the output of one of the samplers in my scope, spitting out values with the channel input set to "GND". Ideally, we would expect the sampler output to be constant, but in actuality it bobbles quite a bit.

TDS644B sampler offsets.jpg
TDS644B sampler offsets.jpg (55.28 KiB) Viewed 3071 times

If these offsets can be measured, stored, and applied to actual measurements, then the scope will display a clean trace. But if the offset correction values are not applied, then the trace will not be correct. In the case of my scope, I got about one division of fuzz on each trace. But after running SPC, the fuzz vanished. (One of the problems in my scope is that the SPC calibration values vanish at power down -- so I am also chasing NVRAM issues.)

Unfortunately, it rather looks as though the sampler chip for Ch 2 of your scope might have a problem that goes beyond the normal offset issue. You won't be aware of the sampler issue at sweep speeds slower than 2 usec/div, for the reasons noted above. But once the sampler is put into the signal path at 2 usec/div and faster, then the artifacts are going to appear. And since SPC does not remove the artifacts, it is possible that the chip is malfunctioning internally.

-hb

robertd
Posts: 3
Joined: November 27th, 2017, 12:16 pm
Country: United States

Re: TDS 644B SPC Cal Failure

Post by robertd » November 28th, 2017, 5:44 am

Wow, this forum is AMAZING for fixing scopes! I've had this problem for almost 6 months now, and wouldn't you know, 4 hours after I post for help, that's when I find my solution...and that's with zero replies! Simply posting on this forum must initiate a solution. Tektronix is doing something right!

I don't have GPIB setup or a console port interface, so I'm limited to poking at the hardware, which in my case centered around a likely nvram corruption error in the log. I removed the DS1486 and DS1250 nvrams, read them into backup .bin files, and installed sockets for testing. Here's what I learned about my TDS644B:

1. By erasing the timekeeper DS1486, I determined that it contains the option list (in the 810h to 82Fh range), the error log (starting somewhere around 92B0h), reference waveforms, and probably user setup data/probe data (I'm not sure on this one because I didn't have any previously saved setups to recall).

2. Installing an erased DS1486 caused SPC error at bootup and all 4 channels malfunctioned just like channel 2 had been (messed up above 5us horizontal time base). This also caused data in the DS1250 to change after bootup even without running SPC.

3. Restoring the DS1486 to original and installing an erased DS1250 caused SPC error at bootup and all 4 channels malfunctioned just like channel 2 had been. This also caused data in the DS1486 to change after bootup even without running SPC. So, DS1486 data interacts with DS1250 data in both directions--probably something the internal librarian does because of crc's I'm guessing.

4. I also tried different combinations of the 2 nvram ICs erased one at a time, then with both erased. In all cases, they both returned to the programmer with at least some data written.

5. With both nvrams completely erased at bootup, after running SPC, both nvrams contained as much data as before they were erased (not the same data, but roughly the same "amount" and in the same locations). I concluded that neither of the 2 nvrams contain calibration data necessary for running SPC because A.) running SPC writes all the data they had before being erased, and B.) running SPC always returned scope performance to exactly what it was before erasing the nvrams.

So, because I could duplicate the problem of channel 2 onto the other 3 channels by erasing/corrupting SPC data, it seemed that I was indeed correct in assuming that the error log was indicating that a corruption in the nvram had caused SPC failure, but since the nvram data was generated by SPC and not a prerequisite to run SPC, that meant that the nvram itself seemed physically faulty.

Conclusion: replacing the original DS1250 solved my problem. SPC ran successfully. All 4 channels function (apparently) normal. So the ACQ hardware was functioning, and SPC was attempting to write the correct data to the nvram, but the nvram was faulty and could not retain all of its data.

Its interesting that the programmer never found a read error during the verification cycle with the DS1250 all the times I wrote to it. I reprogrammed backup image data, I filled it with 00, and with FF, and with "random" data, and it passed verification read every time. I guess the combination of data and address had to be just right to show up as corrupted.

hexley ball
Posts: 16
Joined: October 26th, 2017, 9:44 am
Country: United States

Re: TDS 644B SPC Cal Failure

Post by hexley ball » November 28th, 2017, 4:38 pm

Great work, robertd. It is nice to have definitive information about what happens when one or both of the NVRAM chips gets erased.

I wonder if the reason that you did not see any verification errors during your programming exercises is because the programmer was accessing the NVRAM chip at much less than full speed. The programmer probably operates at "parallel port" speeds; perhaps 100x (or more) slower than what the RAM would see in the actual processing system of the scope.

-hb

robertd
Posts: 3
Joined: November 27th, 2017, 12:16 pm
Country: United States

Re: TDS 644B SPC Cal Failure

Post by robertd » November 28th, 2017, 5:37 pm

My programmer is a Dataman 40pro with USB interface. I must have programmed the DS1250 module at least 25 times with different random data and had no validation failures. Write speed sounds logical.

anros1
Posts: 1
Joined: December 12th, 2017, 4:15 am
Country: United States

Re: TDS 644B SPC Cal Failure

Post by anros1 » December 17th, 2017, 6:13 am

Hi all,

I figured it's best to keep TDS series issues in one thread. I have a very clean TDS620A that I bought from a friend a few years back and has worked beautifully until tonight.

I was probing around on an opamp stage when I accidentally touched the probe off of the palm of my hand, whether coincidence or not the scope flashed onscreen and shutdown. It attempted to reboot but failed every time.

I learned about the infamous electrolytic cap issues after I bought this scope so I opened it expecting the worst. Even under the magnifier I cannot see a single leaky cap or corroded trace on the board, everything is super clean. I spot checked the ESR on maybe 10 caps with my Peak Atlas and every cap was sub 1 ohm.

Is there something obvious that I'm missing? Any ideas appreciated.

Cheers,

Post Reply

Return to “Other or Discontinued Oscilloscopes”

Who is online

Users browsing this forum: fenugrec and 4 guests