TDS784A acquisition board partial failure

[Tantratron]

Hi, I'm trying to restore a TDS784A which can boot without any fault (see attached banner picture) but does fail SPC (see attached picture of error log).

The symptoms of Ch2 and Ch3 are confirmed by simple tests:

• trying to change vertical scale on each channel (1 to 4) will always make attentuator relay click audible (1V to 2V) and 100 mV to 200 mV
• only for Ch2 and Ch3 there will NO trace display if vertical is less or equal than 100 mV
• only for Ch2 and Ch3 will display 1 KHz square signal if vertical scale is higher or equal to 200 mV but trigger will NEVER work
• wether Ch1 or Ch4 will always display traces, good trigger and using Leo Bodnar pulser confirms 1 GHz bandwith

Besides the pciture of error log file, here is GPIB extract from the same log

Code: [Select]

ERROR: Internal adjustment range exceeded,
(1 @ 200c): .9997 => Balance dac #2 (ID #145)
ERROR: Internal\sadjustment\srange\sexceeded,
(2 @ 200c): .9997 => Balance dac #3 (ID\s#146)
ERROR: Characterized values out of bounds, Ch2 trigger gain
ERROR: Characterized values out of bounds, Ch3 trigger gain
ERROR: Characterized values out of bounds, Ch2 trigger gain
ERROR: Characterized values out of bounds, Ch3 trigger gain
ERROR: 2260 Calibration failed, Triggers
Since half of the input channels work fine, do you think it is pin soldering issue on some chips, which chip (see attached 2 pictures of the board) do you recommend to check first ?

Thank you

[TERRA Operative]

How are the relays on the input hybrids? They can and will cause SPC to fail if worn out. Might be worth replacing the full set of 16.

I would swap the hybrid modules between channels 1 and 2, and between 3 and 4. Then reset errors and do a full self test and SPC.

If the SPC errors follow the hybrid modules, you need new relays.
If the errors remain on the same channels, then time to dive deeper..

You can also double check contact resistance with a 4-wire multimeter and activating the relays with a bench PSU manually.

[Tantratron]

How are the relays on the input hybrids? They can and will cause SPC to fail if worn out. Might be worth replacing the full set of 16.

I would swap the hybrid modules between channels 1 and 2, and between 3 and 4. Then reset errors and do a full self test and SPC.

You can also double check contact resistance with a 4-wire multimeter and activating the relays with a bench PSU manually.

Is it possible to test the relays (on off and continuity) without deconstructing the attenuator metal housing module ?

What is confusing or intriguing, both Ch2 and Ch3 have same type of failure (no trace signal when vertical selection < 200 mV and no trigger stable trace when vertical signal selection >= 200 mV). There could be a common point failure, why only relays of Ch2 and Ch3 fail then why trigger fail but signal trace is seen on display ?

[TERRA Operative]

You will have to open the metal housing (It's just a bunch of torx screws and it lifts away), then desolder the BNC connection to the hybrid and the hybrid modules carefully lift out.
There's no other way to test the relays.

Just be super careful that they all line up properly when you retighten it back together so they don't crack. As long as they are seated right, it's fine.
Also be sure to replace the thermal paste on the chip, you'll see it when you take it apart.


I'm not sure re. the trace failure, I'll have to think about it a bit. Maybe something to do with the interleaving between channels or something that is only affecting those two.
The relay test is the easiest to do empirically. It may just be those two channels were used a lot causing more relay actuations which wore out the relay contacts.
I've often seen channel 1 and 2 die mostly due to relays wearing out, but it would depend on scope usage patterns.

It might not even be the relays, but with the acquisition board on the bench, you're halfway there already.

[Tantratron]

P.S. Maybe one explanation could be related upon boot, for a few seconds the 7-segment display shows ".L" as I've mentioned in this other thread. To my best knowledge, the ".L" is not found in service manual

[Tantratron]

I've run the ConsolePort to see of any possible clue, see the log file

Code: [Select]



.Bootrom Header Checksum passed.

.Bootrom Total Checksum passed.

.BootRom Check Sum passed.

.Bus Error Timeout test passed.

Kernel Diagnostics Complete.


Calling SDM (monitor) Routine.


.Enabling Bus Control register. Value = 0x67

.IMR 1 Register test passed.

.Misc. Register test passed.

.Timer Interrupt test (Auto-Vector) passed.

.NVRam DSACK test passed.

.NVRam Write protected.

.Flashrom DSACK and JumpCode test passed.

.Flashrom Checksums passed.

Bootrom Diagnostics Complete.


Transferring control to the Flashrom.

DTesting System Ram.
D-> Console Loopback O.K. <-
.
No ethernet available



 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]   

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]   

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]   

      ]]]]]]]]]]]  ]]]]     ]]]]]]]]]]       ]]              ]]]]          TM

 ]     ]]]]]]]]]  ]]]]]]     ]]]]]]]]       ]]               ]]]]             

 ]]     ]]]]]]]  ]]]]]]]]     ]]]]]] ]     ]]                ]]]]             

 ]]]     ]]]]] ]    ]]]  ]     ]]]] ]]]   ]]]]]]]]]  ]]]] ]] ]]]]  ]]   ]]]]]

 ]]]]     ]]]  ]]    ]  ]]]     ]] ]]]]] ]]]]]]   ]] ]]]]]]] ]]]] ]]   ]]]]   

 ]]]]]     ]  ]]]]     ]]]]]      ]]]]]]]] ]]]]   ]] ]]]]    ]]]]]]]    ]]]] 

 ]]]]]]      ]]]]]     ]]]]]]    ]  ]]]]]  ]]]]   ]] ]]]]    ]]]]]]]]    ]]]]

 ]]]]]]]    ]]]]]  ]    ]]]]]]  ]    ]]]   ]]]]   ]] ]]]]    ]]]] ]]]]    ]]]]

 ]]]]]]]]  ]]]]]  ]]]    ]]]]]]]      ]     ]]]]]]]  ]]]]    ]]]]  ]]]] ]]]]]

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]   

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]       Development System 

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]   

 ]]]]]]]]]]]]]]]]]]]]]]]]]]]       

 ]]]]]]]]]]]]]]]]]]]]]]]]]]       

 ]]]]]]]]]]]]]]]]]]]]]]]]]       

            VxWorks version 5.0.2.

            Stand Alone Rom Version.

            Columbia Proc. Brd. Rev COLOR-2!BIG

            System Ram Size = 2097152, Proc ID = 0x31

            NvRam Size = 131072, Clock Interval = 13108 usecs,

            Bootline Storage Size = 220 stored @0x4000024

            Flash ID = 0x4, Flash Wait states = 3 (Misc. reg = 0x1c),

            KERNEL: WIND version 2.0. GCC COMPILED

            Copyright 1991-2001, Tektronix, Inc.

            made on: Fri Jan 29 13:15:26 PST 1993.

            Copyright Wind River Systems, Inc., 1984-1990

Executing Diagnostics

Start Power-On Diag Sequence
hwAccountant probe routines
  Probe for unexpected pending ints
  Dsp Instr mem size
  Dsp D2 mem size
  Dsp D1 mem size
  Dsy Vect0 mem size
  Dsy Vect1 mem size
  Dsy Wfm0 mem size
  Dsy Wfm1 mem size
  Dsy Text0 mem size
  Dsy Text1 mem size
  Acq number of digitizers
  Acq mem size
  Cpu timer interval uSec
  Cpu Dram size
  NvRam mem size
  Limit to 1GS/sec presence
  Opt Math Package presence
  Opt RS232/ Cent presence
  Opt 1M presence
  Acq Intlv Cal Id presence
  Opt TvTrig presence
  Opt TvTrig index
  Dsy color presence
  Opt floppy drive presence
->   Acq number of user channels
dspForcedBus ................... pass
cpuDiagD2MiscReg ............... pass
cpuDiagDSPIntMaskReg ........... pass
cpuDiagDsyAccess ............... pass
dsp68kMemTest .................. pass
cpuDiagFIFOMem ................. pass
dspRunVerify ................... pass
dspBusRequestTest .............. pass
dspImplicitBusAccess ........... pass
dspTristarMemTest .............. pass
dsyDiagPM100Reg ................ pass
dsyPM011RegMem ................. pass
dsyLCSRamdacRegMem ............. pass
dsyMonoRamdacRegMem ............ UNTESTED
dsyVGARamdacRegMem ............. pass
dsyDiagPPRegMem ................ pass
dsyDiagRasRegMem ............... pass
dsyDiagRegSelect ............... pass
dsyDiagAllMem .................. pass
dsySeqYTModeV0Intens ........... pass
dsyDiagSeqXYModeV1 ............. pass
dsyRastModeV0Walk .............. pass
dsyRastModeV1Attrib ............ pass
dsyWaitClock ................... pass
cpuDiagAllInts ................. pass
nvLibrariansDiag ............... pass
calLibrarianDefaultCk .......... pass
dspForcedBus ................... pass
acqProcThermistor .............. pass
trigGtlRegisterDiag ............ pass
trigBtlRegisterDiag ............ pass
ch1LogicTypeComp ............... pass
ch2LogicTypeComp ............... pass
ch3LogicTypeComp ............... pass
ch4LogicTypeComp ............... pass
ch1EdgeTrigDiag ................ pass
lineTrigDiag ................... pass
dlyTrigDBETrigAfter ............ pass
dlyTrigDBTRunsAfter ............ pass
slewrateTrigDiag ............... pass
trigAttenSerialReg ............. pass
trigPreampSerialReg ............ pass
trigDTCSerialReg ............... pass
trigExtlSerialReg .............. pass
trigDacSerialReg ............... pass
TICountersDiag ................. pass
gtlBigCountersDiag ............. pass
trigBtlConfidenceDiag .......... pass
trigGtlCompRamDiag ............. pass
digRegisterConf ................ pass
digSpecialRegisterConf ......... pass
digPhaseLockLoopDiag ........... pass
digSpeedMemConf ................ pass
digAcqMemPatConf ............... pass
digDataFormatConf .............. pass
digAcqSubSampleConf ............ pass
digAcqPeakDetectConf ........... pass
digAcqHiResConf ................ pass
digAcqMemDataConf .............. pass
digAcqMemAddrConf .............. pass
digInterruptConf ............... pass
digA2DConnectsConf ............. pass
digHFStepConf .................. pass
digDspConf ..................... pass
glitchTrigDiag ................. pass
pulseWidthDiag ................. pass
digPixelatorConf ............... pass
trigNibDiag .................... pass
fpDiagConf ..................... pass
optDiagPM110Reg ................ pass
optDiagFloppyCacheMem .......... pass
optDiagFloppyControllerIO ...... pass
optDiagFloppyDrive ............. pass
optRS232DuartIO ................ UNTESTED
optRS232DuartIntLoop ........... UNTESTED
optCentronCntrlReg ............. UNTESTED
optTv8bitRegDiag ............... UNTESTED
optTvShiftRegDiag .............. UNTESTED
optTvXparentCodes .............. UNTESTED
optTvDontCareCodes ............. UNTESTED
optTvEdgeAndLevel .............. UNTESTED
optTvSyncLevels ................ UNTESTED
Executing Smalltalk

hwAccountant probe routines
  Probe for unexpected pending ints
  Dsp Instr mem size
  Dsp D2 mem size
  Dsp D1 mem size
  Dsy Vect0 mem size
  Dsy Vect1 mem size
  Dsy Wfm0 mem size
  Dsy Wfm1 mem size
  Dsy Text0 mem size
  Dsy Text1 mem size
  Acq number of digitizers
  Acq mem size
  Cpu timer interval uSec
  Cpu Dram size
  NvRam mem size
  Limit to 1GS/sec presence
  Opt Math Package presence
  Opt RS232/ Cent presence
  Opt 1M presence
  Acq Intlv Cal Id presence
  Opt TvTrig presence
  Opt TvTrig index
  Dsy color presence
  Opt floppy drive presence
  Acq number of user channels


Smalltalk/V Sun Version 1.12

Copyright (C) 1990 Object Technology International Inc.

One question I wonder, it is the first time I see part of this Console output reporting No Ethernet available and Development System

Otherwise most test seems to have pass or untested so no clue about why SPC fails with 2 failures...

[TERRA Operative]

The Ethernet was a function that was apparently planned but unfortunately never implemented in the shipped product.
Apparently installing an Ethernet card into the PCMCIA version of the HDD option would be somewhat functional, besides no UI or direct OS support etc, so I think you can safely ignore that message.
It would be cool to get more info to try to get it working with an installable driver (way beyond my abilities though!).


I'm not sure about the Development System message though, I am currently making a debug adapter, so I haven't yet played with this part of these scopes yet.
I'll have something in my hands soon though (I'll make an adapter to use the Option 13 serial port, followed by completing mu Debug-USB adapter later) so I can help compare debug outputs then.


My money is still on shuffling the hybrid modules to eliminate that as an issue.

[Tantratron]

P.S. Maybe one explanation could be related upon boot, for a few seconds the 7-segment display shows ".L" as I've mentioned in this other thread. To my best knowledge, the ".L" is not found in service manual

I think we can rule out the 2 seconds display of ".L" because if connecting a USB-RS232 console port, the ".L" never appears. In fact, the dynamic boot 7-segment display is bit different wether connecting or not connecting a console port on J40.

[Tantratron]

Since I've a spare processor board from a TDS540C (this board runs with 68040 under firmware 5.2e), I've installed it in replacement from the TDS784A processor board (runs with 68020 under firmware 4.2.1e). The idea of this test is to eliminate any root cause from processor board to confirm the issue is in the TDS784A acquisition board.

Everything will boot OK except there will be a new error regarding the EEPROM (crc and other) which is known error because 2 reasons. The first reason is if U1052 and U1055 chips storing EEPROM cal constant to be failed then processor board can read these calibrated values. The second reason is because tektronix has progressively the data structure and crc method of the EEPROM depending firmware and acquisition board design. Since TDS784A seems to have same ASIC as all TDSxxC except firmware of U600 chip, indeed I can still see and trigger the signals so in tehory TDS784A could be transformed into a TDS784C.

N.B. As of today, I do not know the correct math description of EEPROM content versus firmware to transform its data but my test shows again the same SPC error message regarding Ch2 and Ch3.

So for sure now, the acquisition board is partially failed with 2 channels not working correctly wether under firmware 4.2.1e or 5.2e.

By the way, since I had a spare D1 databus board, I've tried but it is not the culprit.

[Tantratron]

My money is still on shuffling the hybrid modules to eliminate that as an issue.

I think to have found as spares some hybrid modules from a TDS520D.

Do you know if hybrid version from TDS784A would be same and compatible versus hybrid from TDS520D (made much later design) ?

How about the AUX1 and AUX2 hybrid versus Ch3 and Ch4 hybrid, are they actually the same so they could re-used or swap ?

Generally speaking, is it possible to inter-mix hybrids within A serie B serie C serie and D serie ?

Thank you, Albert

[TERRA Operative]

There are two types of hybrid module, smaller and larger types.

IIRC, if the hybrids are mounted separately to the smaller acquisition board as on some TDS500 series, they are the smaller type hybrid.
If they are mounted directly onto the larger type acquisition board (as you have), they are the large type hybrid.

There are a few revisions of the hybrids (H, G, etc, it's printed on the hybrid board itself) but all large hybrid modules are cross compatible with each other (and all small hybrids are cross compatible with themselves too), regardless of what the channel type or scope model number is. But you can't swap a large for small and vice-versa.

[Tantratron]

Ok here is a status where first, for sure the hybrids of this TDS784A are large model similar to C serie or D serie.

I've removed the 20 screws to extract the hybrids block where as a reminder, only CH 2 and Ch 3 are failed for some reason. Not sure if this is normal but i've noticed that 3 hybrid are revision G3 and the last one is revision J. From this Oscope, there was not thermal paste trace then since I've a spared TDS520D, I've decided to remove as well its hybrid block. Again so trace of any thermal paste. This time, I see all hybrids are revision J so uniform use of hybrid models.

Question 1: do you confirm they should be always thermal paste and if yes, where exactly having in mind I did not desolder the BNC tap so all hybrids are still in place inside the metal housing

Question 2: do you think one issue could be related to mixed hybrid version G3 with version I

Question 3: on each hybrid, there is a black IC with always marked Maxtek 165-2462-12 (all from TDS520D) whereas the ones from TDS784A are not the same (165-2462-02 and 165-2000-00)

I'm attaching single close view picture of both hybrid blocks (top is from TDS520D and botton from partially failed TDS784A)

My plan is now to install the attenuator hybrids block from TDS520D into the TDS784A acquisition board to see if it solves the Ch 2 and Ch 3 failure.

[TERRA Operative]

If the hybrids physically fit, they are cross compatible. The letter denotes the revision (i.e., J is a later revision than G). Mixing hybrids will be no problem (assuming the hybrids are not faulty, or more commonly, the relays aren't worn out)

The heatsink compound is put on the other side of the hybrid, on the opposite side to the chip where it contacts the metal block. You'll see it if you remove a hybrid.

Be VERY careful when reinstalling to make sure all the hybrids are seated properly. If they are not properly seated in the metal block, they WILL crack!
The best way is to place the hybrid assembly as per your photo and lower the PCB down from above.

[Tantratron]

Ok many thanks TERRA, so now I've installed the complete hybrid module from spare TDS520D inside the TDS784A. So far, I do not feel enough confident to remove each hybrid so basically I've swapped in one time the 4 hybrids to see if this solves the initial double failure.

Indeed as you said, the J serie hybrid seem fully compatible or interchangeable, the TDS784A was able to run fine and this time, I could see the 2 traces (Ch 2 and Ch 3) when voltage division was less than 200 mV. I could not still trigger a 1 KHz signal on Ch 2 and Ch 3 but then after some warm up, I run the SPC which fixed that other problem.

So clearly the initial hybrid 2 and hybrid 3 are failed, probably relay thanks to this macro-test using a spare complete TDS520D attenuator.

Another question: do you know that if changing one hybrid or more requires then to re-calibrate the oscilloscope or the interchange of attenuator hybride does not really imapct the initial calibration ?

Last question: have you repaired any hybrid with faulty relay, is it easy to de-solder one failed relay then re-solder a new relay without breaking the ceramic substrate or do you rather recommend to replace faulty hybrid with one valid hybrid (new or from spare stock) ?

Thank you, Albert

[TERRA Operative]

Swapping out hybrids doesn't require a recalibration, the SPC can take care of it. No problems there.

As for replacing the relays, I made a video!
Just be quick with the soldering iron (some extra flux can help, just be sure to wash it away after) and silver loaded solder might help too maybe? But I always just used standard solder with no problems that I could see.

[Tantratron]

Ok thank, I'll first remove both Ch 2 and Ch 3 hybrid from my partial failed TDS784A attenuator housing (only 2 de-soldering action).

Now I still wonder another question, please see again my new attached picture where I've TEXT commented in RED the channel numbers and put a YELLOW arrow pointing at a strange marking of CH 1 chip.

It is clear that whoever repaired this oscilloscope or maybe at factory production in the 90's, they decided to inter-mix hybrid versions (H2462J versus H2462G3). Let's assume it works (inter-mix or interchangeable) even though this is bit surprising but do you think it is normal that the MAXTEK US chip firmware marking of CH1 is upside down silk written ?

My assumption would be that if this black chip hosting the local firmware to digitally interface with the oscilloscope commanding relays, if soldered correctly then the printed white label should be all aligned the same. However the CH 1 or left one is different orientation... any thoughts ?

[TERRA Operative]

I've never taken notice of the chip. I just swap hybrids as I need to, I don't really care much about the revision either...

It doesn't have firmware, it's pretty simple inside. You can see roughly how it works from the schematics available on Tekwiki for the TDS500 series scopes.

There is no real way to rework the chip, as it is bonded to the ceramic substrate in a way that removal would be destructive, so how it is is how it was manufactured.


The hybrid wouldn't have left the factory, let alone be in service for decades if it didn't work...

[Tantratron]

OK but then depending on the hybrid revision, chip version, how can you tell if it can run up to 500 MHz or 1 GHz. Each hybrid besides the attenuator commanded by relays must have some specific filter or amplifier. So depending on the year of design, the TDSxxx serie (suffix A, B, C, D) not all models would run up to 1 GHz. Just curious to understand why they would be inter-mix or interchangeable on any TDSxxx/A/B/C/D

[TERRA Operative]

All TDS600 and TDS700 series scopes are essentially capable of 1GHz, the limiting factors are personality identification resistors on the acquisition board that tell the processor board what it is, and a capacitor on each input diff pair on the 500MHz model, located between the hybrids and acquisition chips that act as a roughly 500MHz low-pass filter.

All hybrids are capable of 1GHz, and essentially all 500MHz acquisition boards that have a 1GHz equivalent can be modded to 1GHz (This may not be true for the early non A, B, C, D models, I haven't played with them enough to check).

I.e., any TDS754A, B, C, D can be upgraded to a fully functional TDS784 of the same letter type, 'simply' by swapping the ID resistors, removing the bandwidth limiting capacitors, and then running a full cal/adjustment procedure. (500MHz scopes are not calibrated from factory to 1GHz. The scope will work at 1GHz, but it won't be calibrated beyond 500MHz).