Author Topic: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)  (Read 148428 times)

0 Members and 1 Guest are viewing this topic.

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #500 on: April 25, 2020, 08:56:52 pm »

Hello

I really need some help with setting and changing up booting parameters using a console. I have read this whole topic and few others multiple times.
I have TDS7104 that is in almost perfect condition, the power cap battery is new, the unit has an older motherboard with 500MHz celeron.

The 7 segment display counts up to L, and stops there.
The console also just stops outputing data after attempting to attach a HDD.
It is the same if HDD was in or out of the unit. Floppy drive, HDD and CD rom are all detected in bios. If I boot into dos using win98 boot disk I can access floppy and CD drive but HDD C: is not detected.
Hard disk works, I checked it in a PC, and everything looks fine. I did not try to boot from it, but it is certainly not damaged, because I just cloned it to another newer drive and the issue still persists.


Now, the really strange part is that when I use c command in console, and I change boot parameters (for example target name to "target" from this default value given bellow) to the same ones posted in this topic by TiN and others i confirm that they are active and taken using p command. It looks perfect, however when I restart the unit, non of the changes are saved. I am back to the old values
Is there a command to save the given values? Are they saved automatically? I feel like I am missing something obvious. |O

Thanks you in advance.

Here is the dump from the console:

Code: [Select]

UUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUU

romPost: MPC106  Walking-one passed.

romPost: MPC106 configuration passed.

romPost: First PCI access  passed.

romPost: PC87560 Walking-one passed.

romPost: PC87560 configuration passed.

romPost: DEC21554 configuration passed.


romPost: PERFORMING ROM CHECKSUM TEST...

romPost: ROM checksum passed.


romPost: PERFORMING DRAM CELL TEST...

romPost: Enabling data cache...
romPost: Data cache enabled.
romPost: Enabling instruction cache...
romPost: Instruction cache enabled.

romPost: DRAM Cell test passed.
romPost: Disabling data Cache...
romPost: Data cache disabled.
romPost: Disabling instruction cache...
romPost: Instruction cache disabled.

romPost: PERFORMING DRAM MARCH TEST WITH CACHE OFF...
romPost: Writing address to each *address...
romPost: Reading address from each *address...
romPost: DRAM march test passed...

romPost: PERFORMING DRAM MARCH TEST WITH CACHE ON...

romPost: Enabling data cache...
romPost: Data cache enabled.
romPost: Enabling instruction cache...
romPost: Instruction cache enabled.
romPost: Writing address to each *address...
romPost: Reading address from each *address...
romPost: DRAM march test passed...

romPost: PERFORMING DRAM WALKING-ONE  TEST...

romPost: DRAM Walking-one test passed.

romPost: PERFORMING NV-MEMORY WALKING-ONE  TEST...

romPost: NV-MEMORY Walking-one test passed.
romPost: Disabling data Cache...
romPost: Data cache disabled.
romPost: Disabling instruction cache...
romPost: Instruction cache disabled.
romPost: Enabling instruction cache...
romPost: Instruction cache enabled.


romPost: POWER-UP DIAGNOSTIC COMPLETE.

Hel










                            VxWorks System Boot


Copyright 1984-1996  Wind River Systems, Inc.





CPU: Tektronix HPP - PowerPC 740
Version: 5.3.1
BSP version: 1.0.8.0
Creation date: Jun 27 2003, 09:52:32




Press any key to stop auto-boot...
 1
 0

auto-booting...


boot device          : sm
processor number     : 0
file name            : c:/vxboot/vxWorks.st
inet on backplane (b): 192.168.0.1:ffffff00
host inet (h)        : 192.168.0.2
gateway inet (g)     : 192.168.0.1
flags (f)            : 0x1008
target name (tn)     : tds7000
startup script (s)   : c:/vxboot/topscript.hw

Initializing backplane net with anchor at 0x1e80000... done.
Backplane anchor at 0x1e80000... Attaching network interface sm0... done

 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #501 on: April 26, 2020, 07:09:45 pm »
Hello, I just want to give some update on my lack of progress. |O

Inspired by this post and topic from user Jwalling I removed a  powercap and solder a large shunt resistor in series with the battery to measure current draw of the NVRAM chip because I was perplexed by not being able to store any data in the NVRAM, and the boot parameters just reverting to default values.

The current drawn from battery was about 7.5nA, and that sounds too low, but could still be acceptable. The battery was older but not depleted, voltage was about 3.2V.
When unit was powered bat. current was 0.
Jwalling wrote that he measured about 20nA current load when unit is powered off.

Now for the interesting part. I could never get HDD to show me a partition when booting into DOS from a floppy.
I tried to format and create one using FDISK but the process would usually fail. It looks like connection between HDD, CD drive and motherboard is highly unreliable.
I noticed that sometimes bios would not recognize CDDrive or HDD. Sometimes it would find them if I put a little pressure on IDE connector but with incorrect names.
For example toshiba HDD would be shown as "TNSHIBA" and IBM DARA HDD would be shown as "D@R@" (last bit in a ascii char byte obviously wrong).

I will take out riser card and inspect cables, and connector fingers to see if something is obviously wrong. If someone has a hint, any suggestion is welcomed.
« Last Edit: April 26, 2020, 07:12:04 pm by elektropionir »
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #502 on: April 27, 2020, 05:27:17 am »

Well I worked on this whole night, and here is a last shout into the void.
The problem was in the connections between PowerPC and NLX motherboard. Some pads were dirty but main issue was that the riser board was somehow very sensitive to position and angle of the motherboard.
The mechanics of this particular unit were putting a lot of pressure on one side of the board. It took me a while to level everything and secure the boards in place.

Now it boots every time and passes self test. I did not need to play around with boot parameters in console. default values just worked.

I cloned a HDD into an SSD. For few of my instruments I use DeLOCK 62495 adapter and small mSata. This works really well.

Since this unit has an older motherboard i decided to upgrade sometimes in the future to Intel SU810, but I cannot find a bios for that board, even if many members have confirmed that their TDS7104 has that board inside...
I also have a 100MHz FSB Coppermine Celeron CPU that is slightly faster than what TiN originally did with Pentium 3.   1100/128/100/1.75V, number SL5XR.
I expect it should work without mods.

If someone has a bios for the intel board, can you please share it.
 
The following users thanked this post: PrecisionAnalytic

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #503 on: May 12, 2020, 07:21:27 pm »
...and now I ran into another issue with the TDS7104 power supply. The analog power supply (+-15V, +-5V) is dead. The board makes a funny noise, that is hard to locate.

However: I reverse engineered the schematic of the whole analog power supply section.

Temperature protection, over voltage protection, over current protection, everything you can imagine is there.

Cause of my issue is not yet identified, but this schematic will help.

No warantee, that this is 100%correct, use at own risk.

UPDATE: The attachment was not correct and containsed only a single page. Fixed.
« Last Edit: May 12, 2020, 07:24:15 pm by Xyphro »
 
The following users thanked this post: Lahay_Alexey

Offline Lahay_Alexey

  • Contributor
  • Posts: 7
  • Country: ru
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #504 on: May 12, 2020, 08:34:17 pm »
Hello, Xyphro. You did a great job, thank you. I have a request: I am trying to improve the characteristics of TDS7054 to TDS7104 and I need a photo of the acequion board area. Required area in the photo from the attachment:
989940-0
At the end - the report from me.
« Last Edit: May 12, 2020, 08:36:09 pm by Lahay_Alexey »
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #505 on: May 12, 2020, 08:40:44 pm »

How did you know, that my tds7104 is open :-D
Find below.
If you find a way to improve tds7104 to 4GHz analog bandwidth, let me know :-)
Let me know, if you need more photos, can also PM them.
 
The following users thanked this post: Lahay_Alexey

Offline Lahay_Alexey

  • Contributor
  • Posts: 7
  • Country: ru
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #506 on: May 12, 2020, 08:56:49 pm »
Thank you for the photo. Unfortunately, tds7104 can only be changed to TDS7404 by radical replacement of the acquisition board and casing. You can specify the values of the following resistors from the photo:
R781, R791, R782, R792, R783, R793, R784, R794.
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #507 on: May 12, 2020, 09:29:41 pm »
My 7104 was also opened and I was uploading a photo but Xyphro already posted it.

@xyphro

When you say it makes a strange noise, can you be more specific? Is it a click or a buzz? Periodic?
Are you testing the PSU outside the unit? Are other voltages present, from other converters? So we know the issue is not in the input side/PFC.

I ask this because usually such problem would not be in these protection OTP/OVP parts.
With the schematic you posted is a bit hard to tie everything around and primary side of the circuit is missing.
+-15V and +-5V are separate secondaries of trafo T7. There is a peak current control loop from each of the secondaries that probably ties back to primary over opto.

My suggestion if you are confident in troubleshooting SMPS, and have proper equipment, take out power supply, supply it from isolation trafo, it can be a small one since the unit is unloaded.
Start the power supply. I think there was a pinout in the service manual and you can find which pin is turn on pin.
If you have a working scope check the signals from each of the current trafos 1K resistors. It could be that one secondary is overloaded/shorted and it pulls the entire voltage down, or control simply turns off because the peak current was exceed. That would be easily seen on the scope.
That would manifest in a sharp periodic clicking usually, depending on how the overload was implemented in these converters.
If you see nothing, or very low current through the secondaries, you can check directly voltages on the secondary windings in front of current trafos. and go from there.
If you see nothing there too, the problem is on the primary side, and you will have to trace out the primary driving circuit. That is less complex than what you already did.

Good luck.

Edit: Do not touch heatsinks when power is on. They are most likely tied to + or - Vbulk and it can kill you. I am not sure who will read this in the future, so I have to write it.
« Last Edit: May 12, 2020, 09:36:57 pm by elektropionir »
 
The following users thanked this post: Lahay_Alexey

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #508 on: May 12, 2020, 09:32:05 pm »
Thank you for the photo. Unfortunately, tds7104 can only be changed to TDS7404 by radical replacement of the acquisition board and casing. You can specify the values of the following resistors from the photo:
R781, R791, R782, R792, R783, R793, R784, R794.

Send me an email in Private messages I will send you high res photos
 
The following users thanked this post: Lahay_Alexey

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #509 on: May 13, 2020, 04:50:05 am »
Moin Electopionier,

I included the primary side in the schematic. The pdf has 4 pages. Also the current monitoring. The current monitoring is global to the board. Digital and analog supplies current monitors are all tied together, likely to turn of via the power on/of optocopler path.

I began probing when it was connected, im the meantime I have it externally, yes with isolation transformer and proper precautions (when I repaired my 1st smps about 25 years ago, I made that mistake touching the cooling plate :-)). I hooked up loads to the supply.

So... Digital sipplies are all ok, including the +3v3, for that reason I focus on the analog supply path. OTP OVP OCP is not triggering, this globally turns off the suplly.

Focus now is the primary side of the analog transformer.
There is a voltage feedback loop over an optocouplet and +-5V are the ones, that are taken to close the loop.

What I notice here is, that the optocoupler works. After a short amount of turn on and off, it goes to permanent off state.
Still, ths VFB of uc3844 is 2.51V, which is, because the COMP pin has 4.2V and only due to the resistive divider path with the 68K1 resistor you get that voltage.
This looks a bit tight dimensioned, considering, that 2.5V is the nominal threshold here.

Replacing the uc3844 resulted in no change of this behaviour.

Oh... The ISense path shows a slowly changing DC voltage of 300..500mV, which is funny, will look in that domain a bit further. Would not expect DC here, especially not, if the OUT pin is continously LOW.

Edit: The sounds are mainly some clicking and sometimes a (sorry for german: Pfeifen, with about 400Hz frequency). I tried long time to find the location of that sound, but failed (Another incidence of where  a "sound camera" would be cool to have).

Beeing more an RF guy, SMPS are a bit of miracle, allthough I understand the principle well, lacking a bit of experience how it exactly works to the latest details (like the current sensing path), allthough I repaired already quite a few, so any comments welcome.

I attached the primary side of the schematic + notes of some voltages I measured on the UC3844. Whenever I measure a DC voltage with a multimeter, I also checked with the scope, if it is really DC or not. So.. UC3844 is pretty stable in the sense of, that nothing is toggling there.

Best regards,

Kai
Ps: REF shows proper 5V on the uc3844! The big capacitors and PFC are ok and show proper voltages (otherwise the digital supply wom't work too).
« Last Edit: May 13, 2020, 05:04:46 am by Xyphro »
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #510 on: May 13, 2020, 05:10:30 am »
Oh, what I think is funny is the primary switching side of the main transformer.
JP1 is here the first 4 pins of transformer T7.

There seems to be only a single coil, which is between pin 2 and 4. Pin 1 and 3 seem nowhere connected. Still pin 1 is connected to GND. Maybe there is some additional windings as protection layer?

Or is from my T7 a winding dead?
Does anybody have by coincidence the Scope open and can measure T7 resistances on the primary side, to crosscheck? (Pin 1 is for me the one on the primary side, pointing towards the yellow Fuse F51).

Conceptual question: Why is there a high and a low side switch (driven with solated gate drive), while only a low side switch would be good enough and do exactly the same?
« Last Edit: May 13, 2020, 05:14:30 am by Xyphro »
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #511 on: May 13, 2020, 05:39:55 am »
Hello Xyphro

I am happy that you have some experience working on SMPSs and take proper safety measures.
I am not surprised that replacing the IC did not help you, the control ICs really do not fail in these units often. Sorry I missed the page with prim.

I would give you a small advice, when troubleshooting these circuits you must have structured approach. These are very repairable but please take things slow, replacing components on random will always cause more damage then original fault. Do not jump to voltages around control IC right away, because most often the problems are in power section not control section.
1) Check the currents directly on shunt resistors R12. You should see triangular waveform there, and certainly no dc component. Is the trafo being driven? if not then...
2) Check gate driving on the directly on the push pull output before 4x10ohm resistors. What do you see there?

I did not really understand your explanation regarding optocoupler, but here is what you should have.
In normal operation if the measured output voltage is 0V (or much lower than reference) on the secondary side, optocoupler transistor on the primary side should be opened. You should see 0V on VFB pin.If you see higher  voltage there that would indicate that feedback path is falsely telling controller that voltage is too high and it should bring it down and stop switching.
Edit: My mistake, i wanted to say that if opto transistor is OFF you will see some voltage on the Vfeedback pin. That is good. The optocoupler's collector is connected to current source. COMP pin is open collector and is driven from current source. This voltage is compared to internal sawtooth generator to create driving PWM. 0V on comp pin would equal to 0%duty cycle. 2.5 or 2.51 is where duty cycle starts increasing to max of 50%. So all is ok there. Probably the problem is in the driving section. My advice is still valid. Sorry for the mistake.

Can you confirm this?

Update: can you draw the trafo in your PDF that I can see. But open ended windings are not uncommon in SMSPs trafo desing. Those are  sometimes connected to a copper shield and sometimes are a real winding with 20-30 turns called compensation windings that create "fake" capacitance between primary and secondary side winding terminals and compensate injected RF current. It is a mark of good desing, and not damaged trafo.
As you see I design these types of power supplies for a living :)
« Last Edit: May 13, 2020, 06:56:16 am by elektropionir »
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #512 on: May 13, 2020, 05:56:25 am »
Conceptual question: Why is there a high and a low side switch (driven with solated gate drive), while only a low side switch would be good enough and do exactly the same?

We have many topologies and all provide different benefits to designers. Yours is 99% a dual switch flyback. You can see a short explanation here. But in general dealing with flybacks main issue is how to overcome large transients when the low side mosfet switches off and what to do with energy stored in leakage inductance. This can be few watts actually and is usually dissipated and that creates heat and losses. In dual switch flyback that energy is re-used by returning it to Cbulk.
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #513 on: May 13, 2020, 06:15:18 am »

Thanks for your valueable feedback. Especially giving that topology a name for better understanding on my side. It also gives indirectly the hint, that the transformer is ok.

The replacement of the controller was a quick shot, I don't replace more untill I see the problem :-)

Will do the proposed measurements.

Right now there are 2 phases: after power on it seems to switch for some seconds and then stops.
VFB get's too high indeed, for that reason it stops switching.
What confuses me is, that the comp pin is high, stating correctly a too high voltage, but this comp pin is also making the voltage on the VFB 2.51V, which is so close to the 2.5V threshold of the comparator in the uc3844.
The voltage accross the 750Ohm resistor is btw. 0V in that state, so the optocouplet is signalling correctly too low voltage, matching, what I see on the DC low voltage side.

Anyway, more debug this afternoon, based on your proposed measurements. This will only work after power on for a few seconds.

Best regards,

Kai


Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #514 on: May 13, 2020, 06:50:09 am »

Hi Kai, please see edit of my previous post. I made a stupid mistake in my explanation that I corrected. I haven't read my post after publishing.
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #515 on: May 13, 2020, 07:00:35 am »

Could not wait until end of the day :-)

Did some measurements, please find attached.

The ISense behaviour is puzling me. The Current shunt is 0V continously, still ISense shows some voltage. As this is below 1V, I'd expect it not beeing the cause output toggling off, but well....
This seems to come somehow from the transistor circuit connected between REF, RT/CT pin and ISENSE (is this for slow start?).

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #516 on: May 13, 2020, 07:13:19 am »
No issues, thanks for fixing.
It is clearly very visible, that this is your daily job :-)
I hope I show expertize too in showing, that I can measure this without a lot of magnetically coupled spikes on the scope :-)

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #517 on: May 13, 2020, 07:27:37 am »
I know how you feel when you work on something out of your field. I just spent 2-3 days adjusting the drift on my Advantest spectrum analyser and tracking gen hoping that I do not damage it even more.  :)

Kai, these control ICs are very standard and almost everyone has one. I made you a short simulation in LTspice so you can see what you should expect when there is no feedback from opto. LT1244 is the same IC but made by LT/analog devices.
Can you check if you also have clock signal present in Ct/Rt? Something is keeping the IC off, we must find what that is. So far the control voltage feedback looks OK.

EDIT: I see that you measure 0V at clock signals, that is quite strange. Ah, that was pin 4 of the opto, not control IC. That looks good. Check for shorts on the timing capacitor pin4 of control IC, during startup or continuous operation we should always have clock there. I feel like something is loading that point. The NPN transistor should not do that unless it is damaged but voltage also does not match the expected oscillator voltage. There is something wrong there.

What you can do, if it is not too difficult, you can for example remove 47R5 on output pin so the unit does not turn on while you inspect. Tie push pull to ground so we have no accidental turn-ons even if unlikely.
Remove that NPN from timing circuitry and see if you will get correct oscillator voltage. The ic should always have that, there is nothing in the datasheet that would disable the oscillator that I can see.


Another point, DC offset on Isense pin is also quite common. It lifts the input of the current comparator from 0V so that control loop has some space to set feedback voltage lower than Isense and keep duty cycle safely to 0% taking into account offset voltages in IC and any noise coupling into sense pin. Another benefit is a slope compensation that makes these types of converters more stable, but I will not go into that.
« Last Edit: May 13, 2020, 08:12:49 am by elektropionir »
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #518 on: May 13, 2020, 08:17:13 am »
Thanks again. The RT/CT is clearly not showing any oscillation.
I get 2.5V with random noise on it, but not the typical RC oscillation characteristics, as visible in your simulation.

Finally something, where I can check deeper.

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #519 on: May 13, 2020, 09:14:15 am »
From the datasheet i think that if the voltage on the cap does not rise above 2.8-3V the oscillator will not reset, it will not try to discharge the cap.
Usually oscillators in these ICs are just a simple analog comparator with hysteresis. Loading the capacitor point with 2-3kOhm could keep the voltage too low.
Another possibility could be for example that you have instead of 3k65 a lower resistance, or that you have a breakdown in NPN transistor. That would source more than 10mA into Rt/Ct pin and it would prevent IC from discharging the 2n2 capacitor. You can measure this nondestructively by observing charging of Ct during turn on. It should have time constant of stated component values, and if it is too fast or too slow you have your answer.
« Last Edit: May 13, 2020, 09:16:51 am by elektropionir »
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #520 on: May 13, 2020, 10:22:50 am »
Your guess was right: The Transistor was broken - was unable to measure the diode between Base and emitter.

This can only have been a side effect. Now the UC3844 oscillates like in your simulation, but (and a big but): Once replaced, it oscillates for some time and then breaks again. Then I replace it again and it works again for some time.

Also the (audible) noise is back (which was gone before).

I also measured the voltage accross the current shunt. Now I see something there.

What I see is not, what I would expect ideally. I see short bursts every few oscillation cycles, having swinging also towards the negative side (note: the ch2 is set to AC coupling in the attached screenshot, but with DC coupling it looks exactly the same).

I can not fully exclude, if those are measurement artifacts. I soldered a coaxial cable across the current shunt, but maybe still something coupled in here.

Will now probe step by step towards the FET gates, but right now I suspect something wrong in the final stage FET stage.

Best regards,

Kai
« Last Edit: May 13, 2020, 10:44:59 am by Xyphro »
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #521 on: May 13, 2020, 10:50:20 am »

Ok, now we know why the IC did not start. Large positive peaks could kill this transistor by exceeding negative max Vbe that is usually around 5V. That is why i suggested first to remove driving from push pull stage.
I was afraid that underlying problem is still present.

Check the driving signals at the output of push pull and shunt voltage. Does the effect appear at turn on or turn off? I assume turn off.
To make sure that what you see is real, you can short your coax both ends on gnd or high side and see if the effect is still there. If it is it is not real. But i suspect that what you see is real. Shunt resistor is very low values, it would take a lot of power to inject that much noise.

Such high peaks and impulses I would usually see in short circuit conditions either across primary or one of the secondary windings. Measuring output directly on the current transformers can tell you if some secondary is shorted somewhere.
Do you see a ramp current during ON period on the shunt? The amplitude would be much lower than the peaks you show so I cannot tell from the picture.
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #522 on: May 13, 2020, 10:58:13 am »
I know it is a pain, but at this stage to be safe, i would for example remove the fuse in series with high side fet and supply the converter from a lab power supply. Other converters are all fine. but this one i would supply from like 30-40V and a lab PSU with current limit. Check if the body diodes of Mosfets as well as gate resistances are all normal. This should be possible to do in circuit.
It would be much safer for you and for the PSU to find shorts and opens when working with low voltages. You can leave 18V Vcc that is fine.

Disconnect the PSU. You can use also small 9V battery, and a 1k resistor and try to turn on and off each MOSFET, check by measuring Rdson. These are simple steps that will tell you if it makes sense to investigate the circuit in more details, or the problem is obvious. I assumed you already found no visible damage on mosfets. Sometimes that happens.
« Last Edit: May 13, 2020, 11:38:49 am by elektropionir »
 

Offline Xyphro

  • Regular Contributor
  • *
  • Posts: 55
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #523 on: May 13, 2020, 08:17:22 pm »
Hi again,

I tested now:
Body diodes of the IRFP460 (not IRFP640 as I wrote in the PDF)
- present as expected :-)
- RDS(ON) of Q10 is 0.23Ohm and it turns off as expected
- RDS(ON) of Q9 is 0.235Ohm and also that one turns off as expected

Furthermore, I tested the voltage limiting circuit with the 2 18V zener diodes. Also here I see, that it limits correctly to roughly 18.8V. I tested this in both directions to ensure, that both the zener diodes work as expected.

I measured (CR80/81) and also they properly conduct only in one direction and show very low Vf as expected.

I right now now injected 30V into the Fuse (Fuse removed) and also injected the 18V externally.
The UC3844 oscillator still oscillates and the ISense pin still shows the same huge voltage transient. So far so good.
Unfortunately I have killed the NPN transistor FMMT720 of the isolated gate drive transformer driving circuit due to a measurement mistake.
Reichelt order is out, too tired now to replace it with another transistor from a different section of the power supply.

What I think I had seen is, that the amplitude of the transformer (T52), that drives the gates was for one of the final stage FETs about only half as high as for the other ones. Unfortunately seconds after this, I killed the above mentioned transistor.

I will check further tomorrow. Learned a lot today in terms of SMPS testing metothology today, thanks for your hints.

Getting a bit afraid, that the actual transformer T7 is killed. This would likely be the killer criteria and the way forward for me to replace the full power supply. I expect this is a custom made transformer, that is not available off the shelf.

Next steps for me:
- Replace the Gate drive FMMT720 transistor.
- Check, if my observations with the different voltage amplitudes for the 2 final stage FETS is correct.
- Maybe replace the primary side coil from the transformer with a resistor and measure ISense again?
- Continue investigations on the secondary side (but could that kill components on the primary side? Based on the schematic I can first decouple the 15V path, to check, if there is a short present

Best regards,

Kai
PS: Attached how it looks like now. Need to remove a lot of flux at the end, but well, lot's of cables. NICE :-)
« Last Edit: May 13, 2020, 08:27:49 pm by Xyphro »
 

Offline elektropionir

  • Contributor
  • Posts: 32
  • Country: de
Re: Tek CSA7404/TDS7000 repair project (Upgraded memory/CPU)
« Reply #524 on: May 13, 2020, 09:52:27 pm »
Hi again,

I tested now:
Body diodes of the IRFP460 (not IRFP640 as I wrote in the PDF)
- present as expected :-)
- RDS(ON) of Q10 is 0.23Ohm and it turns off as expected
- RDS(ON) of Q9 is 0.235Ohm and also that one turns off as expected

Furthermore, I tested the voltage limiting circuit with the 2 18V zener diodes. Also here I see, that it limits correctly to roughly 18.8V. I tested this in both directions to ensure, that both the zener diodes work as expected.

I measured (CR80/81) and also they properly conduct only in one direction and show very low Vf as expected.

I right now now injected 30V into the Fuse (Fuse removed) and also injected the 18V externally.
The UC3844 oscillator still oscillates and the ISense pin still shows the same huge voltage transient. So far so good.
Unfortunately I have killed the NPN transistor FMMT720 of the isolated gate drive transformer driving circuit due to a measurement mistake.
Reichelt order is out, too tired now to replace it with another transistor from a different section of the power supply.

What I think I had seen is, that the amplitude of the transformer (T52), that drives the gates was for one of the final stage FETs about only half as high as for the other ones. Unfortunately seconds after this, I killed the above mentioned transistor.

I will check further tomorrow. Learned a lot today in terms of SMPS testing metothology today, thanks for your hints.

Getting a bit afraid, that the actual transformer T7 is killed. This would likely be the killer criteria and the way forward for me to replace the full power supply. I expect this is a custom made transformer, that is not available off the shelf.

Next steps for me:
- Replace the Gate drive FMMT720 transistor.
- Check, if my observations with the different voltage amplitudes for the 2 final stage FETS is correct.
- Maybe replace the primary side coil from the transformer with a resistor and measure ISense again?
- Continue investigations on the secondary side (but could that kill components on the primary side? Based on the schematic I can first decouple the 15V path, to check, if there is a short present

Best regards,

Kai
PS: Attached how it looks like now. Need to remove a lot of flux at the end, but well, lot's of cables. NICE :-)


Good work. Primary side looks ok so far. You can check if all the diodes on the secondary are all ok. Usually that would make your short.
I am talking about diodes that come after current transformers BR5,6,7,8,9... If the diode that is placed between ground and Elco is shorted in any of the secondaries it would create such spikes in the off period that you see in primary.
I  suspect a short on the secondary because prim. short would usually be much harder and kill the primary mosfets and fuse first... Also it would be probably more in synch with gate driving signal. And there we see some 1us delay.
That is why I said would like to see shunt voltage/prim current and gate driving to see is the current impulse happens during mosfet ON or OFF period. These are all educated guesses, i cannot be sure before testing.

I still think it is unlikely you killed your trafo. You are correct that it is not of the shelf part. However these supplies are difficult to come by and expensive.
Even if trafo is dead, perhaps we can rebuild it.  We are not there yet. Lets take things slow.

You can for example easily check secondary diodes, and then disable secondaries by opening current trafos. They are connected with just normal wire, you can lift one side and disconnect it. I would first try +-15V and put a small load on +-5V output, and then do it other way around.
Disconnect +-5V and put some small load on 15V to limit flyback voltage...

That still looks ok :) , you can clean it up nicely once we find a problem. Just try not to damage PCB by using too high temperature to solder/desolder small parts. That can really make life difficult.
« Last Edit: May 13, 2020, 09:54:54 pm by elektropionir »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf