Repairing the aquisition-system of the Philips PM3323 Oscilloscope

[SaabFAN]

After about 2 years of more or less passive search, I've finally got a PM3323 here.
Naturally, it has the problem with the noise on both channels and also a probably a bit more concerning problem: Channel B starts wandering higher and higher on the screen after about a minute and disappears completely.

Voltage-Measurements with the Multimeter show the + and -14V-Rails sitting at +/- 13,35V.
The cause for the noise is, according to research on the Net, always caused by dried up electrolytic SMD-capacitors. The way too low voltage is most likely caused by resistive loading of that rail by the broken caps. If not, the repair will become a bit more of a challenge

The caps used in the device are all 10µF/16V electrolytic capacitors with a diameter of 4mm.
I don't have these caps here at the moment and because the holidays start soon, I doubt I'll be getting the replacements before chrismas.
BUT, I have 47µF/10V Tantalum capacitors and about 30 1206-Package ceramic caps with 10µF and 35V rating.
At least the so-called "Driver-Unit" that sits on top of the CRT uses the 10µF-caps exclusively in the power-rails. Unfortunately, only 4 caps have 9V on them. The rest are connected to the 14V-rails.
The big CCD-Unit on the other hand also has the caps connected to the inputs and outputs of some OpAmps. I'll have to see if using ceramic caps with their low ESR will work here.

Other than the CCD-Unit and the Driver-Unit, the PM3323 and the PM3320A are pretty much identical inside. CCD-Unit and the Driver-Unit contain a lot of SMD-Components and the CCD-Chips have been mounted directly on the board instead of being mounted on small daughter-boards and are also of a different type: PM3323 CCDs are ICs in a ceramic package whereas the ones in the PM3320A are mounted in plastic packages.

Once I've taken some pictures, I'll post these here.

After solving the hardware-issues, I'll do the same modifications to the scope that I already did to my PM3320A, which are being discussed in this thread: https://www.eevblog.com/forum/repair/repairing-an-old-pm3320a/

[SaabFAN]

Well, good news: 0805-Capacitors fit perfectly on the footprint of the old capacitors. Unfortunately, they smell like dried fish when desoldering them with hot air. One even popped (a few seconds after desoldering)!
Also bad news, it seems like at least one cap has leaked its corrosive guts onto the board - And of course it did so right next to a Via -.-

Also, as promised: Pictures of the boards (CCD-Unit and Driver-Unit).

[SaabFAN]

I've replaced all the SMD-Capacitors with ceramic 0805 caps and the noise has been reduced significantly. But it is still there and Channel A still has massive problems. Channel B works more or less now (still unacceptable amounts of noise, but I suspect that's coming from the ADC-Board, as replacing the caps on on the ADC-Board my PM3320A, which is in much better condition when it comes to the capacitors, caused a noticeable reduction in noise-floor).

Channel A on the other hand does all kinds of crap:
- Time-Base higher than 200µs: Signal disappears to the top of the screen and no amount of Offset-Voltage is able to bring it down
- Timebase between 200µs and 200ns: Signal shows lines in regular intervals, which scale with the offset-voltage
- Timebase between 100ns and 5ns: Signal looks normal (but I'm not trusting the image on the screen too much in that regard)

First I thought, there could be something wrong with the Frontend, but there's absolutely no change in the behavior when I change the Volts/Div setting. So it has to be a problem related to the CCD-Chips or the transport of the data from the frontend to the ADC.

[tautech]

Does it have a 50 front end.
These are often damaged by to high inputs as usually they're only 5V max.

What about a damaged FET input stage, it's another point of fragility in CRO's.

[SaabFAN]

I've checked the capacitors in the Power Supply and replaced a few that had slightly higher ESR than the ones in my part-bins.

Basically every cap checked out OK. They perform like brand new ones in fact.

After reassembly of the PSU, I fired it up and decided to measure the Clock-Driver unit, as that is one of the hottest running units in the scope.
Once I stuck the probe in there, I accidentally touched one of the metal plates mounted on the PCB, which caused quite a spectacle: Blue arcs, screams from the PSU and then Short-Circuit Protection kicked in.
The plates carry -12,82V!!!!!!! (which is supposed to be -12,7V, according to the service manual).
Result of this little mishap, Channel B now has a similar problem as Channel A does (Video-Link: ) and the -14V Rail is now sitting at -13,06V (before: -13,32V).

I'm not sure where the problem is at the moment. It could be the Clock Driver-Unit, but I'm not so sure and will have to hunt further.

[SaabFAN]

Well, after disassembling the scope, taking out all the boards on the bottom and replacing a few more capacitors, I can say a few things:
- The capacitors are mostly fine (the blue axial ones): They measure all about the capacity they are supposed to be and the ESR is between 200 mOhms and about 3 Ohms.
- The scope has had at least one one previous repair attempt: Several screws are loose and I've even had a few screws just falling out the case - And nobody knows how much damaged those have already caused 
- The scope has seen WATER! There's rust in the mounting of the hinge and the connectors in the Driver-Unit did have green corrosion on them! No big corrosion-spots on the bottom boards though.
- At least one transistor on the A33 P2CCD-Unit is burned out! Once I've reassembled everything, I've turned the scope back on to check if recapping the Management-Unit yielded any success and immediately a BCW33 transistor, connected to the 40V-Rail, started glowing and smoking. I've checked, all the cables are where they are supposed to be and no pins in contact with metal they aren't supposed to touch. The transistor in question is V8323, which buffers and amplifies the CDOT_BO-Signal (Supply: +40V-Rail, Input: OUTO from D8201 (the CCD-Chip of Channel B) via 330 Ohms resistor, Output: Into another amplifier via a 330 Ohms resistor. No leg has short circuit to GND.).
- Voltages are still out, even after replacing quite a few capacitors: +/- 7V @ 7,67V / +/- 14V @ 13,06 / 20V @ 19,5V / 5V @ 5,12V / 40V @ 38,6V, which means something is putting an unacceptable load on the power-rails.

Most troublesome is the burnt out transistor and the fact that there's evidence of water coming in contact with the scope, which makes this waaaaay more than just the capacitor-replacement repair I was hoping for.

[tautech]

Voltages are still out, even after replacing quite a few capacitors: +/- 7V @ 7,67V / +/- 14V @ 13,06 / 20V @ 19,5V / 5V @ 5,12V / 40V @ 38,6V, which means something is putting an unacceptable load on the power-rails.

Which one is the reference rail ?
If more than one rail is out it can be the reference rail has the problem and then all the rails that are referenced to it suffer to some degree too.

[SaabFAN]

Reference-Voltage is 10V, which is spot on. The 5V-Rail is regulated against that 10V reference and and sits at 5.12V.

Also when testing the scope without the P2CCD-Unit, I've seen the +14V-Rail jump up to 14V for about 10 seconds, before dropping back down.

[SaabFAN]

I've continued to investigate the problems of the scope and I've managed to locate massive amounts of noise in the analog circuits of Channel A (something is seriously wrong there). Channel B seems to be doing just fine.
Also some of the OMXXX-Hybrid parts get way warmer than others, also indicating that something is wrong there.

The problem is: All the caps I've replaced so far are fine and the Supply-Rails have very little noise on them. In many cases, I see no difference between the noise floor of my Rigol or PM3320A and the Supply-Noise on the PM3323.
So there must be something wrong with the components themselves.

The biggest problem though is the fact that the P²CCD-Chip of Channel B is broken. I don't know if I damaged it during removal (there are quite a few cables that need to be removed and most of them are connected right next to that ceramic packaged chip) or if it was on it's last legs anyway. I'm guessing the flexing of the board caused it to break.
This is a gamestopper. I've looked on the web and there are no OM3021 Chips available. There isn't even a datasheet available anywhere online.
So my only hope is a donor device with a working P2CCD-Unit. Otherwise, the PM3323 is nothing more than a source of spare parts for my PM3320A

[vanders]

Hello to all,

Have also a PM 3323 scope on the bench, which also has a noise problem, i will make some pictures and come back soon.

[vanders]

Think this is the most common problem..

Start to  change the capacitors first?

[wazoo]

As all of us, change ALL SMD capacitors and check power supply
more than 95% of noise problem is it

SaabFAN has fight    with this equipment and knows a lot   

[vanders]

On which board first ?
Today i opened the scope for the first time..

[vanders]

When i disconnect the marked cables, i get a more or less a noisefree, A and B channel line, so the problem must be found in the pictured board, and start to change the smd capacitors ?

[wazoo]

as a suggestion I will change ALL SMD capacitors and Axials (can change to radial version because of axial are rare today)
you can wait for the power supply section
as other users experience (mine too) ALL board capacitors are "de-capped" because of 30 year instrument and this failure is a "standard"
(bad quality capacitors, less cost, old tecnology, design failure....who knows...)
changing all will renew your intrument for years again  (except if there are other issues  ) 

you can make a simple test desoldering various SMD 10uF caps on P2CCD board and verify capacity
I am 95% sure that will see 0uF   
check PCB around this capacitors and probably will see some electrolite leakage
mine shows the same input "noise" in both channels as yours, and I am changing slowly (because of hobby time  )

you can follow SaabFAN threads and will see his intense research

as you can see on this other thread , last posts
https://www.eevblog.com/forum/repair/repairing-an-old-pm3320a/125/
I have got great difference from starts to today step
I am waiting for new caps to complete repair

[vanders]

Yes thats the way to go, i will order a pack of 50 smd caps ( cost almost nothing )  they will fit right in..

Visual for the other components i don,t see any burnt or leaking components,

Are you changing your caps also by the same smd type?

[SaabFAN]

That you see no noise if you remove the marked cables is caused by the fact that these 4 cables provide the Clock-Signals to the p2CCD-Chips (those fragile ceramic unobtainium beasts of which one died in my PM3323 - BE CAREFUL when removing the board with the vertical amplifiers! The barrier between the P2CCD-Unit and the amplifier-board has springs on it, which send the metal piece flying once the screws are loose. This most likely destroyed the P2CCD-Chips of channel B in my unit.).

If you have another oscilloscope, check the output from the vertical amplifier-board. In my scope, one channel was quite noisy while the other one was ok. And then just keep changing capacitors until the noise is gone (didn't help in my unit, because one of the amplifier-hybrids was creating the noise while the supply-voltages were ok).
Should one of the orange hybrids be dead or damaged, I think it is possible to recreate one on a small pcb with off the shelf-components. Some of them even have their internal schematic drawn in the service-manual.

You also have to replace the capacitors on the Clock-Driver Board, which sits above the CRT.
In my unit I discovered the interesting phenomenon that there were some oscillations happening on that board. If you see the same and manage to find the source, please tell us here

Once the analog boards, the P2CCD-Unit and the Clock-Drivers are sorted and no longer noisy but there's still a lot of noise on the screen. take a look at the ADC-Board: There are also a few caps that needed replacement in my PM3320A. With new ones, noise dropped quite noticably.
Also noteworthy: My PM3320A uses a different ADC, which is housed in a metal can, than the PM3323, in which the Chip is under a golden hood on a ceramic carrier. The ceramic one has more than 6dB more noise than the one in a metal case.


From what I've seen so far in these scopes, basically everything can be repaired, some parts are even still being made and listed as "Active" on the manufacturer's websites, except for a few key components:
- P2CCD-Chips. If they're dead, it's Game Over until a donor-unit comes along! Replacing them with modern components basically means building a 1-Channel 500MS/s Sampler with a 50kS/s Output and a DSP to recreate the quirks of a P2CCD of which there is no datasheets available
- Programmable Logic Sequencers: NOS-Devices available and recreation with modern CPLDs possible, but getting the program out of them requires REALLY expensive special programmers.
- Transistor-Array-ICs - There are a few transistor-arrays on the boards, which are custom philips parts and not available any longer. They could be recreated with tiny SMD-Components, but like the Logic-Sequencers the work to be done would be extreme.
- The CRT (Obviously, although with enough work, a display-adapter could be built with a STM32-Chip to read the display-memory and write the content to an LCD - The Service-Manual goes into quite great detail about how the display-system works.)

[vanders]

Goodevening,

Thanks for the info, will order 100 new smd 10 uf 16 v ( instead of the 50 mentioned before )  when i changed them will give you forum members informed about the progress,

Tnx again

[vanders]

Update about the PM3323, changed all the smd caps, and channel A works now with some problems, line is a bit noisey, and so now and then the line is disapearing at start up when it run some time it stays in place,

Channel B is not working, some noise and more than 1 line, it react a bit on the input, there is a diference between with and without signal on B input,

Maybe someone on the forum has some ideas, for a next step.

[SaabFAN]

Looks like the same problem I'm having with my PM3323: Several lines at the same time on screen.
As far as I was able to diagnose the problem before the P2CCD-Chip of Channel B died, this is caused by the Clock-Driver Unit sending clock-signals at the wrong time to the CCD-Chip, and maybe also by the analog switches, which are supposed to alternate between Signal and GND for the Aquisition and Correction-Stroke when in P2CCD-Mode.

You should check the signals from the Clock-Driver Unit for oscillations. I was able to pick up sine-waves from the driver-unit at about 200MHz during times when the clock-signals weren't supposed to be there.
The result is, that the P2CCD-Chip shifts out analog samples at the wrong time or stores them at the wrong time.
The fact that the waveform of these wrong signals was sinusoid, indicates that the drivers are oscillating on their own, which means that somehow there's feedback to the inputs of the RF-Amplifiers, or the termination isn't working properly.

Have you cleaned the Clock Driver-Board? It might be that there's dirt or residue from leaking capacitors that conducts enough signal back to the input.
It is also possible that there's enough noise on the Supply-Lines to get the amps to oscillate.

[vanders]

Other idea that i have is changing channel A to channel B but the cables with the connectors are to short, to swap them, just to be sure that the problem is not in the clock driver unit.

maybe when dismantling the board from the casing cables might be long enough

Also, the supply lines must be ok i think, as channel A is working with some small amount of noise.

clock driver unit is clean but will check more precise..

[wazoo]

Hi again.

Last capacitors are arrived. ( I have buyed 10 SMD cap less than needed  )
I am happy because after changing all capacitors (except on power supply)
the Philips works great, at least more than 99% 

My brother has checked with his Rigol DSA815TG  SA up to 300Mhz
and signal looks very good and clear
we have had lucky




 

[SaabFAN]

Nice! That's how it is supposed to look like
Either I was unlucky with my unit, or my idea of replacing the electrolytic capacitors with ceramic ones wasn't such a good idea

Now you can head over to my PM3320A-Thread and modify the Processor-Board and RAM to speed it up a little

btw. You should replace the caps on the ADC-Board with bigger ones. Helped reduce the noise on my PM3320A.
Also Philips used 2 different ADCs: One in a ceramic package and one in a metal can. The metal can-package apparently has about 10dB less noise than the one in the ceramic package.

[wazoo]

Thanks SaabFAN

Now osc is in my brother hands and he decide what to do now 
I have archived your post if he want to "upgrade" it 

Thanks again for your hard work
I hope you will find a solution for your osc.

......................
Probably a workplace colleage will bring me another unit   
the good news has flown very fast    thanks to whatsapp 

[vanders]

That looks good. felicidades!!

Question you can answer i think, does your shift knob of the B channel have influence on the screen of the A channel?

What i,am going to do now is printing the calibration part of the manual and check from the beginning, starting with the voltages and so on, there are a lot of pots to do..

[SaabFAN]

Alright, second try!

I've got a new PM3323 with some goodies in the box. Among them a Keithley 6147 Triax to BNC-Adaptor (at least that's what the sticker on the bag says) and two 300 MHz probes.

Unfortunately, the scope was badly packaged: The only shock absorbing quality of the package was the fact that it was in an old cardboard box that was inside another old cardboard box. See pictures for the results 

Good thing I've held on to the other PM3323 - It's Frankenstein-time! 

The "new" scope exhibits the same noise on both channels that's expected from the PM3323.
I've ordered caps (I'm going to use electrolytics this time as I suspect that using ceramics was the reason it didn't work last time) and will replace damaged parts with good parts from the other unit where necessary.
The tube looks good. Sharp picture and very bright. And inside the scope there's no visible buildup of dust. Like the unit sat in the attic for most of its life.
Once I have replaced all the SMD-Caps, I'll post if it works again.

btw. The Triax-Adaptor is for sale.
https://www.eevblog.com/forum/buysellwanted/keithley-6147-male-triax-to-female-coax-bnc-adaptor/

The last picture isn't showing the damage to the shaft of the rotary encoder properly - That thing is bent about 10 degrees and unusable.
The knob beside it bites into the plastic of the front panel and is stuck. The remaining elements of the front panel seem to be ok.

[SaabFAN]

I've replaced all the caps, which included having to repair two traces that had been eaten away by the electrolyte. One pad was in particularly bad condition and I had to resort to "unconventional" methods. Including using some WD-40 to break through the corrosion - I do NOT recommend this method to anyone unless there's no other way to get through a 1mm thick crust of corroded metal (usually there is) - and using tons of flux with the soldering iron turned up all the way. In the end, I managed to remove most of the corrosion and the new capacitor has contact now.
The traces I had to repair were the -7V Supply-Trace to a LM337 that is surrounded by capacitors and is responsible for creating the CCD-Gain reference voltage, as well as a GND-Connection of a cap close to one of the LM317 that create the 9V-Supply for Channel B.

After this treatment, Channel B works pretty well. It's out of calibration, obviously, but I get a clean signal with only very little amount of noise.
Channel A on the other hand still doesn't work. Down to 500µs/div I was able to get a trace on the screen by playing around with the potentiometers (which means that the CCD should work), but there's absolutely no sign of any input-signal being converted.
At faster timebases, which use the "time conversion"-method, the trace jumps up high and leaves the screen at the top.

2 possibilities: A problem with the clock-drivers of the CCD, or the circuits responsible for the leakage correction aren't working.

[SaabFAN]

Well, channel A gives me a trace on screen now.
The problem was that the trace from a 2V Zener-Diode, which is responsible for providing a -2V Rail for the Substrate of the P2CCD-Chip via R8089, was interrupted by corrosion right next to the solder-pad of R8089 (see pictures).

I bet destroyed traces were also the problem with the PM3323 I tried to repair earlier, which then failed rather spectacularly

The symptoms were pretty clearly centered around the P2CCD-Chip of Channel A (with the added problem that my Scope-Probes sometimes had problems breaking through the oxide-layer of solder joints), with Channel B working sort of (Noise and jumps of the trace at timebase-settings slower than 200µs/div).
Using Channel B as a reference, I walked through the entire signal-chain and discovered that the outputs from the P2CCD-Chip of Channel A were at the same DC-Level, but their amplitude was 10x higher.
With everything set exactly the same on the input, I started focusing on the output-side and then the supply of the Chip - I know: Supply should be checked first, but because everything else around the chip worked, I assumed supply-voltages to be ok.

After repairing the trace, Amplitude of the outputs was basically the same on both channels and by following the procedure for the pre-adjustment of the P2CCD-Chips I managed to get the picture on my measuring scope that is specified in the Service-Manual.

Searching in Forums, it seems a lot of people have the same problem (Channel A or B being outside the screen), but nobody wrote about the cause, so hopefully this will help others fix their scopes: Channel A / B being outside the screen = Check Substrate-Voltage - Mine is at -1.409V and pretty much the same on both channels.

Next I'll check why the 14V-Rail is only at 13,15V and if there are other rails that are out of spec. I haven't seen much noise on the supply-rails so far. At least on the P2CCD-Unit the rails seem pretty clean.

What can I do btw. to prevent any further corrosion? I've cleaned all areas with a brush and Alcohol and during soldering I used large amounts of no-clean Flux. Is that enough or are there additional methods I could try?

[tautech]

............
What can I do btw. to prevent any further corrosion? I've cleaned all areas with a brush and Alcohol and during soldering I used large amounts of no-clean Flux. Is that enough or are there additional methods I could try?

Aerosol based PCB lacquer.
Not all rework jobs call for it but you'l find uses for it in many things that are in adverse environments.

I use it for self etched PCBs to protect the copper from tarnishing until you get around to populating the PCB. No worry about removing it as most of them are solder through.
This is what we commonly use in NZ:
https://www.jaycar.co.nz/circuit-board-lacquer-spray-can/p/NA1002

[SaabFAN]

I've managed to get through most of the calibration-steps and get the majority of the adjustments done (I stopped because it is too hot here now to work on this scope).
Unfortunately, the scope still has a 14-Rail that sits at 13,5 V at the PSU, which becomes 13,15 V on the P2CCD-Board.
The 9V-Rails are a bit high (9,30 V), but considering the fact that everything is working at about 50°C ambient temperature inside the case, I think a little bit of drift can be expected by LM317-Regulators and their feedback-networks.

The 0,5 V too low voltage on the 14 V-Rail is probably caused by weak capacitors. Or one of the many 1 Ohm-Resistors in the power-rails has drifted up. The rail isn't watched and  only the 5V-Rail is regulated, and one other voltage monitored.

During calibration I stumbled across what seemed to be a problem with one of the potentiometers: I had it turned all the way in one direction and just barely managed to get the specified 6 div deflection.
But apart from that, I think I can call this repair a success for now
Front-Panel replacement was done in under 5 minutes btw. Much easier than anticipated.