Repairing and Upgrading an old PM3320A - Including adding FFT!

[Stonewerk]

Actually, ~ $100 for a high quality, multi-layered extender card for boards of this size is not unreasonable.

As you're aware, the connectors are not the issue.

A good extender card should have good power/ground planes, guarded signal routing, etc. so as not to introduce further problems into the signals that you are testing/adjusting.

Yes, I've considered this route, including sourcing for the connectors (not an issue) and executing a suitable layout. However, fabrication with a trusted (i.e., quality) vendor does place the price significantly below $100.

And at this stage, I don't want to execute a group fab / sell deal - 'been there, done that, and was really a giant pain in the hintern 

The main issue here - across the pond - it that it's difficult to find really *any* accessories for this 'scope.

I think the best way, and probably the cheapest, is to get some Europa-Format Vero-Boards with provisions for 96Pin-Connectors and build one yourself. Buying such cards is, at least in Germany, crazy expensive (over 100€ for such a card).
I found 96pin-Connectors for 5 to 9€ on ebay.

An alternative, which could also feature measurement-points, would be to design a PCB in Circuitmaker (or any other CAD-Software without limitations on size and Pin-Count) and let one of the cheap chinese manufacturers produce them. Shouldn't be too expensive and you have several boards to test several modules at the same time or sell those you don't need.

[Stonewerk]

Good? Bad?

From my perspective as an engineer, an answer to your question "depends' and can be a bit complex, and my first statement would be an inquiry into the instrument settings, configuration, terminations, exactly how you're making the measurement and interpreting the results. Also, I'd inquire about the type and characteristics of the signals you're measuring, the equipment and environment you're measuring, your test objectives, etc.) But that's not what you want, eh?   

Some people have said that  I look like Yoda, so I'll say, "Not what is good, not what is bad, but what is." The application the suitability is... 

So I'll make some assumptions and answer this way:

• With nothing connected (no signals, probes, etc.), I press "auto" and let the unit autoset.
• After the routine completes, I set the displayed channel (A) to maximum vertical gain - 5mV/div - and minimum timebase (5 ns/div).
  
• Other Vertical / Horizontal settings are unchanged.
• The observed peak variation in the trace height is not significantly larger than the trace itself, (which is around 1/10 division, approximately 0.5mV), therefore something around 0.5 - 1mV peak-peak.
• Enabling channel cursor and select Measure. Amplitude, Peak, the peak value varies from about 0.97mV to 1.17mV.

Is this what you're asking? I'd deem the performance of this parameter is quite reasonable.

The traces on the boards are all pretty thin, so IDE-Cables should be able to handle the currents.

On another note: How much noise does your Unit show, Stonewerk? I get pretty consistent 1,2mV of noisefloor in all timebase-settings at 5mV/div. I don't know if that is good or just at the corner of going bad, but my Rigol DS1054Z has about 600µV of noise at the fastest timebase and increasingly higher noise at slower timebases.

[Stonewerk]

The following is an addendum to my earlier comments, regarding the PM3320A firmware images posted to K04BB by SaabFAN:

I have also made a cursory test of the ROM firmware for the PM8956A (A7 IEEE/RS232 Option Card).

After updating and testing the CPU firmware to v4.1, the existing firmware for the IEEE/RS232  was replaced with the newer firmware. Operation with the new firmware appears unaffected, meaning that original functionality does not appear to be adversely affected. However,, no new functions or options appear, nor have any performance improvements been observed. As before, a comprehensive differential comparison has not yet been performed.

Note that it was not necessary to update the IEEE/RS232 firmware. The original (B744.1, B743.2) PM8956A firmware seems be to compatible with the updated (v4.1) CPU firmware, without any operational or performance issues (with the caveat regarding depth and breadth of testing, etc.)

The check sums for the updated IEEE/RS232 ROMs, posted by SaabFAN:

D112: E709
D113: 7C10

The compatibility of the new IEEE/RS232 firmware was not tested with the older (v2) CPU card firmware.

This completes my initial assessment of the newer CPU and IEEE/RS232 firmware. In summary, it appears that the new firmware is compatible with older hardware as noted (see earlier posts for details) although enhancements and improvements appear to be limited.

I am now operating my PM3320A with the updated firmware for both the CPU and IEEE boards as provided by SaabFAN. If any issues are found (or significant improvements observed), I will post them here.

[SaabFAN]

Yes, that was exactly what I was doing on both units to observe the noise-floor of about 1.2mV.
Given that your unit has pretty much the same amount of noise, I don't think that a complete Recapping of the PSU will be necessary any time soon. But I have a lot of SMD Tantalums available in my part-boxes and from the defective scope I know that these caps fit almost perfectly

About accessories: I think the only accessory worth getting are probes with a Sensor-Ring to tell the scope if the 10x-Attenuator is active or not. And maybe an RS232-Cable. Software, if you can even find some, is probably unable to run on anything but MSDOS, but I have seen drivers for the PM3320A on the National Instruments website. But I guess the prices for Software from that place can pay for an entire lab-setup
Do you have the RS232/GPIB-Manual btw.? It holds detailed information about how to communicate with the PM3320A, as well as all commands available. They even put some examples for driver-programs in there!
I have also thought about using an Arduino to build some kind of a digital Plotter, but stopped that project for the moment, because for now taking pictures with my smartphone is still good enough

[SaabFAN]

I just had enough of modeling and texturing for today and opened the scope to make pictures of the boards.
They are a bit big though, so I had to upload the High-Resolution Pictures on google: https://goo.gl/photos/jmf11acvXfQHRyx69 The pictures attached here have reduced resolution.
Unfortunately, I haven't seen anything that would point towards a hardware-switch for the FFT-Function. The only thing that I noticed is a diode that connects from Pin 9 of D1723 to Pin 11 of D1746.

About the noise: I just recognized that it increases linearly when I increase the V/div. The Rigol doesn't do this, which leads me to believe that the noise comes from either the analog circuits or the Track&Hold-Circuits in front of the ADC.
The Service Manual also doesn't state any noise-standards, but states an error of 2 to 3% to be within spec.

[SaabFAN]

And more pictures.
Again no obvious Hardware-Switch, but as the Philips Engineers decided to put the Trigger Address Comparator (part of the DPU-Control) the MRAM-Unit, I wouldn't be suprised to find the switch on an entirely different board^^

[Stonewerk]

Yes, that was exactly what I was doing on both units to observe the noise-floor of about 1.2mV.
Given that your unit has pretty much the same amount of noise, I don't think that a complete Recapping of the PSU will be necessary any time soon. But I have a lot of SMD Tantalums available in my part-boxes and from the defective scope I know that these caps fit almost perfectly

I agree.... I don't think they'll be any improvement, and the chance of something breaking - becoming broken - rises dramatically   

[Stonewerk]

...
About accessories: I think the only accessory worth getting are probes with a Sensor-Ring to tell the scope if the 10x-Attenuator is active or not.
...

IMO, the probe attenuation display's not a big deal, and you'll likely get along fine without it (i.e., it'll be pretty obvious when readings are 10x or 100x off...) It's set by a fixed resistance between the sensor ring and ground. If you want, I 'll provide details (I need to look it up - I hate quoting off the top of my head and giving wrong values...)

Less for the display, but especially for performance, it's HIGHLY desirable to have probes that are designed specifically to  match the characteristics of the vertical amplifier for which they're matched. That said, it's ridiculously difficult to find '3320 accessories on this side of the pond, unless one's willing to pay out of the traditional orifices... 

[Stonewerk]

Do you have the RS232/GPIB-Manual btw.? I

The part number for the RS232/GPIB interface is (Philips) PM8956A/01 or simply PM8956A (search key). Yes, I have a scan, but I can't remember where I downloaded it from - I thought it was from K04BB, but I can't find it on there, now...

SO I just uploaded the manual to KO4BB. If you can't wait for the document to show up, PM me and I'll happily email you the manual.

[Stonewerk]

I have also thought about using an Arduino to build some kind of a digital Plotter, but stopped that project for the moment, because for now taking pictures with my smartphone is still good enough

Camera/phone etc. is easiest (and quickest), but if you want a plot, it's really easy and you probably already have what you need! Basically, (I'm assuming you're familiar with connecting and configuring RS232 - if you're not - PM me):

Setup Computer:
1. Connect the scope to your pc via the RS232 interface.
2. Bring up a terminal program on your PC - I use Realterm, available on SourcefForge. Configure it to capture data from the port.

The transfer will be HPGL commands, so text capture is fine. I suggest 19.2k 8N1 for the interface parms.

Setup PM3320A:
RS232 interface first:
3. Select Display DISPLAY / OPTION / INTERFACE / RS232-C
4. Config to match PC FRAME, OUTPut Speed, INPut Speed (19k == 19.2k)
Setup Plotter next:
5. Select Memory SAVE/PLOT / SELECT / DIGITAL / PLOTTERS, then choose one of the HP plotters (doesn't matter which...)
6. RETURN / PLOT (This may already be selected by default)
7. RETURN / RETURN (To get back to main SAVE/PLOT menu)

Now you're ready to plot!
On PC:
8. Begin port data capture.
On PM3320A:
9. Select DIGITAL (This is at the top level of the SAVE/PLOT menu, it should display PLOT R0 in the middle of the menu bar)

The plot should begin immediately, with a ***PLOTTER ACTIVE*** at the bottom of the screen, and data should be transferring to the PC.  The scope will display (be patient, this'll take a few minutes) A message will be displayed when the plot is complete.

On PC (when plot has completed):
10. VERY IMPORTANT: End the capture, and SAVE the captured file.

You know have an HPGL file containing your plot. You can view this with an HPGL viewer, such as HPGLView, and or send it to your printer (if it handles HPGL), or do a screen capture from the viewer.
 
The process sounds complicated, but it's really easy and intuitive once you do it... It helps to have the manual, of course....

I've attached a sample plot below - text file from the capture, and a jpg from a screencap of the HPGL viewer. The display is the CAL connected to Ch A. (Quick and dirty - I haven't configured pens, etc.)

[Stonewerk]

...
... terminal program on your PC - I use Realterm, available on SourcefForge. ...
...
....is with an HPGL viewer, such as HPGLView...
...

Sources:
Realterm: http://realterm.sourceforge.net/
HPGL_View: http://service-hpglview.web.cern.ch/service-hpglview/hpglviewer.html

I'm very pleased with the operation of both of these utilities...

[Stonewerk]

... Software, if you can even find some, ...

Controlling the scope through the RS232 interface is relatively straightforward - when you get the manual, it'll become clearer.

I wrote some simple scripts to setup and capture data from the scope via the RS223 i/f, and a utility to convert the data to CSV format to import in Excel. Then I use Excel to perform analysis on the data, such as FFT, etc.

I posted a separate thread on the forum to inquire how others are making use of the IEEE/RS232 interface - no replies, yet, just lurkers...

[Stonewerk]

...
Then I use Excel to perform analysis on the data, such as FFT, etc.
...

I've attached one of my Excel files - this is a 2 channel FFT data analysis, with a sine wave on one channel, and a square wave of the same amplitude and frequency on the other. This is a good example, as it shows how to process data from both channels. There are seven tabs (worksheets) in this file, the first containing the scope display plot, and the last containing the FFT analysis. The sheets in between show the underlying progression... 

I'm posting this to help users realize that they are not limited by the math functions provided on the scope... 

[SaabFAN]

Okay, I'm going to revive this thread to have everything about the PM3320A repairs in one place here.

I'm currently waiting for the first board-revision for my bachelor-project and have some time on hand to do some maintenance on the big boat anchor.

The main problem is: It is drifting A LOT!
Another problem: The GPIB isn't working, but RS232 is running flawlessly (the biggest issue: Get the PC to cooperate ).
Third problem: The clock is wrong several seconds after a few hours and leave the scope off for 3 days, it is out by a several minutes.

The drifting is especially noticeable when the instrument is cold and starts warming up. The high speed mode, which utilizes the CCD-Chips, and the low speed-mode drift apart and also the measured values drift up to 0.7 divisions.
I've used the internal tools to check the alignment of the screen and that stays rock solid, so no problem there.

This means, the problem is sitting on the big main board in the bottom of the unit, which is a really massive board.
By now I've recalibrated the unit about 6 times and it takes only a few days for some drift to become noticeable. The drift from cold to warmed up is there from the beginning, but that's to be expected.

What I suspect to be the source of the problem: The potentiometers. They don't look like they're sealed, so most likely dust and other stuff crept in there over time. Especially because they're right in the airflow-path. They're also really twitchy when trying to adjust them.
Is there some way to clean those with some chemicals/oils, or are carbon track-potentiometers uncleanable and need replacement?

[tautech]

What I suspect to be the source of the problem: The potentiometers. They don't look like they're sealed, so most likely dust and other stuff crept in there over time. Especially because they're right in the airflow-path. They're also really twitchy when trying to adjust them.
Is there some way to clean those with some chemicals/oils, or are carbon track-potentiometers uncleanable and need replacement?

You mean preset pots right ?
Unusual that they impact on drift but yes old ones can be sensitive to adjustments.
When I find this I give them a good workout that helps the wiper contact to the track. Skeletal presets can be a mongrel in this regard and it's not unheard of that replacement is needed.

WRT drift, have you monitored the PSU voltage and ripple as the unit warms up ?
Tired caps ?

[dom0]

carbon track pots can be cleaned with brake cleaner, but not on the PCB. I find it easier to just replace them when I have to get them out anyway.

[bd139]

Interesting thread. I had the predecessor of this unit for a bit, the PM3315. The same CCD capture device in it but was a lot less fancy. Terrible aliasing problems and noise and the mains filter exploded. It drifted too. Never got to the bottom of that. Most complicated scope I’ve ever seen inside. When I got it, the PSU was tripping instantly. Turned out to be one of those blue axial Philips caps on the motherboard. Replaced it with a Vishay/BC unit (direct descendent of the original) and it sprang into life.

Couldn’t stand the bloody thing so got rid of it.

[SaabFAN]

I've spent yesterday troubleshooting the GPIB-Board.
All the electrolytic caps were bad (Capacity about 100µF instead of 220 and between 1 and 4 Ohms ESR), so I replaced them.

Afterwards I ran the built-in tests and discovered that the RAM-Test fails.
So I desoldered both chips and checked them with the TL866 programmer. Both got a clean bill of health. Not sure what's going on there... Serial communication works - I can plot screens and it understands and reacts correctly to the commands i'm sending.
Supply-Voltage on the GPIB-Board is 5.06V, which shows basically no drift from cold to warm.


The other issues, regarding calibration-drift, are probably related to bad caps. Out of curiosity I soldered 330µF caps in parallel to the 100µF caps on the +/-15V Rail on the ADC-Board and got a significant reduction in noise. So I'm going to replace all the capacitors and, if I can find replacements, the potentiometers.

[SaabFAN]

I have replaced all the electrolytic capacitors on the ADC-Board.
Just ONE measured ok (Right capacity, ESR lower than 1 Ohm).

Unfortunately, I managed to install one of the caps (C664) backwards across the negative supply of D616. At first the signal drifted upwards on the screen, but now it is always at the top of the screen and the signal seems to disappear after R719. The Outputs of D616 stay at 1.5V and the ADC sees a constant 7,5V at its input.

I suspect that the wrong negative supply somehow killed the multiplier, which would be quite bad, as I havent been able to find that part anywhere on the net. Not even datasheets are available...

Transistor V621 tested ok and the transistor-tester showed a hfe of about 370.

Would it be possible to replace the multiplier with some other part?

[SaabFAN]

After a day of tinkering, I've managed to get the trace back on the screen - The -14V Rail was gone because a resistor failed open. Most likely because of mechanical stress due to it sitting right next to the Mainboard-Connector.
But there's still no signal going from the CCD-Board to the ADC.

Right now, after unsoldering and socketing the Multiplier-Chip, I'm not so sure anymore that that chip is dead. All the PN-Junctions are ok (0,7V in one direction, open in the other).
I'm not familiar with analog multiplier-circuits, but I think it might be possible that I misinterpreted the measurements and there won't ever be any voltage-change at Pin 17 of D616 because the signal is represented as a current going either into Pin 17 or through V621.
The fact that Pin 15 was constantly sitting at +1,5V might have been caused by the missing -14V-Rail (-14V-Rail was at +2,8V).

I have also discovered that the coaxial cables have seen way better days. The plastic inside feels stiff and can be much easier damaged than the stuff inside the cables I'm used to. I've also discovered that the center connector of the cable sending the CCD-Data to the ADC had broken off and was just barely making contact the entire time. And the shield seems to have never made contact with more than one tiny wire.
So if I'm able to get what's left of the issues with the ADC-Board sorted out, that cable has to go too.

[SaabFAN]

I finally got it working again.

The Multiplier-Chip didn't go bust and the last problem was the coaxial cable that broke about 4cm from the board. 

Now I have to replace the caps on the big board on the bottom and then recalibrate the entire unit. Right now I have 3 divisions difference between direct and P2CCD-Mode.

[bd139]

Keep up the good work  . Sounds like good progress!

[SaabFAN]

Yes, now that I have (kinda) understood how the circuit works, I realized that I chased ghosts for the better part of the day and completely unecessary desoldered the multiplier. 

With a bit of luck, the big CCD-Board and the Managment-Unit, as well as the frontend will react a bit more kindly to the maintenance-work

[Bashstreet]

If not already after visual inspection of the whole scope i would go and measure all the resistors as matter of course in such equipment.
Also i would change if not all at least the smaller caps close to any heat source.

Good quality electrolytics are fine unless you want to spend some money.

Sorry to jump in like this if what i am saying has already been evaluated my apologies.

[macboy]

Interesting thread. I had the predecessor of this unit for a bit, the PM3315. The same CCD capture device in it but was a lot less fancy. Terrible aliasing problems and noise and the mains filter exploded. It drifted too. Never got to the bottom of that. Most complicated scope I’ve ever seen inside. When I got it, the PSU was tripping instantly. Turned out to be one of those blue axial Philips caps on the motherboard. Replaced it with a Vishay/BC unit (direct descendent of the original) and it sprang into life.

Couldn’t stand the bloody thing so got rid of it.

I had never known about the PM3315. That bears little resemblance to the PM3320A. Notably, the PM3320A has aliasing detection that actually works. A warning LED "lights if the sampling frequency is smaller than twice the trigger pulse frequency". Actually, I really like my PM3320A. Clearly, it was a very capable machine way back when, and served me well until I bought a more modern (still >15 year old) Lecroy. I still like its razor sharp CRT and high resolution rendering of waveforms.

Although the PM3320A is rated for 200 MHz and 1.75 ns rise time. I measured mine at 375 MHz (-3 dB point) and < 1.2 ns rise time.  That's nearly double the specified bandwidth! Combined with the 10 GS/s at the fastest time base, it's a capable sampling scope, if you can deal with its limitations (especially 4k sample memory).

I hope SaabFAN is successful in the repair, it will be a useful instrument in the end.