LeCroy 9384L repair (+Parts request)

[shakalnokturn]

I bought this scope on eBay from Germany about a year back.
It was selling as not working and came with the HD01 option, having seen Dave's video I figured it should be an easy repair 
Due to really poor packing (large sealed-air pouches only, in a much too large box) it arrived with a well shattered plastic case, fortunately the front panel survived thanks to the dedicated cover.
While collecting the bits and disassembling the remains I found a gold plated pogo-pin floating around so half expected that it sprung its way into the scope and caused a short on the acquisition PCB.

So I finally gathered some courage to at least test the electronics.
Before powering it up I had a peep in the PS9384 PSU, no burn marks or debris, a close look at the PCB's showed that the Nichicons hat started oozing. It was caught early enough to not be a significant mess. All secondary side electrolytics were replaced on the Power Systems Inc. PSU, and my "Save the hybrid" ® mod. applied. Also checked start-up and bleeder resistors, removed auto voltage selector assembly.

On the Integrated Power Designs PSU, all electrolytics were replaced except for the primary bulk one, TO220 diodes were tightened a little harder.

Lay out on the dining table the scope came up on first attempt, after 3 min. while I was trying to bring all 4 traces on screen there was a sudden crackle in the PSU, scope went dead...

A quick peep showed the 1A fuse (after diode bridge) on the Integrated Power Designs PSU had blown.
The IRF830 was dead, the UC3844 survived though, a closer look showed dust and capacitor electrolyte build up under the MOS, opto and TL431 despite having been blown through with an air compressor.

After better cleaning and reassembly it was running again. All unshielded channels seem to at least be responsive to the CRT flyback EMI.
I've checked / readjusted supply voltages.
RTC battery needs replacing...
Next step is to reassemble it and actually use it to check for hidden faults, for now that means sacrificing my working  9344 for the case 

So I'm either after a spare case at a low price or I'll have 9344 parts for sale soon.

[scopeman]

Nice work Paul!

Sam
W3OHM

[SiliconWizard]

I have a 9354AL and I like it (apart from being pretty chunky and loud.)
I beefed it up to 64MB of RAM (the max I think) and enabled all software options through a programmed GAL. Changed the fan for a much quieter one, but this is still a bit loud...

One part that is still in the works is the replacement of the CRT with an LCD color screen.
Found out that you can get color video output from the video chip (other people have done that), so I soldered wires on the corresponding pins, linked that to an FPGA board, wrote code... so far it encodes compressed video and sends it through USB. That part is finished, but I want to directly drive a TFT panel with that, and that project has been put on hold for quite a while...

[shakalnokturn]

Now reassembled in the 9344's case which is unfortunately brittle and cracking around the handle, I've tried glue for PVC drain pipes, that seems to work better than cyanoacrylate or epoxy.
Both scopes have screen burn but the 9384 was the worst so it has inherited the 9344's front panel and CRT too.
Funnily on first power-up after reassembly it restarted 3 times on the run at the point where it starts displaying the traces, it hasn't shown this behaviour since. The error log shows 3 FP_handle  errors...

Nice work Paul!

Sam
W3OHM

Thanks Sam. Funnily I was expecting a lot more troubleshooting, I'm not complaining but it's a little disappointing in a way that the previous owner got rid of the scope for a handful of electrolytics.
It's a shame there is no external access to the -2V adjustment on the PSU, service manual shows there should be though.

One part that is still in the works is the replacement of the CRT with an LCD color screen.
Found out that you can get color video output from the video chip (other people have done that), so I soldered wires on the corresponding pins, linked that to an FPGA board, wrote code... so far it encodes compressed video and sends it through USB. That part is finished, but I want to directly drive a TFT panel with that, and that project has been put on hold for quite a while...

So you're working on that too, I'm not, I know of four different people who have looked into LCD and/or colour display for the 9300, none of which have shared their findings publicly.
How did you go about the colour information? When I probed around I found that it would be easy enough to give each trace a different colour but more complicated to attribute a given colour permanently to a given trace. The different trace video signals seem to be allocated by order of activation on screen rather than by trace number.
I possibly missed a quite obvious simpler approach though. When I looked into it all the interesting signals (except for pixel clock) could be found on Vision-C GAL

[SiliconWizard]

One part that is still in the works is the replacement of the CRT with an LCD color screen.
Found out that you can get color video output from the video chip (other people have done that), so I soldered wires on the corresponding pins, linked that to an FPGA board, wrote code... so far it encodes compressed video and sends it through USB. That part is finished, but I want to directly drive a TFT panel with that, and that project has been put on hold for quite a while...

So you're working on that too, I'm not, I know of four different people who have looked into LCD and/or colour display for the 9300, none of which have shared their findings publicly.

I'll have no problem sharing that information once I get some time to.

How did you go about the colour information?

As I mentioned, the video chip itself puts out digital color information! It's just combined as a monochrome signal using a single DAC on those scopes, but you can get the digital signals directly on the video chip. It involves soldering tiny wires to the corresponding pins on the video IC. Each pixel is encoded as a 12-bit word containing color information, and basically different parts of the display have a different color. Each of the 4 traces have a different color, so there's nothing to do once you get the pixel data out. Actually, the video chip is probably shared with some scopes of the LC series (I didn't check), or if not, they were at least planning on designing a color scope when they designed this video chip. It's a custom Lecroy ASIC.

You get the pixel clock, HSYNC, VSYNC and 12-bit pixel data from this IC, so it's then a matter of re-encoding that to a usable format. One of the issues is that the video is scanned from bottom to top and then left to right - so you need to "rotate" the image. That's why you need an FPGA and some memory to do this.

I'll attach a capture of an image my system was able to get (yes the project dates back a while...) - as I mentioned, it was encoded and transmitted through USB on a PC, but could stream in real time. Since the image is rotated, for feeding a "normal" display directly, it requires a significant amount of memory.

[edpalmer42]

I like that color screen!

I wondered why there were color settings / tests in the Maintenance Menu.