[Luke] isn’t able to declare total victory yet. His LeCroy 9450 oscilloscope repair project has seen some success, though. The glitchy screen seen above is just one of the problems it had, but has now been fixed. When [Luke] got his hands on it, this was one of three screen states: the other two being normal operation or completely dead. Replacing the screen connector was all it took, so he moved on to the second part.
This one is much less trivial. Only one of the two channels works—which might be the point at which many would abandon the repair—but it’s still a fine single-channel scope. [Luke] continued to trouble-shoot by disassembling the bottom of the case and breaking out the device’s schematics. He traced the circuit and found one module that is suspect (and is looking for help finding a replacement). Unfortunately, the problems don’t end there. Another unknown problem is causing erroneous signals on the displayed waveforms. It’s an odd issue but it really feels like he’s close to solving this one!
Go thru all connected boards, modules, and edge connectors with bright light and cleaning tip of cloth on a spatula. Blow with compressed air and inspect fail-prone solder joints like regulator chips etc. Blow in and work all switches and pots while blowing then do it again is necessary with solvent and blow. Disassemble then reassemble, you will get it.
what he said + cold spray is your best friend….
Man does that glitched output look cool. Straight out of a scifi movie.
exactly what i thought.
I’d dump this crappy 80’s era scope and buy a working Rigol 1052E.
The 9450 has 400mhz bandwidth and you can buy them for under $100.
And my Rigol supports USB drives. I win.
2/10 Troll
Gave you 2 points because I looked up the stats of the Rigol.
50MHz is pretty slow and $330 to boot.
You hack it to 100MHz. Fail.
100 MHz < 400 MHz – primary school fail.. loose one point from your 2/10 Troll.
I said a few days ago that there needs to be a ‘Reddit’ style up/down vote system for comments here. And again, we have a prime example of where it would be good!
Krusty, so you can turn it in to a worthless circlejerk hugfest? Because it hurts your fragile mind to simply skip over posts?
^Exactly. Those comment voting systems are worthless.
400Mhz@400MS/s < 50MHz@1GS/s to a user not extremely experienced though. And this tends to be the problem with 20 year old high end DSOs in general. A sample rate not several times higher than the analog bandwidth, MS/s for MHz, means that there just *can not* be the kind of input filter you need to reliably avoid aliasing effects if used as a general purpose scope. Even at 2MS/s per MHz you would probably need a filter of Order n where n is the bit depth to really be sure – and building an 8th order filter that a) does not mess up your waveshapes by introducing frequency dependent delays and b) will not need manual tuning and retuning is not trivial.
That said, the advantage of some of these older DSOs seems to be higher bit depth and quite a lot of cool factor :) – and that you CAN do UHF stuff IF you are aware of the missing antialising filter.
“[ogap] decided it was too much effort to fix what he had, and that learning something along the way was overrated. It was much easier to just buy a new one.”
Yeah – that’s what we’re on hackaday for.
It is sad how some people really don’t get it.
Dump a LeCroy 9450A and go for a Rigol DS1052??
Thats like dumping a luxury Mercedes from the 90’s with some small dents and buying a chinese two seat car (with USB and a option to ‘hack’ it to go 100) instead.
More like dumping a luxury Mercedes from the 90’s with half of the cylinders not working and radio that only plays gospel.
Hey, there’s still a market for that in some parts of the country!
I have a 9450, a 9414, and a 9424. They all share a lot of parts, and one possibility is to pull the input buffer off the EXT input and stick it on the blown channel. All three are the same. (The 9450B had a much tougher input buffer, but that came at the cost of reduced bandwidth.)
Debugging the power supply and video card is way easier if you make a cable that allows you to remove the power supply from the machine so you can get in there while it’s live. You probably already know to be really careful of the high voltage for the CRT. I’ve gotten replacement video cards off ebay, and I believe you should be able to use the card from almost any of the 94xx machines, although the 9400 and 9410 may be too slow, but the ’14, ’24, ’30, and ’50 should all work.
Do you have any solutions for a LeCroy scope that boots, displays everything just fine, but appears to be ‘frozen’? Does not respond to input controls, except for the system reset (and reboot) combo. No response on the serial port either.
Not off the top of my head. The rs232 and gpib seem, from the hardware, like they should be unrelated to the front panel. There’s a front panel processor that’s an obvious place to look… if everything else worked.
I’ll play around with mine a bit and see if I can come up with any ideas.
Service manual is on elektrotanya (And various other places). IIRC serial port only works for a certain range of GPIB adresses (Either above or below 32). Have not had to debug that circuit so I’m not sure.
RE: Frozen LeCroy scope. My 9450 does that, or rather DID, do that. I’ve upgraded the firmware and the problem seems to have gone away. BUT when it froze, I found that the buttons on the RH side of the screen that turn the channel 1 and 2 display still worked and if I press one of them FIRST, that it unlocked the keyboard. Firmware is in ROM and I got it from Ken, the same guy that sells the LeCroy SW Option Enabling GAL chips on E-bay. Seller ID SA77KM.
Reply to my own post. I checked the Lithium backup battery in my 9450 today and it’s down to 1.52 volts. It should be 3 volts. That may have something to do with the keyboard locking up. Even if it doesn’t, that’s still too low IMO. Also Ken says that the 9450 will fail to boot if the battery is too low, so I’ll be replacing mine ASAP! The battery is a Renata 1000-0B and Mouser and Interstate Battery both sell them for $12.99. NOTE: Mouser’s linked data sheet shows a Renata 1000-1B!!!! The -0B and -1B are different so make sure that you get the correct one! Ken tells me that the Calibration Constants aren’t in RAM and that you can safely change the battery without losing them. (Phew!) The battery backed RAM contains the waveforms stored in memory and Setups so you may lose them.
PS the LeCroy service manual says NOT to place the CPU board on conductive antistatic material when it’s out of the scope. They say that it can drain the battery. Frankly I’m more worried about static damage than draining the battery. My advice would be to lay it on a piece of non-conductive material on top of a properly grounded anti-static mat. I would avoid any material such as plastic that could easily generate static. Something like paste-board would probably be a safe choice.
Sometimes works and sometimes doesn’t sounds like bad connector or solder joints to me. Reflowing everything w/ an SMD hot air unit would probably be an efficient way to fix it if it’s not the connectors. Making extender cables so you can lay it out on the bench is the place to start. That will help rule out faults in the internal cables.
Intermittents are a bitch to isolate. Bending boards gently with an insulated stick will often help isolate the solder joint, but with SMD parts that’s tricky to do without causing collateral damage.
Junking a 400 MHz scope just because it’s not working is pretty wimpy in my book. It may be sensible for a business, but a working 400 MHz scope is no small bit of kit.
I hope it wouldn’t get junked in any case: it’s 80% good, and would sell for a reasonable amount on ebay to someone who has any number of other problems and needs repair parts.
Even the math eproms on these things are worth some money if they have the FFT options.
If the Leiden Makerspace would junk that ‘scope, It would get a nice place in my “lab”. (But I don’t think they will, and I have a good oscilloscope already, though it’s not 350Mhz)
I was troubleshooting a Philips O-scope which had excessive noise on the channels. It came down to two chips, both $800. I choose the wrong one…
From reading his blog it looks like the issue was caused by somebody exceeding the ratings of the input on channel one, frying an amplifier and part of the ADC board. Building a new input stage amplifier might be a fun project though!
btw it would be a worthwhile debugging exercise to dig out a 9-to-25 pin serial cable, a 5v FTDI serial-usb cable or buying a $40 gpib-usb, and running the machine via the interface and grabbing screenshots, to make sure it is the video board not the processor that’s all wonky.
Then you also suddenly have a full SCPI-controlled instrument and that’s a significant increase in its capability. A few lines of python and you can start doing custom trigger patterns.
And once you’ve done that, machines with broken buttons and screens are much more useful, and boy are they reasonably priced.
Video board is OK, that was just a loose connector. One of the ADC boards has a problem yet to be found, the analogue input board (9450_7) has a broken HHZ406 that’s now on the ext. trigger ch, and it developed a new problem (Since the blogpost) causing “noTMS” error on ch2. (So I got the working ADC card from that channel, switched it for the one from ch1, et voilà: ch1 works as it should, without the extra pulses/grass/noise. CH2 however is dead now. I hope and suspect it is just one of the connectors to the adc board or trigger board.)
Ahh, can’t edit my comment. I mean the “NoTMS” error is hopefully only a loose connection. The ADC board itself still has other problems.
Was a loose connection indeed! (Scratch trough a PCB trace, probably caused while placing the cover back on. Patched with a wire, error disappeared. Still extra noise on the signal though. )
If it uses charge plates for horizontal scanning, maybe the plates are causing some kind of ESD that has damage the rest of the circuits for channel one (maybe due to over use of those circuits). Just a thought though.