Out of the Dishwasher...
And back into my laboratory. ;) Today I will post the final details of today's toiling with a Hewlett Packard/Agilent Technologies DSO. (Digital Storage Oscilloscope)
Well, here is a small corner of my lab, most of it is off camera and in worse shape. There has been little time to pick up. Besides, my associate gets a little peeved and says, "I was working on that!" whenever I pick something up and put it away. So there is an excuse. Yes folks, it is a professional mess.
One of the attenuator assemblies was bad. An attenuator allows the scope to automatically detect the best possible settings for signal acquisition, whether it gets passed to an amplifier, or tuned down through a 50 ohm or 1 mega-ohm impedance before being pushed through to DAQ and signal conditioning. The chassis contains the high voltage drive circuits, power supply and Trinitron monitor. These were cleaned out the day before after dumping the works into the sink and scrubbing it all out, then left overnight to dry in the open after a bath of 99.9% pure alcohol and compressed air. It's not just clean, it's sterile!
Here is a closeup of an attenuator assembly complete with relay control wires, relays/switches, signal wires and bypasses...
Once the attenuators are in place, it needs to have the front end put back on. There is a fair amount of electronics in it, previously cleaned out and ready to go. The front plate holds the attenuators in, adds structural rigidity and encloses the monitor so that you do not get zapped when you almost touch it to reach in and push the menu buttons along one side and the bottom of the display.
Here is what it looks like, ready to go, and it's signal wires, screen adjustment lines and attenuator caps in place. We now have a fresh chassis to continue building the guts spread out on my floor back into!
The freshly bathed and sealed backplane was left to dry overnight with three coats of automotive top-coat acrylic. Pay attention to two things:
1) Pin alignment - It is so easy to bend and break the gold pins that tie the passive backplane element to the power supply lines on the other side of the scope. Gently press the fingers through their appropriate slots, double and triple checking before pressing it into place.
2) Do not take your hands off of it! This is where you need three hands, seriously. Flipping the unit over onto it's top will hold the plane in place until you can thread at least 4 screws into it and tighten them down. Then you can turn it back on it's end for the next step...
Next you sandwich the video card and RAMDAC board into place over the passive backplane. Again, use a gentle hand and a keen eye to make sure that you do not bend the pass-through pins. They too are made of a very soft gold and easily deform/break. Put in the last of your screws to hold the whole works in place:
Connect the RAMDAC's RGB output to the driver board for the Trinitron monitor, and be sure to put the fan connectors in the same spot they came off of. I am not certain if the unit controls the fan speed for which section it detects needing more cooling, but why take that chance? The big blue bar is the bus interconnect and retainer for the gold pins that tie the backplane, power supply, and video data channels together.
Turn the chassis over onto it's belly again (upright) and remove the aluminum cross-bar from the outside edge. This will make it easier to get the boards back in and as you add them in - connect the coaxial signal and timing generator cables where they belong.
Here is the oscilloscope's timing generator and sync board. Notice the gold connectors even on the coax ends!
These coax cables *MUST* go back into the original places on the boards or it will seriously screw the machine up. It will not be able to generate accurate results at high frequency samples because of the latency already tuned out of the system - the signals travel down wires that are precisely matched and tuned for known signal propagation speeds. So how does one keep things straight? Here's my secret:
Use a magic marker to label not only which board, but which connector it was already on. I marked the top of the connectors in order from left to right, and then put a board number on the side of the connector before removing it, indicating how to orient the connector too! Kills two birds with one stone. You do not want any sharp bends or kinks here as it either destroys the center conductor or causes a signal reflection down the wire to create a ghost signal. Genius is keeping hard shit easy guys, trust me on this!
Now the front end connections and attenuators are just a scramble of wires. You could technically put them ANYWHERE on a signal input DAQ and pre-amp gain board, but then if the control wires are screwed, nothing works right. All of these wires too bear a mark I can use to easily put them right back where they belong.
About half an hour later, making sure that the right boards got back into the right slots Some of them are redundant for every channel. Putting them by accident into another slot will not harm anything, but then the factory alignment/calibration data held in ROM is no longer valid. The scope must be sent back to factory at that stage to get it reset. So use your marker to ALSO label the boards. The one nearest the power supply/monitor was labeled "0" and then counting up, board 8 is all the way to the outside. Half an hour later, pat yourself on the back for a nasty job well done. You have physically re-assembled your bench scope. It's now back up to it's fighting weight of 65 pounds. Install your aluminum cross-bar and secure the wire tie and you are ready to put the case back on it.
Note the GPIB interface...er...HPIB if you are from the Hewlett Packard camp. Yeah, they invented it, but we have improved on it since then. This allows for communications with other intelligent devices to control them as the scope goes through a programmed series of steps for automating bench testing or assembly line work. Or, you can use a PC with a GPIB...rats, did it again - HPIB (SAME THING!) controller to monitor and pull data in real-time from the scope for remote, and automated operation of the instrument. There is some hefty, twin 120mm cooling as you can see, and this sucker eats 700 watts running at full tilt. The stuff on the top is it's traveling paperwork. US Military service records and inventory tags are present and must stay with the unit until it is finally and completely decommissioned, even out of research/civilian use.
Plugged in and ready to go, she sits with AC applied, main breaker closed. Gives it time to settle up as turning it on right away will usually destroy the power capacitors and filters if it has sat un-used for prolonged periods of time. While you wait, I will let you know now that you will need Quicktime installed to see the video clips I will be uploading in a moment...
Okay, got your Quicktime downloaded and ready? Right click on the two links below to "Save Link As" on your local disc. Fire up Quicktime and load the files into it to see what she did on first power up and boot.
First time in god knows how many years this thing was turned on. Shipping Manifest said it was broken!
Yep, it's broken, but only slightly...it works! From complete failure to a few annoying calibration notices - it's a success.
So it's bitching about not having factory calibration and it's all out of whack. No biggie, time to whip out your calibrated signal synthesizer and give it a *known* signal. Try a few different voltages and frequencies. What may look good at lower voltages and high frequencies may be totally out of whack at lower frequencies and higher voltages - keep that in mind. Putz around with it until you get it to lock cleanly onto whatever you feed it. My choice of feeding while on the bench before it's weaned and sent off into the real world again is a Fluke Digital RF synthesizer. She'll do anything from 10KHz all the way up to 1.7 GHz at a range of 1mV all the way out to a robust 2 volts. It does not take much to tweak sensitive digital electronics!
And the end result? Goofing around and pushing the 'scope all the way out to it's designed limits, you are looking at a "broken" piece of instrumentation accurately plotting a synthesized waveform @ 100MHz & 1.8 volts. Not bad for a grand total of about 9 hours of work - for a little investment of piffling things like Torx drivers, alcohol, simple green and a can of compressed air - you can bring a "destroyed" $60,000 machine back from the dumpsters' hungry jaws.
Next week, I am going to be tearing down an older Tektronix Spectrum analyzer that is usually tuned to a specific satellite in the sky. That will be fun. I will grab photos of that work if I can. That's it for now guys, happy tweaking!
Well, here is a small corner of my lab, most of it is off camera and in worse shape. There has been little time to pick up. Besides, my associate gets a little peeved and says, "I was working on that!" whenever I pick something up and put it away. So there is an excuse. Yes folks, it is a professional mess.
One of the attenuator assemblies was bad. An attenuator allows the scope to automatically detect the best possible settings for signal acquisition, whether it gets passed to an amplifier, or tuned down through a 50 ohm or 1 mega-ohm impedance before being pushed through to DAQ and signal conditioning. The chassis contains the high voltage drive circuits, power supply and Trinitron monitor. These were cleaned out the day before after dumping the works into the sink and scrubbing it all out, then left overnight to dry in the open after a bath of 99.9% pure alcohol and compressed air. It's not just clean, it's sterile!
Here is a closeup of an attenuator assembly complete with relay control wires, relays/switches, signal wires and bypasses...
Once the attenuators are in place, it needs to have the front end put back on. There is a fair amount of electronics in it, previously cleaned out and ready to go. The front plate holds the attenuators in, adds structural rigidity and encloses the monitor so that you do not get zapped when you almost touch it to reach in and push the menu buttons along one side and the bottom of the display.
Here is what it looks like, ready to go, and it's signal wires, screen adjustment lines and attenuator caps in place. We now have a fresh chassis to continue building the guts spread out on my floor back into!
The freshly bathed and sealed backplane was left to dry overnight with three coats of automotive top-coat acrylic. Pay attention to two things:
1) Pin alignment - It is so easy to bend and break the gold pins that tie the passive backplane element to the power supply lines on the other side of the scope. Gently press the fingers through their appropriate slots, double and triple checking before pressing it into place.
2) Do not take your hands off of it! This is where you need three hands, seriously. Flipping the unit over onto it's top will hold the plane in place until you can thread at least 4 screws into it and tighten them down. Then you can turn it back on it's end for the next step...
Next you sandwich the video card and RAMDAC board into place over the passive backplane. Again, use a gentle hand and a keen eye to make sure that you do not bend the pass-through pins. They too are made of a very soft gold and easily deform/break. Put in the last of your screws to hold the whole works in place:
Connect the RAMDAC's RGB output to the driver board for the Trinitron monitor, and be sure to put the fan connectors in the same spot they came off of. I am not certain if the unit controls the fan speed for which section it detects needing more cooling, but why take that chance? The big blue bar is the bus interconnect and retainer for the gold pins that tie the backplane, power supply, and video data channels together.
Turn the chassis over onto it's belly again (upright) and remove the aluminum cross-bar from the outside edge. This will make it easier to get the boards back in and as you add them in - connect the coaxial signal and timing generator cables where they belong.
Here is the oscilloscope's timing generator and sync board. Notice the gold connectors even on the coax ends!
These coax cables *MUST* go back into the original places on the boards or it will seriously screw the machine up. It will not be able to generate accurate results at high frequency samples because of the latency already tuned out of the system - the signals travel down wires that are precisely matched and tuned for known signal propagation speeds. So how does one keep things straight? Here's my secret:
Use a magic marker to label not only which board, but which connector it was already on. I marked the top of the connectors in order from left to right, and then put a board number on the side of the connector before removing it, indicating how to orient the connector too! Kills two birds with one stone. You do not want any sharp bends or kinks here as it either destroys the center conductor or causes a signal reflection down the wire to create a ghost signal. Genius is keeping hard shit easy guys, trust me on this!
Now the front end connections and attenuators are just a scramble of wires. You could technically put them ANYWHERE on a signal input DAQ and pre-amp gain board, but then if the control wires are screwed, nothing works right. All of these wires too bear a mark I can use to easily put them right back where they belong.
About half an hour later, making sure that the right boards got back into the right slots Some of them are redundant for every channel. Putting them by accident into another slot will not harm anything, but then the factory alignment/calibration data held in ROM is no longer valid. The scope must be sent back to factory at that stage to get it reset. So use your marker to ALSO label the boards. The one nearest the power supply/monitor was labeled "0" and then counting up, board 8 is all the way to the outside. Half an hour later, pat yourself on the back for a nasty job well done. You have physically re-assembled your bench scope. It's now back up to it's fighting weight of 65 pounds. Install your aluminum cross-bar and secure the wire tie and you are ready to put the case back on it.
Note the GPIB interface...er...HPIB if you are from the Hewlett Packard camp. Yeah, they invented it, but we have improved on it since then. This allows for communications with other intelligent devices to control them as the scope goes through a programmed series of steps for automating bench testing or assembly line work. Or, you can use a PC with a GPIB...rats, did it again - HPIB (SAME THING!) controller to monitor and pull data in real-time from the scope for remote, and automated operation of the instrument. There is some hefty, twin 120mm cooling as you can see, and this sucker eats 700 watts running at full tilt. The stuff on the top is it's traveling paperwork. US Military service records and inventory tags are present and must stay with the unit until it is finally and completely decommissioned, even out of research/civilian use.
Plugged in and ready to go, she sits with AC applied, main breaker closed. Gives it time to settle up as turning it on right away will usually destroy the power capacitors and filters if it has sat un-used for prolonged periods of time. While you wait, I will let you know now that you will need Quicktime installed to see the video clips I will be uploading in a moment...
Okay, got your Quicktime downloaded and ready? Right click on the two links below to "Save Link As" on your local disc. Fire up Quicktime and load the files into it to see what she did on first power up and boot.
First time in god knows how many years this thing was turned on. Shipping Manifest said it was broken!
Yep, it's broken, but only slightly...it works! From complete failure to a few annoying calibration notices - it's a success.
So it's bitching about not having factory calibration and it's all out of whack. No biggie, time to whip out your calibrated signal synthesizer and give it a *known* signal. Try a few different voltages and frequencies. What may look good at lower voltages and high frequencies may be totally out of whack at lower frequencies and higher voltages - keep that in mind. Putz around with it until you get it to lock cleanly onto whatever you feed it. My choice of feeding while on the bench before it's weaned and sent off into the real world again is a Fluke Digital RF synthesizer. She'll do anything from 10KHz all the way up to 1.7 GHz at a range of 1mV all the way out to a robust 2 volts. It does not take much to tweak sensitive digital electronics!
And the end result? Goofing around and pushing the 'scope all the way out to it's designed limits, you are looking at a "broken" piece of instrumentation accurately plotting a synthesized waveform @ 100MHz & 1.8 volts. Not bad for a grand total of about 9 hours of work - for a little investment of piffling things like Torx drivers, alcohol, simple green and a can of compressed air - you can bring a "destroyed" $60,000 machine back from the dumpsters' hungry jaws.
Next week, I am going to be tearing down an older Tektronix Spectrum analyzer that is usually tuned to a specific satellite in the sky. That will be fun. I will grab photos of that work if I can. That's it for now guys, happy tweaking!