If you are serious on to repair this DSO, here is my suggestion: You got the same model twice? ok. good. measure the third channel ADC on a working one - pin by pin with a mixed signal analysis equipment and see what´s going on and then compare it to the faulty one. May you have to do it on the ram-chips as well, but as jet i do not think so, if the ram is faulty you would probably get this issue at all channels. As the i think the storage and read out order is random within the RAM. They might used DMA for feeding data into the ram and the read out is doing seperately thus the ram would be assigend to a particular channel but this thought would may to far out for this thing.... Inspecting the ADCs would may help in the first place....
Swapping modules just to test would be a nightmare, every tab would have to be unsolder then bent up. I be happy with this knowing channel 3 did not work, even with 3 channels it still 1 more then I have already.
Dave - Before trying to swap parts, desolder the differential output pair between the hybrid and the main ADC, then try to jumper from another hybrid input. You should be able to narrow the fault down to the hybrid or not. The signal will probably look piss poor due to the jumper cables, but if the channel 3 distortion goes away or stays, that’s all you need to know.
Dave... Try feeding channel 3 and another channel with the same signal, and using another scope, probe the hybrid's differential outputs on both channels, just to get sure the hybrits are OK / faulty. Otherwise, looks like a service call directly at Tektronix. Personally, looks more of an ADC problem. Outside of the noise surrounding the waveform, the waveform is very stable. Happy new year 2014 !
It's common to put bypass caps on the bottom of the board near or under the chip to keep traces as short as possible. I've done that in many amplifier designs.
I'd look at the sample SRAM. Those errors look for all the world like bit errors, and the way they move or drop out completely as you change acquisition mode implies that the RAM is involved.
18:58 the black squares are resistors. I'd say its not a layered capacitor since you can see the laser trim lines. I bet the manufacturing process would be very interesting of those hybrids -- specifically how do they mass manufacture them while keeping price down. I'd bet my money on a dry solder joint that's likely electronically floating -- maybe why you're seeing patterns due to nearby data lines. if that doesnt work I'd say something cooked
The black rectangles are indeed resistors. The manufacturing process is quite simple. The resistor material starts off as an ink that is screen printed onto the ceramic substrate (after the conductor tracks have been printed). The ink comes in different ohmic values and by adjusting the shape you can get close to the desired value. The substrate is fired to fix the printed layers in place. Then the resistors are laser trimmed to the final value. One advantage of these resistors is that the completed hybrid can be actively trimmed to get the correct output. The hybrid in the scope does not look as though it would cost that much, maybe tens of £s.
I have one of these scopes at work, so I was excited to see the innards. Those two wires coupled from the mains to the main board are probably for "AC LINE" trigger source. The power supply probably doesn't rely on the main board for 50/60HZ detection.
At 24:00 in the video, yes, that IS a laser trimmed part. This is because the resistor elements are printed right on the board. Never, ever disturb those!
Before doing anything intrusive get another scope out and start probing, comparing working channels with the faulty one using the same test signal. I'd start on the ADC inputs first and work forwards or back from there depending on results. Always best to narrow things down first before spending a of time desoldering and swapping things. Saves spending 4 hours swapping something that wasn't at fault. Totally agree about earlier comments on digital noise from somewhere.
I have one idea, to check if it is and ADC error: As fare as I remember, the four ADCs are multiplexed, when you are using the highest sampling rate. So, when you are using just one channel and a fast time base, it uses two or even all four ADCs for sampling that one channel. If it is an ADC fault, that should give an interesting pattern even when using channel 1.
Dave's said it before, he knows there will be loads of us shouting at the monitor. But that's his off the cuff style that we all love. Scoping the differential outputs of the Ch.3 hybrid and those of another channel with an identical input signal would show whether the fault was in the hybrid or on the main PCB. De-soldering a known GOOD module from another channel is NOT the way to go. Coming back to an unusual fault with a "fresh head" works wonders!
Put a triangle or a square in to CH2 and CH3 and see if the hybrids are putting out the same, than measure the ADC inputs than outputs to pinpoint where its coming from
One problem with solder and SMT is cracks. If you had a cracked connection on a bypass capacitor, you might get a noise-burst waveform like you showed at the beginning of the video. You have to put all connections under at least 10x magnification and check it out.
As mentioned already... I know it's a PITA but I think I'd try swapping over one of the other three hybrids for the suspect one too just to see if it's that. The thing then though is it's probably about a billion quid for a new one, unless you have a donor scope there amongst all the others you have. Great vid
Happy New Year from North Dakota, U.S.A. -22 degrees F here tonight and getting colder as the night progresses. Hope your weather is a bit warmer. Love your channel.
MAXTEK I believe is "Maxim Tektronix". I took apapart a TLA-7M3 logic analyzer modules years ago and some ICs in the same position were branded "Maxim" on one module and "MAXTEK" on the other.
Hi Dave. the hybrid circuit does signal processing and a fault here would be random noise, so not hybrid module fault. the errors appear above and below wave form full display not just in wave form. This would suggest stuck data bit in video display module part of the ADC i.e. you get full scale "noise" regardless of wave form height. Best wishes for new year Barry
Happy new year! Feed the same simple signal on channel 3 and on another channel and compare the output and the input signals of the ADCs. You'll need another fast oscilloscope but I remember that you own one (if you don't borrow one).
What about the Maxtek chip on the hybrid board? Anyone noticed it is BGA connected to that ceramic board? I know first hand those BGA chips do have problems, especially if the ceramic board bends like a bi-metal heating element with different thermal expansion coefficients.... But I don't know what can it do to the input signal to make it so noisy as we see through the ADC of the third channel... What I would suggest if you decide to check on this, is to , well gently since there is a ceramic board underneath, press on the Maxtek chip with a long plastic-rubber tipped thing as not to induce noise while it is on and see if anything changes. Better results can be observed if it was actually on for a while and the chip is warm. Just thought I'd share this, 'cause I thought about it and don't know if anyone debated it. I also would of at least opened it for a quick inspection if I had such a problem, so thanks for sharing. Sharing is caring. :)
Disconnect the differential output and jumper it across to a known good input. No, it won't be controlled impedance and won't "work", but it should rule out the hybrid. I would be investigating some of the data lines from the ADC to the acquisition ASIC / PSU to the ASIC. Good luck!.
Hey Dave, have you tried to measure the differential output of the hybrid module to the ADC with another scope? If the output is clean, you are sure, that the noise is comming from the ADC or the digital part.
Thank you Dave, even your handle removal helped before total frustration and nuking occurred. And a TDS3034B is now enjoying a normal versus intermittent display. Any interest in making a video upgrading Tek TDS3014 and TDS3034's to TDS3054's?
Try to desolder the connections on the hybrid module (the controlled impedance track) between the front end and ADC, and see what happens, if the problem is still there, swap out the ADC chip on channel 2 & 3
Thats my idea upon the swapping is of chips. If it is confirmed that there are no errors in the hybrid modules ALL the following digital logic can be faulty (ADC, fpga/asic or sample memory). To make sure it is not the adc, shortcut the input and look if there is "noise" on a significant bit (assuming parallel output adc) and draw conclusions from there.
Not sure if it's the fluff you are talking about later in the video but at 20:43 on the right EE9842 there is something metallic between the last two visible pins on the right.
There is a big inductor (switching or filter?) close to U660 (channel 3 memory?) Bad decouplers under U660 or U670 depending on which is channel 3? Also check solder joints on these. Happy New Year!
it's interesting that TDS3054 has 5GS/s on all channels, but the persumed-to-be-ADC's packaging as well as heat management doesn't seem to support such high sampling rates... Given that these scopes has a record length of only 10K pts, I am guessing they are not ADCs but CCDs, as in old TDS644A. The rectangle chip in the middle should be the low-speed ADC.
I don't think that a solder joint would do this, if so, hitting the panel should make some difference. I would rather stick a scope in it, go through the front end, and check some signals. Maybe there are some pictures of signals in the service manual, of what should be there.
Dave put an oscilloscope on the output of the differential pair you pointed out or on the input of the chip of two hybrid modules to compare a common signal. It will effect the signal for sure but it will eliminate the source of the error as either before and after the hybrid module. I would focus of the front end either in the hybrid module or even before it as the error before the ADC. Some kind of over voltage or component failure is causing some random noise that the ADC is misinterpreting. That's my two cents.
There was a American company back in the 80's called Gigabit Logic. Everything they made was GaAs, and very fast. Faster than ECL, too. But their product line was limited. Eventually, Tektronix bought the company so they could have in-house high speed custom GaAs parts. The front-end hybrid is most likely a GaAs custom part. The stepped copper on the board is a microwave type part. Remember this is a 400MHz scope, and the converters must run far more than the Nyquist rate. This means the hybrids need clocking at microwave frequencies.
Dave please leave the hybrid modules in place, they are to fragile. If you really want to test if it is in the hybrid modules (which I higly doubt) just swap the differential analog output with a neighboring channel (only 4 wire jumpers)! Yeah there will be a bit more noise and you have to interrupt the lines and make jumpers but it is so much saver than risking to damage 2 hybrid modules.
whayden001 tony abbot, our beloved prime minister(sarcasm, and what you would call a conservative politician), scrapped the nbn(national broadband network) plan to run fiber to the home (at 100mbps) to revolutionize our communication and education infrastructure and stop telecom company monopolization(cough.. telstra) and instead opted to run fiber to the node with a quarter speed to save money, and demonstrate to the rest of the world what a moron he truly is, as aptly demonstrated by the original comment.
alberto gordillo Why would a politician have the say in what telcom companies do? Are the telcom companies reliant upon government money? If so shame on them. I personally feel that government's job is to get out of the way of private businesses. Regulate as necessary and minimally, but get the f' out of the way. Let the market handle itself. We have the same problem and the problem is when government is involved so to are the self interests of politicians. When companies do things on their own they do it to make profits and profits means making something people "want" to buy assuming the people have a "free" choice in what they are buying. Which is the huge debate about Obamacare in the US. What a load of bulls... that is. It isn't mean to help anyone, but those who don't do for themselves and politicians.
politicians don't really have a say in what telecom companies do, which is why we are behind in telecommunications infrastructure compared to the rest of the world and being charged ridiculous prices for basic internet access - here in Australia we get on peak off peak access metered, and we also have our uploads metered against our on peak off peak quota, the government privatizing the telecommunications company telecom had allot to do with some of the woes Australians experience, the government created a giant in a place where there was little competition, so now the notion of free-marketing turns into a monopolized dictatorship by one telecommunications company that owns pretty much all the core infrastructure anyways not too sure on the whole story but you should research a bit before shooting off, some things here in Australia seem to be opposite of what happens in america, that being our liberal party is conservative and our conservative is liberal... i know its fucking stupid lol. the nbn is a voted in government initiative to bring broadband to rural areas, educational institutions and the masses, to put to rest the aging expensive to maintain copper network that telstra runs and is charging us so much to use and replacing it with modern fibreoptics that will future proof our communications network into the foreseeable future and take market control of our core infrastructure away from telstra by obsolescence of their network, the bandwidth is provisioned out to these companies below cost and wholesale for the various telecom companies to provision to their clients.
with telstra a private company, its not as simple as making people want to buy something, many of us cant afford decent internet access, especially in rural areas where 25gb of internet access a month through wireless can cost upto 70 a month with uploads and downloads both counted toward the quota and it being split between on peak and off peak times... we are almost better off with dial up. we have something i think similar to obamacare in Australia its called medicare, and i can tell you its the single most best thing that this country has to offer, it is expensive but it is worth it.
I am a newbie hobbiest electronics person so.... I think with the scope being in a military setting that its possible they were testing Daves flux capacitor design and messed up the AC/DC... I mean ADC by not using one of Daves MFC (Micro Flux Current) devices designed to test these types of capacitors. Seriously.... I wonder about injecting a known signal into all channels and start probing with a different scope to look for differences. Looks like something on that channel may not be decoupled and letting in noise? It would be interesting how the adc itself would produce these results in theroy.
You could swap the hybrid modules Channel 2 & 3. That is a lot of work though. It's probably not worth it economically but from a curiosity stand point it might be interesting. Thanks Thumbs Up.
Hm, the thing that strikes me as odd is that the waveform disappears at certain timebases. What could possibly do that? I also think Davros Idea was pretty clever. It would at least help to pin down the problem!
Is it safe to use a heat gun and reflow some of those chips, or is it too much of a risk, and just sell the item as having one non functioning channel?
The only thing I can think of is to hope that the data interfaces are parallel and try to find the offending bit on the ADC output. Maybe the waveform could give you some clues as to root cause.
Way too much pointless criticism. Boards are supported perfectly when the case is in place. Channel failure is usually caused by someone pumping in a voltage or signal way out of spec. I bought one of these brand new about 10 years ago, and never once had a problem with it.
Probe the output of that hybrid module? Is it trowing a clean signal or some sort of noise getting in from somewhere? Very unlikely but not hard to check.
you could probably swap the channel 3 front end hybrid with one from another channel. if it works, then you know the problem is in the front end. however, i think it's more likely that the problem is in the ADC . that one would be harder to swap though.
The heatsink is most likely tinned copper one, I have seen those used in old computers ATX power supplies. My power supply (old ATX from 1993) has Nippon Chemicon capacitors. Have you checked the passive components? Maybe some resistor is burned out or capacitor cracked and doesn't filter the noise any longer on that channel?
I'd run it open if you can and put pressure on the module and chips. If something is loose, it should resolve or get worse on the screen. Reflow the hybrid if not.
Hybrid Thick Film 'Screen Printed' Multilayer Module. The black parts are printed resistors laser trimmed to value. Very 80's. Nice to see the old school stuff. Love Tektronix gear. It's bananas. :)
I think its interesting that the artefact seems to extend up or down from the waveform differently depending on the waveform's vertical position. i.e. if its all in the top half it goes up from the waveform, if it straddles the midline, it extends both ways. Would that suggest its further along than the analog front end?
Dave, did you run the system diagnostics? They may not be more extensive than the POST, but it is worth trying. Also, there is an error log. They are under Utility/System Config/Diags.
I would first measure the balanced signals going from hybrid to ADC to see if they are clean and the DC bias is similar to other channels. Next I would go through the pins of the ADC and compare to other channels. But most likely there is nothing much to do as one of the chips is likely broken. With good luck some of the bypass capacitors have gone high ESR introducing some glitches.
I'm thinking its a bad capacitor. I would compare the values of the surface mount caps on the hybrid to one that is functioning correctly as well at the values for and continuity of the traces on those boards.
Well that looks like a pain in the *** well either swap the hybrid modules, or try to feed the adc on ch3 with the signal from a known good channel ch1-2-4 good luck dave :)
Why not scope the differential output going into the adcs? compare the frontend output from channel 4 and 3 with same signal and see if you can see any significant differences?
EEVblog Ah I see. Thanks for quick reply. I guess if that's not the issue maybe start swapping parts (adcs)? Seems like a manageable task without too much trouble. Hope you can fix this one up though. Not necessarily to gain a bit more functionality but for youtube videos sake. The last attempted repair with the failed asics(i think) was a disappointment. Understandable since it wasn't an economical repair but none the less. I like to see successful repairs as i'm sure everyone else does too.
Desolder frontend and adc modules in channel 3 and 4, and put them in different order. To me it looks one of the adc digital outputs is wrong (either bad solder joint, or bad decopuling cap, or getting some high frequency noise from analog section). Maybe on adc3 or the glue chip in the middle where it supposedly goes. Also what happens when you ground input 3? With 0V DC do you also see high frequency noise on it? Does the strange signal (best viewed in dot mode), also happens on all vertical scales? (checking small and big signals should rule out issues with noisy relay)
Happy new year! Love this channel and have done for over a year now. Retro teardowns are a favorite but I'm easily pleased so anything goes. My Dad worked for Motorola and still hasn't acknowledged the phone teardown (with it's false aerial). Makes me laugh anyway :) Keep it going for 2014! Big thumbs up from Scotland :D
Can it be lead-free solder joints under the MAXTEK Circuit that has got whiskers between them? Then it might help to blow hot air on the Circuit and heat it up to 180°C. (I repaired a MacBook Pro in this way.) I Think Tektronix use lasers to calibrate the hybrids. They use the laser to cut away parts from carbon trases etc. There are other marks in the hybrid that also looks like burn marks from lasers. But I can't say how they work.
It's a Tek, so cause may be very difficult to diagnose (because Tek's visible quality is so high). Forget it ... move on ...you still have other good channels, so the 'scope is very usable.
It is not worth to swap the channels if you don't have one to replace it anyway, you risk the possibility of breaking some thing else and then the whole scope will be crap, Murphy will get you again. You can also donate it to me, with only 3 working channels........
i petition my ignorant comment and say its a tiny bit of conductive dust between a pair of pins of the ADC, if you look above what is likely the channel 3 ADC there is a lot of dust. its just a guess and im no expert
What happens when you change the coupling mode (DC, AC and GND)? The problem happen only in DC mode? I don't know if in GND mode the RAM memory is used, but if it's and the problem don't appear I think that memory isn't the problem. Good luck. Happy New Year Dave.
I have this kind of oscilloscope. Yesterday something went terribly wrong with the screen. It looks like the sync is broken in some way. I only get random lines and blocks either horizontal or vertical. Any idea if it's possible to fix it?
Haha! I just solved it. I was going to make a video and post to possible get some help on the problem. Just of frustration i bumped the scope in the table while it was starting up and the sync came back :)
If you are serious on to repair this DSO, here is my suggestion:
You got the same model twice? ok. good. measure the third channel ADC on a working one - pin by pin with a mixed signal analysis equipment and see what´s going on and then compare it to the faulty one.
May you have to do it on the ram-chips as well, but as jet i do not think so, if the ram is faulty you would probably get this issue at all channels. As the i think the storage and read out order is random within the RAM. They might used DMA for feeding data into the ram and the read out is doing seperately thus the ram would be assigend to a particular channel but this thought would may to far out for this thing....
Inspecting the ADCs would may help in the first place....
Swapping modules just to test would be a nightmare, every tab would have to be unsolder then bent up.
I be happy with this knowing channel 3 did not work, even with 3 channels it still 1 more then I have already.
yes, i don't have an oscilloscope! Still great to watch the video :)
it definitely looks like digital noise getting into the analog section. Almost as if one of the decoupling caps are open.
feed same signal to 3 and 4 and compare the two with another scope is all i can think off without removing stuff.
Dave - Before trying to swap parts, desolder the differential output pair between the hybrid and the main ADC, then try to jumper from another hybrid input. You should be able to narrow the fault down to the hybrid or not. The signal will probably look piss poor due to the jumper cables, but if the channel 3 distortion goes away or stays, that’s all you need to know.
Cross change the hybrid modules and see if the error moves to another channel...
And maybe the ADC chips if that doesn't work?
Try to reflash the Firmware. Maybe it's Software, a flipped bit in the storage may cause some wired issues.
Dave... Try feeding channel 3 and another channel with the same signal, and using another scope, probe the hybrid's differential outputs on both channels, just to get sure the hybrits are OK / faulty.
Otherwise, looks like a service call directly at Tektronix.
Personally, looks more of an ADC problem. Outside of the noise surrounding the waveform, the waveform is very stable.
Happy new year 2014 !
The floppy is floppy.
Spot on!
It's common to put bypass caps on the bottom of the board near or under the chip to keep traces as short as possible. I've done that in many amplifier designs.
I'd look at the sample SRAM. Those errors look for all the world like bit errors, and the way they move or drop out completely as you change acquisition mode implies that the RAM is involved.
18:58 the black squares are resistors. I'd say its not a layered capacitor since you can see the laser trim lines.
I bet the manufacturing process would be very interesting of those hybrids -- specifically how do they mass manufacture them while keeping price down.
I'd bet my money on a dry solder joint that's likely electronically floating -- maybe why you're seeing patterns due to nearby data lines. if that doesnt work I'd say something cooked
The black rectangles are indeed resistors. The manufacturing process is quite simple. The resistor material starts off as an ink that is screen printed onto the ceramic substrate (after the conductor tracks have been printed). The ink comes in different ohmic values and by adjusting the shape you can get close to the desired value. The substrate is fired to fix the printed layers in place. Then the resistors are laser trimmed to the final value. One advantage of these resistors is that the completed hybrid can be actively trimmed to get the correct output. The hybrid in the scope does not look as though it would cost that much, maybe tens of £s.
I have one of these scopes at work, so I was excited to see the innards.
Those two wires coupled from the mains to the main board are probably for "AC LINE" trigger source. The power supply probably doesn't rely on the main board for 50/60HZ detection.
At 24:00 in the video, yes, that IS a laser trimmed part. This is because the resistor elements are printed right on the board. Never, ever disturb those!
Before doing anything intrusive get another scope out and start probing, comparing working channels with the faulty one using the same test signal. I'd start on the ADC inputs first and work forwards or back from there depending on results. Always best to narrow things down first before spending a of time desoldering and swapping things. Saves spending 4 hours swapping something that wasn't at fault. Totally agree about earlier comments on digital noise from somewhere.
I have one idea, to check if it is and ADC error:
As fare as I remember, the four ADCs are multiplexed, when you are using the highest sampling rate. So, when you are using just one channel and a fast time base, it uses two or even all four ADCs for sampling that one channel.
If it is an ADC fault, that should give an interesting pattern even when using channel 1.
Dave's said it before, he knows there will be loads of us shouting at the monitor.
But that's his off the cuff style that we all love.
Scoping the differential outputs of the Ch.3 hybrid and those of another channel with an identical input signal would show whether the fault was in the hybrid or on the main PCB.
De-soldering a known GOOD module from another channel is NOT the way to go.
Coming back to an unusual fault with a "fresh head" works wonders!
Put a triangle or a square in to CH2 and CH3 and see if the hybrids are putting out the same, than measure the ADC inputs than outputs to pinpoint where its coming from
One problem with solder and SMT is cracks. If you had a cracked connection on a bypass capacitor, you might get a noise-burst waveform like you showed at the beginning of the video. You have to put all connections under at least 10x magnification and check it out.
As mentioned already... I know it's a PITA but I think I'd try swapping over one of the other three hybrids for the suspect one too just to see if it's that.
The thing then though is it's probably about a billion quid for a new one, unless you have a donor scope there amongst all the others you have.
Great vid
That floppy drive lives up to its name! :P
Damn, I missed that pun!
It’s a flapper lmao
Happy New Year from North Dakota, U.S.A. -22 degrees F here tonight and getting colder as the night progresses. Hope your weather is a bit warmer. Love your channel.
swap modules from the other duff oscilliscope ?
Even simpler, just swap modules between channels.
MAXTEK I believe is "Maxim Tektronix". I took apapart a TLA-7M3 logic analyzer modules years ago and some ICs in the same position were branded "Maxim" on one module and "MAXTEK" on the other.
Hi Dave. the hybrid circuit does signal processing and a fault here would be random noise, so not hybrid module fault.
the errors appear above and below wave form full display not just in wave form.
This would suggest stuck data bit in video display module part of the ADC i.e. you get full scale "noise" regardless of wave form height.
Best wishes for new year Barry
Happy new year!
Feed the same simple signal on channel 3 and on another channel and compare the output and the input signals of the ADCs. You'll need another fast oscilloscope but I remember that you own one (if you don't borrow one).
What about the Maxtek chip on the hybrid board? Anyone noticed it is BGA connected to that ceramic board? I know first hand those BGA chips do have problems, especially if the ceramic board bends like a bi-metal heating element with different thermal expansion coefficients.... But I don't know what can it do to the input signal to make it so noisy as we see through the ADC of the third channel... What I would suggest if you decide to check on this, is to , well gently since there is a ceramic board underneath, press on the Maxtek chip with a long plastic-rubber tipped thing as not to induce noise while it is on and see if anything changes. Better results can be observed if it was actually on for a while and the chip is warm. Just thought I'd share this, 'cause I thought about it and don't know if anyone debated it. I also would of at least opened it for a quick inspection if I had such a problem, so thanks for sharing. Sharing is caring. :)
Disconnect the differential output and jumper it across to a known good input. No, it won't be controlled impedance and won't "work", but it should rule out the hybrid. I would be investigating some of the data lines from the ADC to the acquisition ASIC / PSU to the ASIC. Good luck!.
Yay for repair-oriented teardowns!
Hey Dave, have you tried to measure the differential output of the hybrid module to the ADC with another scope? If the output is clean, you are sure, that the noise is comming from the ADC or the digital part.
18:12 Laser-calibrated components, the black mark is where the laser cut the connection.
Thank you Dave, even your handle removal helped before total frustration and nuking occurred. And a TDS3034B is now enjoying a normal versus intermittent display.
Any interest in making a video upgrading Tek TDS3014 and TDS3034's to TDS3054's?
Try to desolder the connections on the hybrid module (the controlled impedance track) between the front end and ADC, and see what happens, if the problem is still there, swap out the ADC chip on channel 2 & 3
Thats my idea upon the swapping is of chips. If it is confirmed that there are no errors in the hybrid modules ALL the following digital logic can be faulty (ADC, fpga/asic or sample memory).
To make sure it is not the adc, shortcut the input and look if there is "noise" on a significant bit (assuming parallel output adc) and draw conclusions from there.
Very tough repair, Dave!
Not sure if it's the fluff you are talking about later in the video but at 20:43 on the right EE9842 there is something metallic between the last two visible pins on the right.
There is a big inductor (switching or filter?) close to U660 (channel 3 memory?) Bad decouplers under U660 or U670 depending on which is channel 3? Also check solder joints on these. Happy New Year!
11:00 a flapping Floppy Drive... brilliant!
it's interesting that TDS3054 has 5GS/s on all channels, but the persumed-to-be-ADC's packaging as well as heat management doesn't seem to support such high sampling rates... Given that these scopes has a record length of only 10K pts, I am guessing they are not ADCs but CCDs, as in old TDS644A. The rectangle chip in the middle should be the low-speed ADC.
I don't think that a solder joint would do this, if so, hitting the panel should make some difference. I would rather stick a scope in it, go through the front end, and check some signals. Maybe there are some pictures of signals in the service manual, of what should be there.
Dave put an oscilloscope on the output of the differential pair you pointed out or on the input of the chip of two hybrid modules to compare a common signal. It will effect the signal for sure but it will eliminate the source of the error as either before and after the hybrid module. I would focus of the front end either in the hybrid module or even before it as the error before the ADC. Some kind of over voltage or component failure is causing some random noise that the ADC is misinterpreting. That's my two cents.
There was a American company back in the 80's called Gigabit Logic. Everything they made was GaAs, and very fast. Faster than ECL, too. But their product line was limited. Eventually, Tektronix bought the company so they could have in-house high speed custom GaAs parts. The front-end hybrid is most likely a GaAs custom part. The stepped copper on the board is a microwave type part. Remember this is a 400MHz scope, and the converters must run far more than the Nyquist rate. This means the hybrids need clocking at microwave frequencies.
Dave please leave the hybrid modules in place, they are to fragile. If you really want to test if it is in the hybrid modules (which I higly doubt) just swap the differential analog output with a neighboring channel (only 4 wire jumpers)! Yeah there will be a bit more noise and you have to interrupt the lines and make jumpers but it is so much saver than risking to damage 2 hybrid modules.
I found the problem if you check 17:46 you can see tony abbot has run pcb trace to the hybrid, but not to the adc
I don't get it. Laughed anyway.
whayden001 tony abbot, our beloved prime minister(sarcasm, and what you would call a conservative politician), scrapped the nbn(national broadband network) plan to run fiber to the home (at 100mbps) to revolutionize our communication and education infrastructure and stop telecom company monopolization(cough.. telstra) and instead opted to run fiber to the node with a quarter speed to save money, and demonstrate to the rest of the world what a moron he truly is, as aptly demonstrated by the original comment.
alberto gordillo
Why would a politician have the say in what telcom companies do? Are the telcom companies reliant upon government money? If so shame on them. I personally feel that government's job is to get out of the way of private businesses. Regulate as necessary and minimally, but get the f' out of the way. Let the market handle itself. We have the same problem and the problem is when government is involved so to are the self interests of politicians. When companies do things on their own they do it to make profits and profits means making something people "want" to buy assuming the people have a "free" choice in what they are buying. Which is the huge debate about Obamacare in the US. What a load of bulls... that is. It isn't mean to help anyone, but those who don't do for themselves and politicians.
politicians don't really have a say in what telecom companies do, which is why we are behind in telecommunications infrastructure compared to the rest of the world and being charged ridiculous prices for basic internet access - here in Australia we get on peak off peak access metered, and we also have our uploads metered against our on peak off peak quota, the government privatizing the telecommunications company telecom had allot to do with some of the woes Australians experience, the government created a giant in a place where there was little competition, so now the notion of free-marketing turns into a monopolized dictatorship by one telecommunications company that owns pretty much all the core infrastructure anyways not too sure on the whole story but you should research a bit before shooting off, some things here in Australia seem to be opposite of what happens in america, that being our liberal party is conservative and our conservative is liberal... i know its fucking stupid lol. the nbn is a voted in government initiative to bring broadband to rural areas, educational institutions and the masses, to put to rest the aging expensive to maintain copper network that telstra runs and is charging us so much to use and replacing it with modern fibreoptics that will future proof our communications network into the foreseeable future and take market control of our core infrastructure away from telstra by obsolescence of their network, the bandwidth is provisioned out to these companies below cost and wholesale for the various telecom companies to provision to their clients.
with telstra a private company, its not as simple as making people want to buy something, many of us cant afford decent internet access, especially in rural areas where 25gb of internet access a month through wireless can cost upto 70 a month with uploads and downloads both counted toward the quota and it being split between on peak and off peak times... we are almost better off with dial up. we have something i think similar to obamacare in Australia its called medicare, and i can tell you its the single most best thing that this country has to offer, it is expensive but it is worth it.
What about exchanging the ADCs of channel 3 and channel 4 to see if the problem then changes to channel 4 as well. I know that would be a lot of work.
I am a newbie hobbiest electronics person so.... I think with the scope being in a military setting that its possible they were testing Daves flux capacitor design and messed up the AC/DC... I mean ADC by not using one of Daves MFC (Micro Flux Current) devices designed to test these types of capacitors. Seriously.... I wonder about injecting a known signal into all channels and start probing with a different scope to look for differences. Looks like something on that channel may not be decoupled and letting in noise? It would be interesting how the adc itself would produce these results in theroy.
You could swap the hybrid modules Channel 2 & 3. That is a lot of work though. It's probably not worth it economically but from a curiosity stand point it might be interesting. Thanks Thumbs Up.
Hm, the thing that strikes me as odd is that the waveform disappears at certain timebases. What could possibly do that? I also think Davros Idea was pretty clever. It would at least help to pin down the problem!
Happy New Year Dave!
Yeah swapping modules with a known good one is all I can think of
Is it safe to use a heat gun and reflow some of those chips, or is it too much of a risk, and just sell the item as having one non functioning channel?
Still worth a fair bit with a faulty channel. Or I could keep it as a lab knockabout maybe?
EEVblog Did I really just call a 500MHz 5GS/s 3 channel DPO Tek scope a lab "knockabout"? *shame*
My idea. if its a suspected bit error, reflow the ADC leads, not the analog hybrid.
EEVblog You might be in possession of the only "3 channel" 500mhz Tektronix scope in the world!
The only thing I can think of is to hope that the data interfaces are parallel and try to find the offending bit on the ADC output. Maybe the waveform could give you some clues as to root cause.
My god, those Tektronix guys sure know what they're doing - scare stuff
Is there any chance one of those RAM chips are faulty - perhaps there's a chip dedicated to each channel?
Way too much pointless criticism. Boards are supported perfectly when the case is in place. Channel failure is usually caused by someone pumping in a voltage or signal way out of spec. I bought one of these brand new about 10 years ago, and never once had a problem with it.
If you can fix it, we are waiting for your video of PART 2 to watch its end.
Probe the output of that hybrid module? Is it trowing a clean signal or some sort of noise getting in from somewhere? Very unlikely but not hard to check.
you could probably swap the channel 3 front end hybrid with one from another channel. if it works, then you know the problem is in the front end. however, i think it's more likely that the problem is in the ADC . that one would be harder to swap though.
A 3 channel Tek scope is better than a 4 channel from most others.
I don't work for them, but have stared at their screens every workday since 1974.
That thing is so much more advanced than my little Hitachi V-212. But at the end of the day they still do the same thing.
Scope the suspect hybrid, with another oscilloscope,
Comparisons can then be made with the good ones !
20:02 hair strand under pcb
I just bought one of these to replace a dead 2265B.
Maybe try tapping around the board while it's running to see if it's an intermitten problem?
The heatsink is most likely tinned copper one, I have seen those used in old computers ATX power supplies. My power supply (old ATX from 1993) has Nippon Chemicon capacitors. Have you checked the passive components? Maybe some resistor is burned out or capacitor cracked and doesn't filter the noise any longer on that channel?
At 11:50 it looked like on of the inductors have a wire missing
I'd run it open if you can and put pressure on the module and chips. If something is loose, it should resolve or get worse on the screen. Reflow the hybrid if not.
Hybrid Thick Film 'Screen Printed' Multilayer Module. The black parts are printed resistors laser trimmed to value. Very 80's. Nice to see the old school stuff. Love Tektronix gear. It's bananas. :)
I think its interesting that the artefact seems to extend up or down from the waveform differently depending on the waveform's vertical position. i.e. if its all in the top half it goes up from the waveform, if it straddles the midline, it extends both ways. Would that suggest its further along than the analog front end?
would love one of these, good diagnostic!!
I would use another scope to compare the output of a good hybrid versus the channel 3 one.
Woo, even at new year good old Dave Jones is in the lab tearing up oscillioscopes
Facinated by that hybrid front end, looks like amazing tech!
Dave, did you run the system diagnostics? They may not be more extensive than the POST, but it is worth trying. Also, there is an error log. They are under Utility/System Config/Diags.
8:10 that's one exquisite handle though..
I would first measure the balanced signals going from hybrid to ADC to see if they are clean and the DC bias is similar to other channels. Next I would go through the pins of the ADC and compare to other channels. But most likely there is nothing much to do as one of the chips is likely broken. With good luck some of the bypass capacitors have gone high ESR introducing some glitches.
I'm thinking its a bad capacitor. I would compare the values of the surface mount caps on the hybrid to one that is functioning correctly as well at the values for and continuity of the traces on those boards.
Love the teardowns. Thanks Thumbs up...
What about swapping ch1 and ch3 hybrids to see if the issue moves to ch1?
Well that looks like a pain in the *** well either swap the hybrid modules, or try to feed the adc on ch3 with the signal from a known good channel ch1-2-4 good luck dave :)
Why not scope the differential output going into the adcs? compare the frontend output from channel 4 and 3 with same signal and see if you can see any significant differences?
Wait for the next video, already shot that...
EEVblog Ah I see. Thanks for quick reply. I guess if that's not the issue maybe start swapping parts (adcs)? Seems like a manageable task without too much trouble. Hope you can fix this one up though. Not necessarily to gain a bit more functionality but for youtube videos sake. The last attempted repair with the failed asics(i think) was a disappointment. Understandable since it wasn't an economical repair but none the less. I like to see successful repairs as i'm sure everyone else does too.
Desolder frontend and adc modules in channel 3 and 4, and put them in different order.
To me it looks one of the adc digital outputs is wrong (either bad solder joint, or bad decopuling cap, or getting some high frequency noise from analog section). Maybe on adc3 or the glue chip in the middle where it supposedly goes.
Also what happens when you ground input 3? With 0V DC do you also see high frequency noise on it?
Does the strange signal (best viewed in dot mode), also happens on all vertical scales? (checking small and big signals should rule out issues with noisy relay)
Terrible aliasing problem with am modulated rf viewed at low sweep rates on this scope.
What is the purpose of the floppy?
My dad designed the SmartDisk solid state floppies :)
agree with @jared reabow, u had a ferked one so switch modules
Happy new year! Love this channel and have done for over a year now. Retro teardowns are a favorite but I'm easily pleased so anything goes. My Dad worked for Motorola and still hasn't acknowledged the phone teardown (with it's false aerial).
Makes me laugh anyway :)
Keep it going for 2014! Big thumbs up from Scotland :D
Can it be lead-free solder joints under the MAXTEK Circuit that has got whiskers between them? Then it might help to blow hot air on the Circuit and heat it up to 180°C. (I repaired a MacBook Pro in this way.)
I Think Tektronix use lasers to calibrate the hybrids. They use the laser to cut away parts from carbon trases etc. There are other marks in the hybrid that also looks like burn marks from lasers. But I can't say how they work.
comments:
Hairstrand on BNC input
Bad relay?
Bad BGA solder joint
Measure same signal on two channels and trigger on the ok channel.
Bad relay is interesting idea, it controls attentuation I guess, and if noisy, it will move / multiply signal by constant factor.
Some day I'll get lucky and get my hands on one of those puppy's. With one channel out I'd be happy with it though :)
It's a Tek, so cause may be very difficult to diagnose (because Tek's visible quality is so high).
Forget it ... move on ...you still have other good channels, so the 'scope is very usable.
Hi Dave love the video, but come on, please at least put the "spoilers" below the "show more" button :)
happy new year
Seems like the problem is a defective chip. Those lines look like a sampling error.
It is not worth to swap the channels if you don't have one to replace it anyway, you risk the possibility of breaking some thing else and then the whole scope will be crap, Murphy will get you again. You can also donate it to me, with only 3 working channels........
Happy new year Dave :-). Now I'm really curious what the error is. Waiting for part 2 and hope it runs better than with the Lecroy :-).
Commenter nr11 in 2014. Happy new year from Norway:)
i petition my ignorant comment and say its a tiny bit of conductive dust between a pair of pins of the ADC, if you look above what is likely the channel 3 ADC there is a lot of dust. its just a guess and im no expert
works and looks wonderful
What happens when you change the coupling mode (DC, AC and GND)? The problem happen only in DC mode? I don't know if in GND mode the RAM memory is used, but if it's and the problem don't appear I think that memory isn't the problem. Good luck. Happy New Year Dave.
australian defence force. Wonder for what they use Osziloskopes there?
Dave, do you still have that scope? If you're not using it, can I have it?
I have this kind of oscilloscope. Yesterday something went terribly wrong with the screen. It looks like the sync is broken in some way. I only get random lines and blocks either horizontal or vertical. Any idea if it's possible to fix it?
Haha! I just solved it. I was going to make a video and post to possible get some help on the problem. Just of frustration i bumped the scope in the table while it was starting up and the sync came back :)