Maybe try what the pros do and use a makeshift clapper board. You can just clap two books together or something. Then you can just slip the audio to match in your editing software
Yet another excellent video from TSP! Thanks! One thing to improve perhaps: when you comment the screen capture (schematics), you should use the same mic as when you are at the bench. Sounds better IMO.
21.5k, 220nF Fc = 33.6Hz, This is really a great teardown repair, I wish you spent more time analyzing the CC/CV control loops, this is a good schematic to those who need a diy power supply.
Exactly! Shahriar, that would be a great topic for a separate video! Thank you for this one. I am also an EE and I've started to do the same - buying damaged units from ebay and fixing them - inspired by you. I am even considering posting a few videos on that as well, despite the great standard you have set! Thanks again :)
@@tfinmoraes I always felt perplexed why people are building diy power supplies from non verified schematics, if we looks at this schematic and remove the ADC and DAC blocks we have a baseline analog power supply with tracking and stability control loops, parts are cheap and available (expect maybe for a transformer), for the digital controls a few 100 lines of firmware code an ADC and DAC of choice, bam! you have a robust diy power supply. but nobody seems to want to do this? I would, but I have too many 2019 project commitments at the moment.
@@yaghiyahbrenner8902I see your point, and given all the proven schematics out there ("Designs by the Masters", as Winfield & Hill would say), it certainly makes sense. But there are at least 2 counter points. One is that it is pretty educational starting from scratch - one does learn a lot going through it. Second, if the final result is the main conern, then for different transistor models (and sometimes even the same model), the feedback loop must be adjusted for rock solid stability/ ringing/ overshooting. That is where a dedicated video from Shahriar would be great. Despite so much material available on that as well, transistor stability is an ever needed subject.
Hah, easy fix! I love it when problems are silly things like this. No part replacement, no confusing debugging. Follow the signal path and there's a bad connection. Kind of unexpected that Agilent would have a QC issue like this, though.
Thanks Shahriar, great Video! I think the PCB might have been contaminated (maybe with grease or something) in that area. That would explain the bad connection and formation of these tiny micro solderballs between components in that area.
Nice video. Bad solder joint on older gear are not that uncommon but it’s usually the cheap Chinese junk. What’s sad is these are expensive well designed power supplies let down by sloppy assembly. Agilent/Keysight outsource a lot of their sub assemblies, and sometimes manufacturing of entire instruments, and this is what happens when corporate greed tries to save too much money.
Nice T/Sing. I think the PC board is probably flexing more at that point. Could be it was dropped a few feet. If I did it, I would put a jumper between the resistor and the PC trace just in case it came loose again. Thanks!
I had a 33401 that had a tombstoned resistor network from the factory. I guess it had a wisp of solder connecting the floating end to the PCB for a while and moving the meter around caused it to open. I removed and resoldered the network and it's been working fine since.
I don't quite understand how they are using one DAC to control 2 isolated outputs? Like the DAC has to be ground referenced to _something_ and in turn both outputs must have some reference to the same _something_ in order to work with the DAC output or am I missing something here?
The DAC is multiplexed. That means the DAC is switched constantly between the different outputs. DACs can be expensive and this was a way to save some money in the design.
@@RyanUptonInnovator That part I understand. However the output of the DAC and also of the 8 multiplexed outputs is referenced to _one_ ground. The outputs of the power supply are isolated (floating) though, so how do you transfer the ground-referenced DAC-voltages to two different, floating outputs? If the outputs grounds are not connected to the same ground as the DAC (which they can't both be, since they are supposed to be isolated), how do they interpret the DAC output?
I was confused by that as well. In the schematics you can see they are using a optocoupler with two photodiodes(HCNR200). The LED is driven with feedback from one of the diodes, while on the isolated side the matched photodiode ensures the voltage will be the same.
Damn, you were so lucky, just one cold joint. I had E3647A where almost nothing worked, couldn't set most of the parameters, I had to replace multiple precision resistors(those green 1206 ones seen at eg. 20:31), around 10 in total, few on the processing daughterboard and I think one or two on the main power board. But then OVP wasn't working I think on the second channel, about it later. At 10:30 you said that it's very unlikely that only one channel of a quad op amp has failed, but well that's what happened in my unit, U25 had one OPAMP crazy broken, sometimes it was working sometimes not(OVP), I spent hours on tracking the issue, once measured it, it was fine, then proceed with other components and of course couldn't fix it because nothing else was faulty. It was a very weird intermittent fault. Since then I don't just assume "Oh those don't usually fail it can't be it" For details about this failure google my thread on eevblog: *Need help with understanding part of the circuit in Agilent PSU to repair it*
Old fart here. Is it possible for you to explain/draw/highlight a little slower? Sort of a half way speed between Mr Carlson and his incessant safety admonitions and the 'just under the speed of light' explanations you used in this video. My vacuum tube era brain just can't keep up.
I've taken to watching me Carlson's videos at 2x speed these days, it makes them mostly watchable. When he starts shilling his curve tracer, capacitor tester, or probe though I just turn the video off. Saves a lot of time. He deserves to make money for his work but over the last year or so it seems like all his "repair" videos are specifically selected to let him advertise those products, which makes them somewhat useless for the rest of us who only have normal gear.
@@witeshade , but Paul Carlson doesn't (that I KNOW of) sell any of his gear... so why do you say this? It does make me a bit creeped out to hear someone say that he "advertises" his "products," since he doesn't sell anything. Not that I wouldn't buy anything he makes. de KQ2E
@@BruceNitroxpro he constantly makes videos where the primary tool he uses is one that he designed and sells access to on patreon. maybe he doesn't sell the product itself, but when the entire video is him demonstrating how that tool makes it so easy to do everything, and then tells you where to go pay money to get the complete plans for the device, that looks like shilling to me. I've basically stopped watching his videos because I am sick of hearing about how fantastic his probe and curve tracer and cap tester is.
The soldering on those three components looked different from everything else. Is it possible someone had replaced them for some reason? Maybe an ECO or something?
I only wish that I could find something this simple to fix my E3646A! I know this is a long shot, but would you mind taking a peek at the video linked here? I have a very intermittent issue where the output of channel 1 will become unsteady, drift downward (slowly at first, then faster, until it reaches a threshold at which point it jumps back (close) to the originally set voltage). I tried to troubleshoot it myself, but I am not an EE - just a hobbyist (Mech E), and I have a hard time getting it to reproduce the problem when I need it to. Most of the time, it seems like it has to be a cold start (not used in a while) for this to happen, and it seems to resolve itself after about 15-20 minutes. Channel 2 is still perfectly usable, but it would be nice to get this fixed so I have a 100% functional and reliable meter. The Power on Full Test passes. Here is the video of the fault happening: ua-cam.com/video/_PwfBVxP9Ls/v-deo.html
The board didn't exactly have a generous pad area for the components. Also, for the price point that these instruments sell at, I'd expect a heavier copper layer. The brief touch you made with the iron to melt the solder suggests a seriously low thermal mass = very thin copper = susceptible to trace breaks. I agree with others that I don't really "get" how a single DAC could run two separate channels. I'll have to look for the schematic of this unit, make a note of the components used, look up their datasheets and work it out that way.
@@aqib2000 Yep. I have an 80W Metcal MX-PS5000 here. I've been "in" to electronics for a while, so I have a good feel for things. If I touched the trace of a resistor on an Apple Macbook, using my Metcal, for THAT period of time, it wouldn't have had time make the solder flow into a reliable joint. That power supply had traces that looked like a highschool kids first PCB etching project.
I think a PSU should always display both the set and measured voltage and current simultaneously, instead of having to press a "Display Limits" button. This has always infuriated me with these PSUs. The worst bit is when you change the current limit and the PSU either can't show the measured current anymore, so you can't see when the current stops rising, or it doesn't show the limit you're setting, only the measured current. I don't get why manufacturers save a few cents by not displaying two extra numbers on multi-thousand-dollar PSUs. It's kinda impossible to find a PSU with a clean output that has a good GUI and doesn't break the bank. The chinesium "ruideng" modules have a great interface, but are of course dirty AF.
To remove the noise from those switching power supplies, all you need to do is add some capacitors and inductors on the output. The Rui Deng DPS5015 Supply and an AC - DC power supply costs less than $100, with 0-50 V at 0-15 A / 0-750 W. For an additional $1000, you can buy a lot of inductors and capacitors... and a gold plated, solid copper, enclosure as well. Even then, it will be smaller, lighter, more functional, and easier to use. Included are USB control from a PC or wireless control from an android cell phone. All of the parts will be soldered to the PCB. To add an "air of legitimacy" you can also probably get a genuine "Keysight" badge to glue on the front for a few dollars on the internet. www.aliexpress.com/store/product/RD-DPS5015-Constant-Voltage-current-Step-down-Programmable-digital-Power-Supply-buck-Voltage-converter-color-LCD/923042_32702714880.html?spm=2114.12010615.8148356.5.7c13411c4za6aA
As always, very ineresting to watch these kinds of videos. I don't know if it's just my PC or not, but the sound and the screen capture are kind of out of sync. Tried different video quality settings, so it's not from UA-cam (this time)
Starting around 11:43, there seems to be a discrepancy between what you are pointing at on the schematic and what you are discussing. You talk about the output transistor while circling an op-amp, for example. I think you had a problem with your screen capture or something. It makes the latter part of your discussion about the schematic really hard to follow.
I have noticed that the frame rate (FPS) in most of your vides while doing screen capture is quite low. It is annoying because the mouse (and other objects) jumps/stutters instead of moving smoothly. I assume you use OBS for screen capture, it should be easy to just increase the FPS to 60 for future videos if possible. The size of the files should not become much larger because there is not much movement on the screen hence the compression will work very efficiently.
You make great videos that I really enjoy watching! But you should really learn how to say the word voltage, since it's used so very often in your videos. It's not voltaesh, it's more like voltedge if I try to spell how it should be pronounced.
It's a shame you let down your considerable intelligence by indulging overpriced low tech instruments. It's a 1200$ unit that weighs a ton when it could cost 50$ and be superior in every way. Weight, precision, noise, power, stability, durability.
@@6alecapristrudel Neither. It's a simple certainty that a superior switchmode design could readily be done for pennies with HF pcb trace trafo. It's just not a challenging power level. Including plentiful filtering and a linear stage if you must. Sharir could easily design such a circuit.
@@danfrederiksen7397 If someone were to make the thing you describe, would they want to sell it for 50 bucks? Given that they're making something better than a thing that sells for so much more?
@@Thesignalpath Notice that he's got a different account now, although the name is the same. I guess his old one got banned or something. The never-ending story of Dan Frederiksen. Always talking angrily on electronics channels about how professional instruments are low tech and overpriced, while never producing anything of his own. Yay!
Sound is about 30sec out of sync for the circuit diagram section.
Damn it! I need to find a better screen capture solution.
Maybe try what the pros do and use a makeshift clapper board. You can just clap two books together or something. Then you can just slip the audio to match in your editing software
@@Thesignalpath Isn't it OBS? I'm not a heavy user, but I've never had a problem with OBS.
as is tradition
ye dud use obs
@@SidneyCritic Will get that. Thanks.
Pooch cameo always makes for great teardowns.
Pooch said the CAT scan results will be in sometime tomorrow.
Hehe. He loves doing CAT scans.
@@Thesignalpath My cat looks almost Identical to him, but mine spends a little too much more time at his food bowl. ; )
An excellent example of fault-finding techniques. Thanks for explaining all so clearly.
Yet another excellent video from TSP! Thanks!
One thing to improve perhaps: when you comment the screen capture (schematics), you should use the same mic as when you are at the bench. Sounds better IMO.
21.5k, 220nF Fc = 33.6Hz, This is really a great teardown repair, I wish you spent more time analyzing the CC/CV control loops, this is a good schematic to those who need a diy power supply.
Exactly! Shahriar, that would be a great topic for a separate video! Thank you for this one. I am also an EE and I've started to do the same - buying damaged units from ebay and fixing them - inspired by you. I am even considering posting a few videos on that as well, despite the great standard you have set! Thanks again :)
@@tfinmoraes I always felt perplexed why people are building diy power supplies from non verified schematics, if we looks at this schematic and remove the ADC and DAC blocks we have a baseline analog power supply with tracking and stability control loops, parts are cheap and available (expect maybe for a transformer), for the digital controls a few 100 lines of firmware code an ADC and DAC of choice, bam! you have a robust diy power supply. but nobody seems to want to do this? I would, but I have too many 2019 project commitments at the moment.
@@yaghiyahbrenner8902I see your point, and given all the proven schematics out there ("Designs by the Masters", as Winfield & Hill would say), it certainly makes sense. But there are at least 2 counter points. One is that it is pretty educational starting from scratch - one does learn a lot going through it. Second, if the final result is the main conern, then for different transistor models (and sometimes even the same model), the feedback loop must be adjusted for rock solid stability/ ringing/ overshooting. That is where a dedicated video from Shahriar would be great. Despite so much material available on that as well, transistor stability is an ever needed subject.
Great job on finding the fault!
Always great to watch your videos, thanks friend!
Alex.
Rosanah is going to be happy :)
Hah, easy fix! I love it when problems are silly things like this. No part replacement, no confusing debugging. Follow the signal path and there's a bad connection.
Kind of unexpected that Agilent would have a QC issue like this, though.
Boy that was an easy fix.Would have been cool to actually see it lifted with a macro lens. Nice job.
cannot wait to see the repair of those other equipment :-)
Fantastic job....cheers.
Excellence video, thanks for sharing the experience.
Good to see the cat scan is still functional 😂
Thanks for this! It was very educative to see the process.
Thanks Shahriar, great Video! I think the PCB might have been contaminated (maybe with grease or something) in that area. That would explain the bad connection and formation of these tiny micro solderballs between components in that area.
Useful video 👍
Nice video. Bad solder joint on older gear are not that uncommon but it’s usually the cheap Chinese junk. What’s sad is these are expensive well designed power supplies let down by sloppy assembly. Agilent/Keysight outsource a lot of their sub assemblies, and sometimes manufacturing of entire instruments, and this is what happens when corporate greed tries to save too much money.
Good Video thanks. :)
Nice as always !
5:20 Ah, the portable static charge generator. Every lab needs one.
Nice T/Sing. I think the PC board is probably flexing more at that point. Could be it was dropped a few feet. If I did it, I would put a jumper between the resistor and the PC trace just in case it came loose again. Thanks!
I had a 33401 that had a tombstoned resistor network from the factory. I guess it had a wisp of solder connecting the floating end to the PCB for a while and moving the meter around caused it to open. I removed and resoldered the network and it's been working fine since.
I don't quite understand how they are using one DAC to control 2 isolated outputs? Like the DAC has to be ground referenced to _something_ and in turn both outputs must have some reference to the same _something_ in order to work with the DAC output or am I missing something here?
The DAC is multiplexed. That means the DAC is switched constantly between the different outputs. DACs can be expensive and this was a way to save some money in the design.
@@RyanUptonInnovator That part I understand. However the output of the DAC and also of the 8 multiplexed outputs is referenced to _one_ ground. The outputs of the power supply are isolated (floating) though, so how do you transfer the ground-referenced DAC-voltages to two different, floating outputs? If the outputs grounds are not connected to the same ground as the DAC (which they can't both be, since they are supposed to be isolated), how do they interpret the DAC output?
I was confused by that as well. In the schematics you can see they are using a optocoupler with two photodiodes(HCNR200). The LED is driven with feedback from one of the diodes, while on the isolated side the matched photodiode ensures the voltage will be the same.
HCNR200 optocouplers. Currently made by Broadcom, formerly Avago, formerly Agilent, formerly HP. Briefly seen at 11:18.
www.broadcom.com/products/optocouplers/industrial-plastic/specific-function/high-linearity-analog/hcnr200
Wery interesting scheme!
Damn, you were so lucky, just one cold joint.
I had E3647A where almost nothing worked, couldn't set most of the parameters, I had to replace multiple precision resistors(those green 1206 ones seen at eg. 20:31), around 10 in total, few on the processing daughterboard and I think one or two on the main power board.
But then OVP wasn't working I think on the second channel, about it later.
At 10:30 you said that it's very unlikely that only one channel of a quad op amp has failed, but well that's what happened in my unit, U25 had one OPAMP crazy broken, sometimes it was working sometimes not(OVP), I spent hours on tracking the issue, once measured it, it was fine, then proceed with other components and of course couldn't fix it because nothing else was faulty.
It was a very weird intermittent fault.
Since then I don't just assume "Oh those don't usually fail it can't be it"
For details about this failure google my thread on eevblog:
*Need help with understanding part of the circuit in Agilent PSU to repair it*
Old fart here. Is it possible for you to explain/draw/highlight a little slower? Sort of a half way speed between Mr Carlson and his incessant safety admonitions and the 'just under the speed of light' explanations you used in this video. My vacuum tube era brain just can't keep up.
I've taken to watching me Carlson's videos at 2x speed these days, it makes them mostly watchable. When he starts shilling his curve tracer, capacitor tester, or probe though I just turn the video off. Saves a lot of time. He deserves to make money for his work but over the last year or so it seems like all his "repair" videos are specifically selected to let him advertise those products, which makes them somewhat useless for the rest of us who only have normal gear.
You can play the video at various slow speeds.
@@Factory400 Cool. Thanks!
@@witeshade , but Paul Carlson doesn't (that I KNOW of) sell any of his gear... so why do you say this? It does make me a bit creeped out to hear someone say that he "advertises" his "products," since he doesn't sell anything. Not that I wouldn't buy anything he makes. de KQ2E
@@BruceNitroxpro he constantly makes videos where the primary tool he uses is one that he designed and sells access to on patreon. maybe he doesn't sell the product itself, but when the entire video is him demonstrating how that tool makes it so easy to do everything, and then tells you where to go pay money to get the complete plans for the device, that looks like shilling to me.
I've basically stopped watching his videos because I am sick of hearing about how fantastic his probe and curve tracer and cap tester is.
Thanks for your great videos! I have a E3646A that is totally dead, only the fan works at times. I am really at a loss where to start. Any clues?
The soldering on those three components looked different from everything else. Is it possible someone had replaced them for some reason? Maybe an ECO or something?
I don’t think this supply was ever repaired. The original factory seal was intact.
The soldering did look a bit janky in that area.
I only wish that I could find something this simple to fix my E3646A! I know this is a long shot, but would you mind taking a peek at the video linked here? I have a very intermittent issue where the output of channel 1 will become unsteady, drift downward (slowly at first, then faster, until it reaches a threshold at which point it jumps back (close) to the originally set voltage). I tried to troubleshoot it myself, but I am not an EE - just a hobbyist (Mech E), and I have a hard time getting it to reproduce the problem when I need it to. Most of the time, it seems like it has to be a cold start (not used in a while) for this to happen, and it seems to resolve itself after about 15-20 minutes. Channel 2 is still perfectly usable, but it would be nice to get this fixed so I have a 100% functional and reliable meter. The Power on Full Test passes.
Here is the video of the fault happening:
ua-cam.com/video/_PwfBVxP9Ls/v-deo.html
Attention to detail.
What a surprise! Another broken instrument fixed easily.
I know! Isn't it great!
That's a really rare problem to have wow
Cold solder in a HP instrument.. something is not right.
The board didn't exactly have a generous pad area for the components. Also, for the price point that these instruments sell at, I'd expect a heavier copper layer. The brief touch you made with the iron to melt the solder suggests a seriously low thermal mass = very thin copper = susceptible to trace breaks.
I agree with others that I don't really "get" how a single DAC could run two separate channels. I'll have to look for the schematic of this unit, make a note of the components used, look up their datasheets and work it out that way.
digitalradiohacker / you do realise those jbc irons have some serous wattage too. Though I do agree about the poor pad spacing
@@aqib2000
Yep. I have an 80W Metcal MX-PS5000 here. I've been "in" to electronics for a while, so I have a good feel for things. If I touched the trace of a resistor on an Apple Macbook, using my Metcal, for THAT period of time, it wouldn't have had time make the solder flow into a reliable joint.
That power supply had traces that looked like a highschool kids first PCB etching project.
What do you use as your PRIMARY standards (assuming that you are doing this with metrological precision)?
How do you find schematics of things like that?
This one was easy, quick Google search and you can find it.
👍👍
Sometimes you need to source a very expensive parts. Other times one cent worth of solder.
I think a PSU should always display both the set and measured voltage and current simultaneously, instead of having to press a "Display Limits" button. This has always infuriated me with these PSUs. The worst bit is when you change the current limit and the PSU either can't show the measured current anymore, so you can't see when the current stops rising, or it doesn't show the limit you're setting, only the measured current.
I don't get why manufacturers save a few cents by not displaying two extra numbers on multi-thousand-dollar PSUs.
It's kinda impossible to find a PSU with a clean output that has a good GUI and doesn't break the bank. The chinesium "ruideng" modules have a great interface, but are of course dirty AF.
I agree. The new Keysight supplies do have that.
It also displays only one channel at a time :). Horrible UX.
To remove the noise from those switching power supplies, all you need to do is add some capacitors and inductors on the output. The Rui Deng DPS5015 Supply and an AC - DC power supply costs less than $100, with 0-50 V at 0-15 A / 0-750 W. For an additional $1000, you can buy a lot of inductors and capacitors... and a gold plated, solid copper, enclosure as well. Even then, it will be smaller, lighter, more functional, and easier to use. Included are USB control from a PC or wireless control from an android cell phone. All of the parts will be soldered to the PCB.
To add an "air of legitimacy" you can also probably get a genuine "Keysight" badge to glue on the front for a few dollars on the internet. www.aliexpress.com/store/product/RD-DPS5015-Constant-Voltage-current-Step-down-Programmable-digital-Power-Supply-buck-Voltage-converter-color-LCD/923042_32702714880.html?spm=2114.12010615.8148356.5.7c13411c4za6aA
@@shazam6274 output capacitance needs to be as low as possible in a power supply
Is video not in sync with audio???
Give it the psu to Dave Jones and he can't repair it... Jajaja
Lol, that kinda became a meme by now.
Well not everywhere...
Wow, so even Keysight makes mistakes too
Awesome! Wonder what caused the dislodgement... Agilent QC? Maybe someone hit it? Disgruntled employee? Aliens?
Based on my observation, the only possibility is aliens.
@@Thesignalpath No, it's lead free solder. It's always lead free solder.
@@Thesignalpath aliens come for our lead.
Aliens?
I noticed a cat acting suspiciously! They always revisit the crime scene.
As always, very ineresting to watch these kinds of videos. I don't know if it's just my PC or not, but the sound and the screen capture are kind of out of sync. Tried different video quality settings, so it's not from UA-cam (this time)
Tape on the transformer obvious...
Starting around 11:43, there seems to be a discrepancy between what you are pointing at on the schematic and what you are discussing. You talk about the output transistor while circling an op-amp, for example. I think you had a problem with your screen capture or something. It makes the latter part of your discussion about the schematic really hard to follow.
I am not impressed with the production quality of the soldering in this unit. I hope this was a Monday's unit.
Cold solder in a HP instrument.. something is really wrong.
I have noticed that the frame rate (FPS) in most of your vides while doing screen capture is quite low. It is annoying because the mouse (and other objects) jumps/stutters instead of moving smoothly. I assume you use OBS for screen capture, it should be easy to just increase the FPS to 60 for future videos if possible. The size of the files should not become much larger because there is not much movement on the screen hence the compression will work very efficiently.
You make great videos that I really enjoy watching! But you should really learn how to say the word voltage, since it's used so very often in your videos. It's not voltaesh, it's more like voltedge if I try to spell how it should be pronounced.
English grammar, vocabulary, diction and accent are superb. Don't mind this comment.
McG
You should learn to be more tolerent.
Comments like this are just background noise.
@@jonka1 You are probably right. I was actually just trying to be helpful but I can see how my comment can come off as slightly intolerant.
It's a shame you let down your considerable intelligence by indulging overpriced low tech instruments. It's a 1200$ unit that weighs a ton when it could cost 50$ and be superior in every way. Weight, precision, noise, power, stability, durability.
Not sure if trolling or just wishful thinking?
@@6alecapristrudel Neither. It's a simple certainty that a superior switchmode design could readily be done for pennies with HF pcb trace trafo. It's just not a challenging power level. Including plentiful filtering and a linear stage if you must. Sharir could easily design such a circuit.
@@danfrederiksen7397 If someone were to make the thing you describe, would they want to sell it for 50 bucks? Given that they're making something better than a thing that sells for so much more?
I think I have tried to explain this to you before. But I see that you are still confused. :)
@@Thesignalpath Notice that he's got a different account now, although the name is the same. I guess his old one got banned or something. The never-ending story of Dan Frederiksen. Always talking angrily on electronics channels about how professional instruments are low tech and overpriced, while never producing anything of his own. Yay!
STOP SENDING MORE THEN ONE EMAIL NOTIFICTION !
This video was posted twice because of an error in the rendering. I don't choose how notifications are sent, UA-cam does.