As the problem of risking a perfectly good PCB during rework seems to affect a couple of users - and hopefully not an increasing number in the future - I thought I'd chime in briefly and share my two cents about how to approach it.
It always amazes me how people risk three-digit-bucks products (computer mainboards, cell phones, audio gear) while desoldering a fraction-of-a-buck component (passives, ICs, interconnections). The way I see it, it's the PCB that should be preserved at all cost, not the component that goes into the bin anyway.
After having this happen to myself a lot - delaminated copper, broken vias, cracked PCBs - I nowadays resort to strictly (and purposefully) destroying the component while going as easy on the PCB as possible.
In most cases, the component in question is either damaged anyway, cheap or won't survive the heat of desoldering - at least not without application of proper specialty rework tools.
Thus, generally speaking, physically decoupling/separating the solder pads from each other is the way to go that I would recommend from my own experience.
An IC is removed by cutting it's legs of (scalpel-sharp knife), then desoldering them one by one.
A through-hole electrolytic capacitor is best removed by pulling the metal can off the electrodes, then desoldering the residual two pieces of wire.
Given how little force/leverage is required in order to desolder a piece of wire either from a pad or out of a hole, the risk of damaging a PCB while doing so is minimal compared to keeping the component fully intact.
Now, with smaller two-pole SMD components, it's different. Trying to crack the ceramic carrier of a resistor or capacitor is very hard on both PCB and tools.
Here I'd recommend applying heat to both pins/pads at the same time, patiently waiting for the solder to start flowing before using any force.
The lead free solder in modern industrial soldering has a very high melting point (and isn't very ductile until just below that). Thus, patience and a powerful soldering tool really are a virtue here.
If the tip of the iron is too small or the power insufficient, a second soldering iron does the trick. For one-of jobs, this can be virtually any tool that won't torch the PCB (even if it's originally intended to solder the gutter).
As a rule of thumb, as long as force is required to dislocate an SMT component at all, it's still too early (read: too cold) to apply force in the first place.
Now, if something bad has already happened to the PCB, not all is lost in every case. Reworking the PCB doesn't have to stop at trying to replace components, it can extend to replacing copper traces and vias.
With a single-layer board, this is straight forward (trace the copper, replace by wire).
With a two-layer board, broken vias can be replaced by fiddling (thin) wire through the broken via and soldering it to the respective copper layers.
With a multilayer PCB, this is - again - different, as only superficial damage can be repaired without requiring extensive knowledge of schematic and circuit.
But most damage on multilayer PCBs will presumably still be superficial (as long as you didn't pry that capacitor through it's holes until the copper came off...).
Again, that's just my recommendation from my own experience.
Every case is different and mileages do vary.
Cheers,
Sebastian.
It always amazes me how people risk three-digit-bucks products (computer mainboards, cell phones, audio gear) while desoldering a fraction-of-a-buck component (passives, ICs, interconnections). The way I see it, it's the PCB that should be preserved at all cost, not the component that goes into the bin anyway.
After having this happen to myself a lot - delaminated copper, broken vias, cracked PCBs - I nowadays resort to strictly (and purposefully) destroying the component while going as easy on the PCB as possible.
In most cases, the component in question is either damaged anyway, cheap or won't survive the heat of desoldering - at least not without application of proper specialty rework tools.
Thus, generally speaking, physically decoupling/separating the solder pads from each other is the way to go that I would recommend from my own experience.
An IC is removed by cutting it's legs of (scalpel-sharp knife), then desoldering them one by one.
A through-hole electrolytic capacitor is best removed by pulling the metal can off the electrodes, then desoldering the residual two pieces of wire.
Given how little force/leverage is required in order to desolder a piece of wire either from a pad or out of a hole, the risk of damaging a PCB while doing so is minimal compared to keeping the component fully intact.
Now, with smaller two-pole SMD components, it's different. Trying to crack the ceramic carrier of a resistor or capacitor is very hard on both PCB and tools.
Here I'd recommend applying heat to both pins/pads at the same time, patiently waiting for the solder to start flowing before using any force.
The lead free solder in modern industrial soldering has a very high melting point (and isn't very ductile until just below that). Thus, patience and a powerful soldering tool really are a virtue here.
If the tip of the iron is too small or the power insufficient, a second soldering iron does the trick. For one-of jobs, this can be virtually any tool that won't torch the PCB (even if it's originally intended to solder the gutter).
As a rule of thumb, as long as force is required to dislocate an SMT component at all, it's still too early (read: too cold) to apply force in the first place.
Now, if something bad has already happened to the PCB, not all is lost in every case. Reworking the PCB doesn't have to stop at trying to replace components, it can extend to replacing copper traces and vias.
With a single-layer board, this is straight forward (trace the copper, replace by wire).
With a two-layer board, broken vias can be replaced by fiddling (thin) wire through the broken via and soldering it to the respective copper layers.
With a multilayer PCB, this is - again - different, as only superficial damage can be repaired without requiring extensive knowledge of schematic and circuit.
But most damage on multilayer PCBs will presumably still be superficial (as long as you didn't pry that capacitor through it's holes until the copper came off...).
Again, that's just my recommendation from my own experience.
Every case is different and mileages do vary.
Cheers,
Sebastian.
The toothbrush alone might not have done the trick, but using isopropyl alcohol (rubbing alcohol) or other PCB cleaning agent might. There's still some residual flux on the PCB that can be removed in order to make things easier.
Both pads of R141 came off (even though the right one in the picture still sticks in place). Both will require rework.
From over here I can't see any vias directly affected by the lost copper. Both pads had traces connected to them, none had a via directly underneath them.
In such a situation I would carefully remove the laquer from the connected signals and find a good spot to solder a thin wire to.
When done precisely, both the via and the trace can possibly be exposed for soldering.
I'd send a message to Hypex' support department, they are generally very helpful in such cases. Maybe they can confirm the degree of damage and some optimal locations for rework.
Good luck,
Sebastian.
I get hiss even without any source connected to my ncore on my 110dB+ compressions drivers (and these are 16ohms drivers...).
Correct me if I am wrong, but reducing the internal gain of the amp will reduce that hiss, whereas reducing the gain before the amp would obviously have no effect (other than reducing any additional hiss comming from the source, of course).
Now about the "why a 200W amp" (100W/16ohms in my case) question...
Well, do you know of any amp that could match the nc400?
The highs are where it shines the most, and the powersupply can be shared, so why choose anything else? (except for the price, of course
)I am still not sure if I will need a supertweeter in my system, but if I do I am sure I will buy another pair of nc400 to power those...
Another candidate would be The Wire LPUHP, but it cannot be get as an assembled kit... (and even the boards are long sold out...).
Thank you all for this constructive help !
=> sek, StigErik : I understood that yesterday, unfortunatly too late !! I hope it will help the ones that need to remove R141 (to optimize the gain chain )
=> pos : the ncore is a very good amp, and YES 100dB + speakers are very happy with it !!
I get some fast support from Hypex. At the same time I get the nc400 at work to look at it under a microscope.
Same conclusion as Hypex : there was a connection under the pad !!!
I managed to fix it, it was not easy...
And now the Ncores are playing music !
I will need 2 more to drive my bass dipoles. I may let R141 this time as they need more power ...
=> sek, StigErik : I understood that yesterday, unfortunatly too late !! I hope it will help the ones that need to remove R141 (to optimize the gain chain
)=> pos : the ncore is a very good amp, and YES 100dB + speakers are very happy with it !!
I get some fast support from Hypex. At the same time I get the nc400 at work to look at it under a microscope.
Same conclusion as Hypex : there was a connection under the pad !!!
An externally hosted image should be here but it was not working when we last tested it.
I managed to fix it, it was not easy...
And now the Ncores are playing music !
I will need 2 more to drive my bass dipoles. I may let R141 this time as they need more power ...
I get hiss even without any source connected to my ncore on my 110dB+ compressions drivers (and these are 16ohms drivers...).
Correct me if I am wrong, but reducing the internal gain of the amp will reduce that hiss, whereas reducing the gain before the amp would obviously have no effect (other than reducing any additional hiss comming from the source, of course).
Now about the "why a 200W amp" (100W/16ohms in my case) question...
Well, do you know of any amp that could match the nc400?
The highs are where it shines the most, and the powersupply can be shared, so why choose anything else? (except for the price, of course
I am still not sure if I will need a supertweeter in my system, but if I do I am sure I will buy another pair of nc400 to power those...
Another candidate would be The Wire LPUHP, but it cannot be get as an assembled kit... (and even the boards are long sold out...).
Maybe a L-pad will be the best solution, this is a clear case of a complete mismatch between amp and driver. Use some high quality MOX resistors to soak the power you do not need for this driver.......
And lose the low output impedance / high damping factor of the amp?
is it? I can see the reasons for why pos wants to use the nc400, and lowering the gain would definitely help in his situation.
For now I don't feel confortable enough with the idea of removing that damned R141 myself, so I use autoformers.
It is a much better solution than resistors Lpad IMHO, for compression drivers at least : the nc400 sees a higher load, the compression drivers sees a decent damping factor, and almost no power is lost in the process. What not to like?
I use those:
eBay | +Jabo+Autotransformer+für+Hornspeaker++Klipsch+und+Altec+Crossover+
But even the cheap inwall models sold at parts epress do well for the high frequencies, for what I can measure and hear, except for a tiny UHF rolloff that is easy to compensate.
It is a much better solution than resistors Lpad IMHO, for compression drivers at least : the nc400 sees a higher load, the compression drivers sees a decent damping factor, and almost no power is lost in the process. What not to like?
I use those:
eBay | +Jabo+Autotransformer+für+Hornspeaker++Klipsch+und+Altec+Crossover+
But even the cheap inwall models sold at parts epress do well for the high frequencies, for what I can measure and hear, except for a tiny UHF rolloff that is easy to compensate.
Distortion? Not likely to be measurable at those frequencies with any half decent autoformer (not to mention the big ones linked above).
non-linearity? The frequency response was unchanged with the big autoformer, and with only a minor rolloff up high with the cheap one.
I have done a lot of measurements on my compression drivers, and could not detect any drawback in using an autoformer vs a lpad.
non-linearity? The frequency response was unchanged with the big autoformer, and with only a minor rolloff up high with the cheap one.
I have done a lot of measurements on my compression drivers, and could not detect any drawback in using an autoformer vs a lpad.
gain resistor removal
The thought of surgery on my NCores scared me at first. What gave me the creeps was the idea of tossing around solder all over the place, bridging all kinds of points, or destroying a random component in the vincinity by ways of excess heat.
After a close encounter with our old friend hum, I did not only guess but know that a 400W amp wired straight to a 30/70W (sine/peak) fullrange driver is dangerous to the latter. Guess I was lucky the aluminium cones did not have a facelift that day. So it was decided, those gain resistors had to go.
First up, UcD400ST. Here it is R3 which needs removal for minimum gain. Desoldering can be messy, so I thought, why not slice the thing in half, remove the debris and leave anything stuck to the pads in place? I even considered the dremel with a cutting wheel on the sole condition that the board leave enough room. Eyeballing the wire cutter, I thought maybe the force pushing the pads outwards might become too great if I just cut through. So I tried to nibble away horizontally, maybe get the upper half to come off or something, and the poor thing just crumbled.
So I took some narrow pliers and just squished to dust what remained of the connection between the two pads. Holding the board at an angle makes the dust fall right off the edge. I gave the board a good scrubbing with a toothbrush and a sweep of the compressed air can anyway, just to make sure nothing builds a bridge that should rather be burnt.
What can I say, it powered up neatly and did not eat the canary speaker after that, and would the latter not be so bad, it would have been music that was played, except at a lot lower a volume.
The problem with NCore and its R141 is that it is a different type of component - not a cuboid of black brittleness, but a ceramic tube, almost like ordinary through-hole resistors, and pretty tough. I had to push a lot harder to get that one pulverized, but it went smoothly as well. While slowly increasing the force applied to the part, it is paramount to not pull upwards or sideways against the board. Holding the amp module by the resistor alone looks brutal, but the weight of an NCore or UcD against the pads does not seem to do any harm.
I expected some remains to be stuck to the pads, but apart from a thin layer of solder they were almost even. Looks like it is possible to get those gain resistors clean off without even heating up the soldering iron Just grab the module by the gain resistor and crush - the latter. Squeeze, don't pull - that's what Americans say, huh.
Hope that helps whoever would like to lower the gain and is afraid of the whole desoldering business.
The thought of surgery on my NCores scared me at first. What gave me the creeps was the idea of tossing around solder all over the place, bridging all kinds of points, or destroying a random component in the vincinity by ways of excess heat.
After a close encounter with our old friend hum, I did not only guess but know that a 400W amp wired straight to a 30/70W (sine/peak) fullrange driver is dangerous to the latter. Guess I was lucky the aluminium cones did not have a facelift that day. So it was decided, those gain resistors had to go.
First up, UcD400ST. Here it is R3 which needs removal for minimum gain. Desoldering can be messy, so I thought, why not slice the thing in half, remove the debris and leave anything stuck to the pads in place? I even considered the dremel with a cutting wheel on the sole condition that the board leave enough room. Eyeballing the wire cutter, I thought maybe the force pushing the pads outwards might become too great if I just cut through. So I tried to nibble away horizontally, maybe get the upper half to come off or something, and the poor thing just crumbled.
So I took some narrow pliers and just squished to dust what remained of the connection between the two pads. Holding the board at an angle makes the dust fall right off the edge. I gave the board a good scrubbing with a toothbrush and a sweep of the compressed air can anyway, just to make sure nothing builds a bridge that should rather be burnt.
What can I say, it powered up neatly and did not eat the canary speaker after that, and would the latter not be so bad, it would have been music that was played, except at a lot lower a volume.
The problem with NCore and its R141 is that it is a different type of component - not a cuboid of black brittleness, but a ceramic tube, almost like ordinary through-hole resistors, and pretty tough. I had to push a lot harder to get that one pulverized, but it went smoothly as well. While slowly increasing the force applied to the part, it is paramount to not pull upwards or sideways against the board. Holding the amp module by the resistor alone looks brutal, but the weight of an NCore or UcD against the pads does not seem to do any harm.
I expected some remains to be stuck to the pads, but apart from a thin layer of solder they were almost even. Looks like it is possible to get those gain resistors clean off without even heating up the soldering iron
Just grab the module by the gain resistor and crush - the latter. Squeeze, don't pull - that's what Americans say, huh.Hope that helps whoever would like to lower the gain and is afraid of the whole desoldering business.
OllBoll, your method looks manageable as well
As far as I am concerned, whenever there is nothing at stake I am quite well in control of my extremities, but once there's a chance for something to break, like really bad, I am bound to mess up. Glass of water next to the keyboard? What spasm is going through my shoulder to move my elbow not past but right through it when there is not even a reason for moving it there to begin with? Lend a hand, or rather sever an arm for that matter?
Just desoldered probably damaged voltage regulators from the O2 board during PSU debugging, now let's go for the replacements. Placed them in the right position and direction ten times at least, right before soldering they fall out, and then I put them in all the wrong ways possible except they wind up on the right side of the board!? Needless to say, desoldering a second time I fried the PCB pretty bad and I yet have to answer swalter's e-mail how it all went. That time I was just a iota of frustration away from kicking myself in the head for being a stupid f'tard...
Needless to say, when it came to the UcDs and NCores, my first thought was to hit Murphy over the head with a hatchet. Avoid whatever can go wrong the stupid way. Which is also why the dremel ultimately stayed in its box that day
"What could possibly go wrong!?"
*wheeeeeeeeeeeeeeeeeeeeeeeeCCCCHHRRRRRRRRRRRRRRRRR*
"WHAAAAAARRRGHHHH!!"
Guess I'm just fsck'd that way. No hard feelings unless it's blatant malice or my own fault
As far as I am concerned, whenever there is nothing at stake I am quite well in control of my extremities, but once there's a chance for something to break, like really bad, I am bound to mess up. Glass of water next to the keyboard? What spasm is going through my shoulder to move my elbow not past but right through it when there is not even a reason for moving it there to begin with? Lend a hand, or rather sever an arm for that matter?
Just desoldered probably damaged voltage regulators from the O2 board during PSU debugging, now let's go for the replacements. Placed them in the right position and direction ten times at least, right before soldering they fall out, and then I put them in all the wrong ways possible except they wind up on the right side of the board!? Needless to say, desoldering a second time I fried the PCB pretty bad and I yet have to answer swalter's e-mail how it all went. That time I was just a iota of frustration away from kicking myself in the head for being a stupid f'tard...
Needless to say, when it came to the UcDs and NCores, my first thought was to hit Murphy over the head with a hatchet. Avoid whatever can go wrong the stupid way. Which is also why the dremel ultimately stayed in its box that day
"What could possibly go wrong!?"
*wheeeeeeeeeeeeeeeeeeeeeeeeCCCCHHRRRRRRRRRRRRRRRRR*
"WHAAAAAARRRGHHHH!!"
Guess I'm just fsck'd that way. No hard feelings unless it's blatant malice or my own fault
The damping factor is very important with cone drivers, with compression drivers it is less important. But what is wrong with an att. network before the driver? Most passive speakers use them to match the midrange and highrange.
- Status
- Not open for further replies.