new killer scope in town - a true game changer from R&S - RTB2002 & RTB2004

[Hydron]

No offence but the roast pork I just cooked looks tastier than your PGA chip!

[nctnico]

Typically I have my hotplate set to 250 degrees and start rework when the PCB has reached 100 degrees. I use hot air set to 330 degrees or 340 degrees at most.

Whow! Is 250°C not a little bit to much? I would worry for the bottom parts in this case. I startet at 150°C and use hot air on top wht 370°C.

No, because there is no direct contact with the plate. Most of the transfer is radiated.

[Robaroni]

Do you, or does anyone else, have picture of the chip before heating it?

What did you expext from a picture? It is a normal 16pin QFN with EP. There is nothing special with this chip. I solderd chips like this many times! The problem in this case is that this chip has a normal power consumption of 1W and for this reason the PCB is full of copper. R&S delivered us a realy good high quality PCB for our money! The only problem they did not use a socket. 

Olaf

I "expext"(sic) to see the location of the chip on the board and any top copper surrounding it, which I did from someone gracious enough to post it!

[bayjelly]

Anyone know what RTB2004 setting(s) and/or signal input value cause the LMH6518 to go into saturation (>=1700mVp-p)?

This would indeed be crucial to know, does anyone here?

[Darkover]

This would indeed be crucial to know, does anyone here?

It happens to me when I was measuring current with a self developed current probe. (DC capable)
The probe has an output (OPV) with 0 to 2.5V.

So for example with a 25mA amplifier-head I am measuring 0-25mA to 0 to 2.5V. I shift the ground line
to the bottom of the screen and 2.5V/25mA was at the top. Then I was measuring a testboard for some
time that draws current and the probe output was probably at 2V, close to the top, but not the top of the screen
for most of the time. And because I was interested in current consumption in different situation I did it
for some time. (more like hours, not minutes)
That means more or less, measuring a DC value that uses the full screen and stays at the top for some time.

Olaf

[Hydron]

The key question is what the PGA sends out, NOT what is on the scope input. There are amplifiers and attenuators in there, both in the PGA and the front end. I'll try and measure the actual output for on-screen traces when I have my scope open.

[alonsojar]

Hi,
Regarding the possible problem with the LHM6518 chip. I'm using sometimes a micsig diff probe DP10013 that has a max. output level of 3V. Am I risking or stressing my scope using it using full screen vertical ranges reaching the probe limits for a few time?
Sorry if this question is not relevant, but I'm just an oscilloscope end user and I'm getting lost about its internals.

J. Alonso

[Hydron]

So the thing to remember is that we're talking about the signal at the output of the PGA to the ADC, NOT what is going into the BNC of the scope. There's a bunch of gain/attenuation and offset applied along the way inside the front end circuitry, so you can't directly determine what is and isn't in the allowed range.

It also depends on the ADC input voltage range, if that's say 2V full scale without any internal gain option then there's a huge problem as every range will fall foul of the PGA limits if you have steady state signals near the top or bottom of the screen. I will be looking at the actual outputs when I have time to open up my scope.

[Darkover]

I guess it is mostly save if the default DC Level is not for a longer time at the upper or lower level in the  selected range when the ground level is on the other screen border. That means in 99% of the normal use you are save.
But of course it may depend on the exact attenuation level like Hydron said.

The main problem is probably you never will _feel_ save. That is a little bis uncomfortable.

Olaf
 

[alonsojar]

Thank you Hydron and Darkover for the explanation!
Now I start understanding that this problem may happen as part of the signal conditioning and it's not directly tied with the actual bnc input but related in a not obvious way with it, depending on the amplification or attenuation internally applied to properly process the input signal.  Wow!
I hope you'll succeed the repair Darkover. I couldn't imagine how difficult could be removing a single chip!!
J. Alonso

[Robaroni]

When I have to remove bad chips with pins I usually cut off the pins at the chip then desolder the remains with my desoldering tool.

I'll have to experiment with hidden pins. I'll probably use a CNC mill to cut the chip down to within maybe 0.1mm of the board and see if it comes off easier.

Just some thoughts.

[bayjelly]

Is it certain that a signal that exceeds the safe output specifications will always also be either on the top or the bottom of the screen (or beyond), in all possible configurations?

I don't know the intricate details about how scopes are engineered, but this seems like an assumption. Is that assumption commonly true?

[Darkover]

I'll have to experiment with hidden pins. I'll probably use a CNC mill to cut the chip down to within maybe 0.1mm of the board and see if it comes off easier.

I don't think this are good thoughts!

1. You don't have any problem with Pin, because the device is a DFN.

2. You don't habe a problem with the pads of a DFN because they are all swimming in melted solder when you try do desolder them. Oh, and a few 0402 and one SOT23-5 around the LMH6518 are swimming, too.

3. The problem is only the exposed pad. And you have to remove the LMH without ANY destroing or scratching of the copper below the DFN because the new chip needs it. The tiny little LMH sucks 1W and it needs a realy good thermal connection and this is especially true for the exchanged part that you will solder later.

Olaf

[Darkover]

I don't know the intricate details about how scopes are engineered, but this seems like an assumption. Is that assumption commonly true?

Of course, everythink is an assumption now! But I think it is a good one.
To move from a dedicated guess to certainties it needs some measuring that give us the numbers. But the playground for the developer of the scope is small because he needs to use the dynamic as much as possible to lower the noise of the frontend and R&S is proud and very good in low noise devices.

Olaf

[2N3055]

Maybe investing in one of those would be prudent:

https://eleshop.eu/jbc-protector-for-desoldering.html

[Robaroni]

I'll have to experiment with hidden pins. I'll probably use a CNC mill to cut the chip down to within maybe 0.1mm of the board and see if it comes off easier.

I don't think this are good thoughts!

1. You don't have any problem with Pin, because the device is a DFN.

2. You don't habe a problem with the pads of a DFN because they are all swimming in melted solder when you try do desolder them. Oh, and a few 0402 and one SOT23-5 around the LMH6518 are swimming, too.

3. The problem is only the exposed pad. And you have to remove the LMH without ANY destroing or scratching of the copper below the DFN because the new chip needs it. The tiny little LMH sucks 1W and it needs a realy good thermal connection and this is especially true for the exchanged part that you will solder later.

Olaf

The pins in the case of a WDFN, like the chip in question, is that the pins are in the form of pads contained under the chip which is plastic.

A simple G-Code program on a precision CNC machine can easily remove the chip and leave the pads much easier to unsolder. Again, as I said, I haven't tried it but I propose it because anything I can do to a PCB to limit heat I do. This is why on even large chips with fine exposed pins I clip the pins at the chip and remove its body which I have done many times successfully.


I understand this may not be a procedure a lot of people can't use due to lack of the required equipment but I put it out here as a possible way to accomplish it. As far a scratching the board I'm sure I can do it without damage to the board or any components close to the chip I'm removing. Obviously I would certainly try it experimentally first.

Here's a picture of a hole in a paper clip I drilled, I might be wrong but I think I can remove a chip from a board without damage or excessive heat.

[hotze]

I wouldn't do this just because of the flying, conductive metal shavings

[Hydron]

I think it would be relatively easy to clean the area afterwards - just make sure nothing can get under the BGAs (they aren't that close), you're going to have flux etc to clean up regardless.

I have seen videos of this very technique being used to remove chips in iPhones etc that would be very hard to safely remove any other way. If I had the tools I'd go for this method myself!

[Robaroni]

I think it would be relatively easy to clean the area afterwards - just make sure nothing can get under the BGAs (they aren't that close), you're going to have flux etc to clean up regardless.

I have seen videos of this very technique being used to remove chips in iPhones etc that would be very hard to safely remove any other way. If I had the tools I'd go for this method myself!

I'm in the middle of a design (aren't we always!) but I'd like to try it and report back. I have some micro-inverters that failed on my PV array with some stubborn chips I might try it on. One of the problems today is lead-free solder with its higher melting point. I think if I did get the chip off I'd use leaded solder for the new one.

The other thing I'm thinking about is finding a way to replace the chip so if I ever have to remove it again I wouldn't have a big problem. I'd have to experiment with thermal heating but I'm wondering if I could simply glue a heatsink on top of the chip and not soldering the center tab which really adds to the difficulting in removing the chip, that would make it much easier to remove. Even putting a heatsink above it and heat paste on the center tab.. Those are things industry shys away from for cost effectiveness that we could easily do.

Just thinking out loud.

[Hydron]

Use low-melt (138C) solder on the replacement? Should be pretty easy to replace in the future with that, though it's not quite as nice as leaded solder to work with otherwise. Even leaded will make a big difference vs lead-free!

[tooki]

I like to say the picture only showed the preheating! I had a second heat gun on top. But it was not strong enought.

Let me share a picture from the actual state of work! 

The board is still okay, but of course the original LHM6518 has no offset problem anymore.

And I would never ever try something like this without a 3D microscope. It is an very interesting experience to look at 3D inside of the LHM6518 when the plastik housing falls apart, but the exposed pad is still there.

The next, but now untested idea is to use a Weller pyropen for heating on top. But I feel a little timidly now.  The pyropen has much more energy and can easy roast the board.

Olaf

Good god, man! 

The problem you ran into, I very strongly suspect, is local overheating of a fundamentally too-cold board.

When preheating, remember that every part of the board you don't preheat is acting as a huge heatsink. So the goal is to preheat the whole board, so the rest of the board isn't sucking away heat like a vampire. With hot air, this means taking the time to gently heat the entire board (I usually focus on the perimeter of the board, since the center will get hot by being surrounded by hot board). Another approach (which is likely the best) is to use an oven. You don't need to heat to close to the melting point.

Most likely, the board is soldered with SAC305 solder (melting point 217°C). If the board is at 22°C, then your localized hot air has to pump in enough heat to raise the joint by 195 degrees. Because of the heat sink effect of the ground planes in the board, this means the center of the component will need to go FAR above 217° for the pad edges to reach the melting point.

On the other hand, if you preheat the entire board to, say, 120°C, then you only need to add 97° worth of heat to the component, and because the temperature difference is smaller, the rate of heat flow will be smaller, too. This approach is much, much easier and gentler to do.

[tooki]

Use low-melt (138C) solder on the replacement? Should be pretty easy to replace in the future with that, though it's not quite as nice as leaded solder to work with otherwise. Even leaded will make a big difference vs lead-free!

Just make sure not to mix bismuth low-melt solder and lead solder. Lead dramatically reduces the melting point of bismuth-based alloys, and can inadvertently produce a weak alloy with a melting point below 100°C!

[Robaroni]

Use low-melt (138C) solder on the replacement? Should be pretty easy to replace in the future with that, though it's not quite as nice as leaded solder to work with otherwise. Even leaded will make a big difference vs lead-free!

Just make sure not to mix bismuth low-melt solder and lead solder. Lead dramatically reduces the melting point of bismuth-based alloys, and can inadvertently produce a weak alloy with a melting point below 100°C!

I'd be careful about the melting point, let's remember this part gets hot!

[Hydron]

Hmm I'd kinda just assumed 138C solder would be fine, but you're right, it's pushing ~1.6W and we're relying on PCB dissipation which will be a few 10s of C/W best case, so close to 100C isn't out of the question!

Maybe best just to stick with tin/lead (that 30-something degree difference to lead-free will be a massive help regardless), will be well and truely cooked (well, maybe not quite as much as Darkover's!) if it gets to that melting point!

[Darkover]

Hi everyone,

I made some progress! The defect IC is removed and the PCB looks good again. I increased the distance of the big heating gun at the bottom a little bit. I heated it to 150degree at bottom and probable 120degree at top.
Than I used a Weller Pyrophen gas soldering iron with the hot air nozzle. This iron hat so much energy that is can easiely toast everything,

But before I solder a new chip I will do some testsoldering with:

1. normal lead-solder paste
2. Alpha OM-520  42Sn/57.6Bi/0.4Ag (peak flow 155°C to 190°C)

Than I will decide how to solder it.