Video Teardown and Repair of an Agilent E4421A Signal Generator [TheSignalPath]

[Hugoneus]

In this episode Shahriar repairs an Agilent E4421A synthesized signal generator purchased from eBay. The unit displays the error messages “UNLOCK” and “UNLEVEL”. The cause of failure is investigated and traced. Alongside the repair efforts, the internal construction and components of the synthesizer are also presented. A short tutorial on the advantages and disadvantages of using a Darlington transistor is also presented as well as the schematic of a simple Darlington based voltage regulator. The documents for this tutorial can be found on TheSignalPath.com

Watch the video here: [54 Minutes]
http://thesignalpath.com/blogs/2013/07/21/teardown-and-repair-of-an-agilent-e4421a-synthesized-signal-generator/

More videos at TheSignalPath:
http://www.TheSignalPath.com

[mikeselectricstuff]

A more detailed teardown of a similar unit :

[KedasProbe]

During the video I was thinking what I would do next.
Adding the power supply on that point that I would not have dared to do. (without schematic)

Do you have an idea about why the darlington failed after probably a long time?

"cat scan"    don't go in the joke business though

[SeanB]

The darlington probably failed because of inadequate heatsinking. It is providing a 1.5A load from a 12V rail and is dissipating around 5W of heat. The die probably is running at around 120C, which is not good for long term reliability.

A new thermal pad and a top of case extra heatsink, or mount a bigger heatsink in the case near it and use 3 flying leads to connect to the board would be a lot better solution. It would be interesting to measure the temperature of the device after it has been on and with the case closed for 20 minutes, it probably will be at a case temp close to 70C, and inside will be more.

[kg4arn]

The darlington probably failed because of inadequate heatsinking. It is providing a 1.5A load from a 12V rail and is dissipating around 5W of heat. The die probably is running at around 120C, which is not good for long term reliability.

A new thermal pad and a top of case extra heatsink, or mount a bigger heatsink in the case near it and use 3 flying leads to connect to the board would be a lot better solution. It would be interesting to measure the temperature of the device after it has been on and with the case closed for 20 minutes, it probably will be at a case temp close to 70C, and inside will be more.

Do you guess that it is a fundamental design flaw or is one of the boards faulty and drawing excessive current?

[Hugoneus]

The darlington probably failed because of inadequate heatsinking. It is providing a 1.5A load from a 12V rail and is dissipating around 5W of heat. The die probably is running at around 120C, which is not good for long term reliability.

A new thermal pad and a top of case extra heatsink, or mount a bigger heatsink in the case near it and use 3 flying leads to connect to the board would be a lot better solution. It would be interesting to measure the temperature of the device after it has been on and with the case closed for 20 minutes, it probably will be at a case temp close to 70C, and inside will be more.

I went back and measured this, it was around 45C. It is directly tied to the case of the unit and there is a lot of metal around it.

[Hugoneus]

The darlington probably failed because of inadequate heatsinking. It is providing a 1.5A load from a 12V rail and is dissipating around 5W of heat. The die probably is running at around 120C, which is not good for long term reliability.

A new thermal pad and a top of case extra heatsink, or mount a bigger heatsink in the case near it and use 3 flying leads to connect to the board would be a lot better solution. It would be interesting to measure the temperature of the device after it has been on and with the case closed for 20 minutes, it probably will be at a case temp close to 70C, and inside will be more.

Do you guess that it is a fundamental design flaw or is one of the boards faulty and drawing excessive current?

I don't think the boards were drawing an excessive amount of current. I can check this with my E4432B generator.

[SeanB]

If it is running at 45C then it is likely the original transistor either was not mounted to the right torque ( either too high or too low, high warps the tab and makes it bow and low leaves an air gap) or it was defective and was only partially bonded internally to the tab so it got hot spots forming on the die which killed it. Also possibly the insulating bush on the mounting screw was too long and while the screw was tight the device was actually not being clamped because the bush was absorbing all the compression force. Also likely is the device was soldered before mounting, and the pressure on the leads deformed the tab and reduced heat transfer ability, or cracked the die loose partially. Only way to see is to decap the device with hot fuming nitric acid and see if the die is loose or partially bonded to the tab. Xray would also work.

Not likely there was a high load, the device would have failed and cooked itself well within the original warranty period or the short would have dropped the power rail to ground with time.

[Hugoneus]

If it is running at 45C then it is likely the original transistor either was not mounted to the right torque ( either too high or too low, high warps the tab and makes it bow and low leaves an air gap) or it was defective and was only partially bonded internally to the tab so it got hot spots forming on the die which killed it. Also possibly the insulating bush on the mounting screw was too long and while the screw was tight the device was actually not being clamped because the bush was absorbing all the compression force. Also likely is the device was soldered before mounting, and the pressure on the leads deformed the tab and reduced heat transfer ability, or cracked the die loose partially. Only way to see is to decap the device with hot fuming nitric acid and see if the die is loose or partially bonded to the tab. Xray would also work.

Not likely there was a high load, the device would have failed and cooked itself well within the original warranty period or the short would have dropped the power rail to ground with time.

Thanks for your contribution Sean. I'll try to see if I can get access to an x-ray machine.

[SeanB]

Post it to Mike, he still has one Xray scanner that works.

[KedasProbe]

What about a current spike due to hot plug-in of a board?
Or an static charge that somehow got on the board before plug-in?

Just thinking of possible ways to beak it.

edit:
If the collector of one of the two transistors would have become disconnected then I assume that would fit the behaviour. (should be easy to check by measuring the base collector current ratio)