Kelvin connection in Eagle

Hi Joshua,

It looks like there are more issues than just your connection widths. If you look you actually have the signal polygon fills shorting with your sense trace. You need to do your footprint and your schematic symbol to allow the 4 terminal connection. So on each side the two outer pad pieces connect to the corresponding main resistor pin and the central pad connects to a separate current sense pin with a separately named net. I'm at work at the moment so can't explain further but if you are still having issues I will try and explain better later.

With regards to the connection, turn off thermals for the polygon and then it will connect fully.

Best Regards,

Rachael

Rachael,

Thank you! Ya I fumbled around with it more and improved the resistor footprint to have pads better defined and that helped fix my problem. I'm just getting started with Eagle so I'm learning all the intricacies of how to properly create a part.

On 08.10.2018 17:06, Joshua Vandermeer wrote:

Hey everyone!

 

I'm still learning eagle, so some obvious things are still difficult for me to wrap my head around, but I'm learning!

I think you are way overdoing this circuit, and at the same time you

weaken it. If the resistor is tightly dimensioned for your power, you

need the full pad both for heatsink and current.

Just route the kelvin probes out on the side of the SMD pads. The point

of the kelvin probes are to avoid the voltage drop in the high current

main path. If you just bring out some probes on the side of the smd's,

there will be close to no current generating voltage drop on those

tracks. If you want max noise immunity, route the two tracks

differential back to the sensing circuit. For even more noise immunity

route them through directly down a VIA to a different opposite layer.

The implementation of the kelvin probes is done at the board layouyt,

not the schematic, but if you need others to do the routing job based on

your schematic drawings, draw it like you would route it and put a

comment in the schematic at the info or guide layer.

autodeskguest  wrote:

 

I think you are way overdoing this circuit, and at the same time you

weaken it. If the resistor is tightly dimensioned for your power, you

need the full pad both for heatsink and current.

 

Just route the kelvin probes out on the side of the SMD pads.

Joshua does say in his first post "My supervisor has asked me to implement the following pad layout for the kelvin connection with this shunt resistor". It might not be what you would do, but it is what he was asked to implement by his supervisor.

Best Regards,

Rachael

On 15.10.2018 11:01, rachaelp wrote:

  wrote:

 

I think you are way overdoing this circuit, and at the same time you

weaken it. If the resistor is tightly dimensioned for your power, you

need the full pad both for heatsink and current.

 

Just route the kelvin probes out on the side of the SMD pads.

 

Joshua does say in his first post "My supervisor has asked me to implement the following pad layout for the kelvin connection with this shunt resistor". It might not be what you would do, but it is what he was asked to implement by his supervisor.

 

Best Regards,

 

Rachael

Fair enough Rachael

If this was me, I would go back to my supervisor and demonstrate my

skills and tell him about this implementations weakness but it may not

be the right thing for Joshua this time. That's up to him.

To add more feedback to his supervisor, Im not even sure if a

manufacturer would like this implementation, due to it's potential

bistable nature. When the lead melts, this got a bistable nature of not

falling to rest on the large pads, but may be pulled to one side or the

other.

Thank you both for the great advice. I agree that the implementation I was shooting for was very overcomplicated. I'm an intern right now and trying to take as much instruction as I can. I was given this document to follow, but there was a miscommunication and he just wanted me to learn from it, not actually implement it.

Optimize High-Current Sensing Accuracy by Improving Pad Layout of Low-Value Shunt Resistors | Analog Devices

Ended up just taking the kelvin connection from the inside of the pads.

On 15.10.2018 17:19, Joshua Vandermeer wrote:

Thank you both for the great advice. I agree that the implementation I was shooting for was very overcomplicated. I'm an intern right now and trying to take as much instruction as I can. I was given this document to follow, but there was a miscommunication and he just wanted me to learn from it, not actually implement it.

 

Optimize High-Current Sensing Accuracy by Improving Pad Layout of Low-Value Shunt Resistors | Analog Devices (http://www.analog.com/en/analog-dialogue/articles/optimize-high-current-sensing-accuracy.html)

 

Ended up just taking the kelvin connection from the inside of the pads.

Interesting experiment there.. Using constant current source may not

reveal all the weaknesses tho. Voltage applied to get to that current

may vary to compensate higher R(solder) on the smaller area pads, and

this may cause undesired effects. Also thermal effects may increase

because of the same problem.

Your conclusion seem to be something what I would do, but this is very

dependant on how the resistor is actually built. Choosing the shortest

path to the actual start of the resistor material would give best result.

Because of the resistor physics, you could also analyze the result on

different fine tuned positions of the resistor, but these are errors

that are hard to control in mass production.

If you really would need a high precision, high yield kelvin probe, some

extra attention to all the side effects may be required, but usually you

end up with some calibration process. This is why noise is more of a

real problem than offset errors.

I had a similar need.

What I ended up doing was to create a net class with special trace widths different from the rest on the board. THen When this is autorouted it always is routed directly to where it should be, nothing shared.

--

To view any images and attachments in this post, visit:

https://www.element14.com/community/message/277798