Voltage drop in a PCB power plane

[AdrianRodz1]

TL;DR Summary

Can I reduce the Voltage Drop by decreasing the copper area on a plane layer?

I'm trying to maintain a spec requirement of less than 3% drop from farthest pin to closest pin from source.
I'used the entire layer for copper but my oart does not require such a large pour accept to handle high current.

So I'm considering there's a trade off that should meet inthe middle of reduced voltage drop w/out losing required capacitance.

 

[berkeman]

Welcome to the PF.

Summary:: Can I reduce the Voltage Drop by decreasing the copper area on a plane layer?

I'm trying to maintain a spec requirement of less than 3% drop from farthest pin to closest pin from source.

Yikes! What the heck kind of currents are you pulling to cause a 3% voltage drop in a plane? What is the application? Is your plane on an inner PCB layer or an outer layer? How many ounces of copper did you specify for that plane?

 

[DaveE]

Summary:: Can I reduce the Voltage Drop by decreasing the copper area on a plane layer?

No. Less Cu = higher resistance = more voltage drop.
The resistance decreases if you reduce the length of the trace (or plane) and increases if you reduce either the width or depth of the trace.

 

[Joshy]

You'll see this formula often, which DaveE described. Some people refer to it straight away "rho L over A" in discussion.

$$R = { {\rho L} \over {A} }$$

The resistance is the resistivity $\rho$ times the length $L$ over the cross-section area $A$. The resistivity is a material parameter you'll get for your metal layer (copper in this case). I forget the resistivity often or sometimes I can only find the conductivity. Resistivity and conductivity are the reciprocal of each other.

Anyhoo: Clearly reducing the length, increasing the width or thickening the metal layer will help you reduce the resistance.

 

[phinds]

Can I reduce the Voltage Drop by decreasing the copper area on a plane layer?

@AdrianRodz1 just based on the fact that you could ask that question leads me to think that if at all possible you should consult with a more knowledgeable Electrical Engineer about your whole project. There might be other issues that you could head off.

 

[AdrianRodz1]

Welcome to the PF.

Yikes! What the heck kind of currents are you pulling to cause a 3% voltage drop in a plane? What is the application? Is your plane on an inner PCB layer or an outer layer? How many ounces of copper did you specify for that plane?

Hi Berkman,
Thanks for the welcoming..
I'm appreciating all the reply's, Thank you all!

My board is stnd .062 w/2 oz planes utilizing multiple layers to supply 100v to a single chip.

I had a choke point on one layer that I missed, so I used some of, one of my gnd lyrs. :) Problem solved

 

[phinds]

My board is stnd .062 w/2 oz planes utilizing multiple layers to supply 100v to a single chip.

But his question was how much CURRENT are you pulling to cause a 3v drop?

 

[berkeman]

to supply 100v to a single chip.

What kind of currents? Can you say anything about the application? Have you looked at augmenting the thick PCB layers with Bus Bars?

[Edit] Oops, beaten to the punch again by @phinds

 

[AdrianRodz1]

But his question was how much CURRENT are you pulling to cause a 3v drop?

I believe the answer was in my reply, was it not? "100V"

 

[berkeman]

I believe the answer was in my reply, was it not? "100V"

 

[AdrianRodz1]

View attachment 259983

appologies, approx 55amp

 

[berkeman]

appologies, approx 55amp

So a 3% drop for 100V is 3V, so 3V/55A = 55mΩ

Using the resistance equation that @Joshy gave you above, can you say what kind of resistance you expect for the combined resistances or your supply and ground return paths?

 

[sysprog]

It might help if we could see a schematic diagram of your circuit. Often a circumvention suffices to allow you to not have to definitively solve the problem. Maybe 'here and there' you could use silver instead of copper.

 

[davenn]

to supply 55A at 100v to a single chip

Far out, what type if chip (IC) is being supplied with that much juice ?

inquiring minds want to know

 

[Vanadium 50]

100V and 55A? That is likely to make all the magic smoke leak out. Once the magic smoke is gone, the chip won't work any more.

 

[Joshy]

It sounds dangerous. I had a curious look at some parts on Digi-Key I don't think these numbers are outrageous for certain applications.

We can rearrange the equation from earlier if you're curious. First: If it were me I would partner with another engineer and my supervisor to get support before trying this in a lab.

$$0.03 V \gt IR \nonumber
\\0.03 V \gt I { {\rho L} \over {A} } \nonumber
\\ \Leftrightarrow { {0.03 V} \over {I\rho} } \gt { {L} \over {A} } \nonumber
\\ \Leftrightarrow { {0.03 V} \over {I\rho} } \gt { {L} \over {Wt} } \nonumber
\\ \Leftrightarrow { {0.03 V t} \over {I\rho} } \gt { {L} \over {W} } \nonumber
\\ \Leftrightarrow { {0.03 V t \sigma} \over {I} } \gt { {L} \over {W} } \nonumber$$

I always forget the resistivity of copper, but its reciprocal is the conductivity, $\sigma$, which is easier for me to remember.

$${ {L} \over {W} } \lt { {(0.03)(100)(2.8 \times 25.4 \times 10^{-6}) (5.8 \times 10^7)} \over {55} }$$

I'm getting about 225, which seems like it won't be hard meeting that particular spec; however: You may want to take a look at your temperature requirements (IPC 2152). I'm sure I'm also overly simplifying this and wouldn't doubt there's a mistake in my quick calculation, and I would guess there are many other important specs. Partner up with someone even if you're confident.

I haven't brought up arcing because I don't know much about it although I think there's something in IPC 2221B about it. There are calculators online... I don't know how reliable it is or aligned with the guidelines: https://www.smps.us/pcbtracespacing.html

 

[Baluncore]

100V and 55A? That is likely to make all the magic smoke leak out. Once the magic smoke is gone, the chip won't work any more.

Maybe it is a power MOSFET switch that is; on, 55A * 0V = 0W; or off, 0A * 100V = 0W.

 

[phinds]

Maybe it is a power MOSFET switch that is; on, 55A * 0V = 0W; or off, 0A * 100V = 0W.

If not, then I think Vanadium has it right

100V and 55A? That is likely to make all the magic smoke leak out. Once the magic smoke is gone, the chip won't work any more.

I mean, seriously ... 5,500 Watts on a PCB? I don't think so.

 

[sophiecentaur]

Maybe 'here and there' you could use silver instead of copper.

. . . . or just use a thick copper strap in strategic places, to bypass the hr path.

 

[sysprog]

. . . . or just use a thick copper strap in strategic places, to bypass the hr path.

I was thinking of the superiority of silver over every other metal in electro-conductivity (even greater superiority in thermo-conductivity) -- you of course have to watch out for how you manage the oxide layer -- In the '70s I and and some friends (we shared some tech gear) had a WW2-era Hammarlund naval ship's radio that had some silver on the board, and that thing, with only its on-board antenna, could get clear signals that had bounced off of the ionosphere from hundreds and sometimes thousands of miles away $\dots$

 

[hutchphd]

I mean, seriously ... 5,500 Watts on a PCB? I don't think so.

We quite regularly run 20A at 120V through tiny strips of metal (called fuses) and think nothing of it. The OP did not say the board dissipates 5000W., although some of his commentary is off-putting. As a MOSFET switch this is not outrageous.

 

[Rive]

I'm trying to maintain a spec requirement of less than 3% drop from farthest pin to closest pin from source.
I'used the entire layer for copper but my oart does not require such a large pour accept to handle high current.

So if I take it right, you both have a maximal resistance and a maximal capacitance requirement, and in your design the two seems to conflict, right?

There are several ways to solve the conflict.
- You can specify thicker copper. That way you don't have to use so wide tracks/planes to have the specified resistance => less capacitance.
- You can add 'busbars' to the design: this come in various forms - basically, an external piece of thick copper
- You can add 'https://hu.farnell.com/harwin/s1621-46/jumper-link-smt/dp/1022336' to the design to reduce resistance: this of course won't work for internal layers

Ps.: well, actually the most trivial solution is to put the two points of interest closer to each other

 

[Baluncore]

I had a choke point on one layer that I missed, so I used some of, one of my gnd lyrs. :) Problem solved

It see that the problem was solved yesterday. Post #6.

 

[artis]

@AdrianRodz1 Sure it's none of my business but still your commentary has intrigued me to understand what kind of schematic is that you are building there?

The easiest way for me to make a pcb trace carry more current is to simply put solder on the trace with a soldering iron , just make a even nice surface and the trace will become some 2x if not more thicker, if you need serious increase then take a bare copper wire and again solder it on the trace. Don;'t heat the trace in one spot for too long as it will most likely cause the trace to come off from the board.

 

[sysprog]

@AdrianRodz1 Sure it's none of my business but still your commentary has intrigued me to understand what kind of schematic is that you are building there?

The easiest way for me to make a pcb trace carry more current is to simply put solder on the trace with a soldering iron , just make a even nice surface and the trace will become some 2x if not more thicker, if you need serious increase then take a bare copper wire and again solder it on the trace. Don;'t heat the trace in one spot for too long as it will most likely cause the trace to come off from the board.

And, when you're by-hand soldering, a little bit of grimy non-pretty excess flux, can be excusable when it's helped to prevent a cold joint.

 

[Baluncore]

My board is stnd .062 w/2 oz planes utilizing multiple layers ...

Multi-layer boards are difficult to modify with solder or links because the internal layers cannot be soldered and most vias do not appear on the surface. The solder mask also makes it difficult to access the surface.

 

[phinds]

We quite regularly run 20A at 120V through tiny strips of metal (called fuses) and think nothing of it. The OP did not say the board dissipates 5000W., although some of his commentary is off-putting. As a MOSFET switch this is not outrageous.

Which is Why I started my comment with "If not ..."

 

[OCR]

Pfff. . . .

Once the magic smoke is gone, the chip won't work any more.

Only a n00b or a luser would ever let his their magic smoke be gone. . . .

.

 

[anorlunda]

I think the OP question has been adequately answered.

Thread closed.