Daniel Giesbrecht, P.Eng.
@giesbrecht.bsky.social
He/Him - Electronics Engineer, Consultant, Owner of Soley Engineering (soleytech.com),
FOSS supporter, high speed digital design enthusiast
Today I tried out OpenEMS. Here's a Gaussian-pulse with a bandwidth of ~5 GHz traveling down a micro-strip transmission line, done in 55 lines of python.
I'm excited to build more complex simulations, like a micro-strip layer transition with nearby vias.
I'm excited to build more complex simulations, like a micro-strip layer transition with nearby vias.
November 12, 2024 at 4:58 AM
Today I tried out OpenEMS. Here's a Gaussian-pulse with a bandwidth of ~5 GHz traveling down a micro-strip transmission line, done in 55 lines of python.
I'm excited to build more complex simulations, like a micro-strip layer transition with nearby vias.
I'm excited to build more complex simulations, like a micro-strip layer transition with nearby vias.
After over a week of prep, I started running an accelerated life test that I’ve been planning for months.
October 21, 2024 at 3:42 AM
After over a week of prep, I started running an accelerated life test that I’ve been planning for months.
My Fluke megohmeter and surface probes arrived today, and I grabbed the first static-dissipative tile sample I could find to try them out. A pass for the tile is 10^6-10^9 Ohms.
October 11, 2024 at 3:55 PM
My Fluke megohmeter and surface probes arrived today, and I grabbed the first static-dissipative tile sample I could find to try them out. A pass for the tile is 10^6-10^9 Ohms.
Sometimes dead-bugging is the best way. 8x 470uF Al-polymer caps on 10 AWG wire.
September 13, 2024 at 5:33 PM
Sometimes dead-bugging is the best way. 8x 470uF Al-polymer caps on 10 AWG wire.
Here's a board I designed for a client a few months ago! It's basically an industrial controller. It's amazing what some lights, a piece of paper and some patience can do. I know very little about photography, but this looks a lot better than the stills I was taking on my phone.
June 23, 2024 at 11:28 PM
Here's a board I designed for a client a few months ago! It's basically an industrial controller. It's amazing what some lights, a piece of paper and some patience can do. I know very little about photography, but this looks a lot better than the stills I was taking on my phone.
Step 0 of learning CFD simulation: learn how to create an air volume around my solid bodies (copper trace on FR-4). It's not _hard_, but there just aren't that many tutorials out there.
May 11, 2024 at 9:49 PM
Step 0 of learning CFD simulation: learn how to create an air volume around my solid bodies (copper trace on FR-4). It's not _hard_, but there just aren't that many tutorials out there.
The $$$$ price-tag of a PDN probe is starting to make sense...
February 8, 2024 at 5:28 PM
The $$$$ price-tag of a PDN probe is starting to make sense...
In ParaView, you can set a unique colour for out-of-range data. This makes it really easy to see what areas of the copper geometry have an unacceptably high current density.
January 24, 2024 at 7:17 PM
In ParaView, you can set a unique colour for out-of-range data. This makes it really easy to see what areas of the copper geometry have an unacceptably high current density.
-83 dB = 1.8 mΩ
1.8 mΩ is still ~ 20% error, but
* I don't have a torque wrench yet,
* I haven't done anything to de-embed the fixture from the measurement,
* the CM choke only has -30 dB of CM insertion loss.
I think if I improve each of these points, I can get a lot closer to 1.5 mΩ.
1.8 mΩ is still ~ 20% error, but
* I don't have a torque wrench yet,
* I haven't done anything to de-embed the fixture from the measurement,
* the CM choke only has -30 dB of CM insertion loss.
I think if I improve each of these points, I can get a lot closer to 1.5 mΩ.
January 21, 2024 at 12:15 AM
-83 dB = 1.8 mΩ
1.8 mΩ is still ~ 20% error, but
* I don't have a torque wrench yet,
* I haven't done anything to de-embed the fixture from the measurement,
* the CM choke only has -30 dB of CM insertion loss.
I think if I improve each of these points, I can get a lot closer to 1.5 mΩ.
1.8 mΩ is still ~ 20% error, but
* I don't have a torque wrench yet,
* I haven't done anything to de-embed the fixture from the measurement,
* the CM choke only has -30 dB of CM insertion loss.
I think if I improve each of these points, I can get a lot closer to 1.5 mΩ.
I spent the day measuring a 1.5 mΩ shunt to make sure I'm on the right track for low-impedance PDN measurements. I soldered it to a SMA fixture and measured it with a CM choke to reduce braid-error.
January 21, 2024 at 12:13 AM
I spent the day measuring a 1.5 mΩ shunt to make sure I'm on the right track for low-impedance PDN measurements. I soldered it to a SMA fixture and measured it with a CM choke to reduce braid-error.
I got my VNA, so I decided to compare the response of this jury-rigged CM choke.
January 20, 2024 at 7:29 AM
I got my VNA, so I decided to compare the response of this jury-rigged CM choke.
New VNA day! 😍
January 17, 2024 at 8:57 PM
New VNA day! 😍
Finally done a design I've been working on for months 😮💨
December 14, 2023 at 6:10 AM
Finally done a design I've been working on for months 😮💨
Here's the measurement of the 5 mΩ (±1%) resistor I made. It starts to become inductive at between 100 kHz - 1 MHz, which is to be expected. At low frequencies, it's pretty flat at 4 mΩ. This is an error of ~ 20%, but I'm actually pretty happy with this result. I'll explain why..
November 19, 2023 at 11:28 PM
Here's the measurement of the 5 mΩ (±1%) resistor I made. It starts to become inductive at between 100 kHz - 1 MHz, which is to be expected. At low frequencies, it's pretty flat at 4 mΩ. This is an error of ~ 20%, but I'm actually pretty happy with this result. I'll explain why..
First, I tried different numbers of turns of 30 AWG twisted pair wire around a ferrite core I had on hand. I stopped at 19 turns because that's the length of wire I pre-twisted 😅
Here's the impact that multiple turns had on CM insertion loss:
Here's the impact that multiple turns had on CM insertion loss:
November 19, 2023 at 11:21 PM
First, I tried different numbers of turns of 30 AWG twisted pair wire around a ferrite core I had on hand. I stopped at 19 turns because that's the length of wire I pre-twisted 😅
Here's the impact that multiple turns had on CM insertion loss:
Here's the impact that multiple turns had on CM insertion loss:
I've been acquiring more lab gear with goal of measuring component and PDN impedance. I don't have all the pieces I need, but I got my Arb Wavegen delivered and I wanted to play with it today and see how far I could push my current setup. I measured a 5 mΩ resistor today.
November 19, 2023 at 11:13 PM
I've been acquiring more lab gear with goal of measuring component and PDN impedance. I don't have all the pieces I need, but I got my Arb Wavegen delivered and I wanted to play with it today and see how far I could push my current setup. I measured a 5 mΩ resistor today.
I spent the morning figuring out a workflow to delete everything I don't need out of my layout, converted the Gerbers to STEP models of the copper, and pruned any remaining cruft in my 3D CAD. Now the geometry is ready for meshing. These are my 5 V rails.
November 13, 2023 at 9:26 PM
I spent the morning figuring out a workflow to delete everything I don't need out of my layout, converted the Gerbers to STEP models of the copper, and pruned any remaining cruft in my 3D CAD. Now the geometry is ready for meshing. These are my 5 V rails.
The current volume density plot makes it really obvious what parts of the copper planes aren't contributing much and could probably be chopped out.
November 13, 2023 at 9:09 AM
The current volume density plot makes it really obvious what parts of the copper planes aren't contributing much and could probably be chopped out.
I successfully burned through importing the mesh, setting up the parameters in Elmer-GUI and getting the simulation to converge. That's it for today!
November 13, 2023 at 9:02 AM
I successfully burned through importing the mesh, setting up the parameters in Elmer-GUI and getting the simulation to converge. That's it for today!
Last time I did meshing, I was using Gmsh. This time around, I decided to give Salome a shot. After 4+ hours of frustration and some key help from a colleague, I've got a mesh that looks suitable and is split into multiple materials.
November 13, 2023 at 3:18 AM
Last time I did meshing, I was using Gmsh. This time around, I decided to give Salome a shot. After 4+ hours of frustration and some key help from a colleague, I've got a mesh that looks suitable and is split into multiple materials.
I have multiple copper planes that are tied together with SMT low-resistance jumpers. First, I'm going to refresh myself on how to simulate multiple bodies with different resistances. I'm representing this 0508 jumper with a simple shape.
November 12, 2023 at 9:20 PM
I have multiple copper planes that are tied together with SMT low-resistance jumpers. First, I'm going to refresh myself on how to simulate multiple bodies with different resistances. I'm representing this 0508 jumper with a simple shape.
I've found that the models from Snap-EDA are generally a lot more representative of what's on the board.
August 24, 2023 at 6:45 PM
I've found that the models from Snap-EDA are generally a lot more representative of what's on the board.
I feel bad criticizing something that's provided to me for free, but the 3D models from Ultra-Librarian leave something to be desired compared to, for example, the part render from Digikey's site.
August 24, 2023 at 6:39 PM
I feel bad criticizing something that's provided to me for free, but the 3D models from Ultra-Librarian leave something to be desired compared to, for example, the part render from Digikey's site.