#3 π My Playbook to Finance your deep tech Startup and Rac with Dowell
The Newsletter of High Frequency people
π Hello! Dr. Molina here! π¨βπ§
Thank you for reading and welcome to the jungle my Newsletter about everything Iβm learning during my journey at Frenetic, and my favorite topic, High-Frequency Magnetics π.
π΅ Disclaimer! Before hurting anyone, I write βοΈ with a touch of humor, and I can make mistakes (I will, for sure), donβt take me too seriously π.
Building a Power Electronics Company
Ten years ago I was working toward my Ph.D. I saw the old methodologies used in power electronics and that motives me to create a company and change the whole industry.
It took me another 3 years to find the moment, but I was decided. During the nights, I read Steve Jobsβs biography of Walter Isaacson to feed my entrepreneurship feelings and show me how important the design is. During that time, a specific problem appears in front of me and I created my first project, the SpCard.
I presented the idea to the University Awards of Startups (ActuaUPM) and I was selected. They gave the motivation with their lectures (Friday afternoons) to build the company.
This was the beginning of the trip of building Power Electronics companies, because I know I can fail but Iβm confident in my long-term vision, which I tune every day, following my references like Bell Labs, Unitrode or Compaq, nothingβs gonna stop me now.
My playbook to finance your deep tech project
Financing the project Itβs very complex for a PhD electronic engineer because our market is not very common in the VC industry and the problems that involve hardware are not easily financed in Europe. You need to convince them about your vision and business potential. I have seen the same problem for other scientific areas.
Some engineers have asked me about how to finance their projects. Here is my playbook.
You need some initial money from your pocket and your relatives. Depends on the type of project, but I would say, enough money for surviving 4-6 months. During this time, you need to minimize expenses and work on the product as crazy. In Spain, between 10-20 k⬠per team member, considering only digital tools, basic OPEX, and living expenses.
In some countries, there are financing lines from the governments as loans with benefits for entrepreneurs. In Spain we have ENISA. With your initial investment, they support you with x2 times your personal investment and you will start returning the money 2 years later without any personal risk. With 2 members, an initial investment of 30kβ¬, you can ask for another 60kβ¬. Now you have 90kβ¬ for the project.
During the first year, get out of the building and discover your customers. if you donΒ΄t feel embarrassed about your product every month, you are late in showing your product to your customers. But the most important task here is to try to understand the real problem of your potential customers.
Once you have cleared the problem you want to solve, you need some metrics of your MVP (or Demo) and conversations with potential customers.
Then, and only then, you are ready for looking for some partners. Yes, partners. Because you donΒ΄t need investors, you need people, willing to invest 200-300kβ¬ in you, not in your product. In my case, I had two initial partners, my first partner was my mentor (and a very good friend), Francisco Berlanga, and the second partner, who also invest 350kβ¬, was JosΓ© Miguel Herrero from Big Sur Ventures.
If you have any project in your hands, donΒ΄t hesitate to write me βοΈ.
Team: Humans
The most complicated stuff is not the technical part, but the humans. You need to hire people, build teams and create a kind of community with culture, rules, and a mission. Your mind will blow up.
Get out of the building
Your ideas will change dramatically once you realize what makes happy engineers happy. Be ready for unexpected feedback. The interviews I had at every conference gave me a lot of tips for thinking about the next big thing.
Building a Sales Process
The first sales are leaded by the founder, but if things are moving forward, you need to hire salespeople and create a process. I have enjoyed a lot this process with my friend and colleague Michael who is since now been invited to write his experience in this Newsletter π. This process is probably the most unknown for technical founders.
Design the Product
Deep tech products are created by Ph.D., MSc, and engineers with very good technical skills. However, no one knows the technology behind google or people donΒ΄t buy a Tesla for the control algorithms.
The Product packaging (Not only physically) is everything, and as Chris Hladcuzk wrote here:
Keep everything as simple as possible. When I say everything, I mean, everything.
Business Model. How can I buy it? What am I buying?
What does the product do?
How can I start using it?
Iβm facing the challenge of building a big Product team. I have hired Xiana, who together with Edu, LucΓa and the Marketing team are building products based on data and customer feedback.
I have learned from books (Inspired is great) and Newsletters like Suma Positiva and Lennyβs Newsletter the basics of Product design. Moving the learnings to the company, we are launching our User Advisory Program, where we will give access to our internal technology to some highly qualified users before being public to get feedback from them.
I have no clue how the product will evolve in the coming years, we have a vision about the future but, the product will be the interface of the engineers to reach the future, therefore, they will decide how the product will looks like.
A Beginnings Tale
I remember the first time I showed to my friend (Who I admire a lot) Angel Riesgo a SpCard (the green board below) and he told me,
-Where is the box to cover these ugly electronics?
-There insβt any boxβ¦Itβs just like this π€.
And now, letβs go to the technical stuff.
Dowell, the first
In 1966, a guy called Dowell published a method that is still used for estimating the Rac/Rdc ratio and estimating the losses. My friend and colleague at Frenetic Alfonso have a lot to say about the evolution of these calculations, but today I will just comment on this classic method.
Basically, with a basic equation, considering the number of turns per layer, conductors diameter, the RDC of those wires, and the frequency of the current, you can predict your Rac and the total losses.
After all my comments about how complex is the Proximity effect and the simplicity of the Dowell model, you could be surprised. Iβm not. We are engineers, we like to build things and see what happens later.
In other editions, I would like to talk about other models and methods of estimating the Rac/Rdc ratio. However, now that we have arrived at the Rac concept, maybe we can answer the original question.
In this article, you can read about the topic and you can also use a spreadsheet for calculating the Dowell value.
Can we measure the Rac?
As you have seen in the first and the second article there are some physical effects that appear between conductors at high-frequency currents. Therefore, when you are designing a high frequency magnetic, the losses in the windings, arenΒ΄t the resistance of the cooper multiplied by the square of the current (current density at DC). These effects (Skin and Proximity) have an impact on the real current density through the conductors. The skin effect tries to concentrate all the current on the surface of the conductor, which selecting the appropriate area, could get in the best case, 80% of usability of the conductor area.
When we consider the impact between layers of conductors, the proximity effect, repeats the same process, induced currents in the reverse direction, reduce the current density in the main direction. This effect changes with the frequency and the current ratios.
In the end, we can measure the Rdc, of the wire and calculate the current density for DC currents and use the rule of having around 5-6 A/mm2. Nevertheless, the RAC is not a static physical parameter, therefore, NO! Rac is not measurable directly as an impedance.
One of my friends, considers the Rac, a concept. Because it helps you to minimize losses, doing a comparison with the Rdc value, which is measurable.
Professor Charles Sullivan has published a method for getting an estimation of the Rac, applying several measurements to the Magnetic component.
One of the main reasons, the Rac canΒ΄t be measured directly is the impact of the core in the measurement. When you measure the Real part of the impedance, you are also measuring the resistance of the core. In his method, basically tries to measure the core resistance, neglecting the wire resistance, using a very thick wire. Later, use this value and the Real part of the impedance, to calculate the Rac.
We have tried to implement this method and itβs very complex, therefore, at Frenetic we have created our own measuring and estimation method.
How can we improve the Rac?
During the publications of the previous post on my LinkedIn, I received a question about how to improve the proximity effect (Thank you Bilal for the question!!).
The comment of Bilal is quite interesting because he also relates the design techniques with manufacturability. John Ryce from the community shared a link with some tips from Dartmouth University with some tips.
Some basics of reducing Rac are:
Select the wire diameter to be relatively lower than the skin effect.
Minimize the number of layers. 1 layers design is perfect.
For more than 2 layers, interleave.
Take care with the faults interleavings.
Explore different wire are vs number of parallels combinations (Increasing parallels doesnΒ΄t correspond directly to decrease losses)
Using the whole coil window doesnΒ΄t mean lower winding losses, because of the proximity effect.
Simulate the losses at Frenetic π.
If youβre finding this newsletter valuable, considerΒ sharing it with friends, or subscribing if you havenβt already.
Sincerely,
Chema π
Nice to see the always nice tips related to losses in the newsletters!. Magnetic components are likely the most complex products to model when we consider passive electronic components (i.e resistors, capacitorsβ¦.. ) due to this effects on wires, core lossesβ¦ Congratulations on the success story you have built with SP Control and Frenetic, from the very beginning to what itβs nowadays. No doubts that thereβs bright future ahead with such a great motivated and professional staff supporting this goal to make Power Electronics more efficient systems.