"Upgrading" a DC-DC Converter / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

"Upgrading" a DC-DC Converter

1 2 next

Move Thread

LAN_403

Inducktion

Thu Mar 14 2019, 06:50AM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

Hi everyone!

So, to upgrade a DC-DC converter;

I already replaced the fets with better rated ones.

Next on the list is the inductor....

The converter runs at 150 kHz, though I can change it to run at 225 kHz. As it stood the original inductor was rated at 15 uH.

If I want to increase the power it can handle, I obviously need to change the inductor to something better.

What core material would be better? Ferrite? Powdered iron? Some sort of fancy alloy?

Should I shoot for higher permeability? Is there a general guideline towards what material to not use?

thanks!

Conundrum

Thu Mar 14 2019, 07:19AM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4059

Good tip, thanks.
My own experiments use a hand made primary constructed from twisted clock wire. Its a pig to wind but worth it!
Incidentally I am using the described modification to build a battery charger so watch this space! Actually more like "More POWAAAAH Igor!!"

2Spoons

Fri Mar 15 2019, 12:23AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

post a schematic.

hen918

Fri Mar 15 2019, 08:10PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

Upgrading a commercial DC-DC converter is basically impossible unless it has been really badly designed in the first place. A MOSFET may seem to be of a lower rating than an alternative, however there are many factors that go into the selection of the MOSFETs and the other components, not just price. For a DC-DC converter, the design process goes like this:
Power & voltage rating changes Topology, component size/package (including transformers, inductors, etc), frequency, current/voltage rating of switches.
Frequency changes Inductance of inductors, turns on transformer, switching component speed, number of cores in high current wire (litz), Gate driver selection and design.
The switching frequency changes everything, from losses to the compensation network in the feedback, if you increase it, you are likely to make things worse. Losses depend on both gate charge and the Rds(on) of the switches, a higher switching frequency is likely to increase peak currents and switching losses, leading to rapidly overheating MOSFETs, MOSFETs that have a lower Rds(on) will have a higher gate charge, lowering conduction losses whilst increasing switching losses.

No (power) inductor operating at more than 1kHz should be powdered iron.
If you really want to go ahead you should increase the frequency and drop the inductance to maybe 10 ÂµH whilst ensuring the resistance is lower. Keep the material the same (N87 or N97 or similar). Don't be surprised if things get hotter.

Inducktion

Sat Mar 16 2019, 01:22AM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

^^^ Is the schematic, albeit it has a tacked on op amp for constant current control as well.

2475e5e321bd0be14c02cef353dbea78

^^^ is the true schematic, though it's really blurry and I apologize. Again, the only difference between the two, is the op amp circuit added on.

As far as upgrading the fets, yes, I know that typically higher rated fets have greater switching losses due to increased gate charge; that's why I tried to get fets with essentially the same switching characteristics as the originals, but better.

The new fets are these, * by Vishay. Gate capacitance of 5100 pF, and the original fets were *, with a gate capacitance of 5000 pF.

Otherwise, the thunderfets are better in almost every way compared to the originals, AND, are known to be genuine.

Additionally, I chose a new inductor material; Magnetics-inc makes a variety of toroids with a material called KoolMu, designed for low loss and high saturation, so I figured it would be a good choice.

2Spoons

Sat Mar 16 2019, 03:38AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

KoolMu is a terrible choice for your application. Compared to 3C90 ferrite, its specific power loss is ten times higher, even for the best KoolMu (125u) - based on 200kHz @0.1T. That means 10x the core heating compared to plain old 3C90.

Use a ferrite core with an air-gap.

Inducktion

Sat Mar 16 2019, 03:53AM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

What type of material should I shoot for? I have some ferrite toroids, I think they use W ferrite (

Or is an air gap a requirement for use with a DC-DC converter?

Edit:

This is the original toroid material:

(under KS090-125A)

2Spoons

Sun Mar 17 2019, 10:00PM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

'W" material has a high mu, so with no air-gap (ie toroid) it will saturate easily. Its really meant for transformers and common mode chokes.
The toroid in your link is a 'pressed powder' type, and so effectively has a distributed air-gap - resulting in less nH/n^2 and higher saturation flux.
If you are trying to upgrade your converter I assume you want more power, so you can expect more loss too - unless you choose lower loss materials. Sendust looks about the same as KoolMu, and both are worse than ferrite (power grades) for power loss. Remember commercial design decisions also include cost as a parameter (sometimes its the highest priority), and that gets balanced against things like efficiency. Your priorities might be different - have a look at TI's webbench power converter designer - it has a knob you can adjust to balance size, cost and efficiency.

Inductors for this type of power converter need to be designed. Start with target output power, then calculate the peak current requirement and inductance you need. Pick a core material suitable for the frequency you will operate at, and that is obtainable in a core shape you like. I would pick a shape that comes pre-gapped if possible as it makes everything simpler - I like the RM am P style cores, but they may not come big enough for higher power systems. Power throughput is pretty much a function of frequency and core volume, so knowing the first two you can gauge how big a core you need (scan though the datasheets). Once you pick a core you can calculate the number of turns, and the wire gauge that will fill the bobbin ( to get minimum DCR). If the wire gauge is bigger than 2x skin depth you should consider multi-filar winding or litz wire, otherwise you are just wasting copper.

Does that help?

types L, R, P, F and T would all work for you. R, P and F seem to be available in more sizes. L looks like its intended for high frequency converters.

I'd say any of the 3C9x ferrites would be OK.

Final choice comes down to price, availability and any special requirements.

Inducktion

Mon Mar 18 2019, 12:32AM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

That all makes sense, thank you. My biggest issue is just trying to find where to buy the ferrites and cores; I've been looking on eBay since I can't seem to find them on arrow or newark reliably. Is there a website or someplace I should be looking specifically, or am i just not finding the right category on arrow/newark?

2Spoons

Mon Mar 18 2019, 02:34AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

those are nz versions of US sites but i'm sure you'll get redirected. Search 'ferrite' then pick the category that looks most like 'cores' while avoiding anything that is for EMC or RF

I used these searches on ebay
ferrite core E -emi -rf -supression -toroid -filter -supressor -toroidal
354 results

ferrite core rm -emi -rf -supression -toroid -filter -supressor -toroidal
107 results

ferrite core p -emi -rf -supression -toroid -filter -supressor -toroidal
286 results

You may also want to search by ferrite type instead of core shape eg N87 (TDK/Epcos)

Your search-fu needs practice, young grasshopper.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob