24-42V to 12V regulator - any advice?

Sounds like a rather clear case for a synchronous buck converter.

Eeeuw. I was already used to toroids back then and >100 kilohoitzes. Pot cores were ok as well for larger stuff, cost less.

Vishay-Dale has very large ones so I don't see a power limit here. But I see an EMI nightmare if you did that :-)

Use a "shielded" inductor. Essentially they are glorified pot cores.

Nah, use a modern switcher, 500kHz to 1MHz, automotive grade. For example, the LTC3810 comes in 100V grade. No catch diode needed :-)

to

V.

.)

st. =A0

or

of

le

he

the

at

sgs L296

I like the idea of extra efficiency from that...

Thanks for your suggestion!

However, I am feeling a bit intimidated, in comparison to what I had in mind, by the largish component count including 3 external MOSFETs, combined with that IC costing more than the one that I had in mind.

Your suggestion appears to me to place a higher value on efficiency than my client here and I am giving. Reducing power consumption by a couple of watts appears to me to matter little for something mostly used while being powered by automotive power (of oddball-higher voltage).

What I had in mind is LM2576, 12V-out, HV version. The inductor gets somewhat substantial, but I like the much lower external component count, especially the much lower external semiconductor count.

My thinking now is to take on a couple watts more heat production and beefier versions of 3 components. I currently favor that over adding 3 external MOSFETs, 1 external diode, 2 external capacitors, and 5 external resistors.

Best Regards,

Thanks for the suggestion.

However, that costs more than what I had in mind (LM2576-HV), by more than the cost of a beefier inductor that LM2576-HV requires. Cost of parts and assembly matter more to my client here than energy efficiency does.

Furthermore, I am feeling intimidated by the 46V maximum input voltage in some oddballish automotive applications with typically-42V. That would be not only the mostly-stalling-short-of-fruition 42V cars, but also the oddball "36 volt" (~42 volts with engine running) trucks or whatever with

Any comments on LM2576-HV 12V, or suggestions on improvement therefrom?

(Reliability and cost of parts/assembly largely matter more than energy efficiency. Smaller "regulator product" size is 1st consideration after cost and reliability, likely assisted somewhat by improved energy efficiency so as to produce less heat.)

What are the physical restraints?

Size?

Case material?

Temperature?

Lower efficiency means more heat. The higher component cost of a synchronous switcher can be less than the heatsinking for a lower efficiency design.

The switcher will be all surface mount, presumably using automated assembly. Heatsinking means manual assembly which is $$$.

Preferably 75 mm by 44 mm by 23 mm, not an absolute requirement.

Trying to use ABS plastic. I might change the plan to extruded aluminum if everything fits in a box small enough to cause overheating if the box is plastic.

This would be used in occupied vehicles. I would say up to 50 C ambient.

Yes, I am thinking all surface mount. I figure an LM2576-12 with 1.5A load current will produce 1.2 watts of heat with 24V input voltage, somewhat less at higher input voltages. That sounds doable to me with surface mounting, with thermal vias and ~10 square cm of copper on each side. However, I seem to think total heat production of the circuit is going to be about 2-2.5 watts, likely needing an enclosure somewhat larger than the 75 x 44 x 23 mm that I said above if the enclosure is plastic.

That was just a suggestion, a chip I have used in designs. There are cheaper sync buck chips out there. LTC chips have the advantage that you can simulate everything, but of course that has its price.

It's not just power savings, it's also the avoidance of large heat sinks or forced air cooling. Typically you don't have a lot of space on vehicles.

I wouldn't want a device designed for 40V max in a vehicle. And it violates your spec of 42V even if some datasheets may put abs max at 45V.

Just don't do it with a 40V device :-)

For various reasons I am not a fan of chips with integrated power devices at such large loads. Maybe someone else has an idea for 100V device here. 24V looks like a European truck. Mind load dumps and similar effects. Otherwise the new regulator could already be dead by the time that big Diesel starts rumbling.

Load dumps and transients are a real issue. I would use MIL-STD-1275 as guidance for what to expect.

My brother in law is doing a truck supply at the moment and is using the LM5085 which is specd at 76V abs max. He has some input circuitry to protect that from transients.

SNIP

Look carefully at the schottky diode junction temperature.

Can you get the heat away from the various components in the space?

I just plugged your requirements into National Webench. I think I would be looking at the LM5088. 92% efficiency and estimated cost of under $3. The chip is rated at 75V.

Yes, I am giving consideration to heat. I am expecting up to 1.2 watts and usually not much less to be produced by the regulator IC, as well as a total of 2-2.5 watts of heat to be produced by the whole circuit.

And I'm going to do it without a fan. I'm trying to do my heatsinking with 2 layers of copper and thermal vias, with all components being surface mount.

LM2576 HV version is rated 60 volts max.

Yes, I am a bit concerned about that. I'm giving a little thought to doing something about that...

I can get the heat out. I will test for being successful at doing that. How small a space I can get the heat out of is the question...

Yes, I was figuring on LM2576 minimum frequency of IIRC 42 KHz, rounded to 41.6667 KHz for period of 24 microseconds.

With the voltages and currents being considered, the inductors needed for that are looking like they are not cheap little jellybeans. With everything else being equal, I would like to go higher in switching frequency than that low one by a factor of maybe 3-4, 5 if switching losses don't get to be too much of a factor.

THANKS!!!

I must have been thinking too much out of a box. The name of that box is "12V fixed output voltage".

This does appear to me to be better than LM2576. 92% efficiency for my application sounds to me accurate. With LM2576, it would have been 4, maybe 5% less. (Optimistically, depending on losses in the inductor.) The difference is at least .7 watt of heat, which will probably matter to me somewhat.

The external component count sounds to me somewhat high, but the external semiconductor count appears to me higher than that of LM2576 by only 1 MOSFET. It sounds to me like the cost will be only slightly higher than that of LM2576 - for .7-.8 watt less heat, a smaller inductor, and

I could be returned to being able to use a low cost open drum core inductor for this - component cost could actually decrease slightly. Then again, I may want an aluminum box instead of a plastic one because an open drum core inductor with this much voltage and current could need shielding!

On the other hand, does anyone here have experience with powdered iron toroid inductors with lower turns count and higher input voltage? As in experience with eddy current core loss? Or, am I going to be experimenting here and bringing news to everyone else here?

(I have already been through a battery charger project where I tried maybe 10 different inductors of open drum core and shielded ferrite styles, looking for lower loss and less heat and less of problems with saturation. I settled on an open drum core one - thankfully, less expensive. That one has an aluminum housing around it thankfully!)

Best Regards,

I am not limiting myself to ~150-250 KHz. I am planning to run the numbers, including switching losses based on switching times that the datasheet for your suggested LM5088 and the one for the MOSFET that I end up choosing indicate as realistic. I have dealt with 500 KHz before, in a battery charger using in part a buck switching voltage-and-current regulator. If 250-500 KHz allows me to use smaller and less expensive components without too much increase of switching losses and any other worsening-with-higher-frequency losses, I will go that high.

This has me thinking about 15-21 years ago, when semiconductors were more expensive, ICs were less advanced, and things were a little slower. Meanwhile, ferrite and copper were cheap in comparison. I made several voltage boost circuits, one to drive a fluorescent lamp that tolerated DC fairly well, and a few for xenon strobes. I minimized total losses with magnetic components on the largish side, and switching frequency so low as to be audible.

I did usually push up the switching frequency to "reliably ultrasonic" for most of these...

One project that I reminisce about is the capacitor charger for the "HSS1" strobe. That involved converting 12V DC to 900V DC, with output power around 10-20 watts IIRC. The "HSS1" strobe is a published project, easily findable with search engines. The transformer just about fits inside a ping pong ball IIRC, and total turns count of the 2 windings is

So now, things are smaller and faster...

And now, I am considering 150-250 KHz, possibly more, and I have already done a 500 KHz switching frequency project involving over 10 watts of output (and voltage bucked by a factor around 3).

One thing coming to my mind is that skin effect has to be a serious consideration for ripple current when frequency gets that high. That sounds to me like 6 of one, vs. half-dozen of something else. However, I am somewhat optimistic that shorter winding wire length means less winding resistance as switching frequency increases, to extent outweighing increase of skin effect.

I could go for more inductance than needed, involving more turns, to get core losses down and ripple current down. But then again, energy efficiency matters in this project mainly for reducing heat production. Meanwhile, higher value inductors have a high rate of having more DC resistance - more heat from I^2*R loss on the DC current.

That sounds to me like experimenting with a dozen different inductors and a few different switching frequencies to see what actually works best, and what costs the least to work acceptably by a good margin.

I did ~10 different inductors in a buck switching project with output over 10 watts before... Measured current draw from the power source and temperature of the inductors...

There are plenty of off the shelf inductors these days, I would only roll my own as a last resort.

For the inductor I would go to

Their inductor finder is useful and gives pricing.

I would look at one of their shielded inductors. I just looked at

Webench suggests a Bourns SRR1260 and Coilcraft have a direct alternative to that as do Wurth.

If this is in Europe where trucks have 24V buses then it is a good thing that he has OV cut-off up front. 76V on its own ain't enough.

Truckers can be pretty brutal. In the colder parts of North America they have been know to do the "revers push". Trucks are left idle at the restaurant. If an engine does stall it may not be possible to start it, on account of the low temps. Then another trucker backs up against the trailer and pushes the truck with the stalled engine until a big black plume comes out the exhaust and the engine rumbles to life.

A Russian told me they have other methods. They take some wood scraps from the truck, place them under the engine, light the pile, then light a cigarette. The fire warms up the engine so that it can hopefully turn over and start. So don't put this electronics gizmo in the engine compartment if the trucks are sold to Russia :-)

For the size you gave 2.5W sounds like too much. Unless it is mounted in a well ventilated area.

WRT to efficiency, I don't know how that is with cars/trucks but in aircraft we turn every gram of kerosene around and around, see if we can shave the power consumption some more.

But remember, the generated thermal energy ultimately has to migrate somewhere. Something has to take it out of there.

For a 24V truck system that still isn't enough. You'd need a separate over-voltage cut-out circuit before that or your client/employer might receive tons of warranty claims.

There are only three things you can do. An OV protector circuit, use a PWM chip that can take a much higher voltage, or use a regular PWM chip but supply if with a separate little regulator. Then pick FETs that can take 200V or so. That makes for a device of cast-iron robustness.

Thanks for the stories of what truckers do - maybe to achieve some badly-needed amusement?

No, my 12V-output regulator gizmo is not going in the engine compartment while a Russian builds a fire underneath. At least, not unless my gizmo is seriously abused. It is to be used primarily in the driver/passenger compartment (powered through the cigarette lighter jack), secondarily in repair shops powered by the battery via clip leads.

:-)

Was just to show what goes on in that market. Once I was asking a service tech why two caster legs on our ultrasound machines often look so scuffed up. "People stand on those while pushing the unit down a hallway".

If it goes into the automotive world and into export, it's almost guaranteed that it will be abused.

Ok, then at least you won't have heat issues. Whether it survives well electrically depends on whether or not they start the engine with this device plugged in.

I thought that I did say somewhere in this thread that 2-2.5 watts could be a tall order for the package size that I said I prefer. I am leaning towards upping the package size. What I will do is extensive severe testing to make sure that nothing comes close to overheating in adverse thermal conditions.

I do thank Raveninghorde for his suggestion that appears to me likely to reduce the heat production by around 30% without much cost increase, despite much higher component count. Even with this, I am allowing myself to have test results pushing me to upsize the overall package from 75 x 44 x 23 mm.

I seem to think that 1/10 of a mile per gallon difference in fuel efficiency is considered tiny by most car drivers. Heck, I seem to think American car drivers think 1 MPG difference is not much, sometimes not even bothering to keep tires topped off to achieve that.

In an extreme case, difference between 49.9 and 50 MPG at 20 MPH with

At this rate, I doubt a watt or 2 is likely to be considered significant.

That is something that I am very much aware of. I have a lot of experience here. My 2 biggest clients somewhat like me calling myself "Dr. Heatsink", and the one having more hands-on experience at my guidance is impressed by what I predict for things overheating.

I am not committed to the package size that I mentioned before, since I consider it optimistic.

Raveninghorde's suggestion that involves an external MOSFET does appear attractive to me here. It appears to me that I can find a "D-pak" or a SOT-223 one, surface-mountable, that will produce less than a watt, maybe as little as half a watt, of combined conduction and switching losses. That sounds to me easy to dissipate with a few square cm of copper on top and bottom with some vias.

Yes, I am a little worried about that... Much of this product's usage will be in 24V trucks.

Thanks for the suggestion!

I am not yet committed to Raveninghorde's alternative to LM2576, though I have a liking at this point to use that with some sort of overvoltage and reverse voltage protection.

I think that's also last resort. I did do a product that is now on the market with a custom one - and that one is in for a "running change".

Thanks for the suggestions! So far, I have been looking only at what Digi-Key stocks. I have been thinking that if something is reasonably available, Digi-Key would have it. Since I have only checked that far yet in terms of such higher currents with 100's of microhenries that LM2576 requires, I have seen mainly powdered iron toroids that look a little substantial in size to me. And, I wonder about how much loss powdered iron has with the voltages, currents and frequencies that I was considering.

At this point, once I breadboard this, I will be trying 5 or 12 different inductors to see what actually works best. I've been through that once before; I can do that again.

Best Regards,

- Don Klipstein ( snipped-for-privacy@misty.com)