more X-rays Silvertel PoE Supply

Any discernible numbers on the chips underneath?

I added a couple of pix to the folder.

The so8 ic is an Onsemi STX9B, but I haven't been able to track down a data sheet.

I've been doing my own PoE experimemts and so far nothing makes sense.

Probably something like this

What's not making sense?

The x-rays are fun, but are admittedly more useful for an opaque potted module or an IC or discrete semi. I can prety well trace the Silvertel without them.

I have a gigantic scary Carribean guy who does the xrays and I wanted an excuse for taking him to lunch, which turned into an interesting hike around The Bernal Cut. He trashed the xray system software, and had to rebuild it, and now it can do movies, which is great.

I'd like to have some products that can be powered by PoE or a wart, and may want to put two 24-volt PoE supplies in series to get 48, so I need to understand how they work. The xrays just help a bit.

(When burials were outlawed in San Francisco, a railroad was constructed to Colma for all the dead bodies and headstones. That forced the Bernal Cut to be dug with people and mules. Colma now has more dead people than live ones. Some parts of San Francisco are paved with old headstones that nobody wanted to pay to ship.

If Ag9900 just works then why would you care how?

I did PoE with LTC4267 some years ago such that 12V could also be supplied from a wart. Fully isolated with a PA1138 transformer.

These days I'd look at MP8017

Can I overpower it with a wart? Does that need another diode?

Can I run two 24v Silvertels in series to get 48?

And I do care how things are designed.

I guess we could. All sorts of objects.

I'd like to do one of my hands, which I think has a foreign object inside, but the machine is too heavily interlocked.

I had an uncle who got a foot xray for arthritis and they saw a small nail inside his toe. He had no idea how it got there.

Maybe. See other discussion in this thread.

I don't see why not since the data sheet says isolated. It shouldn't be hard to verify that.

Depends on whether you can fully reverse engineer an Ag9900, in which case you might either be satisfied or horrified.

Batteries are isolated, but a series string can destroy one cell by back-biasing it.

Some power supplies will latch up, or not start up, in series with others. Lots of suplies won't start (and may fail) if their input is pulled negative.

As the number of items in a design increases, the probability of some stupid error approaches unity. It's usually time efficient to be super careful, as opposed to spinning a few PCB revs.

Why do you need 48V ?

One of these will produce 48V or near enough, but not isolated of course.

Making it 48V shouldn't be hard, but may require a custom transformer.

Would be nice to have a simulation model of MP8017.

To drive loads! To sell more power supplies!

It ocurred to me that if I use two Silvertels with outputs in series, their inputs have to be in parallel. They will probably confuse one another negotiating with the PoE switch.

So I guess customers need to have a 48v wart if they want 42 volt outputs.

If that's the spec then good luck.

Each one would have to be connected to an independent PoE switch port.

You're not thinking of connecting two of them to the same port are you?

I'd consider not using the Silvertel things at all but it may come down to economics rather than electronics.

The spec is anything I want it to be. We are considering posssibilities and will have a hard schematic in a couple of weeks. Some confusion is beneficial in early design stages.

The box will only have one RJ45.

You are very snarky.

They are under $10, not worth the engineering to save a few dollars. Even with a heap of engineering, it would be hard to beat $10.

Time to market is worth a lot too.

No.

I want to get to at least, say, 38ish to simulate aircraft DC power, which is in theory 32 max.

Didn't someone else mention that word here recently? Actually I'm just trying to figure out what you're trying to do with PoE.

Ag7100 will turn 24V into isolated 48V but I don't see why you would want to use PoE to power equipment needing 48V.

Confusion is never beneficial. Indecision is a necessary part of the process of getting to the point where you can make a decision, but you do have to understand all the elements you are trying to pull together before you can make the choices about about how you pull them together.

Most people have toenails. :-)

It was the steel kind, pointy on one end and a head on the other.

Uncle Billy ran the biggest nail distribution business in New Hampshire, but still....

Create a product line of aerospace instrumentation, small boxes that can be placed right next to loads or motors or sensors and need only a cat5 cable for power and computer interface.

The current practice is to put a PXIe or VME crate in a rack and interface it to a computer (maybe $10K) and then buy modules and then design and buy fabulously expensive custom cables out to the field. Tha cables can cost more than the things that they connect. And rack space is often very valuable.

The power supply module is just the test bed for the first box in the family. Most modules would only need a few watts, but this one allows some thermal characterization too.

We're about to go public with the concept. Wanna see the intro?

That's not a PoE pickoff. I guess we could get 24 volts from PoE and boost that up to 48, maybe with some different brick, but the hassle and cost accumulate. If a user needs more than 20 volts output or so, they may have to buy a 48v wart. Higher voltage warts are common nowadays.

Customers might appreciate it and buy some.

That's exactly what I'm doing. Stock parts and an RP2040 CPU chip.

There are *so many* Raspberry Pi kids out there!

Now if I could build the hardware part of the product with Legos...