Ethernet-UART module made by USR and a little about how to use it

I've run into situations more than once where controlling devices directly over USB from a computer is easy to set up, fairly reliable, and inexpensive, but still has its drawbacks. The main ones are that you're tied to the computer and that simple adapters don't provide galvanic isolation. This adapter was bought to solve all of those problems.


It all started when, during a battery endurance test, I needed to set up a load somewhere without a computer. Of course, I could have set up another computer, on top of the ones already running around the clock at home. And it's not even about the cost of a computer; the whole setup just looks a bit... awkward.

After looking around online, I found a solution: build a module into the load that would let me connect it to my home LAN. Since I've run cables not just to every room but also to the storage room, where I planned to put the load because the temperature there is stable, this option made everything much simpler.

To be fair, there's a simpler option: buy RS232-TCPIP converters. I use Moxa units at work, in a two-port configuration. They're great pieces of hardware, but unfortunately they cost so much that you could easily buy a couple of electronic loads for the price of one converter.


So, I found one, paid for it, received it, and have even put it to use—but more on that later.
The title lists a price of $7.10, but the seller charged for shipping, and it wasn't cheap. I used some kind of coupon when I bought it, so the total came to $10.86. That was fine by me, although I later found a better deal. But I'm getting ahead of myself again.

It arrived in a padded envelope. The module itself was carefully wrapped in a sheet of soft material. One pin was slightly bent, but otherwise everything was fine.


It looks just like the seller's picture. There are other modules in the same form factor on AliExpress, some even cheaper, but I was specifically looking for one made by USR. You can tell them apart by the marking on the connector and the arrangement of the components.


Everything is very compact and neatly laid out. One side of the board has a pin header for connections. Near the Ethernet jack there are just two pins; they're not connected to anything and are there to help secure the module to the main board and in the connector.
The module can be powered from either 5 V or 3.3 V.



There are two chips: an ATSAMD20E18 microcontroller and a DAVICOM DM9051NP Ethernet chip. The galvanic isolation transformer is inside the Ethernet jack.


There's nothing else to inspect on the outside. To see what the module can do in software, I need to connect it to power and the LAN at the very least. And to avoid making two trips, I'll test it right away with the electronic load it was intended for.


The load's board has a connector with RX/TX, ground, and 12 V brought out to it. Since the module needs 5 V, I had to bodge in a 7805 regulator in place of the diode that fed 12 V to the connector.


I connected it up. Surprisingly, nothing burned out, and when I plugged in the LAN cable, the module's LED started blinking merrily.


To continue, I downloaded the software from the manufacturer's website. Strictly speaking, the program isn't really necessary; it's only needed when working with a computer, specifically to emulate a virtual COM port. But that's exactly what I needed.


Program settings. As I mentioned above, I needed the program to emulate a COM port and launch minimized in normal mode.


There's an option to configure the emulated COM port manually, but for some reason I couldn't get it to work properly. I created the port, but got no further: my software could see it, but no data was being transmitted because there was no connection to the Ethernet module.


The problem was easily solved by launching Smart VCOM. The adapter was detected automatically, and the program assigned it the next available port on the computer.


The only snag was that, for some reason, everything started working properly only after I added two ports—in effect, repeating the operation shown above twice. At the same time, the second port immediately switched to Connected.


The electronic load software detected the port and connected to it without any problems. That was exactly what I needed.


All the module settings are available through the web interface, and there are quite a few of them.
1, 2. Local port statistics and parameters.
3, 4. UART settings
5. When switching to UART settings mode, more information is displayed briefly. It doesn't mean anything to me, but perhaps someone will find it useful.
6. Password, login, module name, and other settings.


Advanced settings. The same thing happens here, by the way: the full information is displayed only briefly (second screenshot).


I connect a battery to the load to test it in action.


The battery's test discharge and charge went perfectly. Essentially, everything worked just as it did when connected via USB; if there is any response delay, it's practically unnoticeable.



Meanwhile, the software reliably counted the information packets sent and received.


You can start monitoring, but personally, it didn't tell me anything. The main thing was that everything worked reliably and smoothly.



That's all I have to say about the module. I just don't know what else to add, so I'll move on to the modification—or rather, the rebuild—of the electronic load.
You've probably seen it many times in my reviews. It's an extremely useful device, but it does have some drawbacks. Two of them are the most critical for me:
1. There are only two control options: local and via USB. The first can't be used for software tests, while the second creates a galvanic connection to the computer.
2. Since a switching power supply is used most of the time, there are problems when capturing oscilloscope traces of the devices under test.



Of course, I could have tried to fit everything into the original case, but it's simply too small, and besides, I just don't like it. It's basically a standard power supply case, slightly "modified" for this application. Since it hadn't really been modified, a problem eventually appeared due to thermal cycling: hot air doesn't escape well. As a simple fix, I recommend cutting out at least part of the metal cover near the connectors.
You can see in the photo that, in addition to the connections through the connectors, there are also wires soldered directly to the board. I was trying to fix the current calibration drifting. As it turned out, the problem wasn't a poor connection but the control board bending. If you flex the board a little, the calibration returns to normal. Go figure :(


Given all of the above, I decided to get rid of the original case altogether. To do that, I had to take the whole thing apart and remove the internals. I also ended up with a few spare fasteners.


Because of the case design, the display was connected through an adapter board, which I had to remove and later put to a different use.


While I was at it, I picked up various connectors, terminals, temperature sensors, and other odds and ends at the Voron store. I found the Z1 case at home, and bought the transformer on OLX.


I desoldered the original display connector from the adapter board because the case isn't very tall, and with the adapter in place, nothing would fit.
I had another case in mind, slightly smaller, but decided that everything probably wouldn't fit in it—and as it turned out, I was right.


One of the key steps is preparing the front panel. I usually draw it in FrontDesigner, then mark up a draft with all the key currents and dimensions, transfer it to the plastic, and cut it out with a disc cutter. First the rectangular holes, then the round ones.


Next, I remove the burrs and uneven spots left by the milling cutter. I print the final panel template with a few millimeters of margin around the edges, put a strip of ordinary tape across the top, and stick the paper onto the front panel.
Self-adhesive paper would be much more convenient, but I only have a couple of sheets at home, and I haven't been able to find any for sale :(
Tape comes in clear and slightly yellowish varieties. For obvious reasons, clear is the better choice.

I wanted to buy caps in three colors, but the store only had two. It also turned out that the button plungers were a little too large for them. I also bought connectors there, both banana jacks and microphone-style ones.

Sorry about the photo quality. Most of the pictures were taken while I was sawing, drilling, and so on.


Here are the sockets I bought, compared with the medium-sized ones from my previous review. I considered using the large, cheap ones, but decided to keep them as donors for the devices they're currently used in, and not buy any more in the future.


The front panel is practically finished. Drawing it took quite a while: I wanted everything not only to fit and be convenient to use, but also to look neat—a sort of “techno design.”


The back is much simpler: an opening for the hot air outlet and a power cable connector. The Ethernet module is mounted very simply: I soldered a couple of pieces of fiberglass laminate directly to the connector housing, drilled holes in them, and tapped them M3.


One of the key parts to upgrade is the transformer. Of course, it would have been easier and cheaper to install a switching power supply, but I wanted full galvanic isolation, so I bought two transformers. I originally wanted to buy two 2×15 V units, but the seller accidentally sent one as requested and the other as a 2×12 V unit. It's not a deal-breaker, but it's a bit inconvenient, since this setup won't let me charge a lead-acid battery. On the other hand, there's an indirect benefit: it'll be harder to “kill” the lithium battery protection circuit. I'll probably talk about that in the next review.


The transformers cost 200 UAH each, about 7.5 dollars, and I can already hear the question: why didn't I buy the familiar TPP transformers? The ones I found simply wouldn't fit in the case height-wise. These are rated at 2×15 V, 3 A, or up to 90 W. They're wider, but not very tall.


The transformers' weight and key dimensions. Also, the 12 V transformer has an output current of up to 3.5 A, while the 15 V version has a fuse on the wire.


The seller said the transformers were made for Sven speakers, from older batches that still used proper copper rather than a cardboard-and-foil mix. I didn't check that this time; I'll save it for the next review. But in my opinion, at least the wire gauge looks fine.


The winding on the second transformer, which is the one I'll use in this load, is almost impossible to see.


The no-load current is rather high, around 58-62 mA. The unloaded voltage is slightly above the rated 15 (30) and 12 (24) V. The mains voltage was 222 V when I took the measurements.


After an hour of operation, the transformer temperature is around 40-45 degrees. I think that's a bit high, but since my device has active cooling, I can just ignore it.


I brought up the transformers for a reason. First, I like them, even though they were originally designed for a different purpose. Second, while their winding voltages are convenient, they have some quirks in my application. Let me explain.

There are several ways to rectify AC. First, the bridge rectifier: simple and convenient, but it requires a 12-15 V winding (circuit A).
The transformers shown above have a 24 or 30 V winding with a center tap and are better suited to producing a split supply, which is useful for amplifiers. As I mentioned above, these transformers are from (or rather, intended for, since they're new) Sven audio equipment. (circuit B)
I had to use a full-wave rectifier, which is simple but has a major drawback: the transformer windings are effectively used only 50% of the time, because at any given moment only one half of the winding is under load. (circuit C). Of course, you'll say that this type of rectifier and the same principle are used even in high-power switching power supplies. That's all correct, but there's one catch: the winding resistance of a switching transformer is radically different from that of a conventional iron-core transformer, and as a result, that resistance starts to have a major effect.

To use the entire winding in my case, I would either have to install a bridge rectifier and load the whole winding, but then the voltage would be 30–40 V, not 15–20 V. Or I could disconnect the windings and connect them in parallel, which is what I plan to do in the next modification. It will be easier to access the winding there, too. (Diagram D).




While figuring out the layout, I thought about the best way to arrange the components and decided to do it as shown in the photo. The fan now cools all the internals at once, and the hot air is expelled outside. Of course, I still need to measure the transistor temperature under full load, but I think it turned out no worse than before. If necessary, I can simply replace the fan with a more powerful one.


While replacing the wires going to the transistor, I happened to notice that its leads had been left at full length. I'm more used to shortening them by about half.


I connected most of it and powered it up just to check that the whole setup worked at all.


Next, I connected the remaining wires and installed a protection relay. I mentioned before that this model of electronic load has a design flaw: when it starts up, the charger is already enabled. If the power supply can't start the controller properly, it will keep charging the battery indefinitely. This problem occurs with relatively low-voltage batteries, such as Li-ion batteries, especially when they're fully discharged. The relay switches on after a short delay, and its contacts are connected between the charger's output and the battery. Newer loads have a relay controlled by a microcontroller. I even found which pin controls it, but unfortunately, the firmware is old and doesn't support this feature.


The transformer's secondary winding is connected using the original connector. I thought that would look neater.


The Ethernet adapter board is connected to the main board with a four-wire female-to-female cable and plugged into the electronic load's original socket.


You could say it's almost done.



The final touch for this stage of the modification. You may have noticed the four-pin connector on the front panel. Since I often test batteries, and to avoid having wires in the way all the time, I decided to connect the battery holder through a separate connector. I plan to do the same with the other load.


Now I can get back to the long-term tests :)


While I was preparing the review and modifying the load, I had time to order and receive five more adapters. I liked them so much. I ordered them from another seller, and buying 5 brought the price to about $7.50 each, including shipping.


So, friends, I can say without a doubt that I'm happy with these adapters. They may have some drawbacks, but I haven't come across any so far; everything works great. The connection is as simple as it gets: RX to TX, TX to RX, 5 V, ground. The software found the board automatically and created a virtual port. That's it!
Now, here's what I've gained and what I plan to improve in the future.
1. Added control over the local network. It's convenient—the load is no longer tied so closely to my workbench.
2. This also provides galvanic isolation in the control circuit, so there won't be any more ground loops or other problems.
3. Added a transformer, so an external power supply is no longer needed.
4. The power supply is now galvanically isolated too, so I can safely use it with an oscilloscope.
5. Added protection against incorrect startup when a battery is connected.
6. Installed an extra connector for the battery holder, which makes things much more convenient.
7. Arranged the buttons and connectors properly, with the positive terminal on the right and the buttons in the -/+ order, not +/-

What I planned to add
1. Two fan speeds, using thermal switches I bought for the purpose. I haven't gotten around to it yet. What's more, I put a 300 Ω resistor in the fan's power circuit, so now it runs quietly all the time and switches to full speed when commanded by the load.
2. Manual switching of the supply voltage between 12 and 20 V. But since I used a 12 V transformer, I haven't done this yet. I'll probably add it to the next load.
3. Not an addition, but a major necessity. After several years of use, the control board has suffered from constant thermal cycling, and now its zero point has shifted, i.e. For example, instead of a 1 A charge/discharge current, we actually get 1.02 A and 0.98 A. If I flex the circuit board slightly, the readings return to normal depending on how it is flexed. I tried reflowing the solder joints, but that didn't solve the problem, and I don't think calibration will help.

Things I forgot and will add later
1. A connector for an external, high-power 24 V power supply. It's needed for working with high-voltage batteries; still undecided for now.
2. A galvanically isolated control input via a USB-UART converter.


So, it's still a work in progress, but I'm already happy with how it's turning out.