Quoted from megadeth2600:I don't mention that because there are many, as in hundreds, of off the shelf switching regulators out there to choose from that will easily accept 48V and regulate it down to 5V @ about 80% efficiency with a cost no more than $2.00 USD.
Here are 17 from Texas Instruments that could have been designed into any Stern node board that required 48V in and 5V out with a max output current of 2.0A ... these are the results of a cursory search too ... I'm positive there are several more ... and this is just TI. Vendors like ST Microelectronics, Analog Devices, Maxim, etc all have their own regulators available that meet or exceed requirements. I'd guess that there are at least 250 devices out there on the market that will be available until at least 2030. More if you don't care about a part's end of life. Also, many of these parts are footprint compatible ... you have a wide range of choices.
https://www.ti.com/power-management/non-isolated-dc-dc-switching-regulators/step-down-buck/buck-converter-integrated-switch/products.html#p238min=1.5;48&p238max=48;100&p634min=-3;5&p634max=5;92&p451max=2;2.5&p212max=11.704312114989733;100A Vin/Vout difference is only considered for efficiency purposes when evaluating a switching regulator ... you tend to only insure you are not exceeding a maximum input voltage. Vin/Vout difference is more of a linear regulator constraint (and a 36V difference in that case would be horrible) ... however, a switching regulators can still regulate 5V from a 48V source with 50% efficiency if the output current is in the microamp range. Of course, current demand will be much, much higher (which results in a much much higher efficiency).
As an example, Stern's 520-6976-72A node board uses a Monolithic Power Systems MP24943DN ... it has a max input voltage of 55V which is plenty of margin. Again, only efficiency is impacted when comparing the output voltage to the input voltage (this is not a linear regulator). Looking at the curves on the datasheet linked below, the node board will have a conversion efficiency around 75% to 80% assuming an output current of about 0.5A to 1A.
The regulator also has output overvoltage protection as well as a thermal shutdown mode. If something goes wrong with a linear regulator, you're out of luck until a fuse blows ... you tend to destroy other parts before the fuse opens (mainly driver transistors ... and possibly a pre-driver or logic IC). The switching regulator simply turns off.
https://www.monolithicpower.com/en/documentview/productdocument/index/version/2/document_type/Datasheet/lang/en/sku/MP24943/document_id/1966I remember there being problems in the early days of Spike, but I haven't heard such complaints myself for a few years now. That isn't to say that there isn't a problem ... I could very well be wrong. I just haven't seen many people complaining about node board failure. I think those early problems were due to vibration and connector issues and not the regulators themselves.
I'll agree with that. It's all of the linear voltage regulation/conversion after the transformer that I dislike. These days, it is a waste of power and money to convert mains voltage to 48VDC solenoid voltage using linear conversion and regulation.
That is after 10 years of continuous, 24 hours a day, 7 days a week operation with near or at zero mean time between failure. That is a pretty good torture test. They don't get downtime ... at all. Some places will operate the equipment more than 10 years since the telecom grade core and edge routers routers are designed to be scalable and require a substantial investment up front.
While I sound like a Stern fanboy defending their electronics, I am anything but that. They disgust me anymore. However, I don't think that their electronics will be a problem two decades from now. I suspect mechanical engineers and machinists will be saving the pinball machines of today in the future ... I find the quality of Stern's basic mechanical assemblies to be garbage anymore. I have never had more issues with coil stops and flipper links in my life. If they're cheaping out on that stuff now, wait until they apply the same kind of idiocy to assemblies that are specific to a particular machine. Their profit margin on pinball machines is ridiculously high ... they can *easily* afford to maintain acceptable quality ... however, they don't ... they seem to cut costs and raise prices on everything anymore. Disgusting.
This is a super interesting discussion, I would actually love to have failure data on recent build of node boards. There is an on going debate where people claim "Stern Spike system reliability is crap", and i'd like to have factual data to assess whether this is right or wrong
it is not in the interest of Stern to have self dying machine. They need to fix their reliability, as it has a financial impact, so I doubt they just left it "like this", but I could be wrong.