DIY High Power Lab Bench Power Supply

[D4no0]

I was thinking for some time to build some high power lab bench supply. I found a good source of power supplies (power supplies used for severs, they usually cost about 30$ on ebay and usually rated 1000-1200w). The next step is somehow to control the current and the voltage, so here I wonder what would be the best solution since you cannot find buck/boost converters handling such currents. I am a newbie in this field so please don't be too harsh on me.

When I built my small bench supply (120W), I used a LTC3780 and replaced the variable resistors on the board with ones that can be turned by hand, works like a charm even to this day. Also commercial solution is not feasible since it costs 1000$+.

So I wonder is it possible to make a DIY solution(under 200$ would be perfect)?

[magic]

Maybe start with modifying some wall wart to understand the principles. Some folks in the past considered modding ATX supplies too, but I don't recall any complete build.

The voltage part is generally easy, most of the time there is a circuit on the secondary which senses the output voltage and signals the primary PWM controller to slow down when the voltage gets too high.

Beware that increasing output voltage may not be safe or perhaps even possible. Those things still use transformers and the rule with transformers is that for more secondary voltage there will be more primary voltage one way or another, no exceptions.

For current, you would add a shunt (if you are lucky, one is already built-in for overcurrent detection) and add some comparator to detect when current is too high and signal the primary-side controller. Or perhaps exploit the built-in OCP comparator, if available and usable.

Such contraption may not have the same response speed and general performance as a real lab PSU. At 1kW there is a lot of potential for magic smoke, or worse if you stick your fingers into the primary.

[wasyoungonce]

Silicon Chip DIY variable 45vDc 8 A linear PSU good enough

http://www.siliconchip.com.au/Issue/2019/November/High+performance+linear+power+supply+–+part+two+++?res=nonflash



Part 1 here:

http://www.siliconchip.com.au/Issue/2019/October/45V%2C+8A+Bench+Power+Supply+to+build?res=nonflash


That’s part 2. Search site for part 1. Edit see above trouble with 1st link

[Gladius]

So I dont know what exacley you are looking for but having DIY several powersupply solutions in the past I found that there are some quick and easy of the shelf solutions that worked great for me.
Depending on what you need, how high your voltages and currents have to be, how stable and how accurate etc there are different solutions.
The allround easy, relieable and fairley cheap thing I have been using for a good 3 years now are Little chinese modules by the name of DPS50XX where XX is the number of amps. Ill include a link for the 50V DC 20A one plus case and powersupply down below.
The work reliabley are accurate and come with a bunch of cool features. Word to the wise though NEVER connect the output and the input or you will blow the module.....
Also if ever you feel like you need a porteable solution slap some (protectet) 18650s out of an old battery pack along with a decent boostconverter in the case and you have a porteable powersupply for tinkering on the road. Thoug for this project I suggest one of the lower power but smaller form factor modules.

If ever I am mad enought to do stuff that is higher voltage and current I have a big ass DIY full bridge rectifier with smoothing and a big variac. Hope I dont have to tell you but damm be carefull it can and will kill you. I am an professional electrician and I have a 10mA RCD and a super fast blow breaker just in case I do something stupid. Aside from protective gear of cource. Add to that a voltage meter and a amp clamp and you are good to go for the rough stuff.

Also a quick and dirty solution  but this time without much controll. Old PC powersupplies. They usualy have 3.3V 5V and 12V  can deliver around 20A (depend on the supply) and have overload and short cirquit protection.

Hope some of that babbeling was usefull and if you have any more quiestions just ask.

DPS5020 Module:
https://www.aliexpress.com/item/32849728737.html?spm=a2g0o.productlist.0.0.423e70a1fqme5X&algo_pvid=7767424e-0380-45b8-ab8f-a1bd2da3972a&algo_expid=7767424e-0380-45b8-ab8f-a1bd2da3972a-3&btsid=5054abd1-23c0-4087-8278-7c7427d751fa&ws_ab_test=searchweb0_0,searchweb201602_4,searchweb201603_52

Case:
https://www.aliexpress.com/item/1000004752402.html?spm=a2g0s.9042311.0.0.27424c4dmEjfGd

Powersupply:
https://www.aliexpress.com/item/32905696401.html?spm=a2g0s.9042311.0.0.27424c4dmEjfGd

Also if you need to step u the voltage (Input to module must be higher than 50V) though take the 1500W rating with a grain of salt
https://www.aliexpress.com/item/32964442336.html?spm=a2g0o.productlist.0.0.656862f7ECcwvT&algo_pvid=1defd118-f8ac-4c53-a252-7ca9c6a98297&algo_expid=1defd118-f8ac-4c53-a252-7ca9c6a98297-0&btsid=9c9850a3-18f8-42fd-bddb-6244570834c7&ws_ab_test=searchweb0_0,searchweb201602_4,searchweb201603_52

[poorchava]

Using a complete ac/dc power supply is a good idea, since this is the most difficult part to get right. A 48V output would be the best, since you probably will only step it down, not up. High power step down is just a half bridge, inductor and some capacitors. High power step up generally involves a transformer (not that hard to do, but more complex than step down).

What I would do:
-find some good PWM source, preferably fast conventional uC (like STM32H7 value line) or something with HRPWM capability (STM32Gx, STM32F343, Infineon XMC1xxx, XMC4xxx). TI TMS320 family also have that but they are a complete pain in the ass to develop for and I would not advise them for someone without lots of previous microcontroller development experience.

-make a gate drive circuit: either an integrated bootstrap driver or a gate drive transformer + dual low side driver

-make a mosfet half-bridge using something rated at ~200V. In general you could go lower with VDS, but this involves more trial and error and fine tuning the circuit.

-calculate capacitors and inductors for worst case conditions. Formulas are available in most datasheets of integrated synchronous buck converters. Formulas scale to very high power levels. The inductor will most likely have to be custom-wound, but this is not difficult at all. Sendust or similar material toroidal cores ara available without any problem, and most manufacturers make available step-by-step instructions on inductor design for most general topologies. Also TI hjas an excellent PDF series "Magnetics design" (google for that term)

-make some sort of feedback loop. In general the theory says you need to calculate transfer function, use Z-transform and loads of other stuff. In reality for majority or use cases a P or PI regulator works fine and you can tune it by trial and error or using one of popular autotune algorithms (eg. Ziegler-Nichols method).

I have built several 1000W+ single-phase step-downs that way. You can also make it multi-phase, which is just 2 or more smaller converter connected in parallel and driven out of phase with each other (180* for 2ph, 120* for 3ph, 90* for 4ph etc).

Unless you need to fit in specified volume or enclosure and therefore require high frequency to miniaturize components, something like 80...120kHz fsw should be a good starting point.

It does sound scary, but it is not. It gets more complicated when you have more constraints like volume, emissions, cost, DFM and so on, but something that is good for a one-off lab power supply is quite easy to build. Also, stupid shit starts to happen when you get into stuff like 5V@300A, but that is a different story altogether.

Some time ago we have managed to fit 3x synchronous step down, 2.2kW each in about 100x170x100 volume (including heatsink, excluding ac/dc), powered from 48V, running at 200kHz, with silicon semiconductors only (GaN is yet very expensive and problematic in implementation and SiC makes sense above ~800V). Unfortunately I'm not at liberty to post photos of that.

By the way... high power is a relative term. I work for a specialistic T&M equipment company and we have some jokes going around about how Insulation Tester guys consider 1Meg resistor "almost a short" and from microohm meter people you could hear phrases like "...u know... some large resistance... like 4 miliohm". Same as motherboard and microwave guys sosider everythin under 1GHz "almost DC". Ask EV charger company about something "low power" and they you'll get something like "uhmmm... under 15kw, right"?