First "Real" Project: 0-60V 10A Buck Lab PSU. Am I about to blow something up?

[Samir003]

Hi everyone!

I’m new here and relatively new to electronic design. I’m embarking on my first "serious" personal project outside of the basic stuff I did in school.

I’m trying to build a CC/CV Lab Power Supply capable of 0-60V and 0-10A.

It all started because I looked at the prices of decent units on Amazon/Aliexpress and thought: "I’m almost an engineer, I can surely build something better for the same price or less!". Well... I don't believe that anymore, but the schematic is already done, so there's no turning back now.

My design "philosophy" (or constraints) were:

• Cheap-ish: Using common parts.
• Low BOM count: Consolidated as much as possible.
• THT only: My SMD soldering skills are... a work in progress.
• Pure Analog/Logic: No MCUs (keeping it old school).
• High Power: More punch than the typical 30V/5A entry-level units.
• Goal: Actually work without catching fire.

The design is based on a TL494 for PWM control and an IR2110 for the high-side gate driver. It’s a Frankenstein mix of YouTube tutorials, Gemini AI suggestions, my old textbooks, and some trial and error in simulation.

I’ve spent a little bit of time cleaning up the schematics and trying to follow best practices, but since this is my first design with more than one IC, I’m sure there are some rookie mistakes hidden in there.

I’m not looking for perfection, just a decent, robust source that I can use and abuse for future projects. I’ve attached the PDF with the schematic design.

Any feedback, suggestions, or "stop right there, you're going to blow up your house" comments are more than welcome!

Thanks in advance!
Samir

 Buck6010-LabPSU_Rev1.pdf (1394.19 kB - downloaded 14 times.)

Edit: I removed the images because they made it difficult to read in the forum.

[max.wwwang]

Not sure if this is something that happens only to me, but when an inline picture is of big size, it becomes a pain reading the post and the picture ...

[Whales]

I had to edit the forum CSS to see the images:   img { max-width: 100%; height: auto; }

@Samir003: Please attach a PDF of the schematic and remove the images.  People are going to run away if they can't see your stuff.


It looks like you have a big external 60V PSU?  What is it?

Have you run any full closed loop transient simulations on your design? I see you noted "open loop" on one of the graphs.  I worry the design will be unstable because I see little in the way of stability compensation components, and IIRC the TL494 datasheets do not go into much detail of this side of things (they had a different audience compared to modern small buck IC datasheets).

If you want a software suggestion then QUCS-S has a TL494 model and its transient simulation capabilities are alright, but you will have to substitute your output mosfet with something approximately similar and substitute the gate driver with a voltage-controlled-voltage-source.  It should still yield some interesting stuff 

You will want to simulate your design under both voltage-controlled and current-controlled conditions, because both loops behave differently.  You will also want to test it at high duty cycle (>50%) as well as low duty cycle (<%50) as low tends to be easier/more stable than high in buck control loops.

[Samir003]

Thanks for the suggestion about the images, I thought it was only happening in my browser.

The PDF is already available in the post.

My plan is to use a generic commuted power supply (link for a Mexican distributor).

I know that due to circuit losses, if the power supply only delivers 60V, I won't be able to get 60V at the output, but I'm counting on these power supplies being somewhat adjustable and I hope to get a slightly higher output voltage to power my circuit.

I haven't run any closed-loop simulations; I was hoping it wouldn't be necessary since I'm just learning how to use ngspice within KiCad and it's not very user-friendly, but I suppose I can try.

Thank you so much for the suggestions; I'm going to start working on the simulations you recommended.

Is there anything else I could improve about the schematic or the operating principle of my circuit, or does it look good enough?

[Whales]

Yeah most spice doesn't tend to be particularly user friendly.  I'd try and claim that QUCS-S is better, but I'd be lying, I'm just used to it.

> I haven't run any closed-loop simulations; I was hoping it wouldn't be necessary
> [...]
> Is there anything else I could improve about the schematic or the operating principle of my circuit, or does it look good enough?

You can't design high speed closed loop systems like this.  You will be in for a lot of frustration.

You have to use one of the following to get them to behave stably (ie not blow up): numeric analysis, simulation, copying of an existing proven design or trial and error.  Trial and error is a great solution for many systems that only contain opamps that can't blow themselves up, but your circuit has large transistors switching 10's of amps at 60V that will blow up within milliseconds if something goes wrong.  These are very frustrating to do trial an error on without first choosing one of the other options.

If you have only built small buck converters using datasheets that provide step by step equations then you have been shielded from the difficulties of "generic" SMPS design (you will note the IC's start to be called "PWM controllers" instead of "buck regulators").  The TL494 is a generic controller that provides some very basic building blocks to a variety of DC-DC circuits.  It does not, however, provide easy stability solutions.  Instead the IC designers of the TL494 expected the circuit designer to provide the expertise in stability.

[Psi]

TL494 was a nice chip a long time ago.
Personally I would have use something more modern with a fet bridge and sync rectification, but on the other hand, the TL494 is nice old tech that requires you get into the technical side and understand more about how things works. It doesn't hold your hand as much as common chips. So it's good for learning, which looks to be your goal.

And if you have "0-60V 10A TL494 based Buck converter" on your CV when looking for a job that's probably going to be more impressive than if it was using a modern IC.   So i say 

[Samir003]

[...]
You can't design high speed closed loop systems like this.  You will be in for a lot of frustration
[...]
You have to use one of the following to get them to behave stably (ie not blow up): numeric analysis, simulation, copying of an existing proven design or trial and error.
[...]

Okay, I'm going to try to do a simulation, but instead of using the full models and all the complexity of the switching, I'm thinking of using an Averaged Model. Would this be sufficient, or is it really better to do a full simulation of all the ICs?

Also, I apologize for the format and my replies; I'm learning to use this forum and it's very different from what I'm used to on other sites. I imagine this was the standard for forums when this one was created.

[Samir003]

TL494 was a nice chip a long time ago.
Personally I would have use something more modern with a fet bridge and sync rectification, but on the other hand, the TL494 is nice old tech that requires you get into the technical side and understand more about how things works. It doesn't hold your hand as much as common chips. So it's good for learning, which looks to be your goal.

And if you have "0-60V 10A TL494 based Buck converter" on your CV when looking for a job that's probably going to be more impressive than if it was using a modern IC.   So i say 

Well, you've given me a good justification for using it.

The truth is, I chose it because it was the only IC that fulfilled the function, it came in PDIP-16 packaging (easy to solder), and it's also sold at the electrical supply store near my house.

[Psi]

Yeah, your options are very limited if you need it to be in a PDIP package.
All the really modern stuff is DFN/BGA/QFP.