Mechs were something I struggled with a ton for a really, really long time, and there were a number of reasons that contributed to those difficulties, so I'll basically address this in three (hopefully short, as my spouse is eager for us to go to dinner) sections:

Spatial Reasoning and Intentional Markmaking

I've been drawing regularly since I was 12 or so, but I wasn't really learning in any structured manner - at least, not until I was 23, and more meaningfully when I took a number of very helpful courses (the parts relevant to this section of my answer here being what served as the basis for Drawabox). While I started on paper, once I moved to digital when I was 14 when I got a very, very tiny non-display wacom tablet (smaller than the "small" sizes they sell now), it ended up being what I used for a good while. What I didn't realize was that, while the tool certainly doesn't make the artist, some tools can contribute to making certain things a lot more difficult than they need to be - and drawing on a tiny tablet made any sort of intentional, clean, and purposeful mark making a big challenge. Being the beginner that I was, I assumed I just wasn't suited to that, so I gravitated more towards digital painting, which was more forgiving on a small tablet, and assumed that this was just what I was more "meant" to be doing.

Unfortunately whether you intend to draw or paint or whatever else, intentional markmaking is important and lays the bedrock for developing one's spatial reasoning skills, which in turn are critical to creating anything from your imagination. Note that when I say imagination, I don't mean working without reference - I mean when you're drawing anything where the resulting form you're after doesn't already exist to be copied from a single source. So, for example, the demo in this video I did for Proko's youtube channel on the topic of Aphantasia, where I'm drawing a griffin based around tigers and pigeons, I may be using reference but the decisions being made are my own, from my own imagination. The references are just a tool to help fill in for information I don't yet have in my visual library, or at least not in as detailed a form.

Being able to consider how the things represented in those reference images exist in 3D space, how they can be rotated and manipulated - and not just in the two dimensions the reference image literally is - is critical to being able to use these representations of information to help flesh out our own designs, and again, clean and intentional markmaking lays the groundwork for developing those spatial reasoning skills.

Design

One of those very helpful courses I took was "Intro to Form Language" with John Park, who incidentally is extremely well respected for his mech work (among other things), and at the time he'd recently come off working as one of the lead designers on Hawken. The course itself explored what tools we have at our disposal when it comes to thinking about design, at a very base level. So you might have reference images, but those reference images can be broken down into visual components like the shapes and forms they're composed of, as well as the proportional relationships between those elements and how adjusting those relationships can result in very, very different impressions. I remember one of the exercises we did had us draw the same object from as many different references as possible (I drew pages upon pages of '67 Shelby Mustang GT500s), and then towards the end he had us modify those proportions to see how we could take that existing thing and make it feel like a fundamentally different thing - I found it delightful how making the cab taller made the coolest car in existence look like it was fit for clowns.

Design concepts like shape and form language, as well as proportion, play a critical role in how we approach inventing our own designs from other things, and how it's not just what pieces we pull from elsewhere that dictate a design, but it's perhaps more about the way in which we connect them, the way we fuse them together, and all of the relationships we opt to create, that really constitute the final design.

This class played a massive role in changing how I considered the way I drew... well, anything. As implied earlier, I have aphantasia, which means that I can't "see" the things I imagine in my "mind's eye". It's all just blank, and so I went into all of this with the expectation that I would never be able to be a concept artist, that all of my designs and ideas would forever be uninteresting, boring, and uninspired. But in truth, it turned out that I just didn't know how to approach developing those skills, and moreover, I wasn't aware of which skills themselves needed to be developed. I was more predisposed to looking at art as this one big thing, when in truth it's a multitude of individual skills that overlap and play off each other in different ways. You can be incredibly skilled in some areas, but utterly baby-like in others, simply because it's not something you dug into yet.

As a side note, another thing this course introduced was the idea of balancing detail - that packing in as much detail as you can, wherever you can, is not inherently a good idea. Rather, we have to consider how we balance areas that are specifically lacking in detail ("rest areas", where our eyes don't have to be bombarded with information) and areas that have more detail packed into them ("areas of interest") and how we can use that balance to guide the viewer's eye through an object or a scene and convey what it is we need the viewer to understand.

Source Material

The last thing that contributed to my ability to draw mechs and mechanical structures was what kinds of source material I was paying attention to. Often times students will think, "well I wanna draw mechs, so I should go and look at mechs other people have drawn" - and that makes sense on its surface. But what other people draw is already derived from the things they've studied, and while some of that may well be other artists, it's also backed up with studying a lot of real things. Those real things can range from big mechanical structures like engines, existing robotic arms and other articulating components, and really just heavy equipment in general, to the smaller, more genericized elements from which the larger structures are composed (hinges are a big one, but you've also got things like basic pneumatics and hydraulics, actuators, etc.) - and it's this latter group that is going to be more important.

While the former group with its more "complete" things carry a lot of value in showing you how things can be put together to create something larger, it's understanding how those smaller pieces fit together, what they do, and how they do it, that will really have a bigger overall impact on what it is you design, and the novel details you'll end up including in those designs. Hinges were so big for me when I was taking that class because there were so many different ways in which the problem of "how do we create a pivoting joint" could be solved, each with its own different kind of named hinge. And of course, the better we understand the hinges that exist, the more we're able to invent our own to suit our needs - and it's with the backing of those real examples of problem solving that we're able to create designs that feel real.

Now I'm still not specifically a mech artist, since I mostly gravitate towards fantasy, but here are some examples of what the above helped me produce:

• This mech was something I drew while taking the course I talked about above. It was something I started drawing late at night just for the hell of it, with no planning or intent. I say this because as someone who at that point still felt like mechs were so far outside of what I'd be able to draw that I'd been consciously avoiding them, it was a big surprise to me to just see all of those skills I'd been learning come together and show that what had seemed impossible beforehand was just a complete misunderstanding of the skills required, and how to develop them. This ended up being used in one of my final project illustrations for that class.

• Those same skills came in very handy years later when I was working as a concept artist, and was tasked with designing vehicles for fundamental redesign of a game the studio I was working for had already developed (in effect we were reskinning it, as it wasn't especially interesting as it was, despite being fun to play). These aren't mechs by any stretch, but being novel vehicles they work off a lot of the same core concepts: 1, 2, 3, 4, 5.

Anyway, I hope that helps answer some of your questions, or at least points in a direction as to what to seek out to address them.