In Part 1 of this article, we discussed the layout of the custom exhaust for the SR-1 race plane, using 3D printed parts to establish angles and pipe lengths before cutting metal, as well as sourcing some of the specialized components. This article looks at the actual weld-up.
Get Bent
Once the complete exhaust had been assembled from PVC and 3D printed bends, I was able to order straight tubing and mandrel bends. These were trimmed to the dimensions established by the 3D printed proxies. I conferred with Specialty Products Design (SPD) during this process to make sure that what I showed up with at their shop would adhere to best practices—basically minimal gaps at weld joints and a logical order of operations for tack and final weld. Initial fit-up of parts was accomplished with a mix of hot glue and foil tape to hold everything together. In some locations, I attached “fingers” to the engine support frame to hold tubing at a precise angle.
I’ll note here that the mandrel bending process tends to flatten the tube a bit, and it can help to squeeze the long axis to get the tube back to round again where it butts to the straight tubing, which will be perfectly round. Being a couple hundredths of an inch out of round is not a big deal; sometimes trying to squeeze a mandrel bend back to round just seems to end up making it more square than round, so this is definitely a case where better is the enemy of good enough.
Watch any video on welding stainless steel and they’ll make a big deal about cleanliness. Part of cleanliness is keeping contaminants out of the shaping process. This can be accomplished by having a separate set of grinders and files for stainless steel and aluminum; sometimes this is for galvanic corrosion issues, but here it’s about avoiding contaminants in the weld line. There didn’t seem to be a strong consensus on sanding belts, but I decided to err on the side of caution and changed out my sanding belts and deburring drums to silicon carbide (as opposed to the more common aluminum oxide, which some folks claimed could contaminate the stainless steel) for the duration of this project, and only used them for stainless steel. All that said, I seemed to have been more concerned about all this than the welders themselves.
Once all the tubing and mandrel bends were cut and sanded to shape (including a 45° bevel at the butt, as requested by SPD—see pics), the final prep work on my end was to deburr the butt weld edges, polish the weld zone with Scotch-Brite, and do a last, thorough cleaning inside and out with acetone. Parts were then kitted up and the engine loaded into the car and taken to SPD.
SPD was fine with me assisting the welder during the weld-up—which made sense, since I had prepped all the parts and knew how it all went together. Not all shops work this way though; some do not want customers in their work areas. And to be fair some folks just want to drop an engine off at a shop, let the shop do all the prep work and tube shaping, and come back when it’s all done. That had been my original thought too. But as soon as I started designing the system, I realized that due to my very specific goals on how the system was to be routed and clearance issues, it would be better to maintain control over the whole process.
Working with Kyle Knisley, the welder SPD assigned to my project (he had also done my collector, so I felt comfortable with his work), we unloaded the engine/support assembly from the Forester and forked it to his workstation. I then briefed Justin with an overview of the exhaust layout, the prep work I’d done, and any concerns I had. This is standard operating procedure for me for any project where I am bringing in helpers or experts: I’ve typically spent days or weeks prepping materials, and pre- and post-briefs are important to make sure everyone is on the same page, understands the task, and concerns or feedback can be voiced for my own professional improvement.
Welding the sockets to their mandrel bends, and the EGT bungs, was the first step, since this could be accomplished on the bench (no orientation or clocking was required for these welds). Once the bench welds were finished, we tacked the pipes for the Number 1 and Number 2 cylinders, since these could be (somewhat) easily removed from the engine and collector after tacking.
To explain further: Remember the pipes are slip fits on both ends (exhaust port and collector inlet). Because the Number 1 and Number 2 cylinders have forward-facing exhaust ports (vs. Number 3 and Number 4, which are aft-facing), and because the turbo collector also faces (approximately) forward, pipes 1 and 2 can be pulled forward and will exit the exhaust port and collector inlet at the same time. Once removed, pipes 1 and 2 were fully welded on the bench and then slipped back into their respective exhaust port/collector locations.
Side note: When (tack) welding on the engine itself, ground as close to the pipes as possible (we used the collector/turbo flange). Clamping elsewhere, such as the prop flange, can cause potential problems if the ground path results in arcing.
Pipes 3 and 4 are another story. The sockets come out of the cylinder exhaust port by moving aft, but the turbo collector end of the pipe—as we’ve already established—removes pulling forward. Therefore, to remove pipes 3 and 4, the turbo must be unbolted from the collector, removed, and the collector pulled aft in order to give the pipes enough space to slide aft out of the exhaust port. This meant that once runners 3 and 4 had been tack welded, I needed to remove the turbo in order to remove the pipes so that Justin could finish the butt welds on his table.
Because the turbo/collector was serving as a fixed reference/end point for the runners, I wanted to weld as much of pipes 3 and 4 in place before removing the turbo, since once the turbo was removed, I was going to lose that fixed reference. Now, we could have completely tacked pipes 3 and 4 together, removed the turbo, and done the final welding, but I had two concerns: 1) That we might accidentally break a tack before we had a chance to weld, and 2) both tacking and welding can significantly warp the runner.
For these reasons, I had Kyle do an initial set of tacks at the cylinder and collector ends (these runners each had three butt welds: forward, middle and rear). This allowed us to remove the runners as fore and aft halves (they weren’t joined at the middle butt yet), weld them, and reassemble them to the engine/collector. As suspected, the weld process slightly threw off the last (middle) butt joint, but we were able to remove the pipes and “massage” them with a few whacks of a dead blow hammer to realign them. We then performed the last middle-butt tack welds to join the fore and aft halves, and then removed the turbo in order to remove the assemblies for final welding. Of course, the same concerns about breaking a tack or a warped runner still applied, but in this case, we were only dealing with a single joint, not three, so the likelihood of a tack break or tolerance stackup error was significantly reduced.
Once everything was welded up, we inserted all the pipes into their cylinders, then wiggled the collector onto the four pipes (even though everything was tacked in place, the inherent stresses of the weld-up mean that you still end up with a bit of warping and misalignment compared to the “dry fit”). The collector went on without too much trouble, and with a bit of finagling we were able to tighten up all the exhaust flanges. Kyle then welded wing tabs onto the collector and pipes—these are bolted together to prevent the collector from vibrating itself back off the end of the pipes. At this point, our work at SPD was finished and we loaded the engine back into the car.
Turbo Hanger and Downpipe
Back at the shop, I had two remaining tasks to complete the system: the turbo support hanger and turbo downpipe. First was the hanger: Although once assembled the four pipes have no problem supporting the weight of the turbo, under operating conditions (extreme heat, vibration and G-loads) cantilevering the heavy turbo off the pipes is asking for cracked pipes.
My goal was to support the weight of the turbo, i.e., in the Z-direction (up/down). Loads in the fore/aft direction are trivial, so the stresses are easily handled by the pipes. In fact, I would like the turbo to be able to easily translate fore/aft to allow for the change in pipe length as they change temperature. Side-to-side loads are also expected to be small but are vibratory in nature and so deserve attention. I expect these side loads to be shared between the pipes and the intercooler. The intercooler is attached to the engine mount adjacent to the turbo and will react to these side loads vis-à-vis the compressor outlet hose. All of this remains untested—it’s a new design after all—but I’ve followed best practices and studied other installations. We will monitor the exhaust system and nearby components closely during flight testing to make sure any issues are addressed.
The hanger is simply a section of aluminum tubing with rod ends threaded into both ends. Use of rod ends facilitates mounting, permits movement of the turbo in the X-Y plane, and eliminates any bending-induced stresses from the mount or turbo so that the tube is in pure tension. A laser-cut stainless steel bracket serves double duty as the hanger tube connection point to the turbo and replaces washers for the hot-section clocking bolts. I chose this location as it is not uncommon to use the clocking bolts in this manner (the OEM Rotax does this), and it also supports the turbo close to its center of mass, which minimizes bending stresses on attached components.
The final section of the exhaust system was the turbo downpipe, and this needed to wait a few months until I had progressed the firewall forward and cowl design. Once I knew exactly where I wanted the exhaust to exit the cowling, I repeated the above process of creating a 3D printed mockup to check for fit and clearances. I then ordered mandrel bends and a flange, and cut and prepped parts for the welder as I did for the runners.
Stock turbo flanges are available from places like Summit or Burns Stainless, but I wanted a modified profile so had mine cut by SendCutSend. By this time, I had found a marine shop nearby with welders experienced in welding stainless steel. These guys work on 2,000+ hp cigarette boats, so they kind of chuckled when I told them this was for a 155-hp engine.
So what are we talking in dollars for this custom exhaust? The custom collector from SPD was the most expensive individual part at almost $1,000. Edge Performance kitted up a set of exhaust sockets, flanges, high-quality EGT probes and bungs, which added $800. After that, the tubing and mandrel bends ran a few hundred dollars and the SPD welding was less than $2,000; the downpipe parts and welding added another $500 or so. All up I don’t think I spent more than $5,000 for the entire exhaust (Note: I kept the turbo from the original, so that’s not included); an OEM Rotax turbo exhaust will run you about the same. I doubt there are any significant weight or performance differences, so unless you have a packaging issue like I did, the stock exhaust is the way to go.
The real cost of a custom DIY exhaust is your labor, since the way to save money on a project like this is to do what takes the most time: Figuring out the layout and then cutting and fitting the tubes and bends. Packaging-wise, I’m really happy with having moved the turbo to the aircraft centerline. Dealing with the radiated heat from the exhaust and turbo is going to constitute a fair amount of work, and I’ll cover that in a future article.
Brazing Turbo Oil Lines
The turbo is lubricated by engine oil and requires a pair of feed/return hard lines. Because the turbo was relocated, new hard lines were required. Three of the four terminations are brazed banjo fittings, so a jig was fabricated to ensure that the fittings were clocked precisely to the tubing. Thanks to my machinist friend John Lynch for helping me out with my first brazing project.
Hey Eric! What mechanism are you using to control the wastegate on your engine? Thanks!