Saturday, September 26, 2020

Problems With Your Kernel Boot in VirtualBox?

I had a boot problem with my SilverHawk Linux build.  It runs under VirtualBox, and the "init" for the boot did not load.  It actually resulted in a Python stack trace that had so much information, it scrolled past what was needed, and you can't scroll back.  There was instantaneously a need to log all kernel boot information.  Here's what you need to do in VirtualBox.

  1. Open the settings for your virtual machine.
  2. Click on "Serial Ports"
  3. Place a check mark in "Enable Serial Port".
  4. Select "COM1" for the port number (unless you have another one, obviously).
  5. Select "Raw File" for the Port Mode.
  6. In the "Path/Address", you can then specify the file you want it logged to.

Once that is done, you can then add a kernel command to redirect output to your boot kernel command line in your Grub/Grun2/ISOLinux boot program.

serial --speed=115200

Now when you boot your system, you should get a new file (overwritten) that contains the boot sequence that can be used to debug your boot problem!

Sunday, September 20, 2020

Carburetor - Adding Ported Vacuum

On my carburetor, there is no ported vacuum for the vacuum advance on the distributor.  Many hot rod folks say to just use manifold, but manifold and ported are actually inverted.  Using manifold vacuum for your vacuum advance would mean that you get MORE vacuum advance on idle than you would on open throttle.  So, I decided to remedy that by installing a nipple on my Holley carburetor fuel metering panel.

You have to make sure that the fuel metering panel supports it, and that the carburetor has the input, too.  Ported vacuum is truthfully the vacuum produced by the venturi effect, which comes from the carburetors venturi (not all carbs have them, but mine as a venturi carburetor does [surprise, surprise]).  It also had the vacuum channels required to support venturi vacuum, so I know I'm good.

I was actually surprised to find that the vent hole where the vacuum nipple attached was not sealed.  Venturi vacuum would pull air directly from the outside.  That was more shocking.  Most folks drill out the metering panel a little bit more and install a 1/8"NPT nipple.  I wanted a more "stock" look, so I measured a few things.

I used some brass hex rod that I had laying around, and drilled it out, then shaped it on the lathe for the nipple that I had on my old carb.

Parting it off, I had a functional vacuum tube.  The fuel-metering-panel end had a taper so that it would be a press fit into the existing hole.

It was then only a matter of pressing it into the fuel metering block and ensuring that it did not cover the hole into the venturi vacuum chamber.

Now to get the carburetor installed.

Custom Elbow Spacer for Car

 As I'm nearing getting the Corvette fully operational, I found with this new engine block that I needed a customized elbow for the oil pressure sensor.  On this new (old) block, the oil pressure port is sitting on the back, top edge right by the intake manifold.  With having the Holley Contender put onto the block as the manifold, I suddenly had the need to redirect the oil sensor in a different direction and away from the HEI distributor.  So, I made two of them (the first one turned into a goof).

The first one I made from a block of aluminum.  I ensured I had a flat and parallel surface or two to use as a marking surface.

I got them marked up.

I drilled the through-holes for the elbow first using a 7/64" drill bit, broke that off, and re-did it using a #18 drill bit.  This is sufficient because the sensor only has a 0.020" hole for the pressure to make it through, so a #18 is plenty of surface area without going too large on the hole. I also used a 3/8" end mill and bandsaw to remove extra stock so I didn't have such an interrupted cut on the lathe.

Once I had the two drilled holes connected and cleaned, I chucked it up into the lathe.  I centered it using the #18 drill bit to make sure I was axially in line (not a rotating part, so even if I'm off 0.005" there isn't a problem as long as I don't puncture the side or have weakness there).  Then I could cut a flange and thread.

I went to install it (I never even cut the upper, female thread before this point).  Turns out, it wouldn't fit.  It was about a half inch too short to clear the intake manifold.  [sigh].  Round two - ding! ding!

This time, I took a chunk of aluminum round bar.  This was the easier way to do it, anyway.  I used the lathe to drill the first hole (much longer than the first time, because I can cut stock off if needed).  Then I turned it down for the shoulder, and subsequently turned down the threaded portion and installed the thread.  After that, I put it in the mill and put three flat faces on there (two so I can use a wrench to get it tight enough, and one for the face that will have the intersecting hole).  I did the parallel faces first, followed by the intersection next.  With the last face done (and while it was still in the mill), I used a 5/16" end mill to put a hole in it, then a 3/8" part way in, followed by a short stint on a 7/16" end mill.  See, this one has to be a 3/8" pipe thread, which has a taper, and the different sizes will get me close enough to use a pipe tap.  I then started the pipe tap using a straight-shank center in the same 3/8" collet, and put the tap in place and began cutting.

I think this iteration is going to work just fine.

Sunday, September 6, 2020

Mad Scientist

 I have my old PC I built in 2004.  It still runs awesome (it has a copy of Windows Vista that I don't boot into).  Recently, the memory crashed, so I ordered some new RAM for it.  While it was out, another project I've been procrastinating started.  I wanted to quiet it down a bit so I started working on my liquid cooling setup.  First, I got all the parts together and started assembling :

My first implementation had the liquid split before each CPU.  I'll tell you in hindsight that this is a bad idea.  I was lucky enough to not have the system fully installed before I realized that.  If the liquid splits, one of those tubes could have an air pocket that prevents liquid from flowing there.  That means you'd have one CPU not getting cooled.  Again, bad idea.  Here it is, replete with a ton of air pockets that I could not clear out because of the arrangement :

My second iteration cycles just fine.  I pipe one quad-core CPU output as an input to the other CPU.  Yes, that second CPU will run hotter.  But with the speed of the liquid, it should be minimal.

Once I validated that I have a functioning closed-loop on the liquid, I started looking at the wiring.  I need to power all three fans on the radiator, and I need to power the pump.  That means I need a molex splitter and a molex-to-3-fan adapter.  Those have been ordered, and should arrive some time this week.  Once those arrive, it will be time to actually bolt it all together.  I need to bolt the radiator and the pump to the chassis, and then I can find a way to anchor my ugly, cheap reservoir (yes, that reservoir is a pickle jar with one tube extended all the way down to pick up the water/additives).

Again, I don't care how this one looks - I'm not going for "style" points here.  If I did, I'd have used a red-dye instead of a blue one.  It is simply a proof of concept before I do the other liquid-cooling setup.  That one is also an older machine (again, a 2003 dual CPU [each being a quad-core Xeon]).  That one is specifically intended to be a work of art.  Reservoirs will be glass jars with custom lids and fittings turned on the lathe, copper and brass manifolds all out in the open (no chassis other than a carbon fiber back plane to hold it all together, think "exposed steampunk").  I am getting close to crossing that one off my list since I'm now taking my first steps.

Good News and Bad News 2020

 I have good news and bad news.

That's not true.  I have great news and okally-dokally news.

Before I could get it to run on it's own, I had to drop the 18 PSI fuel pressure down to 8.  I installed a fuel pressure regulator for that, and then it only took an hour to get it "running".  The Corvette has now been started on it's own (no more pouring fuel down the carburetor throat).  The timing was 4 degrees after top-dead center.  It makes a difference when you advance it enough.  Right now it's at 8 degrees before top dead center - I need an additional 4 degrees advance, but it still runs.

Now for the okay-dokally news.  There are a number of issues I need to work through before I run it again.

First, the radiator hose is 1.510" on the inside diameter.  The radiator is 1.131" on the outside diameter.  Yes, it didn't just "leak", it flooded my garage.  That's okay.  I just need a Gates 26390 hose reducer.  It should go from a 1.5" size to a 1.25" size.  That problem is "solved", but is waiting on parts.

Next, I need to work on the fuel delivery.  The engine runs best with the idle mixture screws in all the way (barely seated).  That's not right (hear me out - it's really not).  It means I have fuel coming from somewhere.  The main jets don't engage until you have a little more pressure (your power circuit for fuel).  So, it's not my jet sizes.  The throttle plates are probably going to need adjusting,  to get the transfer slot mostly covered.  I know the secondary throttle plates are good.  The primaries I cannot see, so I'm going to have to take off my carburetor again.  If I am going in there, I probably need to do some work on my metering panels.

I want to (while I have it off) check that I have the power valve blowout protection, and get some new power valves in there to make sure that isn't my problem.  The symptoms align with the power valve being blown.

I need a ported vacuum for the distributor vacuum advance.  I need to install that to my fuel metering panel.  It looks like it has everything there (channels, etc).  I think it will be just drilling the fuel metering panel out and installing a hose flange for the distributor.

A little more information.  The intake manifold, 14057053 GM CFM 43, is for some 350 CUI blocks.  It's a spread bore intake.  I have a spread bore adapter for a square-bore carburetor.  Note, I still have my "Holley Contender" intake manifold from my old 350 CUI engine.  It is a dual-plane 300-38 (701R-38) "dominator" intake manifold.  If I remove my carburetor to handle the power value blowout protection, I'll probably swap that in.  Here's the "Contender" information :

The carburetor is a LIST-7010 carburetor (4160-style).  It has a secondary metering panel installed to it, center-hung floats (fully adjusted now), and should flow 780 CFM on it.

This thing should be able to idle smoothly.  Let's get started tuning it!