I have had a go (again) at tuning my car since - although it seemed to run
ok - it had something of a reluctance to accept large throttle openings at medium revs. For example on the M6 at 55mph I would have to change down to 4th to hold my speed on the climb up to Saddleworth when returning from the Dark Side.
Firstly I decided to check out exactly how it was going with a few tests:
At low revs, any throttle setting was fine: It ran smooth and pulled well.
At medium revs on a small throttle - it was also fine, but at medium revs on a medium (or higher) throttle it was flat: i.e it didn't want to pick up, the more you open the throttle the flatter it seemed.
Interestingly, at high revs any throttle setting was fine: Once past 3500 rpm it pulled really well
My first thought then was timing. Since fitting MegaJolt I had never had a serious go at optimising it since right out of the box with my guestimated maps, it worked pretty well. And anyway- without a rolling road the differences that small changes make are really hard to detect. So a couple of days were spent in a cycle of web-searching, changing the map, and road testing. A lot of frustration ensued as I just couldn't get rid of the flat-spot. After two days I was pretty certain that timing wasn't the cause of the flat-spot. Not saying that the timing is right, mind - just that it's not the cause of the flat spot.
So I started looking at the LPG setup...
My car has the most basic setup you can get - a vapouriser, a manual power-valve and an annular mixer. According to the internet, there should be three adjustments available:
Idle bleed screw: a simple needle valve that allows an amount of gas to bleed into the system continuously - this is to compensate for there being virtually no vacuum at the mixer at tickover and does not pass enough gas to enough to effect anything else. It is quite common (depending on physical variations in setup) for this to be entirely closed. This looks like an old carburettor mixture screw and sits on the side of the vapouriser.
Bias control screw: This controls the amount of spring pressure holding the main gas feed valve (in the vapouriser) closed against the force generated on the diaphragm by the mixer vacuum. The more this is screwed in the more vacuum it takes to open the valve and get gas. It is a hex-headed bolt set near the idle bleed screw.
Power valve: A simple valve which progressively blocks off the pipe which feeds the gas from the vapouriser to the mixer. It obviously effects how much of the vacuum at the mixer is 'seen' by the vapouriser.
Having got this straight I then started looking for instructions on setting up an open loop system.
First point is that you really need some sort of exhaust gas analyser, and preferably one that could be used while driving so that you can check various engine loadings. Now I had already gone part way towards this last summer when I installed a lambda sensor and constructed a small indicator circuit - however it appeared at the time that the circuit did not work correctly, so assuming I had got it wrong (it was my own design) I abandoned it. Anyway, now my thoughts turned to it and the possibility of either getting it working or at least of savaging some bits, so I located it (under the settee!!! ???) and set-to testing it again. Guess what? It works! It always did. All that was wrong was the setting of one of the trimmers... D'OH!
So I wired the display to the lambda and started it up. Setting the display unit up proved troublesome - I didn't have a reference voltage and I didn't know what the mixture currently was, so what I ended up doing was running it on petrol and setting the trimmer so that the indicator would go from rich to lean under various conditions. Still not sure how close it is but at least I now have a rich-lean indicator.
Now for the gas. Always lean. That's it - always lean. Tried the power valve all over the place - always lean. Tried the idle bleed all over - always lean.
I couldn't try the bias valve because my unit didn't have one. But wait-a-mo... From my new found understanding (?) of the workings of the vapouriser - it HAD to have one. A good root-around with a mirror and torch found nothing like the pictures, but in the place where I had expected it to be (remember it should be a 10mm hex-head adjuster sticking out the side) I found what looked like a counter-bored hole with an allen bolt sunk well down. I had seen this before, but since it was tight down I had assumed it was a fastener and not an adjuster - but with my new-found knowledge of the internals of the vapouriser - I realise that I was wrong. I gave it a bit of welly and it came loose, and a couple of runs in and out freed it up to the point where it now felt like an adjuster. A quick try with the engine running show that it definitely has an effect.
So now I had all three adjusters available and some basic understanding of what they do, and a working exhaust gas analyser.
Here comes the science bit:
An engine at a given revs consumes a given amount of air.
Twice the revs requires (more or less) twice the air.
The amount of air directly effects the velocity of the air through the mixer.
According to O-level physics, Bernoulli's equation shows that (within limits) the pressure in the venturi will be inversely proportional to the square of the velocity.
So the vacuum produced by the venturi (mixer) in the gas supply pipe is proportional the the square of the revs.
The amount of gas supplied by the vapouriser is roughly proportional to the vacuum that it sees from the engine.
From the above we can see that without the 'power-valve' the mixture would go horribly rich at the higher end of the revs. The power-valve is also subject to Bernoulli's equation, but works in opposition to the venturi and so tends to make the amount of gas supplied more linear with regard to the amount of air being drawn into the engine.
So - it now becomes clear that the idle screw can be forgotten about until everything else is done, the bias valve adjusts the mixture across the range, and the power-valve compensates for the non-linearity of the mixer.
I warmed the engine thoroughly and set to:
Since my gauge was showing lean at all times on gas I started by screwing out the bias adjuster. Unfortunately it just stayed weak and after a few turns it actually started to run rough.. So I set it as rich as I could and went for a test-drive. The only way I could get a rich indication was right at the top of the revs (5000rpm plus) on a wide open throttle. Previously I had not been able to get a rich indication at all on gas so this was actually an improvement!
Two days of screaming back and forth around local roads and adjustments in lay-bys later and I have only improved a little on this position: At tickover the mixture wobbles around lean-rich-lean. When driving, under all conditions except flat out at high revs it shows lean.
The effect on performance has been dramatic though - especially in the 2000-3500rpm range where it now pulls really quite well (although remember that I have not driven any other V8s so I don't really have a 'control' sample).
So finally - why am I unable to get the mixture right throughout the range? Well I suspect that the vapouriser is a little the worse for wear. I think that the non-standard bias adjuster may be leaking air into the system when screwed out which might explain why I can't get it to go fully rich and I am not sure what the rest of the internals are like. I have drained the 'heavy-ends', but cannot fully service the unit without a gasket set which would cost £25, and with a new unit costing less that £70 and the fact that mine has obviously been bodged (the allen bolt), I have decided that once the Christmas spending has worked it's way through, I will replace it. I am also hanging my nose over a Blos LPG carburettor - which replaces the mixer with a vacuum-controlled variable choke system (like an SU carburettor) which removes the need for the power-valve, but that may have to wait a while since there are many other things I need to sort first.
For now though, it is running a LOT better than it was and (hopefully) will be rather more economic than it has been.
If anyone is still reading - thanks! Hope you enjoyed it
Roger