Turbo Time! Strip down and Mod of Turbo Gauge

Background

I bought a Greddy Turbo Gauge off ebay a while ago it if in my Landrover. The engine is a 200TDI, which is a 2.5 Litre  Turbo Diesel.

However, When turbo's go, they go VERY badly and are expensive, so this gauge is more to check on how well it is rather than for tuning purposes. A new turbo is £200 for a second hand one and £400 nearly for a new one! A refurb. kit and rebalance is £100, so I'd rather catch it before it goes for a refurb. than when it goes and cough up for a new one! Observing a loss in boost pressure could be an indicator.

Test and Inspect

So i got this gauge a while ago and notice that the guy had cut the back-light wire off on it near to the nub practically. Of what wire I could clip onto, I stuck  12V on it and sweet-fanny-adams, no back-light!

I then got a mate to check the gauge worked, whilst I blew into the pipe inlet, needle moved when I belw so at least it's primary function worked!

I extended the wires on the back-light first, as this was an easy job, and plus I need to reference which one was positive.

Disassemble!

To quote a certain Mad Aussie Bloke (+EEVblog) :

"Don't turn it on! TAKE IT APART!"

I began to dismantle:

I undid the nut holding the inlet on the back of the turbo, in hoping to loosen something up. No luck! All I succeeded in was to knack up the needles default position. So this meant I had to cut off the crimped on collar holding the front screen in place.....I'll have to come back and think of a way to put that on later.

With the glass out and the nut on the inlet off, I slide the gauge face out of the casing, Inside was a circuit card with 2 blown filament bulbs soldered onto it (why, why would you solder a device that you know will blow and needs servicing to a PCB!) and on the back of the gauge face, what looked like the insides of a grandfather clock!

Old Skool Brown and copper PCB!

WHY! WHY, solder these on?! Use a F*@king socket!

Pretty Primitive when you look at it!

Inner Workings

After a closer inspection, I managed to work it out.

The thick bent piece of copper is a hollow and is connected to the inlet by a small piece of copper pipe. One end of the bent copper is soldered shut. The closed end of the copper is connected to an arm, the other end of that arm has a toothed rack, that moves a toothed spindle. The spindle is connected to the needle on the front of the gauge and a coiled spring, so the needle can return to its rest state.

When air pressure is applied to the inlet, it inflates the bent copper pipe slightly, this cause the bent copper to expand by increasing its radius (basically trying to straighten itself out!). This pulls on the arm which moves the needle.

There are a lot of small movements involved, that are scaled up to big movements (needle). It could be said that this is a mechanical amplifier!

I managed to reset the rest position of the needle by squeezing the the bend copper pipe slightly to run the rack off the spindle allowing the needle to move freely. Then just resetting the needle so that when the rack ran back onto the spindle, so that it rested as near the 0 mark as possible.

Modifying the PCB 

So the first thing was that stupid filament bulbs come off, with a hot soldering iron:

I had a couple of spare white LED's lying around at home. These will do to replace the bulbs.

Now a few calculations to work out the current limiting resistor:

I got the Vf & If from hooking one of the LED's up to a PSU with a low current Trip and carefully cranked up the voltage with the fine adjust:

So why have I hooked up the LED's in series? Well given that my input voltage is much greater than Vf of each LED, I can do it this way and only have to use 1 resistor.

"But you could do it in parallel, and not hack up the PCB as much!"

Yeh, but that's the wrong way of driving an LED, working below shows why:

Both solutions pull twice the current. Yeh I know its gonna be supplied by the landy's battery and alternator, but that's no excuse to be inefficient! The top solution is not good as I know these 2 LED's are not exactly from the same silicon die! Hence it can't be ensured that each LED will pull 10mA each, and as such their forward voltages will "fight" each other.

If you wan to drive more than one LED at a different current, then the second solution is more suitable.

The top Solution should NEVER REALLY be used!

With components sourced, I modified the LED's so they could be soldered on at the same height as the bulbs:

After some track cutting on and soldering of a few link wires, I got the PCB how I wanted it:

Testing

I set the Benchtop PSU as follows:

So this is basically simulated running off the battery, I set the current trip to the max current limit the LED's can carry, so I don't blow them in the event of a fault.I know they can carry more, but I designed them for 10mA, so more than 20mA, I know somethings VERY wrong!

Flicked power on a VIOLA! Not too bad!

And current consumption is near as damn it spot on to calculations!

So, cranked up the voltage to simulate the alternator running.

Flicked power on. Hardly much difference......

I did expect a current rise of this proportion, so all is as I expected.

So all in all, I'd say that was working properly!

Re-assemble

I popped the PCB back in its place:

Popped the dial back on and bolted the inlet to the back of the case. I then hooked her up for a showcase!

OHHH! Pretty!

You'll notice the bright spots on the dial, this means the LED's are too close to the dial, not the worst thing that could have happened!

That's good! But Have you knacked it by taking it apart?

Well, I  looked at the needle in the mirror whilst I blew into the inlet. The needle moved into the positive pressure range, whether by the right amount I don't know ( I blew to about the 0.4kg/cm mark). If this is wrong then I would think the spring needs adjusting.

The main thing that's a concern is how I'll get the glass cover back on now the collar is mangled! Super glue and a G-clamp is looking a likely solution.

Once I've sorted that, then its ready to mount on the centre dash console with another project I have in the pipeline....watch this space!