Today I want to relay a couple of interesting things that I discovered while examining a malfunctioning Parker SCJN-600-01 pressure gauge - the ubiquitous workhorse of hydraulic technicians all around the globe.
This model "wears many brands", and today I got a good chance to poke inside it and I want to tell you my thoughts and findings.
Why the heck did I decide to gut an expensive pressure measuring instrument? Simple. A friend of mine asked me to take a look at one of his gauges that stopped working. Oh boy, I am so glad he did!
So - the cause of the failure became apparent as soon as I opened the gauge and looked at the pressure transducer:
The tiny golden wires connecting the strain cell to the PCB pads got dislocated, and obviously, no pressure reading is possible with the disconnected wires. I am suspecting the transparent glue is supposed to be covering the wires as well, so this may be a defective transducer.
But why stop here? Since the gauge is already open - let us study its design. First of all - the heart, i.e. the pressure transducer. No markings on it whatsoever, but I would say that it looks very much like a Wika TTF-1 thin-film pressure transducer, at least judging by the picture that they put in the datasheet. I did contact Wika to see if they could somehow supply such a transducer, but of course, I got the reply that the minimum order quantity is 1000 units. This means OEMs only. How sad.
The transducer is very robust (unless someone yanks the wires off of it), and also stable. I have many of these gauges in the shop working problem-free for years.
I disconnected the transducer from the board (still have to see if I can find a way to weld these tiny wires), and connected an external set of resistors in a bridge configuration to run some tests and see how the instrument functions.
Before all - a word about the plastic battery holder that is screwed on the PCB. It's terrible, in my opinion! Obviously, it serves its function, but there's absolutely zero protection from a leaking battery, something that can short the PCB and render the device useless. I would love to see at least a plastic shield between the battery holder and the PCB. I do see the commercial reason behind this.
Another thing - this model is more recent, and it houses SMD components on one side of the PCB only, which is a big step up from older models, that used a more complicated design, and had the analog parts covered with blobs of black or blue goo, most likely to protect the design from "intruders" like myself. Luckily this PCB is completely exposed, as all the boards should be, and is very easy to inspect.
The power comes from the L6920 step-up converter which boosts the battery voltage to a stable 3.3V. A common solution for battery-powered devices. The main micro-controller is the MS9S08LC60 in an 80-pin package. It has an 8-bit CPU, and once again, is a nice solution for a battery-powered device with a custom segmented LCD. By the way, the board crystal connected to the MCU ticks at 32.9 kHz. I suppose this is one of the reasons this system is so power-efficient.
Now the analog part really surprised me. First of all - the transducers bridge is AC driven directly from micro-controller pins! You can't go simpler than this. Here are the scope shots. (The bridge driving ends in relation to GND).
As you can see, the frequency is 100 Hz, but the duration of the pulses is only 1.3 milliseconds. At least this confirms the 10-ms scanning rate mentioned in the datasheet. Driving strain gauges with AC is a good way to cancel the offset drift, by the way.
The bridge output is connected to MCP6071-based instrumentation amplifier, which has its output offset to about 1.6 Volts using the AD5228 trimmer, which can be set by the MCU. I didn't see any communication between the MCU and the trimmer though, at least in this particular gauge.
Here's the sketch that I made:
This is what the amplified signal looks like (at 600 bar):
As you can see, at 600 bar the reading is about one volt above and below the 1.6V offset. The amplified signal is fed directly into the MCU, which means that it uses its built-in 12-bit analog to digital converter.
This actually raises the question of the gauge's resolution, because it uses the 3.3V for reference, which means the swing of 2 volts at maximum pressure doesn't seem to utilize the full span of the ADC, which in its turn means that the available 4096 counts are reduced to about 2500. This still gives the gauge steps of 0.3 bar, which I guess is perfectly usable for the 600 bar range.
Now, as usual, the bullet-points/ideas:
See - digital pressure gauges are simple!