Insane Hydraulics

Site theme image

The Analog Front End of the Cheapest Digital Pressure Gauge

I received another "economical" digital pressure gauge kit this week. Three gauges in it, once again Ritherm (zero affiliations), and it is supposedly more advanced than the very basic 3305 I reviewed the other day. As usual, I will be opening the gauges up and running some pressure tests on them - I like to have an "educated opinion" about diagnostic tools and know what I can (and can not) recommend them for. It will all be revealed pretty soon, but as I was looking through my teardown of the 3305 (which will serve as a baseline of sorts when I will be evaluating these new gauges) I realized that I didn't measure the voltage that the resistive bridge/AD8628 op-amp combination was putting out, and this is an important piece of information for any strain-gauge based measuring system. Today I want to correct that. Put some numbers on things, as I like to call it.

So, I soldered a couple of wires to the gauge's PCB (large components make soldering a piece of cake), connected them to an oscilloscope, and then pressurized the gauge with my trusty Pressure Maker II:

But before I reveal the voltages, let me clear out another omission of mine and show you the electric diagram of the gauge's front end:

As I said before, I really like the application of an auto-zero amplifier - it makes the system very simple. I can't tell what the R16 is doing here, seems unnecessary to me, but the rest of the schematic makes sense - simple and functional - a non-inverting configuration with a gain of 22. The bridge-driving SOT-23 transistor is marked with "H16", but I didn't manage to find a matching model for it that made sense, however by the way it is connected, it has to be a P-channel FET.

And now - for the voltage values (the oscilloscope was hooked to the right side of the R17). Here are the shots at 0 bar:

The MCU powers the front-end circuit up twice per second in 30 ms bursts, and the zero bar output of the operational amplifier "hangs" around 25 mV. There's a lot of 50 Hz noise being picked up here, but it disappears as soon as the op-amp "wakes up".

Now, here are the values taken at 600 bar (I decided not to push my test hose to the "sweating limit" this time):

The 600 bar bumps the output voltage to 107 mV, which means the full scale (1000 bar) value is 178 mV. The bridge is excited with about 3V, and the amplifier gain is 22, so the pressure traducer sensitivity is around 2.7 mV/V, which is a thick film territory. I guess the reddish tracks of the transducer cell (closeup pictures from the last post, if you missed it) are indeed copper.

I also saw that every time the op-amp is powered up there's a 1 volt spike at its outlet that lasts for 250 micro-seconds. Just an interesting fact:

Now I can see why the gauge has only 1000 counts. With about 1.3V fed into the voltage reference pin, and less than 200 mV signal span, it's definitively taking full advantage of the 16-bit Sigma-Delta ADC of the MSP430F4250IDL microcontroller.

Hmm... Now that I said it - I wonder if I can turn this gauge into a 100-bar gauge with a 0.1 bar resolution if I increase the gain and trim the Vref down (just for the heck of it)?.. Sounds like a fun weekend project, doesn't it?