Insane Hydraulics

Site theme image

Exploring the Design of a Low-Cost 100-bar Digital Pressure Gauge

Today I am tearing down the 100-bar digital pressure gauge from the three-gauge Ritherm kit that I tested last week. I truly enjoy doing this, by the way - it's like solving an intricate puzzle - you have fun and learn a ton at the same time.

Dismantling this gauge was a bit tricky because I didn't manage to remove the front cover. I probably could force it - but I didn't want to scratch or dent the housing, and I didn't fancy making a proper jig for that either, so I pulled the internals out "through the back door":

There are four soft silicone-insulated colored wires running from the pressure transducer to the PCB, and a fifth (black) one jammed under the nut of the transducer housing - presumably for "shielding purposes". You can immediately tell how simple the system is - there isn't even an operationál amplifier "in sight" - just the Holtec chip, a couple of power components, a couple of dozen passives (literally -12 resistors and 12 capacitors), an LCD screen, an LED back-light, a couple of push-buttons and... that's it.

I don't mind simple - simple is good! I'll give you my thoughts on the electric schematic and the components in a minute, but before that - let's have a look at the "heart" of the gauge - it's pressure measuring cell, which can be easily removed from the (surprisingly thin-walled) stainless steel housing:

Wow! It is an 18mm concave film ceramic pressure sensor! As you can see, it "faces" a Teflon ring with an o-ring seal inside it, and is held in place with a threaded bronze ring.

The strain gauge transducer measures 9.9 kOhm across the classic full Wheatstone bridge, and I would say that it looks a lot like the A-type AC18C from IntelliBee, although you can never be sure who makes what when you deal with Chinese factories, who are like OEMs for OEMs for OEMs... If we assume it is, then it is supposed to have a typical output of 3mV/V, with ≤±0.3%FS overall accuracy, and less than ±0.1% FS repeatability - which is not bad at all!

And now, let us have a closer look at how this gauge transforms the signal from the cell into the numbers n the screen. When you have a double-sided PCB with such large components, coming up with an electric schematic is a piece of cake:

So, what do we have here? The power-delivering part is simple - there's a diode-less step-up converter (a SOT89-3 package marked as C3 OL, maybe a PAM240?) that boosts the battery voltage to 3V, and then an LDO that drops it to 2.8V (SOT-23, marked as 54FK, according to the AliExpress this would be the ME6206P282MR, possibly a clone of the Diodes Incorporated AP2138 or any other similar LDO). This is a straightforward and perfectly functional arrangement to make sure that you get all the juice out of your batteries.

The "brains" of the gauge would be the 64-pin Holtek HT67F5660 chip that combines a 24-bit Delta Sigma A/D converter, a programmable gain amplifier, an 8-bit microcontroller, and a fully integrated LCD driver in a single package. OEMs use these chips pretty much for anything that needs to read something analog and then display it on a segmented LCD - think weight scales, thermometers, etc... It is truly amazing how much stuff they can fit into a single microchip these days. And for cheap - at the moment of this writing Hotlec USA website lists this IC at $2.15! This is, most likely, for complete reels, but still - 2 bucks for a Delta Sigma ADC that can, according to the datasheet, read a thousand plus samples per second with 14 effective bits is a bargain in my book!

The bridge output goes directly into the Holtec via a couple of 1k resistors - there are pads for two filtering capacitors, but no capacitors, there, strangely. The only thing that I can't explain would be the bridge driving part of this circuit, which remains a complete mystery to me. The excitation of the bridge is, as far as I can tell, performed from the VOUT pin (1), which goes through a SOT-235 IC that I didn't manage to identify (marked as AAQ4). The chip lowers the 2.8V of the VOUT to 1.6V, but the positive reference is fed from a voltage divider paced before the "mystery chip", which I can't really explain. I poked about the board for a long time trying to figure it out and triple-checked the connections, but I can't see why they did it the way they did. The lower end of the bridge stands on two paralleled 2.7k resistors, instead of being tied to the ground - can't see why they did this like that either, but hey - it works, doesn't it? All I need to know is how fast the bridge is "tickled" to see if I can call this gauge a proper gauge or not.

And let me tell you... Oh boy... It doesn't look too good. This is what you get at the VOUT (which is also connected to a wire pad on the PCB marked as V+):

And this is what you get on the bridge:

This is "no bueno" at all. Less than two reads per second! And the ADC is, indeed fast, because the 1.6V excitation shot lasts only for 9 ms. This makes no sense to me. Obviously, this is a low-cost system, but it appears to have all the right components to be much better than what it is! It can be faster, and it can have better resolution too (which is especially shameful for the 400-bar gauge from the kit with its 1-bar step).

So, once again I tell you that, in my opinion, these gauges are only good for static and quasi-static pressure measurements.

I like the simple design, I think that the ceramic pressure cell is great, but I believe that the firmware is underdeveloped and can be better. If somebody managed to give the 400 bar gauge 4000 counts, 100 reads per second, then average every 20 reads for a 5 Hz numeric read-out, and every 5 reads for the 20 Hz bar graph - then it would be a stellar gauge for the price! Right now it is just a steam gauge with a marginally better resolution. Here's how would I sum this "build" up in one image - take a look at this picture, taken from the inside of the housing, and appreciate the "peculiar" alignment of the holes in the acrylic panel with the membrane buttons:

- Does it function?

- Yes, it does!

- But is it good?..

- Good enough... But only for a fraction of pressure-measuring tasks.