After I determined that Parker ServceJunior pressure gauges (and their clones) are solid pressure gauges with respectable accuracy, I set out to verify
a) if my wireless gadget (the latest prototype board) can achieve similar accuracy with common pressure transmitters, especially in the low range, and
b) since we do use and sell quite a lot of pressure transmitters ourselves - I wanted to check how different models and brands compare to each other.
Before I even begin the discussion, I feel that I need to address the matter of terminology first. There is an opinion that a pressure transducer sends out a voltage signal, while a pressure transmitter sends out a current signal. This opinion has its grounds, but in real life, the terms industrial pressure transmitter, pressure transducer, and even pressure sensor are often used interchangeably, so I am not going to nit-pick the words and, like all normal people, will use these terms as I please, meaning, essentially, the same thing - a piece of hardware which connects to a pressure source on one end, and outputs a conditioned analogue signal (be it current or voltage) from the other.
Now, with this out of the way, let's continue.
For the last 5 years, my go-to pressure transmitter has been the Wika A-10. These transmitters are relatively cheap, we can get them pretty fast, and the model that I've used the most is the 400 bar version with the 4...20 mA two-wire interface. Just a couple of months ago one of our clients had one of these sensors damaged on their raise-boring rig and contacted me for a replacement. It turned out that we didn't have suitable Wikas in stock at the time, but we did have some cheap Fox TR49.2 transmitters, and specs-wise they looked like they could be used problem-free. Maybe a little less accurate, but still a 400 bar 4-20 mA pressure transmitter is a 400 bar 4-20 mA pressure transmitter, right? That's what I thought before this test.
That time I got a proper Wika sensor for that client, because they found time to wait for the customary model, and now that I know what I know, I'm glad that I did. So, for those of you who use or plan to use pressure transducers in their hydraulic systems, the following information can be of interest.
My DIY wireless transmitter uses the Bluetooth module's built-in 12 bit ADC and a very simple circuit to convert the analogue signal coming from the sensor, so this test would be a great way to verify if this arrangement is up for the task and if its non-linearity and resolution would be causing any issues.
Another very important thing I wanted to test was the ADC result interpolation done in the software. My current version of the App takes in only two data points to create a reading profile (a data point is the ADC integer output and the corresponding pressure value) and then does simple linear interpolation to calculate the result in the required units - like drawing a straight line on a plane from the coordinates of two points. (You can see the profile data that I used for each sensor in the top table rows). I wanted to see if the two-point profile is enough (although I already suspected that I should refactor the code to use more data points, simply because nothing in this life is perfectly linear anyway, and having an easy way to compensate for it in the software seems like a big plus). On the other hand - doing a test with linear interpolation of the ADC result would allow me to check how good the linearization of the signal is done inside the sensor.
So, I armed myself with the PressureMaker (if you're not familiar with the Pressure Maker read my previous post about Parker gauges), the latest prototype PCB of my wireless transmitter, chose the best of the three Parker pressure gauges from the previous test to serve as my reference, and gathered the following representatives from the pressure transducer community:
The good and expensive pressure transmitters from Hydac
Hydac - HDA 4746-A-250-000 - 0...250 bar, 4...20 mA two wire, typ. accuracy ±0,25% FS, typ. repeatability ±0.05% FS - thin-film strain gauge on a stainless steel membrane
Hydac - HDA 4746-B-250-000 - - 0...250 bar, 4...20 mA two wire, typ. accuracy ±0,25% FS, typ. repeatability ±0.05% FS - thin-film strain gauge on a stainless steel membrane
These are supposed to be very high-quality transducers, and they are expensive - with the price floating around the €300 range.
My go-to Wika transmitter, of course:
Wika A-10 - 12719341 - 0...400 bar, 4...20 mA, non-linearity BFSL - 0.5% FS, non-repeatability 0,1% of span, thin film sensor, sold at about €100, give or take.
Cheap pressure sensors from Fox, which, supposedly, use a ceramic sensor element:
FOX - TR49.2 - 0...400 bar, 4...20 mA two-wire, less than 1% FS linearity and 3% thermal drift
FOX - TR49.3 - 0...400 bar, 0...10 V, less than 1% FS linearity and 3% thermal drift
You can find these sold at €60-70, even cheaper.
And, last but not least, Gems Sensors transmitters
Gems Sensors 3100B0600S01B000 Compact Pressure Transmitter, 0...600 bar, 4...20 mA, 0.25% FS Accuracy
Gems Sensors 3100S0400S01B000 Compact Pressure Transmitter, 0...400 bar, 0...10V, 0.25% FS Accuracy
Their specs are great, they are sold more or less for the price of Wika A-10s, but I found them really cheap on Ebay. I got the 400 version for €25 and the 600 one for €35 (plus shipping). They were new to me, but the brand has existed for a long time, and the specs looked really good on paper, so I was hoping that this would turn out to be a steal (a spoiler - it did!). Normally these are sold at about the same price as the Wika A-10s.
And now - for the test results. First - the low range test:
|Bourdon Tube Gauge, 0...6 bar||HYDAC 4746-A-250-000, 250 bar, 4…20 mA, (-1324 = 0 bar, 1625 = 250 bar)||HYDAC HDA 4746-B-250-000, 250bar, 0-10 V (-2020 = 0 bar, 1853 = 250 bar)||WIKA A10, 400bar, 4…20 mA, (-1320 = 0 bar, 1630 = 400 bar)||GEMS 3100-S0400-S01B-000, 400 bar, 0-10V, (-2027 = 0 bar, 1836 = 400 bar)||GEMS 3100-B0600-S01B-000 600bar, 4…20 mA (-1329 = 0 bar, 643 = 400 bar)||FOX TR492, 400bar 4…20 mA (-1188 = 20 bar, 1106 = 350 bar)|
As you can see - all of the pressure sensors behaved surprisingly well. That is - all except the Fox 0-10 V model, which, for some reason wouldn't respond to such a low pressure, and therefore is not mentioned in this table. What a shame, Fox, what a shame!
The Hydac 0-10V sensor, was spot on, and I guess that's what you would expect from a €300 pressure sensor, but the cheaper Wika A-10, and both Gems behaved quite well, too. I was really surprised that even the 600 bar model from Gems managed to read such low pressures with acceptable accuracy. The cheap Fox with 20 mA output did OK as well.
Now let us move to higher ranges, I am putting all the charts together here, and my comments are placed below. You may notice that there's a word "boom" in one of the table squares. I explain why at the very end of this post.
|SCJN-600-01-MP||HYDAC 4746A 250bar 4…20 mA (-1324 = 0 bar, 1625 = 250 bar)||mA|
|SCJN-600-01-MP||HYDAC 4747B, 250 bar, 0-10 V, (-2020 = 0 bar, 1853 = 250 bar)||V|
|SCJN-600-01-MP||WIKA A10, 400 bar, 4…20 mA (-1320 = 0, 1630 = 400 bar)||mA|
|SCJN-600-01-MP||GEMS 3100S-0400S-01B-000, 400 bar, 0-10 V, (-2027 = 0 bar, 1836 = 400 bar)||V|
|SCJN-600-01-MP||GEMS 3100B-0600S-01B-000, 600bar, 4...20 mA, (-1329 = 0 bar, 643 = 400 bar)||mA|
|SCJN-600-01-MP||FOX TR492, 400 bar, 4…20 mA (-1188 = 20 bar, 1106 = 350 bar)||mA|
|SCJN-600-01-MP||FOX TR493, 400 bar, 0-10 V, (-1990 = 10 bar, 1736 = 400 bar)||V|
The Hydac - both of the sensors always fell within 0,1 bar from the Service Junior. They are good! And also - the 600 bar range Service junior is good, apparently. Don't forget that I am comparing a 250 range instrument to a 600 bar one! I wouldn't accept anything less from a €300 pressure transducer, though.
The Wika - at less than half of the price of the Hydac, and with a bigger range - it showed incredible accuracy - almost always within 0,1 bar from our reference - and only deviating about 0,3 bar in a mid-range spot. Impressive!
The Gems - I was expecting slightly better accuracy from the 400 bar model, but I am nit-picking here. There's a dip about 1 bar mid-range, but it's within the specs, and can be compensated for in the future version of the App, which will include multi-point reading profiles. Not bad, especially given the €25 I paid for this transducer. Now, the 600 bar model exceeded my expectations. This is the only 600 bar range sensor here, and it always landed within 0,1-0,2 bar from the reference gauge reading. Perfect!
And now for the Fox transducers (a sigh...). Truly, I am disappointed with both versions. The transducer with the 20 mA output is usable in the sense that I did manage to come up with a reading profile that approximated the displayed pressure value within 0,2-0,6 bar to the reference, which is OK. What is definitely not OK, however, is the output current signal, which is 5% off at the end of the scale. 5% off corresponds to the error of 20 bar, which is unacceptable, even if you consider the maximum error (they don't give you the accuracy, but rather nonlinearity in the catalog, so, I guess - you can't call it "lying", but I consider it "cheating"). In any case - to make this sensor usable a software connection is a must, and this means that this model can't serve as a simple drop-in replacement for another pressure transmitter. The only good thing here is decent repeatability and the cheap price.
The 10V model didn't fare well either. Below 8-10 bar the reading was simply not usable or nonexistent, and below about 30 it was oscillating all the time (hence the asterisk symbols in the table). Again - at the end of the scale (at 400 bar ) the sensor was outputting only 9,68V instead of 10,00V, which is about 3% lower than it should have. Not great for a sensor in 2019, even for a cheap one. Mid-range error after calibration was about 2 bar in certain places.
I know that these are cheap sensors, so I can't bash them to the ground, but again - I didn't expect them to be perfect, I guess I just wasn't ready to find them that bad. The absence of reading in the lower range and 3% inaccuracy makes this transmitter unusable with my wireless transmitter for diagnostics, but I guess it can serve an application somewhere if acquired cheap (like really cheap).
Regarding pressure transducers with analogue interfaces:
One should be careful when replacing one pressure transmitter model with another even if it has the same pressure range and uses the same signal interface.
Regarding my wireless transmitter project:
Both 4...20 mA and 0..10V sensors can be used with the wireless transmitter for remote pressure monitoring with good precision. I did find that the 10V interface had a slightly better resolution, and also drained less current at high pressures, resulting in longer battery life.
In a system where precision and linearity issues can be compensated for by the software, a pressure sensor's repeatability is more important than accuracy. I am sold now - I definitely will refactor the app to use multiple data points in the reading profiles.
Regarding the tested sensors:
Hydac - great performance, big price tag. Use them if you need superior accuracy and performance and don't mind paying the premium. As a diagnostics tool (coupled to the wireless transmitter) these sensors are overkill.
Wika - good performance, reasonable price. They are called general-purpose for a reason - use them for everything, they are good. Perfectly usable in the wireless pressure gauge system.
Gems - good performance, reasonable price. Just like Wika, but slightly more compact. Taking into account the money that I paid for mine - they are the clear winners here. If you can get these cheaper than the Wika - you should use them. Again - perfectly usable in the wireless pressure gauge system.
Fox - cheap price, cheap quality. Cant' serve as a drop-in replacement without correction. The version with the 10V interface was unusable below 30 bar. I guess they are ok for certain applications, but only if bought at a very economical price, otherwise any sane person should use a Wika A-10 and be done with looking for sensors. The 20 mA version could, actually, be used in the wireless gauge, the 10V version that I tested was an absolute fiasco.
It looks to me that both the Bluetooth module's built-in 12 bit ADC and the very simplistic approach to receiving an industrial sensor reading work just fine! So, I guess - there is no need to re-engineer the hardware part. But I do see now that re-factoring the App to allow for multi-point reading profiles is very beneficial.
The last sensor that I tested was the Gems 600 bar one, and right as we were logging the very last pressure reading, the driving nut of the Pressure Maker broke in half. The timing couldn't be more perfect, in fact - it looked like magic. As soon as my wife hit the enter key the nut snapped with a bang! (this is why there's the word boom! in the mA column - I didn't get time to log the current). Clearly, it was time to go home. So we did. Spending hours in a hydraulic shop logging pressures is not the best way to spend a Saturday evening for a married couple.
The Pressure Maker has already been welded. It will be fine...