AquaComputer MPS 400 Flow Sensor Review

[stren]

The full detailed 5 page article can be found on my website. Here is a summary:



The MPS 400 is the first flow rate sensor for water cooled PC's that uses a differential pressure sensor that has been integrated into a single silicon part. The advantage of such a design is that there are no moving parts that can break down and possibly stop the flow of coolant. It is also exceptionally small and can read an external temperature probe as well as having it's own internal temperature sensor and Alarm function.

Flow

The flow rate sensor should produce a temperature compensated voltage that is linearly proportional to differential pressure. This is converted by a DAC and communicated to the PC by a microcontroller via USB. The microcontroller also takes the "raw" DAC codes and plugs them into a calibration table which spits out a flow rate. Calibration tables are provided for 8 and 10mm tube, but not the common 7/16" and 1/2" ID tube (10mm and 3/8" tube are very close in size). You can even supply your own calibration table. The sensor is configured, measured and logged by AquaSuite, which is free and downloadable from AquaComputer's website. The sensor itself has high accuracy if enough averaging is used. Aquasuite however does not seem to average and instead has a secret sauce "filtering" algorithm which seems to just take the lowest value recently seen. By doing some manual averaging I compiled my own calibration and compared it to the default AquaSuite settings and 10mm calibration table:



By default AquaSuite rounds to the nearest 0.1GPM (although this can be changed in the settings) hence you can clearly see the steps that are due to that. However there is still a large source of error that is due to the calibration table matching poorly, and the error also falls off a cliff after 1.5GPM because the calibration table ends even though the DAC is still outputting useful data. If we plot the error of the two calibrations methods vs my +/-0.07GPM accurate King Rotameter:



Then we can see that with "manual" averaging and a good calibration file then the MPS flow sensor is *very* accurate indeed up until it's *real* range ends at ~2.1GPM. However the supplied calibration file is not only a very poor match for this sensor but also ends unnecessarily soon.

Temps

The MPS 400 also includes an internal temperature sensor and an input for an external temperature sensor. The internal temperature sensor is not very close to the water and so is not only slow to react but has a large amount of error. To test the external temperature sensor I bought an AquaComputer in line temperature probe and I measured very similar results to my Dallas "one-wire" temperature probes:



This really shows just how bad the internal temperature sensor is. Again plotting the difference between the two sensors and my reference dallas probes shows that the external probe is very good:



Again most of the external error is due to poor calibration. I noticed a bump on both sensors at 30C and this turned out to be a rounding error in AquaSuite. AquaComputer are fixing this is in the next release of AquaSuite.

Alarm

The MPS 400 also has a configurable 3 pin "fan" header that can be connected to the motherboard or used to light or extinguish an external LED (not supplied). When connected to the motherboard and configured properly the sensor reads a fixed RPM. If connected to the CPU fan header then you will know quickly on boot if your flow is zero.

Summary

A very good flow rate sensor with a lot of potential let down by poor calibration. If you're willing to calibrate it yourself then it can be very accurate, if you're not then take the value with a grain of salt as it's main value will be measuring a relative change such as "has my loop become clogged because the reading has dropped by 1/2. Ignore the internal temperature sensor, it's too far away from your coolant to be of much use. However the external temperature sensor is a useful input to have. Similarly the alarm feature is a nice bonus too. The MPS400 is however pricey, currently retailing at $68 in the US this seems somewhat high given the poor accuracy and poor support for larger tubing sizes, if AquaComputer can fix these things then I'd say it's good value for money.

[lowfat]

Does the actual flow meter cause any pressure drop?

[lowfat]

Edit: Double post.

[stren]

Does the actual flow meter cause any pressure drop?

I haven't measured it, but it would be about the same as a small barb or a small section of tube as there's nothing really in the way (unless you fit the flow restrictor plate which you're supposed to for 8mm tubing and cal file). I'll run it when I do the next lot of blocks

[Shoggy]

Nice review. I already saw it on your website

One thing is a bit misleading:

AquaSuite, a program available for download from the AquaComputer website, then takes the "raw" DAC codes and plugs them into a calibration table which spits out a flow rate.

The software does not process anything. It is all done in the sensor. This is also necessary because if you connect it to an aquaero it will require the final flow value.

The flow sensor can also be connected to an Aquaero controller unit via the AquaBus connector (cable not supplied)

The supplied 3-pin cable can be alo used for this but only when you connect the USB cable too because otherwise the sensor will get no power. To use it without USB the optional 4-pin aquabus cable is necessary.

[stren]

Nice review. I already saw it on your website

One thing is a bit misleading:

The software does not process anything. It is all done in the sensor. This is also necessary because if you connect it to an aquaero it will require the final flow value.


The supplied 3-pin cable can be alo used for this but only when you connect the USB cable too because otherwise the sensor will get no power. To use it without USB the optional 4-pin aquabus cable is necessary.

Thanks shoggy I'll get that fixed

[Conumdrum]

Your saying unless you do the 5 gallon bucket flow rate method to set a baseline you can't calculate flow?

Not sure what you mean.

[Avathar77]

How does this compare/perform against the "Aqua Computer High-Flow Flow Sensor" ?

[stren]

Your saying unless you do the 5 gallon bucket flow rate method to set a baseline you can't calculate flow?

Not sure what you mean.

You can use the supplied calibration files and it will give you a reading of the flow. However that reading may not be that accurate (in my case ~0.2-0.3GPM off). If you want more accuracy you need to calibrate, either with multiple runs using buckets to calibrate different points of the curve, or by calibrating to a known good flow meter. King meters can often be had on ebay for $40-60. I thought about starting up a calibration service for these, but I think the amount of time it would take me is no actually worth the amount I could charge.

How does this compare/perform against the "Aqua Computer High-Flow Flow Sensor" ?

Martin reviewed that one here, and from the look of his data it was much more accurate, although it seems the range is much less useful. It also has moving parts and it makes some noise so it's therefore a trade off of what you want. If you already had one and wanted to move to this then you could also try and use it to calibrate this one.

[Jakusonfire]

Can you not just share a calibration curve for 7/16 and 1/2 inch tubing? or does it change somewhat loop by loop?

I'm a bit sick of Aquacomputers small ID focus for their products.

[Shoggy]

You would have to use exactly the same connectors with the sensor to get comparable values. The same size alone is not enough.

[Jakusonfire]

Thanks Shoggy, I guess I'll stick with my mechanical meter.