Resonate frequency of what?

100 Hz resonates from 50 Hz (50 * 2 = 100, see? ;) ).

Yes, but what would be in his system that would produce (only) a 100Hz harmonic when driven at 50Hz?

From the pump rotation. I had same results with a spectral analysis of different pump sounds to catch the fundamental frequency, some are 50Hz and others are 100Hz, it depends of the rotation speed and number of coils. Eheim 1046 was exactly 50Hz and Eheim 1048 was 100Hz. Think how about the rotor is driven and you'll understand, it's not an continuous and smooth rotation but a lot of push-pull movement when North/South magnet poles are near coils, this produces a sudden torque and a tiny vibration. This is undetectable for an human (except the vibration with the pump in the hand), but with a lot of pressure measurement per second this phenomenon is easily catched :)

Calculate the mean value of the whole signal will give you the mean pressure, your choices are limited. This inconvenience is flatten if you use some water manometers, inertia of the water column will make its job (50 or 100 Hz is far too quick to get a variation of the column height) and the pressure value will be very stable.

This 100Hz variation is flatten too when pressure is measured far from the pump (WB in his case) because pressure transmission have a finite speed in water and turbulence near the out of a waterblock (if measure is taken here) give also a small variation of static pressure too. You mix this 2 phenomenons at a same time and you'll get a noisy and distorted signal at the end.

Ah makes sense. Thats pretty impressive that his setup can catch those oscillations in torque.

I forgot to talk about the number of channels or blades (depends on rotor topology), the flowrate isn't really "smooth" because of the finite number of blades, it's like a discontinuous movement. The more blades we've got (infinity is one limit in centrifugal pump theory), the less variations in flowrate at a final result. I think the frequency of this phenomenom is higher than the other at 50 or 100 Hz. If rotor has 6 blades at 3000 rev/min, this give 300 blade passages par second in front of the out of the pump, so 300 Hz frequency for (little) flowrate variation.

Thank you Roscal, for the detailed explanation.

Yes, it's impressive to be able to catch that variation. I'd also be looking for a 200 Hz pulsation, if the equipment permits.

Another question... Is it necessary to have the dP meter in the same loop as my water block test bench? I would MUCH rather have it on a separate test loop dedicated to pressure drop tests. That way I can do radiators and other things that create pressure drops and leave my water block test loop alone.

No, it's not necessary.

You'll want to characterize the performance of the block, but also the flow characteristics, but they can both be done seperately.

FYI, the only difference is that the water temp is going to be slightly different, but that wont have a measurable impact.

Thank´s. It was only to compare my calibrations. I´m testing four wb´s. Here I posted some graphs.

For example the following graph I did it for accident, but I liked the result after seeing, very symmetrical

Yellow is the flow in l/h, dark blue is a pump-out pressure en mH2O. The line brown is the restriction in watts between the pump-out, flow sensor and exit of the block, and the rose line is a pressure drop in the wb.

http://personales.ya.com/cuasar/vari...alpresion1.jpg


I don´t have any filter. In this case, I prefer that the sign goes direct to the ADC.

I am using this piezometric tube and a small diameter pipe of silicon to connect it with the sensor.

http://personales.ya.com/cuasar/vari...ion/tpiezo.jpg

http://personales.ya.com/cuasar/vari...on/tpiezo3.jpg

Since Roscal has explained, they are the possibilities that exist but very possibly a mixture exists of both though the type of engine prevails with his number of couples of poles.

In this type of engines, the frequency of nourishment and the number of couples of poles they determine the speed of the impeller of the pump. All those more couples of poles it has, minor will be the number of revolutions.
For example for 1,2 and 4 couples and 50Hz, the revolutions would be of 3000,1500 and 750rpm. rpm´s = 60 x frec / pairs poles
It is possible that this engine has only a one couple and that every 10msec, the rotor turns half a return coinciding with the 3000 rpm's, but I am not sure of it, is only a conjecture and I did´t do any more checkings after doing the graph

In this graph you can see the functioning of a dc pump, with a wired rotor. The waves are not so clear. This wave might be owed only to the spades of the rotor.

http://personales.ya.com/cuasar/vari...on/totpres.jpg

http://personales.ya.com/cuasar/vari...n/totpres2.jpg

Another curious graph would be this. It is taken across a pipe of silicona of 10mm of inside diameter with a length of 2.55m. I measured when I power on and power off the pump every 1msec.
A delay exists is 50 msec aprox, brought between the sensors. In these conditions, the front of pressure would move to an approximate speed of 51m/seg.

http://personales.ya.com/cuasar/vari...on/velpres.jpg

In the left corner, can be see the value in msec

http://personales.ya.com/cuasar/vari...n/velpres2.jpg

Interesting measures, delay show the (small) shockwave, quite like a "water hammer". You should test if sensors have exactly the same response when there are very near each other (<2cm with your copper pipe) with no delay.

Yes. It is a big and noisy bomb. 2.7mH2O and 2000l/h. The impeller is not firmly fixed the rotor, has a light roominess in the draft, it might influence something also helping the rotor to acquire more initial acceleration

Given the incomprehensibility of the water expects for me a minor time or even nothing from delay, but the measurements say the opposite thing to me. Also think that is possible that the pipe of silicon was expanded the sufficient thing to create this delay.
I will try to test today with the piezométrics pipes as unitedly as possible and also I will use a pipe of copper that I have hereabouts.

Nice work.

Thank´s Incoherent

And I´m sorry Jaydee116, I feel to have occupied you message


I have worried a little plus this time and have fit the values to 0 by the very slight pressure differentials due to the assembly, little mmHO to appreciate better the differences. Practically any altitude, few mm of difference between both sensorial is appreciable varying the hydrostatic pressure.

I have tested uniting both tubes more than I have been able being the piezométrics tubes to 3.5cm and the results have been these.


http://personales.ya.com/cuasar/vari...n/velpres6.jpg
http://personales.ya.com/cuasar/vari...n/velpres9.jpg


It is possible to be appreciated that the delay no longer is measurable by the system and that the waveform is very similar. He is peculiar that at the beginning and at the end of the wave, forms a slight emptiness when the restriction and therefore the static pressure is low.
It can be due to the phase angle between the rotor and stator produced for the reason that comments that the impeller and the rotor are not firmly united and is possible that at the moment in which it must make force suddenly, it can will be out of phase slightly. In the synchronous motors it is not difficult that it happens. Or simply it is due to inertias of I eliminate although they would seem to me too fast.

Memory that the circuit is united to a deposit opened to height of the ground of the room
Also, whichever minor is the restriction and therefore smaller the generated pressure, greater is the initial tip of pressure in relation to the following ones. According to the collected data, the maximum obtained pressure 2314mmHO and minimum of -489mmHO has been.



In the following graph, I have measured the pressure differentials between the ends of a copper tube of 1.67m of length and 10mm of internal diameter.
It is possible to be verified like indeed that delay is not due to the possibility that I commented of which the silicone tube expanded minimally producing that delay.

The delay appreciated is 33msec (5224 redline-5257msec blueline) with which the speed of the pressure front becomes to locate in 50,6m/sec. The speed is practically the same one. It is to hope that with another pump and other diameters that speed changed.
The difference of this graph with respect to the previous one would be in which no longer the initial emptiness by the greater restriction of the copper tube takes place, but in return the final emptiness is different. The emptiness indicated by the red line is always positive and the one of the blue line is alternating. It seems by ricochet like a species due to the mass of water or to the same pump.
The greater pressure indicated by negative the blue line as much as positive would indicate the direction of the final of the wave, would indicate the direction of the water obstacle.

http://personales.ya.com/cuasar/vari.../velpres11.jpg
http://personales.ya.com/cuasar/vari.../velpres10.jpg

Many of these things do not influence directly in the refrigeration, this is circumstantial and it worries to me plus seeing as temperatures, pressures and flow are interlaced, but they are peculiar details of knowing and appreciating.
The studies that could be done with the suitable sensor, humidity, vibrations, magnetic field can be innumerable.

Luckily I can put images, if it had to express to me only with words I believe that it would not be understood to me. :rolleyes:

Facinating stuff.

I am really sorry that I don't have time to look at this closely.

I would observe that if you are seeing pressure waves, they should be travelling at the speed of sound in water (1500m/s?) . Or ??.... ... I don't know if the speed would be affected by the wall type???
A curious scenario, being a closed loop.

Cool.

Keep up the good work.

Nice work dnkroz, thx for the test with very close sensors, no delay shown here and same response, very good.

Pressure wave (in case of water hammer) is directly linked by the wall type. The more rigid the tube is, the quicker the "shockwave" travel in this tube up to a limit = speed of sound in water (~1420m/s in normal conditions) like you said Incoherent. You could have an idea of this speed using the ALLIEVI equation with a k coefficient which depend of the tube material. Here it's a tiny pressure wave, not a real and dangerous water hammer :D

For oscillation in the sensors responses, it's quite normal because sensors need a delay to find their equilibrium (given in their datasheet normally, for example 10 ms) when you stop flowrate (inertia of water volume plays a role too). Their measures can't be instantaneous and they could resonate a little (what is their mechanical resonation frequency?) before giving you the "real" pressure value after xx ms. Your sampling rate seems to be very high, how much samples/sec? Your curves are very noisy, you should try with a lower rate, like 1 measure per sec, it's far enough for our application, transients curves are useless for us. But it's always interesting to see others phenomenons to know well your measurement material and its capabilities ;)

It depends of sensors obviously, yours are "cheap" ones with a special technology (piezo) and got their own pros/cons. I worked with high pressure sensors for military and civilian applications (A380, rocket...) to calibrate them within shock tubes in order to get the true response, and for that, sampling is done at ultra high speed ($$$ material) to get millions of data in only some milliseconds (violent shockwave is generated to hit sensor), but these sensors aren't cheap at all obviously :D

No worries the thread is open to all stuff on pressure measurement stuff. :) Good work!

Thank´s jaydee116.



It did not think to me that it go to interest so much the graphs. When I did the first tests it does already a time, I did not give much importance him, but obvious he is interesting.


If they are interested I can continue testing. I have other three sensors at the moment installed in the PC, if I can I will try simultaneously to test with the 4 sensors to different distances and possibilities.

Also to the next one instead of leaving to the pump working several seconds, I will try that only the water hammer takes place to see what it happens. And I will try to make more things to see than it is what it happens, to give a blow in the silicone tube etc.


The collection of data I have it selected for every millisecond, although normally for temperatures and pressures I use readings every second. But also I have tested in 5msec and also enough noise exists.
I have not proven to raise more, but the resolution will lower enough. A second would be very many time, would not see anything. Ground not to use the system for this type of measurements and I have not worked the form to avoid it.
I can try to use another tool of software to test.

It is necessary to consider that the system that I use.
The data logger itself, (the rest of the box has electronics of conditioning signal and power supply), is really simple. It was damaged to me and when repairing it I could see that all the components are SMD, has very few and main and most expensive one, the ADC of 10 Bits, only costs 4€. It does what can.

I do not believe that the offered answer is influenced by the mechanical resonance of the sensor, has a maximum response time of 1msec, and also that the time stabilization I don´t believe isn´t very different from that. In addition, 2.55m of length for 50msec and 1.67m of length for 33msec of delay agrees in proportion.

I am not sure of which can be the reason or the set of them that cause it, and until point the effect hammer it can influence the measurements in a circuit of this type or a pump, inertias.

The advantage like you have commented of the piezoresistives sensors is that they have a very fast answer but also they are more sensible to the noise. I had of series 26PC without amplifying and ratiometric output, based on an enough bridge of Wheatstone and I had problem of noise/stability with him. It is of more delicate use.



Here all the channels at the moment configurated, 11 and the rest virtual for several calculations can be seen. Obvious, normally nonworking with all simultaneously.

Some do not work well, because two pressure sensors are desconected for for the current test. I have installed a Aqua Computer Culpex Pro to test.

If are interested I can do commentaries on them, but in another message.

http://personales.ya.com/cuasar/vari...tacion/com.jpg

Also a long time ago I tested with sensors of vibration placing them in the pump, but I did not finish testing serious. Concretely I used this sensor.

http://personales.ya.com/cuasar/vari.../vibracion.jpg

Also I have another one of mechanical pressure to do a balance to me for little weight with enough resolution. In that I have thought to use for the test bench of water blocks. It is an interesting field.

I believe you, know persons that are employed at A380 :)

Slightly offtopic:

Would one of these be good for differential pressure measuring?

Or would a digital one be superior?
If so could anyone give me a link where to buy one relatively cheap?

No, not for watercooling purposes. In watercooling you'll be nowhere near the 200 psi that meter measures up to, and with that range you are loosing resolution in the 0-100kpa range you should be looking at.

Nevermind the range, that's not the exact model I mean.

Highest range you'll ever need for watercooling is about 15PSID, right?

Here they are the graphs that I commented with 4 sensors. I have had just a short time frees. I hope not to tire.

This circuit don´t have pump and deposit.
It is formed by three sections of tube of 1.33m without air, only elastic water :) between which and in its end I put the sensors. The circuit would be the following one:

(close tube)sensor1(blue)-tube1-sensor2(red)-tube2-sensor9(rose)-tube3-sensor11(green)(close tube)

When I press with the finger in the center of the second section(tube2), between sensor 2 and 9, is generated the following graph. To the being a symmetrical circuit, the answer of the sensors goes simultaneously.
I tested pressing in another asymmetric place, but I lost the test.

http://personales.ya.com/cuasar/vari...4sensores1.jpg

http://personales.ya.com/cuasar/vari...4sensores2.jpg

In this circuit is the following one:

close tube-pump-sensor1-tube of 2m-sensor9-close tube. Without deposit

This time I only make work the pump only a moment of time. Without deposit, the oscillation is less is cushioned and lasts a little more.

http://personales.ya.com/cuasar/vari...oscerrados.jpg

This is a graph again of the obtained resolution, in this case the sensor and the tube is on the ground of the room.
The variation of 7mm H2O (0.0099 PSI) corresponds when I place a pencil (7mm of diameter) under the set sensor/tube and the hidrostatic pressure is modificated. Other times indicated 6 to me and other 8, but the majority has been those 7mm. Without not run pump, it´s easy to see. The 97mm H2O is exerted by the height to which the rest of the circuit is located.

http://personales.ya.com/cuasar/vari...cion/lapiz.jpg

Impressive data! You really know your way around pressure testing!

Hi bigben2k, I am glad that you like

Having means, the imagination makes the rest.


In these other graphs the increase of temperature can be seen the increase of pressure in the system in spite of a stabilized flow.

This experiment I made it from measures of pressure different from other previous ones and in other conditions. The cause did not know. This could be the cause.
It could get to false measurements with the relation pressure/flow.
Although if it is taken for the calculation of the flow or the restriction the pressure differential, would not have of supposing no problem if the differential pressure stays.

The beginning begins in 23.03ºC (CPU OUT 6) and 0.143 mHO (the restriction is low, exit of the pump, "PRESION 1") and finishes in 27.96ºC with a pressure in the same place of 0.416mHO for a period of 3000segundos (50min) and 1000 takings of measures.

That is to say, an increase of pressure of 0.273mHO, due to an expansion, had as well, to an increase of the temperature of water/air of 4.93ºC.

I have not proven mathematically if the practical results have a theoretical confirmation, but those have been the tests. I hope are certain.
Would be interesting to prove with water to temperatures between 0-4 ºC to measure its contraction between this one rank of temperatures.

In these tests, I did not install the radiator and the deposit is hermetic. Without air in the system or very little, the increase of pressure is appreciable.


http://personales.ya.com/cuasar/vari...pres-tempb.jpg


But in this other the deposit was full only until half. With air in the interior, no longer the increase of pressure is appraised.

http://personales.ya.com/cuasar/vari...res-temp3b.jpg


In this, when the pressure is increased with the temperature, I open to the deposit lowering the pressure "presion 1"

http://personales.ya.com/cuasar/vari...res-temp4b.jpg