MFMP: Automated experiment with Ni-LiAlH

  • Can ,

    That will probably be a natural part of an actual experiment since the Ni is a magnetic material. My plan of the moment is to step and regulate the temperature as before, but at each step have a "hold, wav excitation, hold" pattern. This will allow the effects of the .wav excitation to be evaluated at cold temperatures where the effect should be no more than heating and then see how the effect changes at more elevated temperatures.


    The first calibration will be done without a magnetic load in the reactor cell (alumina instead). This will evaluate the effect of the magnetic stimulation on the thermocouples and instrumentation. I also need to make sure that the coil is not going to get too hot and smoke. If the coil is getting too hot, changes may be required to the convection around the coil.


    Can you clarify what information you would be looking to gather during the test you propose? Magnetic losses will be material and temperature specific.

  • BobHiggins

    The information I'm looking to gather is how a "worst case scenario load" composed of a solid core of high magnetic permeability would affect the system during the excitation part, in particular below the Curie temperature. I realize that it would not be high priority relatively to your experiments.

  • Can

    So, when you say, "affect the system", are you looking for real temperature changes from magnetic heating, false temperature changes due to temperature sensors being affected by the magnetic field, obscure magneto-caloric heat pumping effects (where would the heat go?), true or phantom radiation effects, or what? I am just trying to gauge whether the system is capable of detecting anything more than a temperature change for such an experiment.

  • BobHiggins

    With affect the system I mean any change or apparent anomaly that could be interpreted as due to LENR, whether transient or not, during a real experiment with hydrogen. This could include both actual changes due to real effects (e.g. induction heating of the iron piece under testing) or electromagnetic interference with the temperature sensors.


    Admittedly, maybe it's a too broad plan for your purposes since you're only testing Ni powder primarily above the Curie temperature.

  • After a number of false starts, the pre-calibration with the .wav excitation has started. There were a number of program modifications to get everything working as desired. I haven't setup the google drive folder for this yet, but I will do so soon and add photos of the setup. The test will run for ~48 hours like the last calibration cycle. Each step will have a 30 second stimulation burst at about 5 amps sinusoid peak to the coil. This is about 25 watts of RMS power into the coil and will produce sinusoidal magnetic fields in the center of about 140 gauss peak. In the initial testing, temperature measurement anomalies were not seen, but they are not ruled out and are to be examined in the calibration data.


    It seems like the instrumentation complexity has gone up substantially, but that is just subjective.

  • PreCalibration was terminated at 11:40 pm Mountain time. There appears to be an occasional oscillation of the audio PA. There may be a ground loop - playing the .wav file causes an oscillation to appear sometimes immediately after the .wav file stops playing. I probably need to provide a load resistor to the output of the PC's sound card output. I am also seeing some anomalous behavior of the PID control that causes the temperature control to have about 500 seconds of confused control after the end of the .wav input. The control temperature was not affected by the magnetic field, but the PID control loop became momentarily confused. I will probably have to adjust the PID constants to fix this tomorrow - I suspect the derivative coefficient. I don't want this instability to be confused with XH in the aftermath of the sine burst application.

  • Today I am going to add a resistor to the input of the audio PA to hopefully snub the tendency of the system to oscillate after playing the .wav file. The output impedance of the sound card in the PC (headset port) is low and there is no reason to leave the amplifier input high impedance.


    Also, in the light of day, thinking about the failure of my sequence of "HOLD, WAV, HOLD" commands, I have a different thought. The HOLD command uses a PID algorithm to lock in on the correct heater power to keep the temperature at the setpoint. At present, every time the HOLD command is entered, it begins with the I (integral, a summation of errors) set to zero. This is the only memory in the algorithm and that is getting reset for the second HOLD even though it is at the same setpoint. That was apparently responsible for the second HOLD command's driving the heater the wrong way after the WAV disturbance. It took 1000s to settle out after that. Today I am going to modify the HOLD command such that the integral state is not reset. I want the sequence of "HOLD, WAV, HOLD" commands to only create a quick tiny bump in the settled state, otherwise it could appear as a phantom XH following the WAV command.


    If you want to see the anomalies, the data is here on Google drive: https://drive.google.com/drive…B6V0FvQTBFdkU?usp=sharing

  • Can

    Thanks. I don't even have that graph yet. In this, you see the first 2 excitation currents behaving normally. The third current shows the oscillation - I had to manually shut OFF the amplifier to get the oscillation to stop. I tried manually switching to the second channel and the oscillation started back up. The fourth excitation is missing because the audio PA had the input on channel two and the output taken from channel 1 (IQ deficit due to the late hour). The last excitation before I terminated the experiment had reduced current for some reason. Did the resonant frequency change due to the temperature? - something to be checked. At lower temperatures the heater coil is magnetic and could have a slight effect on the resonant frequency. At 300°C tube temperature, the heater coil is hotter, the curie point could be affecting the relative permeability of the Kanthal heater wire. Because the heater coil is bi-filar wound, the coil-to-coil coupling is nearly nil.


    If you want to see the control anomaly I described, zoom into that last period at 300°C and add in the heater power to the graph. You will see that the after the excitation, the temperature is above the 300°C, yet the control increased the heater power! The thermocouple was reporting the temperature high - so this was not a field induced measurement error. The computer is basing its control on that exact temperature that is being plotted, not some other not-shown measurement.

  • Post by Eric Walker ().

    This post was deleted by the author themselves ().
  • Alan Smith

    This test was not fueled - it was a dummy run for calibration. It shows internal heating from the time varying magnetic field which is a good sign that the stimulus is a significant excitation. So, when it is is operated with a fuel, is it reasonable to anticipate the possibility of excitation at the fuel.


    The concern is that the control system behaved badly due to the re-start of the HOLD command, causing the thermal heating from the excitation to create a bigger and longer lasting temperature bump than should have occurred. This could look superficially like XH in a fueled system (even though the power control would have shown it not to have been XH). I want to fix this before a fueled run.

  • Can

    Can you expand the Y of the temperature around the excitation so you can see 1-2°C shifts in the temperature after the excitation and zoom in from just left of the excitation to about 800s to the right of the excitation? It may take a graph stack to do this. What you will see is the temperature settled very close to 300°C before the excitation, a slight increase in temperature during the excitation, and then the increase in heater power being applied after the excitation despite the fact that the temperature is already high. This heater increase causes a bigger bump in temperature than is actually caused by the excitation.


    In a fueled run, if the excitation stimulates XH, you should actually see the heater power going down after the HOLD is re-started, to compensate for the XH being generated internally - the control loop holding the temperature constant by adjusting the heater input power.

  • Can

    If you zoom the time scale around the excitation and the subsequent increase in heater power, you will see that the increase in heater power occurs after the excitation has completed with time between. It is a bad choice for the heater to be increased and this occurred at the beginning of the new HOLD command.

  • Well, today I think I have ironed out the oscillation problem and the software bugs. From my changes and tests today, I expect to have far better lock-in on the setpoint temperature and proper PID operation on a subsequent hold after the .wav excitation. I am going to try to setup a calibration run tomorrow, but it may be an abbreviated one to lower temperature until I am sure how the epoxy filled coil can handle the temperature.

  • Today I have a short calibration test running - up to a core temperature of 600°C. I want to take some baby steps and look for problems in the system and how the field coil is surviving the temperature in the environment. I may have to circulate air to prevent problems. I could do this with an aquarium air pump and blow air where the coil gets hot. For now, its temperature is being measured. The data files are going here:


    https://drive.google.com/drive…VRSjZkLXgyOTA?usp=sharing


    Some files are already there. This run will finish by about midnight my time.

  • This last calibration run is complete and its data is on the Google drive in the folder listed above. It was a stepped exercise from 100°C to 600°C in 100°C steps. Each step included PID hold at the temperature before and after the .wav stimulation. The problem with the anomalous behavior of the temperature control after the step has been fixed, but the experiment was not without incidents. At 400-600°C, the power amp oscillation returned and would have continued indefinitely (I think) unless I turned OFF the power amplifier and turned it back ON. Even when turning it back ON, I had to wait about 60s OFF or the oscillation would resume when the amp was re-powered. The amp is not getting hot - the only thing getting hot is the coil (not even the series capacitors). I also noticed that the current in the coil was declining with temperature. As a next step, I will probably make a longer lead for the amplifier output (it is only about 16" now) and relocate the amplifier farther from the coil. The oscillation is serious because it continues to add heat to the system and could mask observation of XH if oscillation occurred following the programmed excitation.


    Another anomaly was that on one of the steps, the power supply did not respond to a change of voltage command. When I saw that (about 800s late), I adjusted its knob to put it at the correct voltage and the system resumed operation as normal. That is the first time this malfunction has been observed. I will re-examine the code because I thought I had checks in place to cause re-sending of the command if the voltage was found operating at the wrong value.


    At 600°C the hot upper corner of the coil reached about 90°C and there was no evidence of melting or smoking from the epoxy (yet).

  • Here is data from the latest run. The vertical dashed lines denote where the calibration endpoints have been sampled (average of preceding 100 samples). Overall it seems that the calibration is about the same as the last time.




    BobHiggins

    Please also add information on how to correctly decode the newly added columns.

  • Can

    The columns should be identified in the column header in the .csv file. Column 7 or G (column label 4, CoilTop) is J-type measurement of a temperature in the top corner of the field coil. Column 19 or S (column label 16, Neu CPS) is the counts per sample from the neutron counter - similar to the GM, and spectrometer ROI counts per sample.


    It is interesting that the calibration has changed so little compared to the reactor without the field coil. I will post some pictures today of the setup.