Interesting hybrid device (MP108)...I just looked up its data sheet. What is the value of the current limit resistor you are using in your application?
The 50Ω load suggests 200W of power at the 100v RMS you mention but the safe area of operation data for the MP 108 suggests this is not achievable although short pulses are probably possible.
I have done experiments with "active servo feedback" using audio amplifiers...essentially placing a current sense resistor on the cold side of the load with a differential op amp across it to get a ground referenced voltage that is proportional to load current. In audio this is useful because the speaker can act as a generator and distort fast transients, these distortions appear in the load current waveform and thus the voltage waveform derived off the cold side current sense resistor. The distorted "current" waveform can then be fed back into the inverting input of the power amp to provide a correcting signal. All of this was working at under 200Hz, so a far cry from your 500 odd KHz. And my load was inductive whereas yours is capacitive and your current sense resistor is on the "hot" side of the load making it more difficult to extract a ground referenced representative voltage of it so you can see how "imperfect" your "pulse-on, pulse-off" transients are.
You need to do some experiments to see how the current sense resistor effects the load waveform. I would try measuring the load voltage waveform...directly across the piezo with a storage oscilloscope and capture the "on" and "off" transients....if they are "clean" then your system has high damping factor and good impedance matching, however if they are "dirty" with overshoot and phase shifts or waveform distortion , then something is missmatched or your damping factor is to low. Try changing the value of the current limit resistor...say to twice the value to limit the current ti half as much...and see now the switching transients are effected as well as the main body of the pulse.
One thing I will say from experience as I have done a lot of R.F. stuff as well as Audio...your circuit will need to be very "TIGHT". Although (by RF standards) your frequency is not that high...AT THIS POWER LEVEL, 100-200watts...you will need to lay it all out as if it is a circuit operating at ten times this frequency. It will need to be on a groundplane, all DC rails will need to be decoupled with 100nF film capacitors with the lowest ESR and shortest leads possible RIGHT AT THE RAIL PINS OF THE MP 108, bigger non-polarized bypass capacitors, (1uF) will need to be placed right behind the 100nFs and any electrolytics right behind those...keeping all "short and sharp". Same applies at the output and around the current sense resistor..follow the MP 108's application notes very carefully...this is not small signal stuff where you can be a bit "sloppy" . The voltage feedback will need to be taken right at the transducer, if it is a long way away from the MP 102 at the far end of a transmission line... a separate transmission line from directly across the transducer will need to come back to the feedback resistor. If you wish to avoid all this transmission like "guf" with all its impedance matching issues and reflected wave problems, you will have to place the MP 102 fight at the piezo transducer with the shortest connections possible.
Pay careful attention to your "Earthing" too...keep all your "input" and small signal Earthing away from your power supply and load Power Earthing. All the cold ends of those many bypass capacitors in parallel need to be as close to one point at is practical or a very solid ground plane needs to be used. The cold side of the piezo load also needs to return to this same point with as "short and thick" a connection as is possible. If you do all this, a lot of the "noise" and nasty, dirty transiets will disappear and your sonar pulses will become sharp and well defined...and your echos will contain more data about the bottom of the pond than the slopiness of the driving circuit.
I will ask this "important" question again... are you a PhD student? Is this part of a PhD thesis?
The reason I ask this question is because, in my past experience with PhD. students...PhD students are not experienced practical electronics technicians and they do not generally know or understand all the nuances of electronics, particularly power electronics such as this. I have known of people with PhDs who cant connect a car battery to a charger correctly which already had one battery charging on it (he connected his battery in anti parallel and it exploded in his face)...their PhD research has rendered them SO SPECIALIZED in one tiny area of just one field that they loose the most basic common sense.
Cheers and Good Luck
The 50Ω load suggests 200W of power at the 100v RMS you mention but the safe area of operation data for the MP 108 suggests this is not achievable although short pulses are probably possible.
I have done experiments with "active servo feedback" using audio amplifiers...essentially placing a current sense resistor on the cold side of the load with a differential op amp across it to get a ground referenced voltage that is proportional to load current. In audio this is useful because the speaker can act as a generator and distort fast transients, these distortions appear in the load current waveform and thus the voltage waveform derived off the cold side current sense resistor. The distorted "current" waveform can then be fed back into the inverting input of the power amp to provide a correcting signal. All of this was working at under 200Hz, so a far cry from your 500 odd KHz. And my load was inductive whereas yours is capacitive and your current sense resistor is on the "hot" side of the load making it more difficult to extract a ground referenced representative voltage of it so you can see how "imperfect" your "pulse-on, pulse-off" transients are.
You need to do some experiments to see how the current sense resistor effects the load waveform. I would try measuring the load voltage waveform...directly across the piezo with a storage oscilloscope and capture the "on" and "off" transients....if they are "clean" then your system has high damping factor and good impedance matching, however if they are "dirty" with overshoot and phase shifts or waveform distortion , then something is missmatched or your damping factor is to low. Try changing the value of the current limit resistor...say to twice the value to limit the current ti half as much...and see now the switching transients are effected as well as the main body of the pulse.
One thing I will say from experience as I have done a lot of R.F. stuff as well as Audio...your circuit will need to be very "TIGHT". Although (by RF standards) your frequency is not that high...AT THIS POWER LEVEL, 100-200watts...you will need to lay it all out as if it is a circuit operating at ten times this frequency. It will need to be on a groundplane, all DC rails will need to be decoupled with 100nF film capacitors with the lowest ESR and shortest leads possible RIGHT AT THE RAIL PINS OF THE MP 108, bigger non-polarized bypass capacitors, (1uF) will need to be placed right behind the 100nFs and any electrolytics right behind those...keeping all "short and sharp". Same applies at the output and around the current sense resistor..follow the MP 108's application notes very carefully...this is not small signal stuff where you can be a bit "sloppy" . The voltage feedback will need to be taken right at the transducer, if it is a long way away from the MP 102 at the far end of a transmission line... a separate transmission line from directly across the transducer will need to come back to the feedback resistor. If you wish to avoid all this transmission like "guf" with all its impedance matching issues and reflected wave problems, you will have to place the MP 102 fight at the piezo transducer with the shortest connections possible.
Pay careful attention to your "Earthing" too...keep all your "input" and small signal Earthing away from your power supply and load Power Earthing. All the cold ends of those many bypass capacitors in parallel need to be as close to one point at is practical or a very solid ground plane needs to be used. The cold side of the piezo load also needs to return to this same point with as "short and thick" a connection as is possible. If you do all this, a lot of the "noise" and nasty, dirty transiets will disappear and your sonar pulses will become sharp and well defined...and your echos will contain more data about the bottom of the pond than the slopiness of the driving circuit.
I will ask this "important" question again... are you a PhD student? Is this part of a PhD thesis?
The reason I ask this question is because, in my past experience with PhD. students...PhD students are not experienced practical electronics technicians and they do not generally know or understand all the nuances of electronics, particularly power electronics such as this. I have known of people with PhDs who cant connect a car battery to a charger correctly which already had one battery charging on it (he connected his battery in anti parallel and it exploded in his face)...their PhD research has rendered them SO SPECIALIZED in one tiny area of just one field that they loose the most basic common sense.
Cheers and Good Luck
Last edited:
