I need to replicate the way a magnet is being energized by a controller, this magnet is part of an electrovalve used in a high pressure pump working with a 12VDC power supply.

The following figures show the waveform read by my portable hantek oscilloscope, which was connected to my laptop on battery power, therefore the osc's is floating in relation to the power supply. I first took measurements using one probe, and then another reading using two probes.

Figure #1 shows how the signal starts by delivering brief pulses of +120VDC to the right cable at 28 kHz (the osc's ground is connected to the left cable), then it goes into a steady +12V signal, then near the end it start pulsing again down to 0V, and finally, a brief -75V signal shows up.



Figure #2 shows the same measurement, but this time one probe was connected to the right cable, and the second probe to the left cable, while the osc's ground was connected to ground of the 12VDC power supply.




This is important: Using my multimeter, with the pump turned off, I found that neither one of the two wires is attached to the power supply's ground. This tells me that we're dealing with at least three separate circuits that produce three voltages: 12, 120 and -75V

The magnet's data is as follows.




As can be seen, the first thing that intrigues me is the initial pulsed 120V signal, since it peaks at that value at first, and then it gradually decreases to around 90V, this maybe tells me that a capacitor is being discharged into the magnet's coil.

Let me also say that I'm perfectly capable of programming and designing an MCU interface that will produce the necessary pulses at the right frequency and duration.

It is for the following four points that I'm requesting this forum's help.

1. Calculate, if possible, the required current demanded by the coil in each stage. If this is too complicated, then I'll just measure the current using an ampsense sensor connected to my scope and then get back to this forum with the info.
2. Develop a (preferably transformerless) 120V power supply capable of delivering the first series of pulses.
3. Develop another power supply capable of delivering the final -75V pulse. This might have already been done, since people in this forum helped me design one several months ago, capable of continuously delivering 100 mA
4. Develop a circuit with a bunch of mosfets that will be controlled by my MCU that will orchestrate the switching between these power supplies to produce the aforementioned waveform.

I promise to do my homework and study all the recommended documents, and do the necessary research and calculations that are required of me.

So, step one:
How do I calculate the required current demanded by the coil in each section of the waveform, or should I simply go ahead and physically measure what's happening, or both?

Thank you all in advance for all your help.

 

Without knowing the coil inductance (which varies depending on the position of the valve core), core position and speed I don't see how the currents can be calculated. I think you'll need to measure them.

 

Calculate, if possible, the required current demanded by the coil in each stage. If this is too complicated, then I'll just measure the current using an ampsense sensor connected to my scope and then get back to this forum with the info.

The data you've presented so far has no bearing on the current in the coil. For that you'd need to know the impedance of that coil. Bottom line, unless you have more information on the coil, it may be simpler to just measure the current.

[*] Develop a (preferably transformerless) 120V power supply capable of delivering the first series of pulses.

You do know that's a magic term around here, right? But anyway the pulse looks to me a lot like something I was working on, to pulse a 1:10 transformer at 12V to generate a 120V ringing.

[*] Develop another power supply capable of delivering the final -75V pulse. This might have already been done, since people in this forum helped me design one several months ago, capable of continuously delivering 100 mA.

A pulse is also easy with an inductor.

[*] Develop a circuit with a bunch of mosfets that will be controlled by my MCU that will orchestrate the switching between these power supplies to produce the aforementioned waveform.

Well OK, but this task would be a lot easier if you knew what the valve was really looking for. Replicating your measurements is two steps away from the "ideal" waveform. One step is the imperfection of the circuit in replicating the ideal waveform, and the second step is your measurement of it. Knowing the ideal you are shooting for might open up more options for creating it.

 

The data you've presented so far has no bearing on the current in the coil. For that you'd need to know the impedance of that coil. Bottom line, unless you have more information on the coil, it may be simpler to just measure the current.You do know that's a magic term around here, right? But anyway the pulse looks to me a lot like something I was working on, to pulse a 1:10 transformer at 12V to generate a 120V ringing.A pulse is also easy with an inductor.Well OK, but this task would be a lot easier if you knew what the valve was really looking for. Replicating your measurements is two steps away from the "ideal" waveform. One step is the imperfection of the circuit in replicating the ideal waveform, and the second step is your measurement of it. Knowing the ideal you are shooting for might open up more options for creating it.

First, let me thank you profusely for taking the time to look a this thread.

Now to answer your questions.

1.- I thought I had posted all the needed info on the coil, impedance and all. But I guess that what I'm going to do is measure its current along with its waveform. I'll be doing that next week.
2.- I was not aware that "develop" was a sacred word down here, I'll be more careful next time I use it. But to put things in context, I meant to say I develop, and not we develop. I'll do all the hard work, and you point me in the right direction, if you'd be so kind.

When you said "...to pulse a 1:10 transformer at 12V to generate a 120V ringing." you meant you think the initial 28 Khz pulses are due to ringing and not digitally generated?

 

Without knowing the coil inductance (which varies depending on the position of the valve core), core position and speed I don't see how the currents can be calculated. I think you'll need to measure them.

Thank you Alec, I'll do that next week and get back with the info.

 

Wow! Pretty fancy. Maybe I can drop so possibilities and see what you and others think.
The one thought I have is the 120 volts is maybe from a DC to DC converter with current limit resulting in the decay of the voltage during that time as the current builds in the inductor. Then the 12 volts to complete the pull in of the valve and finally pwm to provide holding current. The voltage spike at the end is probably inductive kick when the coil is turned off. I guess you might do it this way to give it a huge pull in to overcome some pressure then reduce the power in the coil so it doesn't have to be to big.
Just my guess. Current would be good. Maybe it is not as high as I'm thinking.

 

Wow! Pretty fancy. Maybe I can drop so possibilities and see what you and others think.
The one thought I have is the 120 volts is maybe from a DC to DC converter with current limit resulting in the decay of the voltage during that time as the current builds in the inductor. Then the 12 volts to complete the pull in of the valve and finally pwm to provide holding current. The voltage spike at the end is probably inductive kick when the coil is turned off. I guess you might do it this way to give it a huge pull in to overcome some pressure then reduce the power in the coil so it doesn't have to be to big.
Just my guess. Current would be good. Maybe it is not as high as I'm thinking.

12V PWM is easy... and yes, I had also considered that the -75V is inductive kickback from the coil, though it's very clearly limited and maybe there's a TVS working at that. It's the initial 120V @ 28 kHz that I find most intriguing... gonna have to wait and see what the current measurements tell me.

 

The waveform is similar to the typical current profile of a fuel injector as shown below in an extract from the LM1949 datasheet, with an initial peak current then a holding current.

 

The waveform is similar to the typical current profile of a fuel injector as shown below in an extract from the LM1949 datasheet, with an initial peak current then a holding current.
View attachment 79981

Interesting, this is not for a fuel injector, but I'll look into it, thanks

 

When you said "...to pulse a 1:10 transformer at 12V to generate a 120V ringing." you meant you think the initial 28 Khz pulses are due to ringing and not digitally generated?

That's what I meant, yes, but it's simply a wild guess. I doubt it needs to pulse, that's just an artifact of how the peak was produced. I think ronv and alec-T are on to the concept of what you need.

 

That's what I meant, yes, but it's simply a wild guess. I doubt it needs to pulse, that's just an artifact of how the peak was produced. I think ronv and alec-T are on to the concept of what you need.

So maybe it's more of a current controlled device rather than a voltage controlled one?

 

It's electromagnetic. It's all about current.

 

It's electromagnetic. It's all about current.

That small comment of yours is of enormous help to me, thank you.

 

Yep, Amp turns.
Are you going to build the driver?

 

Yep, Amp turns.
Are you going to build the driver?

That is affirmative... and I'll build it soon and quick... hope I can count on your help.

 

I'd consider using an LM1949. A fuel injector is, after all, a solenoid valve.

 

I'd consider using an LM1949.

That's definitely on the table... I've already taken a look at its datasheet, but I'm gonna have to wait 'till next week when I take current measurements to decide my next course of action, thanks.

 

Alright, I'm back... this took longer than expected since the machine has been working nonstop and I didn't have any access to it until today.
Remember, what I'm trying to do is replicate the waveform used to control a high pressure valve that is actuated using an electromagnet.

I took more readings, and this time I measured the voltage waveform as well as the current. I used an amploc AMP25 sensor for this purpose. The datasheet says that it has a sensitivity of 37mV/A, so here's the image of both waveforms together. Volts are shown in red, while current in green.



This image clearly shows that current starts to build up until it reaches a peak of 8.44A, after which it goes down and stabilizes at 3.374A. I'm under the impression that this is more of a voltage controlled device, but I might be wrong of course. Also I find it strange that the negative voltage pulse at the very end does not cause negative current. Perhaps the purpose of this pulse is to accelerate the discharging of the coil. In fact, I'm almost sure that's the case, just look at how slowly the current goes down after the initial train of high voltage pulses is finished and a constant 12V is delivered. This current discharge is far slower than the one at the very end, when the final -75V pulse takes place.

So the three possibilities I'm considering are:

A) Build a variable power supply whose output voltage can be controlled by interfacing it to a microcontroller.
B) Build three power supplies, one working at 120V, one at 12V and the last one at -75V
C) Build one power supply working at 12V, with two sets of pre-charged capacitors, the first set would be at 120V, and the second at -75V. I'd be switching between them using three mosfets.​

I don't like plan B, since it would be cumbersome and more expensive that there's need. And maybe I could go for plan A, except I'm feeling a little intimidated by the challenge. And plan C is even more complicated than plan A.

What should I do?

 

Do you have the inductance and resistance of the coil?

 

The coil's resistance is 0.62 ohms, and it's inductance is 1.0964 mH, which was measured with an instrument that excited it with a 1 Volt sinusoidal wave at 1 Khz