Electromagnet waveform

ronv

Joined Nov 12, 2008
3,770
Alec,

I think the controller is a black box. But if it can fire every 20 ms and draws 9 amps form the 120 volts for .5 ms. it is still pretty stout for a high voltage supply.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Is Image 02 what you are referring to as the 'third image' ?
Both scope channels are set to "x100". 100 times what?
Does the machine include a SMPS generating the high voltage from 12V, or is the 120V mains-derived ?
My apologies... it seems that I swapped image02 and image03 in my explanation
The channels are set to x100 due to the type of probes that I used, which are for voltages higher than 30V. The voltage scale is what really counts, which is set to 50V and 20V per division, depending on the measurement being made.
The machine works at a standard 220VAC, and has a large 12V power supply. The 120V are generated from that power supply, using a boost circuit.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
I think the thing in question is where the circuit ground is.
Here is one with the basics like your pictures.
OMG! this is beautiful... I'll start studying it right away!
And yes, the controller is completely encapsulated and cannot be opened without damaging it. That's why I've had to start from almost scratch and rely entirely on the scope's measurements.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Your circuit emulates the coil's voltage very closely, even the fast shutoff at the end. It's the current going through the coil that draws my attention, it's very jagged and not smooth during the plateau like my original measurements. I'll try to tweak it some more... see what happens.
Another two interesting facts:
It seems that the circuit is short-circuit protected, and won't get damaged if both cables at the coil's connector are shorted. Only a small spark will arc between them when this happens.
If I were to touch both cables, either when they're connected or disconnected from the coil, I'd get a very small and mild shock.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
@ronv, I'm beginning to think that the best approach to solving this problem would be to build the last circuit you've proposed, tweak it, and see how it behaves.
The coil's inductance was measured using a different waveform and frequency than the one being used to operate it, so its inductance could be different from 1.1 mH (although I wouldn't know by how much). And that could mean that the simulations are off the mark. We need real interactions to to infer a realistic scenario.
 

ronv

Joined Nov 12, 2008
3,770
@ronv, I'm beginning to think that the best approach to solving this problem would be to build the last circuit you've proposed, tweak it, and see how it behaves.
The coil's inductance was measured using a different waveform and frequency than the one being used to operate it, so its inductance could be different from 1.1 mH (although I wouldn't know by how much). And that could mean that the simulations are off the mark. We need real interactions to to infer a realistic scenario.
I suspect your right. The inductance probably changes as the plunger pulls in and things like that.
Are you good at microcode? I'm not but I have a few ideas for the PWM of the high current pulse. It seems like maybe they are trying to control the timing very close. Is that the case? What does it dispensed, Is the amount set by time energized? Does the setting on this machine need to match others?
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
I suspect your right. The inductance probably changes as the plunger pulls in and things like that.
Are you good at microcode? I'm not but I have a few ideas for the PWM of the high current pulse. It seems like maybe they are trying to control the timing very close. Is that the case? What does it dispensed, Is the amount set by time energized? Does the setting on this machine need to match others?
Not to brag, but I'm excellent at microcode :cool:. Regarding electronics, that's the area in which I need no help. So no, generating the 28 kHz pulses and the rest of the switching involved doesn't scare me one bit. I'm an expert on the 8051 architecture, and don't feel attracted to the raspberry or arduino fads. This is because I write all of my code in assembly, and design all of the digital circuits in my projects myself... But I'm rather weak when it comes to analog stuff. I just hope I can eventually return you the favor somehow, you've helped quite a bit already.
Anyway, the pump dispenses a special type of oil for a continuous process, and yes, the dispensed amount depends on the amount of time the valve's energized. And no, the machine's setting does not have to match other's.
I'm going to study your last circuit more carefully, and choose the right mosfets (I think you mentioned on earlier posts that they weren't matched to the circuit), then I'm going to search for possible 12V to 120VDC converter circuits and I'll get back to you, hear what you have to say.
Thanks again.
 
Last edited:

ronv

Joined Nov 12, 2008
3,770
Alright! :D

I give you my idea anyway. I think what I would do is generate a table in code that represents the ideal current wave form. Since there are only about 12 pulses so it's not a big one.
We then need to add a small current sense to measure the current. You can then compare desired to actual and change the duty cycle to get back on the curve. This will make it insensitive to changes in resistance from heat, varying voltage, etc.

And yes, I didn't pick mos fets or set the gate voltage correctly for the 120 volt one.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Alright! :D ... I give you my idea anyway.
I think it's a great idea... though I have little familiarity with op-amps and/or comparators (I suspect they'll be involved for the current sense somehow) I trust I'll have your guidance for designing things properly.

I found two circuits for generating 120 VDC from 12 VDC:

12V-to-120V-dc-dc-converter.png


dcdc120.gif

Source pages for each can be found here, and here.

I personally like the one that is not using the 555 better, since it's output has small value caps... although a special type of transformer needs to be made! ... so maybe for starts the one using the 555 is simpler and easier?
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
I think it might be easier to just build a little AC supply with a budget regulator. Something like this:
With a transformer like this.
http://www.hammondmfg.com/186_187.htm
So you're saying I should build a simple supply with its primary being fed from 120VAC, and focus my efforts on the driver circuit, instead of making things too complicated at first?
I can see that your circuit delivers a steady 5 amps, and the coil peaks at about close to 9 amps, for a very short time of course. Wouldn't that be a problem?
 

ronv

Joined Nov 12, 2008
3,770
It's interesting. A lot of the 9 amps comes from the 12 volt supply that is turned on at the same time and has no resistance in series. So the 120 volt supply only delivers about 4.5 amps for 0.5 ms out of each 20 ms. So the average is pretty low (.5/20 * 4.5) and is mostly supplied by the caps in the supply.
 

Alec_t

Joined Sep 17, 2013
14,280
I agree with Ron re the 120V supply. Since the machine uses 220VAC mains, a step-down transformer plus rectifier bridge would be simpler than a SMPS.
One difficulty with implementing a design with high-side switching of the 120V to the coil is the P-FET gate drive arrangement. I realise Ron's sim is just an outline, but note that M1 wouldn't really appreciate a Vgs of 120V ;).
You also need to look at power handling and efficiency in a practical circuit. For example, R1 is dissipating an average 35W (cf the coil's 9W) ! Can that be reduced by a PWM system?
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
I agree with Ron re the 120V supply. Since the machine uses 220VAC mains, a step-down transformer plus rectifier bridge would be simpler than a SMPS.
One difficulty with implementing a design with high-side switching of the 120V to the coil is the P-FET gate drive arrangement. I realise Ron's sim is just an outline, but note that M1 wouldn't really appreciate a Vgs of 120V ;).
You also need to look at power handling and efficiency in a practical circuit. For example, R1 is dissipating an average 35W (cf the coil's 9W) ! Can that be reduced by a PWM system?
Thanks for your input, and yes, I think I'll build ronv's suggested power supply instead.
Could a driver like this one be a better option for M1 and/or M2?
 

ronv

Joined Nov 12, 2008
3,770
The good news it the duty cycle is very low for the high voltage circuit - only 0.5 ms out of 20 ms. So average power is not very high.

I've been trying to figure out why they use such a high voltage and the only thought so far is that the inductance of the coil goes way up as the plunger goes into the coil. Did you measure the inductance with the plunger in and out of the coil?? Could it be 10 times higher? Might be...
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
The good news it the duty cycle is very low for the high voltage circuit - only 0.5 ms out of 20 ms. So average power is not very high.

I've been trying to figure out why they use such a high voltage and the only thought so far is that the inductance of the coil goes way up as the plunger goes into the coil. Did you measure the inductance with the plunger in and out of the coil?? Could it be 10 times higher? Might be...
The inductance was measured without the plunger installed. Only the coil was tested, and not the entire assembly, since the lab equipment used for this purpose did not allow it.
Perhaps if we were to build a circuit specifically for this purpose? Or maybe if I were to build your driver circuit its behavior would tell us a lot more about the valve's real inductance?
 

ronv

Joined Nov 12, 2008
3,770
Aha, I think that solves the mystery. :D

At some point we will need to know what the current waveform looks like as the solenoid actuate, but it looks like you have that capability.
I have some honey dos, but back later.

Let me see if I can't come up with the rest of the circuit.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
Aha, I think that solves the mystery. :D

At some point we will need to know what the current waveform looks like as the solenoid actuate, but it looks like you have that capability.
I have some honey dos, but back later.

Let me see if I can't come up with the rest of the circuit.
Thanks... I'll be waiting for your response, since I'll be ordering all the parts from Digikey as soon as the circuit is complete. Although I do have lots of other components available, I'd like to keep improvisation on this project at a minimum, at least on the beginning phase.
 

Alec_t

Joined Sep 17, 2013
14,280
An opto-isolated driver similar to one in the link (albeit for DC not AC) looks ok for M1.
Inductance will certainly increase when the pintle (plunger) is present and will change with its position. Another variable, difficult to model with any accuracy, is the (relatively minor, I believe) coil current change due tothe emf generated when the pintle accelerates and decelerates.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,218
An opto-isolated driver similar to one in the link (albeit for DC not AC) looks ok for M1.
Inductance will certainly increase when the pintle (plunger) is present and will change with its position. Another variable, difficult to model with any accuracy, is the (relatively minor, I believe) coil current change due to the emf generated when the pintle accelerates and decelerates.
Thanks Alec_t, your opinion is thoroughly appreciated. What about M2, is it ok being driven as it is?
 
Top