Proportional valve coil driver selection


#1

Needing help to select a driver to power up an electrically-powered hydraulic proportional valve from 0-12 VDC. The coil draws ~2.5 Amps at 12 V, yet when I construct a simple emitter follower of a Darlington pair rated at 4 Amps with varying input voltage from a potentiometer, the hydraulics do not react until the pot is almost fully open, then the hydraulics move an inch or so and the transistor smokes out. When the coil is hot-wired direct to the battery the hydraulics move correctly but way too fast. Please help me select a device robust enough to construct a voltage follower that can reliably run this coil.


#2

Hello @DixonS

I think I need some more information to try to help you. Do you have a data sheet for the coil? Is current proportional to voltage for movement? 12V/2.5A = 6V/5A?

I am trying to understand what happened in the failure. Was there more than 4 amps running through your transistors?

Can you show me the current schematic that you are using?


#3

Hi, Robert_Fay, and thanks.

The coil is from like 1983 so there is a slight chance the company that used it in their hydraulics would still have a data sheet available; I’ll call them tomorrow.

I doubt it would draw twice the current with half the voltage but don’t know enough about coils to imagine how that could happen, but, the coil runs an hydraulic device called a proportional valve, which essentially controls the fluid flow based on the current through the coil.

The simplified hydraulic circuit is just the proportional valve in series with an on-off valve that runs the hydraulic cylinder. The on-off valve is very similar to a relay, except the plunger opens and closes an hydraulic valve rather than an electric switch. The proportional valve is the speed control: the higher the voltage across the coil, the wider the opening for the fluid flow.

When I hotwire the proportional valve coil directly to the 12V battery, the hydraulics work fine but way too fast. I imagine I could make it a 3-speed with a few switches and resistors for a 3-level voltage divider, but the infinite speed selection of a potentiometer would be way better, and safer.

So, I wired it up as a simple voltage follower with the center tap of the pot connected to the base pin of an NPN Darlington pair rated at 4 Amps (ON Semiconductor BD681STU), the 12V supply on the collector, the emitter on top of the coil, the bottom of the coil to ground.

I turn up the pot S-L-O-W-ly, waiting for the hydraulic ram to cre-e-ep into motion (like it used to with the original circuitry), but nothing happens until the pot is almost fully up, when the ram suddenly responds for a moment then the transistor fries.

I gather it’s maybe because there are no diodes and resistors to protect the transistors from whatever feedback or spikes the coil produces, so I have ordered some TIP120 power Darlington transistors, which have the resistors and diodes and can supposedly handle 5A continuous, 8A surge, so maybe that’ll do it. If not, then I need to know what ever will.

I’ll figure out how to scan in and attach the drawings from the manual, but they are only logic diagrams and probably not very useful. The schematic for the proposed repair (pencil on scratch paper) is just the switches to select which cylinder to activate with a potentiometer-controlled voltage follower running the proportional valve coil (for speed control).

Thanks very much for your help, anticipate hearing from you soon and ordering the recommended circuitry.


#4

@DixonS I suspect that the main problem with your circuit is one of thermal management; you’d need to attach the transistor to a rather beefy heatsink for that approach to work.

Assuming your proportional valve is basically an electromagnet vs. spring sort of affair, the coil resistance is probably about 12v/2.5A=4.8Ω. In order to apply (for example) 6v to the coil (at which point 1.25A of current flow would be expected) 6v would also have to be dropped across the transistor. P=V*I, therefore 1.25A * 6V=7.5W is getting burnt off in the transistor. Unless you’re talking an automotive 12V (which is closer to 14 with engine running) in which event your worst-case output would probably be @ 7v, and you’d be burning closer to 10.2W in the transistor.

Sticking with the approach you’ve started with, (there’s more than one way to approach the problem…) I’d suggest a much beefier transistor like a 2SD2083 and a thermal management solution with no more than 8°C/W or so of thermal resistance, more or less depending on your specific conditions. Note that the mounting surface of that transistor is likely connected to one of the terminals, so if you’re going to use a vehicle frame or some such as a heatsink, an electrically insulating thermal interface material would be needed.

There’s an article describing the process of making basic thermal management calculations here.


#5

Thank you, Rick_1976. I have just ordered 3 of your recommended devices to try out, and will figure a way to sink the heat.


#6

Got the drawings and sketches together but DigiKey won’t allow new users attachments to new users.


#7

So, about other approaches: an off-the-shelf PWM device that needs no software would prob’ly get better control but it looks like the original manufacturer used this same “brute force” method, so, yeah . . .

Doesn’t mean I have to.


#8

In terms of other approaches, I’d suggest you look at controllers for brushed DC motors. 1286-1169-ND is probably the most suitable item we have in stock here at Digi-Key at the moment. Thanks in large part to the FIRST robotics program, quite a few similar products designed for 12V systems are in commercial production. Such products are typically controlled via a hobby servo-style PWM interface, which would probably call for some sort of 555 timer circuit if you’re software-averse.


#9

I would love the most elegant solution but unfortunately I don’t have the capacity to put all that together quick. Just need a reliable black box that’s 0-12V low-current in, 0-12V high-current out. Don’t really care if it’s BJT, FET, op-amp, whatever, just so I can plug it right in and get back to work, and the controls aren’t too jerky. Thanks.


#10

If you’ve got transistors on order already, they’ll get to you before anything else will. Give 'em a try…


#11

Partial success! All valves have speed control. Because of differences in the hydraulic circuitries driving the loads, like a check valve acts like a diode and so forth, one cylinder is more controllable than another.

The main thing is, the proportional valve coil doesn’t move the plunger at all until the potentiometer is almost full on, then the cylinders extend way too fast. But, back off the potentiometer and the cylinder slows down. Unfortunately, only about 5° of turn on the pot covers the entire range from fast to stop, so it’s difficult to attain the exact right speed.

The potentiometer is 10K and I’m guessing the cylinders act that way because of the base current going through the top half of the potentiometer, so, is there a a certain ohmage more compatible with the 2SD2083? Also I need it in a push-button switched potentiometer, so the cylinders stop when I let go.

Basically, it needs a circuit that can muscle that coil plunger open a tiny crack at a time over most of the potentiometer’s range, not open it all at once at the top and only run the coil plunger over 5 degrees of turn.

Do you think a different pot will do it, or should I switch to a different driver?

Thanks


#12

It’s not entirely clear what’s giving rise to that behavior; might be something with the electronics, might be some characteristic of your valve. I’m still working completely from imagination as to the nature of your system, so it’s difficult to say what might or might not work with confidence. (did you ever get a datasheet for that valve? if so, try sharing again…)

In the interest of gaining more info, I’d suggest grabbing a multimeter and checking a few things;

  • disconnect the coil and record output voltage from the driver circuit as a function of pot position. if you don’t get a linear relationship (e. g. 1/4 output at 1/4 turn, 1/2 @ 1/2, etc) between the two, there’s a good chance that either I’m imagining your circuit incorrectly or that something’s amok with the wiring.

  • Re-connect the coil and repeat the above measurement process. Note the point at which motion of the cylinders begins. If the output voltage of the circuit becomes grossly non-linear with a load (coil) attached, that’s instructive also.

  • If it’s not too much trouble (you’d need two meters…) record the current through the coil as a function of the applied voltage, again noting the point where motion begins.

If your valve has significant stiction or other non-linear behavior, it’ll be difficult to get a nice control behavior if it doesn’t provide you with some sort of feedback mechanism. But again, I’m still doing a lot of guessing here. The above measurements should help get a better handle on what’s going on, and hopefully avoid a few cycles of the buy-it/try-it cycle.


#13

Thank you, Rick_1976. The test with the coil disconnected (emitter floating) already took place, the error from “trial and error”. Though no meter was connected, a meaningful result was obtained: when the potentiometer was almost wide open, it smoked out, destroyed. Don’t know if it hurt the 2SD2083, just threw it away and hooked up a new one with the new pot.

The circuit is the simplest voltage follower from “Active Devices 101” textbooks: 12V power supply with the positive side hooked to the top of the potentiometer and the collector pin on the 2SD2083. The center tap on the pot goes to the base pin. The negative pole of the power supply goes to ground, along with the bottom of the pot. The emitter pin hooks to the top of the coil, and the bottom of the coil goes to ground.

Impossible to screw up, I thought, until I forgot to hook up the coil, smoking the pot.

I’ll connect it with a load other than the coil and make the voltage/current vs pot position charts and let you know in a few days.

Data sheets for the coil are not available. The data sheet for the 2SD2083 confuses me, not knowing half of what they’re talking about. One table says it needs 24mA base current to switch on, then there’s a graph of Ic vs Vce with base current curves from 1.5 mA up to 30, where it’s clearly on. If it truly needs 24mA to switch on, then one would have to sweep the 10K pot pointer way up where it’s at 500 ohms top, 9500 ohms bottom, putting 11.4V from base to ground. Would maybe a 1K or 100 ohm pot allow enough base current with the pot pointer nearer the bottom? If so, I might nit have to make those charts, difficult in cold weather.

Thanks.


#14

If I understand correctly, your circuit looks something like the below:
image

If this is correct and the pot burnt out when the indicated coil resistance was left unconnected, it’s a pretty good indication that the actual circuit in front of you differs in some way from the concept of it that is in your mind and/or on paper; you ought not be burning out pots in such a circumstance. Bear in mind that the metallic tab on the transistor is in all probability connected to either the collector or emitter (the datasheet doesn’t appear to specify, which I’d consider a noteworthy omission…) and that this would likely cause problems if whatever you’ve got it fastened to is not electrically isolated.

The 24mA figure in the “typical switching characteristics” chart represents a test condition used when measuring the indicated switching times in that chart, not an operational requirement for the device. Think of that transistor as a sort of proportional valve unto itself; the amount of current entering the collector being a factor of hFE greater than the amount that one dumps into the base. It’ll work down to microamp-level.

Double check your wiring, being particularly careful about the metal tab, and making sure that that “floating” emitter wasn’t in fact shorting to ground.


#15

Ok I’ll ohm out one of those Darlingtons to see if the metal plate is connected to a pin but, at the time the pot burned out, the transistor wasn’t mounted. The metal plate was just floating in the air. Now, with the behavior described earlier, it is mounted on the frame. My phone won’t pull up your image so I’ll have to wait 'til I get home and try it on the computer.


#16

So, the collector is wired to that backplate, like you said. Fortunately, it was electrically isolated from the frame (heat sink paste and paint as insulator). Probably I’ll mount it on a heat sink away from the frame anyway.

Your drawing is exactly the circuit I’m using, yes. So, I brought the coil home and set up a simulation with a car battery. Hooked it up with a 5-ohm resistor load, then with the coil load. Put a receiver hitch pin through it for a steel core.

Measured a bunch of parameters every ⅛ turn of the pot: potentiometer resistance, I(Load), V(BE), V(Tap [pot pointer]), and V(Load) for the resistor load, all those plus magnetic field strength for the coil load.

Measured magnetic field strength by counting how many chainlinks would hang from the hitch pin without letting go.

For both the resistor and the coil, electrical parameters except V(BE) are mainly linear, a little raggedy at the ends. The only non-linear curve is the magnetic field strength, which takes off kindalike a diode curve b’tween ⅜ & ½ a turn, reaches half its peak (33 links) at ¾ turn, peaks (62 links) at ⅞ turn, and falls back to 55 links at a full turn. Eerily similar to that ram, which would only move when the pot was pointed b’tween ¾ & 1 turn.

So, before I go repeat these measurements out in the freezing air, do you think this coil was designed that way or is it fritzed out? Because if it’s designed that way I can stretch out that response range to cover the whole potentiometer by just sticking a resistor in after the pot. But if it’s fritzed out I’ll have to poke around and hunt one up, unless you can cross-reference for Hydraforce coils with your database. Hydraforce 12 VDC, 65G7112, 4399.

Thanks.


#17

It’s not immediately apparent to me why you’d have that sort of non-linear behavior with the magnetic aspect when everything else (coil current particularly) seems to be tracking nicely. It might well be a sign of something gone amok, or it might operate that way for some deliberate reason–I can’t tell you one way or the other with confidence.

But so long as everything else is indeed tracking nicely and not letting the magic smoke out or otherwise showing signs of distress, adding some resistance to your pot to get the sort of control fidelity you’re after could be a workaround.

Note that if you don’t have a fuse in the supply lead to this system already, it would be a very good idea to add one. And I’d definitely recommend something more reliable than a layer of paint as an electrical insulator between the transistor tab (at nominal 12V) and a grounded frame, since a short circuit there would break things in a very big hurry…