Smart Solenoid Drivers

Go To Last Post
19 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, I wish to make a smart solenoid driver.
I was going to use some standard relays, but after some research I think the benefits of a smart driver are worth it, these being:
1) Less mechanical wear from ramping up and down the current (soft on/off).
2) Reduced power consumption (holding current only needs to be 1/10th of switching current), thus a much smaller power supply is required.

The solenoids draw 1.25A at 24V. Now I just need a method of allowing less current to flow to the solenoids. I have tried using a Power Darlington pair (MJ3000) in a TO-3 package mounted to a large heat sink. It got very hot and unstable.

I was thinking of a switch mode supply to drop the voltage, but then because I need to drive 12 solenoids, I am reluctant since then I would need 12 large capacitors and 12 large inductors (keeping component count down is critical!).

I found this beauty: VN380 (SMART SOLENOID DRIVER
SOLID STATE RELAY)
http://doc.chipfind.ru/pdf/stmic...
Basically it preforms all the smart functions described in 1 & 2 above. It seems like a rare item. So I am looking for an alternative common part (if it breaks down, I dont want people to battle to repair it).

Any idea of a general route to try, or a component that would make my life much easier. The best I can think of is passing a PWM micro controller output through a low pass filter (DTA conversion), then boost the signal through a voltage follower, and then onto the gate of a MOSFET which would drive a solenoid. I am not to sure about MOSFETS being half on. As far as I know they are more suited for ON/OFF applications.

Can you think of something better? I know you smart people can! Thanks for any advice.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

search for 'pwm solenoid driver' to get these.. http://focus.ti.com/lit/ds/symli...

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

bobgardner wrote:
search for 'pwm solenoid driver' to get these.. http://focus.ti.com/lit/ds/symli...

Those devices are really neat. Checked all my local suppliers, none of them stock them. When you live in a 3rd world country its hard to get anything out of the ordinary. Thanks for the suggestion though, its a cool little IC.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Who cares if the local distributer stocks them? I believe TI is trying real hard to sell their products in ______ or wherever you live (note: in the profile, you can fill in a field with your city and country, which will be helpful to other forum readers). The TI rep ought to be able to get samples. The local distri will order them if you are going into production. Thats how Capitalism works.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

bobgardner wrote:
Who cares if the local distributer stocks them? I believe TI is trying real hard to sell their products in ______ or wherever you live (note: in the profile, you can fill in a field with your city and country, which will be helpful to other forum readers). The TI rep ought to be able to get samples. The local distri will order them if you are going into production. Thats how Capitalism works.

I set up my location in my profile. The reason I care wheter my local distributor carries them is because the the machines I make are usually installed in different cities. And if it breaks it would not be me who fixes it. And production could stand for days while they wait for an exotic part to be flown in. And since I am only making one of these devices, I have no bargaining power with the distributors.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think the idea is have a hi voltage to hit the solenoid real hard to get it to move, then turn on a second output with a hold transistor that runs from a lower voltage. This could be done with 2 mosfets, 2 avr pins, and of course, 2 power supply voltages, which you might have already, like 12 and 5.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Or some clever use of switches and capacitors.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just use a micro with pwm to switch a mosfet. You already have the inductor - the solenoid. So ,use the pwm to modulate the current through the solenoid. Make sure you have a catch diode across the solenoid or some other form of snubbing. You could use a 555 timer ic if you really wanted it simple.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

did you ever think of using a latching solenoid? it uses a neodimium magnet to hold the plunger in after energizing the unit then you hit it with a reverse pulse to unlatch the plunger as the plunger has a spring on it to unlatch.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
Just use a micro with pwm to switch a mosfet. You already have the inductor - the solenoid. So ,use the pwm to modulate the current through the solenoid. Make sure you have a catch diode across the solenoid or some other form of snubbing. You could use a 555 timer ic if you really wanted it simple.
Kartman to my rescue again!
Yes it took me awhile to realise how silly I was being. After a lot of experimentation, with everything over heating, I finally realized PWM is the way to go, since the inductor will keep the juice flowing anyway.

So I have been testing the PWM route (refer to the schematic below). And I have a problem. The output from the optocoupler rises slowly (R2 cannot be decreased to spent up the switching, because then the ouput does not swing all the way to ground). I think this long period between switching is what is making the MOSFET overheat.

I added a push pull transistor pair to try and speed up the process. This helped, but then the output from the transistor pair is 24V. And the datasheet for the MOSFET says VGS should not exceed 24V, so thats why I added the potential divider network R3/R4. I set the resistor values as low as possible for 1/4W resistors. This is a horrible solution, as the gate current is now just a small tap off of the transistor output, resulting in slow switching times = very hot MOSFET.

This circuit also uses many components (undesirable when you are making 16 of them).

I spent hours looking for buffers that are 24V compatible to no avail. It seems all IC's die at 20V. I only found some current sinkers like the ULN2803, but that has a slow turn on time and cannot source current. Can anyone see a solution that is way more elegant?

Attachment(s): 

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jpmayn1 wrote:
did you ever think of using a latching solenoid? it uses a neodimium magnet to hold the plunger in after energizing the unit then you hit it with a reverse pulse to unlatch the plunger as the plunger has a spring on it to unlatch.

Sounds fantastic, but also sounds pricey. I shall research them. The guy I am making this circuit for doesnt have much money, and doesnt want to part with much $$$$. Thank you for the suggestion I will look into it!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Optos are a curse! I avoid them where possible. If you need to do 16 drivers, it would seem reasonable to use another AVR (or a number of AVRs) to do this. You can then talk to this AVR via an opto if you so desire using serial. If you choose your mosfet correctly, you should be able to use a logic level mosfet driven directly from the AVR port pins as long as you don't make your PWM frequency too high or use too large a mosfet (larger mosfets have a higher gate capacitance).

With a bit of lateral thinking, you should be able to use the internal oscillaotr in the AVR and devise a serial protocol that tolerates the frequency variation. You should be able to keep the number of components small using this technique.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
Optos are a curse! I avoid them where possible. If you need to do 16 drivers, it would seem reasonable to use another AVR (or a number of AVRs) to do this. You can then talk to this AVR via an opto if you so desire using serial. If you choose your mosfet correctly, you should be able to use a logic level mosfet driven directly from the AVR port pins as long as you don't make your PWM frequency too high or use too large a mosfet (larger mosfets have a higher gate capacitance).

With a bit of lateral thinking, you should be able to use the internal oscillaotr in the AVR and devise a serial protocol that tolerates the frequency variation. You should be able to keep the number of components small using this technique.


Optos although a pain to work with, I think they are worth it for the extra level of protection. I know you really dont like them though! When I looked at a PLC as you suggested, all the inputs and outputs had optos!!

That would be a clever little circuit you suggested, but the problem is, then I would have to add a 24V to 5V switchmode supply.

I kept trying some different things until I came to two solutions that both work. The MOSFET stays at room temperature switching 2Amps at 2Khz for both solutions. For future referance, anyone looking for direction refer to the schematic below. The tranzorb D2 should be replaced with a Zener diode of about 12V (I didnt have any zener diodes lying around), and then R4 decreased accordingly.

I think I am going to go with the second solution (Op amp one) as it only requires 1/2 an op amp and 2 resistors!

Thanks for all the help guys, especially you Kartman!! :D

Attachment(s): 

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Most opamps don't like purely capacitive loads, and a gate of a MOSFET is one. And most opamps can't deliver a lot of current for fast switching.

I think this is a quite an complicated circuit, why just not use a logic level MOSFET that can be connected to the AVR directly?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Most opamps don't like purely capacitive loads, and a gate of a MOSFET is one. And most opamps can't deliver a lot of current for fast switching.

I think this is a quite an complicated circuit, why just not use a logic level MOSFET that can be connected to the AVR directly?


Well the only logic level MOSFETs I could find from my supplier are these:
http://www.mantech.co.za/Stock.a... (only the ones with prices I can get)
After downloading and reading the datasheets for them, they do not seem to be very advantages to me, e.g. the IRLZ44N has a high input capacitance of 1700pF. Whereas the MOSFET I am using (IRFZ24) only has an input Capacitance of 365pF (although its Ron is higher by the same ratio. But since I am using a PWM signal I am assuming most of the power dissapated is from the switching, not the on state).

Do you think an AVR driving a 4x larger load is going to last longer than an op-amp driving a 4x smaller impedance? I am not trying to be funny, maybe you could recommend a common logic level MOSFET that you use (I have only found the IRL family of MOSFETs.

The reason I am using an opto coupler is because the inputs are from places in a factory, and it is to protect the delicate micro controller from the big bad world outside. And the Microcontroller is going to be a TQFP64, I dont ever want to have to attempt to unsolder it!!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Most opamps don't like purely capacitive loads

Thanks for your comments, I have been thinking/researching this point. I tried placing a 100E resistor in series with the output of the OP-AMP to the Gate, I hope it makes the op amp happier! It did make the MOSFET run a little hotter though. I tried a 10E resistor and that that kept the MOSFET turning on hard, but then I dont know wheter it will make a noticable differance to the opamp?

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You could try one of National's "Simple Switcher" series.. These are very compact switchers with most of them integrated switches, that usually only require a small inductor and diode plus a couple resistors and caps. National will send samples for most, so it doesn't get much cheaper than this. What's more, some of them have programmable current limiters...

http://www.national.com/analog/p...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Don't rely on the optos to protect you from the big,bad world. They're only used for galvanic isolation and will only provide limited protection from ESD and fast transients. So, don't ever think of them as a magic bullet.

2kHz pwm is a bit slow - the major reason for higher speed is to avoid audible buzzing. So that's why you'll normally see PWM frequencies of 10kHz and higher.

Personally, I think the op-amp solution sucks. I use Sharp PC900 optos for my work with is usally isolating serial comms. These devices have an i.c in them to take care of the opto receiver. These could probably run a logic level mosfet directly. My suggestion of having another AVR to do the pwm didn't find favour with you? Having a micro on the 'hot' side is not an issue if you opto isolate the input side so that it doesn't become the 'meat in the sandwich' so to speak. Your mosfets are just as succeptable to transients etc as your AVR is, so you need protection whichever way you look at it.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is what I have come to prefer to optos for isolation. Offers much better response, uses much less power, and operates at 5v/3.3v with up to 90Mbps data rate. 25KV/uS transient immunity, UL certified for 2500Vrms for 1 minute, CSA certified, etc.... Sets you back about 5$ for a quad.

http://www.analog.com/static/imp...