Bilateral switch for PWM

[nose]

Hi all,

I'm currently hacking an old arcade machine to work with an Arduino. It has some 24v motors that draw 0.5 amps and I want to pulse them with PWM to control their speed.

I've identified the 2 wires that control the motor - and obviously these work in either polarity to make the motor go forward or reverse. I want to just cut one of the wires, and rapidly open and close this circuit to control the speed.

My problem is that a reed relay is not fast enough, a bipolar junction transistor only allows current to flow in one direction, and an analog switch or optoisolator just cannot handle 0.5 amp load.

Does anyone have a suggestion for a bilateral, fast switch that can handle 0.5 amps and 24 volts?

Many thanks.

 

[Harald Kapp]

You could use a Photomos relay, for example: http://www.panasonic-electric-works.com/peweu/en/html/photomos.php?

Whether it is fast enough will depend on your timing requirements, see datasheet.

Harald

 

[nose]

Thank you Harald.

These looked promising but the switching speed will need to be μs rather than ms. Some solid state relays may work for me?

 

[nose]

Would I be barking up the wrong tree by looking at FETs or TRIACS?

 

[duke37]

If you put bridge rectifier in the motor lead, then the current can be made to run in one direction. Thus you can use a FET. You will have to have a floating ground.

 

[nose]

Hi Duke, and thanks.

I need to keep the 'feature' of current running in both directions to keep the machine running properly.

Is it possible to run current through a FET in both directions? Or am I off the mark here?

 

[duke37]

You cannot use a fet in both directions but you could use a n channel and p channel for each direction.

My suggestion was to rectify the current so that the polarity at the fet was the same whichever way the motor was running. The fet would go between the + and - terminals and the motor wire between the AC terminals.

 

[Harald Kapp]

This is what duke recommends:

The current ouside the bridge rectifier can be bidirectional (even AC). but the MOSFET will see only DC.
There's one issue with this circuit, however: the control voltage is between gate and source and therefore is connected to the current carrying voltage. Depending on the exact circuit you most probably will need some kind of potential isolation between the controlling circuit and the power circuit.

Haralk

 

[duke37]

Thanks Harald, I will have to learn how to draw a diagram easily one day.

Since PWM is being used, there will be some voltage across the fet when it is turned off at all levels of demand. This can be rectified and fed to a capacitor to provide a power supply to drive the fet. A resistor and optical isolator would then complete the job. At high motor speeds, the voltage may not be sufficient since it will be the difference between supply voltage and motor back EMF. Some experiment would be needed.

A small pulse transformer would drive the FET but I am not into transformer design.

Duke

 

[nose]

Wow, thanks Harald.

It is a DC current I'm dealing with here by the way, I should've mentioned that. Before I look into this rectifier set up, would something like this suffice? I'm wondering whether this is actually bilateral? http://www.farnell.com/datasheets/88977.pdf

 

[Harald Kapp]

From the elctrical parameters: yes - 60 V / 1 A
From the timing parameters probably not. Turn-on + turn-off time make 2.5ms. Thus the max. frequency of the PWM signal could be 400 Hz but without a chance of actually doing PWM (because these are min. times). You'd need some headroom for the PWM.
If your motors are happy with, say, 100 Hz PWM, then the timing would be:
Period of PWM signal: 10 ms
Unusable on-off times: 2.5 ms
remaining time: 7.5 ms
If your PWM has 500µs resolution, this results in 15 steps. It's up to you to decide whether that is enough.
But normally a PWM for a DC motor would operate in the kHz range, preferably > 15 kHz. Reasons:
1) at frequencies < 1 kHz the motor might be fast enough to actually follow the on-off of the supply and may noticeably jerk.
2) at frequencies > 20kHz the noise from the PWM is no longer audible (at least to humans - bats are another story)
On the other hand: the higher the frequency, the higher the switching loss. You'll have to compromise.

Maybe you can get hold of on one of those:
http://www.supertex.com/pdf/app_notes/AN-D26.pdf

Or google "isolated MOSFET control" - sorry, I'm not a specialist on that topic.

Harald

 

[Harald Kapp]

I have worked out dukes proposal a bit more:

The extra diode and capacitor will supply a voltage from the voltage across the MOSFET during its off time. This voltage is used by the upper optocoupler to drive the gate, turning the MOSFET ON.
Tu turn off the MOSFET, the upper optocoupler is disabled (no LED current) and the lower optocoupler is turned on. thsi will turn the MOSFET OFF.

Be sure not to turn on both optocouplers at the same time. This would
a) discharge the capacitor
b) burn the optocouplers since the load current would then try to flow through them.

Harald

 

[(*steve*)]

Another option would be to use an H bridge to determine the direction and PWM either by another mosfet in either of the supply rails or by clever manipulation of the H bridge.

 

[nose]

Wow, thanks Harald. That looks doable and addresses everything Duke mentioned, perfect.

Right, I'll have to do some homework and figure out some suitable components. I have some of these opto isolators kicking around, they might be my starting point.

What do you use to make your nice diagrams, by the way!?

Thanks again,

Pete

 

[Harald Kapp]

Hi,
I'm using Microsoft Visio. I could have used Eagle, too, but for such simple schematics Visio is good enough.
The couplers you reference have ~5 µs combined rise and fall time. So they are 500 times as fast as the photomos relays. Good for >10 kHz PWM. Hard to find faster ones.

Reagrds,
Harald