High voltage transistors & flyback diodes for relays

Thread Starter

leonhart88

Joined Feb 23, 2007
118
Hey everyone,

Just a quick question about a transistor I am using. I have these dual NPN high voltage transistors (STD845DN40). The datasheet says they can take up to 400V across the collector and emitter. I'm just curious if I will still need a flyback diode to protect the transistor from solenoids and their back EMF.

Thanks,
Philip
 

Ron H

Joined Apr 14, 2005
7,063
The peak flyback voltage is difficult to predict.
Why do you want to eliminate the diode? Is it to save parts, or to speed up solenoid dropout time?
 

Adjuster

Joined Dec 26, 2010
2,148
Drop-out time can be safely minimised by replacing the simple freewheel diode by a combination of a diode and a resistor (less ideal), or preferably a combination of an ordinary diode and a Zener, or a dedicated transient suppression device like a Transil. Obviously all the parts used need to be rated for the coil current, at least on an intermittent basis

The coil turns off with a predictable voltage across it, which can be less than the driver breakdown but far more than a simple diode drop. The coil current thus decays more quickly. Some relay manufacturers discourage the use of "plain" anti-flyback diodes because the slowed current fall can lead to long and less predictable drop out times. Rarely, this might lead to permanent sticking on, although it is moot whether any relay which does this is satisfactory.
 

John P

Joined Oct 14, 2008
2,025
Maybe the zener diode could be placed between the collector and base of the transistor. That way it would keep the transistor conducting until the voltage dropped enough to stop the diode conducting, and the power dissipation would be in the transistor rather than the diode. It seems workable, but does anyone ever do it that way?
 

Adjuster

Joined Dec 26, 2010
2,148
Maybe the zener diode could be placed between the collector and base of the transistor. That way it would keep the transistor conducting until the voltage dropped enough to stop the diode conducting, and the power dissipation would be in the transistor rather than the diode. It seems workable, but does anyone ever do it that way?
I would not really recommend it, as for a start the input impedance to the transistor would be strange during turn-off. This would certainly be a bad idea in the case of a weedy little CMOS gate driving a power MOSFET, where the coil current suddenly becoming available to the gate circuit might let the magic smoke out of the driver. There might be also a question of the gate (or base) safe voltage being exceeded, perhaps on a negative back-swing if the zener voltage was big.

One might add a series resistance and diode voltage clamps to protect the gate/base and the driver, but is it worth all the bother?
 

John P

Joined Oct 14, 2008
2,025
But surely with a bipolar transistor (which we're told this is) the base voltage would be self-limiting, i.e. Vbe is generally less than a volt. As for input impedance, I don't see how it matters, as the driver has turned off at that point. One issue is that the driver mustn't sink current when it's in its off state. That would steal the current from the transistor base and thus defeat the protection.
 

ErnieM

Joined Apr 24, 2011
8,377
leonhart88: A solenoid looks like a big fat inductor so when you turn it off the voltage will keep rising until it finds a place to dump the current. It usually has no trouble making several hundred volts, or enough to breakdown your 400V transistor and permanently damage it even during the first time you turn it off.

Maybe the zener diode could be placed between the collector and base of the transistor. That way it would keep the transistor conducting until the voltage dropped enough to stop the diode conducting, and the power dissipation would be in the transistor rather than the diode. It seems workable, but does anyone ever do it that way?
Simon says don't do that. The whole point of turning off the base drive is to turn the transistor off. Your way you get a possibly predictable (but I wouldn't want to do it) turn off where you are basically attempting to switch the full relay current from the collector into the base: obviously with base current the collector is still on somewhat, so you are just slowing down the transistor turn off and probably dissipating more power in that zener then if you place it with the diode as Adjuster stated. His way is the standard way the entire industry does it.

I know it's standard, my company makes suppressors for several major relay manufacturers and they all use a zener and a diode, typically using a 36V zener and 1000V PIV diode.

I did a detailed analysis of this situation here.
 

SgtWookie

Joined Jul 17, 2007
22,230
[eta]
Ernie, didn't see your post before I made mine.

Why don't you copy that whole post you referenced and put it in a blog entry? (I should be doing the same thing...)

Just out of curiosity, I threw together a simulation of John_P's suggestion along with three other methods in LTSpice. The coils all have the same properties; 0.2H, 24 Ohms, 100pF parasitic capacitance, for 500mA peak current when the supply is 12v. The voltages on each collector is plotted, along with the inductor current. Since only the time during transistor turn-off is of interest, the plot starts 2mS prior to turn-off, and stops 12mS after turn-off.

I used the BD139-16 because they're fairly inexpensive, have a Vceo of 80v, Ic of 1.5A and a minimum hFE of 100 - but I'm pretty much "cheating" using an hFE of ~27 for the base resistor calculation.

At any rate, the collector of Q1 shows a lot of ringing (~2x the solenoid supply) for a relatively long period of time. This is due to the L and parasitic C of the inductor; only snubbed by the parasitic resistance of the inductor. The ringing could be problematic for emissions.

The collector of Q2 shows a lower current decay time and far less ringing using the back-to-back diode and Zener.

The collector of Q3 shows no ringing at all, but a rather long decay time.

Q4 shows the relatively long decay time one would see with just a diode across the inductor.

Of course, these are just simulations based on a number (lots!) of assumptions, which may not be correct; and the parameters would need to be changed for each individual situation. There is really no "one size fits all" that will be optimal. Only after testing with real-world parts will one be able to get valid answers.
 

Attachments

Last edited:

John P

Joined Oct 14, 2008
2,025
Thanks for running the simulation, SgtWookie. It looks as if the practical problems with my design show up when the inductor isn't just an inductor! Maybe an RC snubber would help to deal with a secondary effect like that--presumably the major part of the energy stored in the inductor gets burned off in the transistor. I claim that the advantage to my idea is that the zener can be small, as it doesn't dissipate much energy but simply controls the transistor.
 

SgtWookie

Joined Jul 17, 2007
22,230
Thanks for running the simulation, SgtWookie.
My pleasure. As I implied, it may be "garbage in, garbage out" (my simulation, not your idea) - as I made some assumptions that could be far off base. For instance, a relay coil would likely have a much higher resistance, and who knows how much parasitic capacitance. The resistance part would be easy to check, but inductance/capacitance would require a bit more than a multimeter to test. Feel free to download the simulation and experiment with it in LTSpice; it should run as-is even with a fresh install.

It looks as if the practical problems with my design show up when the inductor isn't just an inductor!
Well, as I mentioned - not all parameters are modelled; this is just an approximation, and numerous shortcuts were taken to get it to run quickly.

Maybe an RC snubber would help to deal with a secondary effect like that--presumably the major part of the energy stored in the inductor gets burned off in the transistor.
Watch that parts count climb!

I claim that the advantage to my idea is that the zener can be small, as it doesn't dissipate much energy but simply controls the transistor.
I see that. The power dissipation in the Zener is low, and the transistor power dissipation jumps. The big question mark is how fast will the transistor respond; it was already fairly well turned off when the rising spike hit the Zener voltage, turning the transistor back on in the linear region. I didn't model ANY of the other parasitics; for example the inductance of the wiring between the solenoid, Zener, transistor base, etc. That's part of the "shortcuts" I mentioned above.

Your mileage may vary, considerably.
 

Ron H

Joined Apr 14, 2005
7,063
This is a useful discussion, but we still don't know why the OP wants to eliminate the snubber. I was going to suggest the diode+resistor solution in my first post, but I thought maybe he was just trying to save parts.
 

Thread Starter

leonhart88

Joined Feb 23, 2007
118
Hey guys,

Sorry I did not get back to this post until now...I've been very busy the past couple of days/weeks.

Ron is right, I was just wondering if I could save parts and space, as I am building a relay circuit to drive lots of solenoid valves (through-hole parts). I usually use 1N4005s for the flyback. I think those are rated up to 600V. I thought that since the transistors were rated up to 400V, those diodes might not be needed. However, I wasn't so sure because I know the flyback helps direct the current into a "loop" until it is dissipated. I decided to put them in anyway just to be on the safe side.

Solid state relays would have allowed me to minimize parts/space...but there are various reasons why I do not want to go with solid state.

Thanks for the replies!
 
Top