I plan on putting a triac across the contacts of a 10 amp SPST relay to greatly extend the life of the relay contacts. The pair will be controlling a purely resistive heater, which will be cycled frequently. The power dissipation in the triac will be excessive without the relay. The triac will be controlled via an MOC3010 optocoupler, and the relay has a 5V coil; both will be controlled by a uController.
I simply plan on turning on the LED inside the optocoupler, then a few
100ms later, turning on the relay. Then to turn power off, I turn off the relay, then a few 100ms later I turn off the optocoupler LED. The only thing I'm not sure about is how the triac will behave when the relay opens. Will the triac immediately conduct? It seems like it should. Any other problems you can see with this scheme?
A 10A relay is fairly small, so it should close in a few ms. Opening time can be speeded up a lot by suppressing the back-emf using a zener or TVS instead of a simple diode.
I recently reduced the opening time of a 5V coil 16A/250V relay from
6.5ms with a diode to 1.6ms with a 27V SOT23 TVS. The part I used was MMBZ27VCLT1G but MMBZ27VALT1G is just as good. I paralleled the two diodes in the package. This arrangement passed elevated temperature endurance testing at a test lab at full resistive load for
100k cycles with random phase switching at 6 cycles per minute.
Switching such a relay about 2ms before zero crossings might work rather well.
You replied just in time. I was getting ready to post about this interesting suggestion by Mark.
Eliminating the triac/optocoupler completely is very appealing of course. Cost is no issue since this is for a home project, but a relay only solution should be much more durable. My initial thought was that the relay set/release times are too unpredictable, but your comments make me think otherwise.
So, I put a standard zener across the coil pins, in the same direction as a standard diode would be, or doesn't it matter? Is the breakdown voltage important? In other words, what is the ideal zener breakdown voltage for a 5V coil; as low as possible, or somewhere above that? Of course the driving circuit would have to withstand whatever breakdown voltage was chosen.
The next question is when to actually activate the relay. Since I'm using a microcontroller anyway, I thought about some fancy options. My first thought was that maybe the relay set and release times will vary with age/temperature/mood. So....I thought, each time before the relay needs to be turned on:
I use a second relay (which will be used as a sort of fail-safe disconnect anyway) to disconnect power to the contacts
Use the uC to turn on the main relay and time how long before the contacts actually close using some sort of monitoring circuit
Open main relay
Close the fail-safe relay
Close the main relay using the delay time obtained after zero crossing has been detected
Since a power transformer will be used to power the microcontroller circuit, the easiest way to monitor the AC for zero crossing would be through the secondary, but the transformer could be introducing some delay too. Or, I use an optoisolator with the input connected to the
Motorola (now On Semi) has some old app notes on doing this. I think you should turn on the triac driver before de-energizing the relay, and keep it on for a few AC cycles.
Also, think about what happens when the relay fails to close (and eventually it will). Also think about what happens when the triac fails to open (and probably it too eventually will).
Try to minimize the time the triac conducts. You may not get overheating per se, but the thermal cycling will eventually kill the triac.
Not the MOC3010, that's not zero crossing. But I should have been more clear. When I mentioned using an optoisolator to detect zero cross, I wasn't referring to a triac output like the MOC3010; I was referring to a standard transistor out.
I think I just got a better idea. A bigger triac I don't believe will fix the problem, since as I understand, they all have about 1V across them when conducting. This will dissipate the same amount of power no matter what size the triac is.
Here's what I do, it's simpler, safer, and more flexible: I use two transistor-out optocouplers (4N27 or similar), one monitoring the AC line, one monitoring the AC controlled by the relay.
The first opto tells the micro when the AC supply has zero crossed. The second tells the micro when the relay contacts actually closed. There will be software calibrations that indicate the delay between AC line zero cross and setting/releasing relay. When the relay is set or released, the 2nd opto will monitor if the contacts closed/opened near zero crossing. If not, it will adjust it for next time. Fully automatic, self adjusting.
In addition, the 2nd opto will be useful to verify the relay is functioning.
You replied just in time. I was getting ready to post about this interesting suggestion by Mark.
Eliminating the triac/optocoupler completely is very appealing of course. Cost is no issue since this is for a home project, but a relay only solution should be much more durable. My initial thought was that the relay set/release times are too unpredictable, but your comments make me think otherwise.
So, I put a standard zener across the coil pins, in the same direction as a standard diode would be, or doesn't it matter?
You need the zener *AS WELL AS* the diode - if you only have the zener it does exactly the same as the ordinary diode, put the zener the right way round so it zeners the back emf and its forward conduction will short the relay drive - you need to still have the original diode in series with it to prevent that.
The basic theory is simple - just the diode shorts the back emf which pushes current back through the coil which delays unlatching.
The zener allows most of the back emf to dissipate without shunting back emf current back into the coil by only clamping the back emf to a safe value for the transistor/chip driving the coil.
I'd choose a more reliable relay. What happens to the triac when the relay fails? Is there a thermal shut-off for the heater?
--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------
Unless the zener is connected beween ground and the collector of the driver transistor, in which case you don't. However, the SOT23 dual TVS can be used in a series configuration where one diode acts as a zener and the other is used like a conventional package.
The ideal zener breakdown voltage is a compromise between opening time, breakdown voltage of the driver transistor, peak power dissipation in the zener diode and possible emc issues. About 5x the coil operating voltage worked well for me with a 5V coil. The opening time was within 10% of that obtained with no transient suppression at all (achieved by using another relay to drive the first one.)
There will be variation from one opening to another. Expect at least
+/- 5% variability for a new relay and more as it wears out.
Why the need for a ZENER?, Energy from the collapsed field is reversed In polarity and a standard method of a SI diode should handle that just fine and produce the reverse field current in the coil that you're looking for to neutralize the mag field and may even reverse it to bouce the contact away a little faster.
We've made this type of soft start for relays before how ever, you do need some delay with the off cycle of the triac to remove the arc on the opening of the contacts. CLosing is not a problem since the Triac is coming on sooner than the contacts can.
I concur on this point. My relays seem to be polarized. If I apply current to the coil in the wrong direction, they do not engage, however they pull the same amount of current regardless. And, with a regular 1N4004 diode across the coil, pull-in is about 5ms, open is about 2.7ms.
I can't find any information on how this type of relay works. All I find is that some have an integrated diode and that's why polarity is important. In case anyone is curious, part # is Omron G6C-1114P-US-DC5
The approximate method is used in configuring electronic brushes for fast high-current multi-phase transfer switches, though there is usually a control interface monitoring current in these, to ensure correct timing and potential degrees of coherrence.
A triac will naturally turn off at the zero current crossing after the drive is removed, providing that re-applied dv/dt or junction temperature is not exceeded.
At turn-on, for possibly inductive loads, gate current is normally maintained until conduction is achieved. Your static gate drive would serve that purpose.
An optical triac gate drive will require some headroom, so a small voltage rise across the compound part, in the microsecond (or gate pulse rep period) range, must be expected as the mechanical contacts break first in the contact opening sequence.
Resistive loads don't normally present a particular hazard to properly constructed relay contacts. Though there's an inductive component in both interconnection and the commercial source itself, it's usually nothing that an appropriately sized snubber can't handle. A snubber would be needed to protect a triac anyways, if it were used.
Reliability and life are only increased if the additional complexity isn't detrimental to the whole. A properly heatsunk triac might be a more reliable solution. When calculating efficiency, don't ignore relay coil power consumption or gate drive circuit losses.
The advantage of the zener is that it dissipates the energy stored in the relay more quickly than a plain diode. After the drive is removed, current keeps circulating through the clamping device and coil for a time which is greatest if the voltage is clamped to a low voltage.
Here are some real test results on a 16A relay with 5V coil (HF115F/
005-1HS3AF). The data are a little sparse as I concentrated the testing on the conditions I was most interested in at the time.
Condition Duration of significant coil current flow Range of opening times for multiple tests No suppression not measured 1.56ms (only a few tests) Single shunt diode
15ms 6.56 -
6.60ms Two series shunt diodes
10ms 5.34 -
5.60ms
1k5 parallel resistor time constant 40us, 110V peak 1.84 - 1.94ms
13V zener across driver
1.2ms 2.08 -
2.22ms
24V zener across driver
0.4ms 1.64 -
1.73ms
37V zener across driver
0.2ms 1.54 -
1.61ms
While it might not appear to matter whether the opening time is long or short, this does (according to some relay manufacturers' app notes) translate into a variation in the velocity with which the contacts open and hence the duration of arcing.
The endurance data quoted by many manufacturers is usually based on tests with NO coil suppression. Some manufacturers specifically warn that a simple shunt diode may reduce relay life.
Can you re-post this, in readable format, or verify that I've correctly re-written your findings below? The data looks interesting, but line wrap makes it difficult to figure out the correlations. My attempt at correlating is below.
Ed
Condition Duration of Range of opening significant coil times for multiple current flow tests
No suppression not measured 1.56ms (only a few tests)
Single shunt diode 15ms 6.56 - 6.60ms
Two series shunt 10ms 5.34 - 5.60ms diodes
1k5 parallel resistor time constant 40us, 110V peak 1.84 - 1.94ms
13V zener across 1.2ms 2.08 - 2.22ms driver
24V zener across 0.4ms 1.64 - 1.73ms driver
37V zener across 0.2ms 1.54 - 1.61ms driver
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.