Finding part to switch 120 VAC @ 2000 Hz in < 1ms

Semiconductor
1

Hi,

I am looking for a part, (possibly an SSR) that is able to switch 120 VAC @ 2000 Hz.
This sine wave AC is produced by an inverter for EL wire.

My switching signal is up to 3.3 V, from an ESP32.

The response time should be less than 1 ms, ideally less than 0.5 ms, and the switching frequency should be able to go as low as 10 milliseconds, i.e. turning on and then after 10 ms turning off, and after another 10 ms turning on again.

The part should be able to withstand continous operation under these conditions, for at least 8 consecutive hours.

Thanks for your help.

1 Like
2

How much current needs to be switched?

IIRC, a TRIAC could do the job if the current isn’t too high.

1 Like
3

I have used TRIACs in the past, but those seemed to not tolerate that high of an AC frequency, and they broke. The current is 120 mA max.

1 Like
4

I know most TRIACs are only for sub 200HZ but I thought there were some low power units that worked up to about 10kHz.

However it’s been over 30 years since I worked with TRIACs.

AFAIK in the past many AC SSR’s were just TRIAC output optocouplers, could be that there are MOSFET based AC switching SSRs available .

1 Like
5

Hello Thokari,

Let me echo @PaulHutch, suggestion. This application is best solved with a thyristor, either an SCR or Triac.

The challenge is to find a device that will meet your needs in an inexpensive easy to use package. Know that this class of semiconductors is available in small packages such as a TO-220 all the way up to single devices larger than a human fist that are capable of controlling thousands of amperes with kilovolt systems. In fact, I once worked on a 15 thousand horsepower system that featured banks of SCRs.

Please help me understand your application. You stated 2000 Hz on a 120 VAC system. If I understand correctly, that’s 0.5 milliseconds switching on system that has an AC frequency of 16 ms (assume 60 Hz). What am I missing?

As for parts to interface a 3.3 volt logic level signal to 120 VAC, I recommend something in this class of solid-state relays. The DIN rail will provide a good anchor to keep things safe.

If that doesn’t work - perhaps it’s too slow - you may need to do some design work using parts such as this random phase triac driver.

Please let us know if we can assist.

Best Wishes,

APDahlen

Schematic of random phase triac driver.

2 Likes
6

The AC is produced by an EL inverter, such as this one
https://www.reichelt.de/de/de/inverter-12v-110v-fuer-folie-bis-300-cm--el-inverter-03-p60209.html?r=1

It produces about 100 to 120 Volts AC, at a frequency of about 2000 Hz, which reduces as more EL wire is attached, down to about 1000 Hz, these values are estimates, but it is important to know that the frequency might vary, also during operation, when individual channels are switched on and off, which is what my circuit should be designed to do.

So I want to switch up to 8 channels of EL wire on and off. Each channel is up to a couple of meters of EL wire, which draws about 15 mA per meter, so I am usually below 60 to 90 mA, so I am assuming 120 mA to include a safety margin.
The interval for the switching is in the range of 10 milliseconds, preferably it could go down as low as 5 milliseconds. The application is music visualisation of a sampled audio signal, right now with a sampling window of 16 milliseconds.

The switching also has to be responsive, which is why I am asking for a response time, preferably, below half a millisecond. I realize the human perception can only detect 10 to 20 ms, however since the audio sampling takes 16 ms already (and should not go much lower), the visualisation is already that far behind in time relative to the audio signal, so every additional millisecond counts here, which is why I need it to be as low as half a millisecond.

I have used this specific part in the past, but they kept breaking, presumably because the TRIACs where not specified for that high of an AC frequency.

I assume that is also the reason this part has been retired.

Thank you for your help, and let me know if there are any other questions.

2 Likes
7

Now I understand. Thank you, Thokari,

Today I learned something new.

A quick search suggests that Spark Fun was the primary supplier for the sequencer. At this point, they appear to have discontinued all EL products.

I’m beginning to see your problem. A search reveals that many companies provide inverters, however I was unable to locate any easy-to-use micro to EL boards.

The best I can do for you is recommendation you repair the units or design your own. The board (partial schematic below) uses common components:

Partial cchematic for the SparkFun EL sequencer

Sorry I couldn’t be of more assistance. Perhaps another DigiKey community member can offer an alternative.

Best Wishes,

APDahlen

Helpful videos:

2 Likes
8

@APDahlen I have plenty of inverters at home.
I also have plenty of said optocouplers (MOC3063S) and TRIACs (Z0103MN), as well as having soldered this circuit myself already several times, as well as having ordered replicas from China.
The problem with these specific parts seems to be that the TRIACs will easily fail, as they seem to be specified for an AC frequency of 60 Hz, a fact that I only learnt after many unsuccessful tries of making this circuit work in the long run, and actually looking up the TRIAC’s specification myself.

So I am looking for different components that are specifically designed to operate under these conditions, or at least have a reasonably high chance of tolerating them.
An AC frequency of 2000 Hz is very unusual, to the point that few components seem to even consider this in their specification. This is the reason I am asking about this here.

1 Like
9

Hello Thokari,

Excellent!

We need to know the failure mechanism. Possibilities include:

  • excessive power dissipation
  • an intermittent or sustained over current
  • excessive speed dV/dt or dI/dt

Each failure mode would suggest a different solution.

  • excessive power dissipation – inspect board for signs of heat. Run the board, turn off power, and see what is hot
  • an intermittent or sustained over current – track channel failures to see if there are problematic EL strings
  • excessive speed (dV/dt or dI/dt) – substitute parts designed for high switching speeds – which in this case implies higher voltage of the ability to handle inductive loads. This higher voltage triac may be a solution. Please double check the pin out before purchase.

Finally, guard against counterfeit products. This design appears to push semiconductors to the edge of their performance. Inferior substitutions are not welcome.

Again, sorry I couldn’t give you a fast or easy solution.

Best Wishes,

APDahlen

10

Hi Thokari,

I looked through specs of various dual-MOSFET output SSR’s to see if any of them might be an option, and the vast majority are too slow. However, it looks like the AQY210S might be a possible solution for you to consider.

image
image997×203 11.3 KB

Its worst-case turn-on time is 0.5ms, with a typical time of about 0.25ms. This puts it in the borderline range for your application. Turn-off time is faster, with maximum time of 0.2ms, so that’s less of an issue. It’s rated for up to 350Vac (peak) and 120mA (peak). It’s far too slow to manage zero-cross turn-on and turn-off for your 2kHz signal, so that may still cause some issues. Perhaps applying a snubber circuit might help with that.

image

11

Hi @David_1528,
thank you for your suggestion.
I have a question about the caveat you are mentioning, regarding zero crossing.
As I understand, this is about protecting the SSR from voltage spikes.

EL wire, can be modelled as a capacitor/resistor in parallel, with 6 nF and 150 kOhms per meter.
Now my question is, how much wire could I connect safely, without causing a voltage spike that would exceed the 350 V rating of this part, when the SSR happens to switch off at peak voltage?
Does that question even make sense? I am not an electrical engineer, merely a hobbyist, and it is a long time ago, that I dealt with anything like this, so any help is appreciated.

12

Hi Thokari,

One thing you could try is to use the MOC3063, that AP mentioned, and plug the EL wire directly to it, i.e. having no discrete TRIAC. The MOC can handle 100mArms, thus would suit to your specs and is most probably fast enough.

E.g. Vishay seems to offer even beefier opto-triacs that have turn-on times in tens of microseconds.

Some sort of snubber circuit may indeed be needed, although being zero-crossing circuit, otherwise local radio amateurs may pay a friendly visit.

Cheers, heke

2 Likes
13

Hi Thokari,

I was assuming it would be more of an inductive load rather than capacitive. Under those circumstances, switching off is not as much of an issue as is switching on, and the issue is more likely a current spike rather than a voltage spike in that case.

Due to the rise time of the switch being no shorter than about 50us, I don’t think it would have a huge current spike - probably in the 20mA - 30mA range.

However, I do like @heke’s find of the IL410-X007T. It handles 300mA and 600Vac. The only question is whether the zero-cross function would work at that frequency. Looks like a promising option.

14

Yes, I guess @David_1528’s concern of the functionality of the zero-crossing functionality is fair. The datasheet is not very vocal about that. The dV/dt rate at the zero crossing for pure sine wave is roughly 2.1V/us (120VACrms, 2000Hz). The datasheet declares that the “Critical rate of rise of voltage at current commutation” is 7-8V/us, thus not that far. As could perhaps assume that the target application is not mission critical, such narrow margin may be tolerable.

Then there is that HV850 chip, that can drive EL wire. It can be pulsemodulated with a digital signal.

Cheers, heke

15

I wish this info had come at the very start of the thread when I had some free time. With a circuit that simple it shouldn’t take that many hours of math and analysis to rule in, or out, a design flaw as the problem. With the math worked out if it’s a design flaw the math will determine if there is viable solution to the problem with available components.

1 Like
16

@PaulHutch I didn’t want to mention the part I had tried in the past, I only did because confusion about the requirements came up. I wanted an unbiased answer to the problem purely based on the stated requirements. I had the fear that the conversation would derail, and people misunderstading me, because that has happened in the past.

2 Likes
17

@heke All EL-Lamp Driver-ICs offered on Digikey have at most 1 khz, but that is just barely enough for for EL wire, which needs upwards of that frequency. I have tried one of those, and it only lit up the EL wire to about half brightness. They are designed for “EL panels” which require higher voltage, but lower frequency.
Another issue with these Driver ICs is, that they don’t respond quickly enough. I takes multiple passes through the inductor coil, that has to be attached, and this decreases the response time of the circuit.

1 Like
18

HI @Thokari

Thank you for providing very insightful information on the subject.

So, it sounds like that maybe the best option is to go with some zero-crossing TRIAC based solution, or having some more exotic stuff, like floating IGBT inside a full bridge rectifier.

Would guess that the cleanest solution is to solve the switching problem right where the AC voltage is generated, not after. That would mean e.g. a high voltage DC source that is steered to EL wire with a full H-bridge configuration (or a half-bridge and split capacitor virtual ground). The bridge timing would come from an MCU or such. Gets complicated…

Cheers, heke