Transformer as MOSFET relay driver

Thread Starter

hrs

Joined Jun 13, 2014
394
Hi,

I've made an audio amplifier that pops 1 or 2 seconds after power off, probably due to DC off-set or something like that. Now I want to make a MOSFET solid state relay based on this article by Rod Elliott, specifically the version with a transformer in it.

He suggests using a 500 μH inductance. I don't understand the importance of that other than that it should be too high or it will have a high impedance at high frequencies. Anyway, I tried to do the numbers and came up with the following, hoping to get away with 14 * 7 * 5 mm ferrite toroids because they're cheap and small:
outer diameter: 14 mm
inner diameter: 7 mm
height: 5 mm
relative inductance: 200 (assumption)
square wave frequency: 100 kHz at 15 Volts peak to peak
primary turns: 20, maybe 21 of AWG25
turn ratio: 1:1
ΔB = E / (4 * n1 * A_c * frq) = 0.1 T (I read somewhere[1] that 0.1 T is typical for ferrite)
L = μ * window_area * n1**2 / magnetic_path_length = 60 μH
Do you think this would work to drive a MOSFET? How do you calculate how much current it can handle?

About the relay, it has "clamping diodes" D4 and D5 that seem to do nothing. Is it that fast switching of the relay might cause high voltage spikes and that the diodes clamp it to either rail voltage?


[1] http://ecee.colorado.edu/copec/book/slides/Ch15slides.pdf
 

Attachments

ronv

Joined Nov 12, 2008
3,770
Hi,

I've made an audio amplifier that pops 1 or 2 seconds after power off, probably due to DC off-set or something like that. Now I want to make a MOSFET solid state relay based on this article by Rod Elliott, specifically the version with a transformer in it.

He suggests using a 500 μH inductance. I don't understand the importance of that other than that it should be too high or it will have a high impedance at high frequencies. Anyway, I tried to do the numbers and came up with the following, hoping to get away with 14 * 7 * 5 mm ferrite toroids because they're cheap and small:
outer diameter: 14 mm
inner diameter: 7 mm
height: 5 mm
relative inductance: 200 (assumption)
square wave frequency: 100 kHz at 15 Volts peak to peak
primary turns: 20, maybe 21 of AWG25
turn ratio: 1:1
ΔB = E / (4 * n1 * A_c * frq) = 0.1 T (I read somewhere[1] that 0.1 T is typical for ferrite)
L = μ * window_area * n1**2 / magnetic_path_length = 60 μH
Do you think this would work to drive a MOSFET? How do you calculate how much current it can handle?

About the relay, it has "clamping diodes" D4 and D5 that seem to do nothing. Is it that fast switching of the relay might cause high voltage spikes and that the diodes clamp it to either rail voltage?


[1] http://ecee.colorado.edu/copec/book/slides/Ch15slides.pdf
I never have been able to get the calculations to come out right, so I would use the circuit based on capacitors. :D
Having said that I would think it would come out close to 500Uh just by looking.
I think the diodes protect against the inductance of the speaker.
 

Thread Starter

hrs

Joined Jun 13, 2014
394
The article recommends against the capacitor based solution because amplifier oscillation might interfere with it. I considered the opto-isolator solution until I looked at the IC prices.

Maybe I'll just buy some wire and toroids and see what I get. But what is the significance of the 500 μH value?
 

ScottWang

Joined Aug 23, 2012
7,397
If the circuit put a zener at there, then the Q2 already lost it's function, because when the Vgs only provide 0.7V then the Q2 will never turn on.

This kind of circuit trying to use the body diode of mosfet to flows the current when the mosfet in the turn off situation.
 

Thread Starter

hrs

Joined Jun 13, 2014
394
There is no turn-on thump, only turn-off thump. The current goal is to disconnect the speaker from the amplifier as I switch it off though a turn-on delay is probably nice to have. I made a provision on the power supply before the filter capacitors so I can detect switch off immediately, the thump follows a second later.

There is discussion on the web about 'muting circuits' and 'speaker protection' and it's ambiguous to me. I would guess that speaker protection also mutes but muting does not protect speakers when the amplifier has a DC fault though this distinction is never made in any of the material that I've read.

Do you want to switch the speaker on/ off when the amp is powered up, whats wrong with a de_thump relay circuit?
For one I've read that it's hard for a conventional relay to break a DC fault at high-ish voltages. Additionally I thought it might be an interesting exercise to make solid state relay.

What type of circuit would you suggest?
 

Thread Starter

hrs

Joined Jun 13, 2014
394
If the circuit put a zener at there, then the Q2 already lost it's function, because when the Vgs only provide 0.7V then the Q2 will never turn on.
I don't understand :) I would expect that a voltage of 12V will develope over the zener but not higher. The function of the zener would be to prevent Vgs getting too high. But maybe I don't understand how a zener actually works.
 

Dodgydave

Joined Jun 22, 2012
11,285
I made one using a simply changeover relay 12v dc fed from a 555 timer circuit, the relay contacts would swap the speaker for a resistor of same value as the speaker,

the relay common goes to the amp output,
the nc to the resistor, the no to the speaker, so on power up the resistor is connected then after a second or two the speaker is connected, and disconnected on power down.
 

ScottWang

Joined Aug 23, 2012
7,397
I don't understand :) I would expect that a voltage of 12V will develope over the zener but not higher. The function of the zener would be to prevent Vgs getting too high. But maybe I don't understand how a zener actually works.
Sorry, it's my fault, Q2 and zener are ok.
My right eye just turning to the white ball from the red ball, but it is still not complete heal ... :(
 

Thread Starter

hrs

Joined Jun 13, 2014
394
No problem and good luck with your eye!

I decided to simulate it in LTSpice and see what happens. It appears to work but changing the inductance value doesn't seem to make much difference.
 

Thread Starter

hrs

Joined Jun 13, 2014
394
I think it is just a value that will keep the current in the timer from being to high. The impedance of the coil is directly related to the inductance.
Ah, so it's deliberately high in order to introduce impedance. So if the current draw turns out to be to high I could add some turns or possibly a current limiting resistor.
 

ronv

Joined Nov 12, 2008
3,770
The article recommends against the capacitor based solution because amplifier oscillation might interfere with it. I considered the opto-isolator solution until I looked at the IC prices.

Maybe I'll just buy some wire and toroids and see what I get. But what is the significance of the 500 μH value?
these are kind of cute. At $8 maybe about the same as a DIY one.
http://www.mouser.com/ds/2/281/kdc_nme-34022.pdf
 

ScottWang

Joined Aug 23, 2012
7,397
No problem and good luck with your eye!

I decided to simulate it in LTSpice and see what happens. It appears to work but changing the inductance value doesn't seem to make much difference.
Thank you and don't forget to attach the asc file.
 

Thread Starter

hrs

Joined Jun 13, 2014
394

Attachments

Thread Starter

hrs

Joined Jun 13, 2014
394
After extensively failing I finally managed to measure the inductance of a slightly larger toroid than proposed in the opening post at ~350 μH. This was done by looking at the input/output voltage ratio of an LR filter with a ~500 kHz square wave applied to it. The formula used[1] is for a sine wave but the order of magnitude should be ok.

The process brought up some questions.
I have been unable to see a phase shift in voltage across the resistor with respect to the voltage across the inductor when excited with a square wave. Is this to be expected or did I do it wrong?

I tried to see the phase shift on a small transformer that I took from an old alarm clock. The reasoning was that the inductance of that transformer would be so high that phase shift would be in reach of my audio frequency sine wave generator. And indeed there is some phase shift but not that much at 50Hz. Which brings on the next question: at what reactance(?) would you typically operate a transformer? Is this determined by the load impedance and required current?

[1] http://www.dos4ever.com/inductor/inductor.html
 

ScottWang

Joined Aug 23, 2012
7,397
square wave frequency: 100 kHz at 15 Volts peak to peak
It should be 0V to peak.
Why it used 0V to 15V, can it be change?
The original voltage and waveform.

ssr-hrs_waveform and circuit.gif


Since the saturation voltage of Vg=10V, so I changed the input voltage from 15V to 9V and the R1=1.2K, the left side of R1 around 13.5V and the output voltage amplitude of sine wave is the same.

ssr_hrs_9Vi_ScottWang.gif


When I changed the input voltage from 15V to 6V and the R1=100Ω, the left side of R1 around 9V, the D1 zener diode can be take away, but I still kept it and replaced with 10V and the output voltage amplitude of sine wave still the same.

ssr_hrs_6Vi_ScottWang.gif
 

Attachments

Top