2 added 95 characters in body
source | link

Use opto isolator on the PWM pin of the IR2184

Note that this answerthis answer was made prior to the OP showing a picture of the magnetic coupler he proposesproposed (thisnote this is not an opto coupler despite the question explicitly stating "opto-isolated" in the title).

PWM operating frequency is going to kick you in the teeth unless you are running sub 10 kHz or you can find a really good opto-isolator that has fast rise and falls times.

A hand waving typical rise and fall time for an opto might be about 5 us and this will be into a quite low load resistance of about 100 ohm. Into 1 kohm it could rise to 20 us or more. Now, if it takes 20 us to deliver a PWM "edge" then you don't want to be switching too often or the rise/fall time inefficiences will be warming things up. So maybe 20 us represents 5% of the period of your PWM.

If you work that out you'd choose a PWM frequency of 2.5 kHz because 2.5 kHz has a period of 400 us and 5% of that is 20 us. But it's a little worse because there are two edges to consider in each PWM cycle so, you'd choose 1.25 kHz as your PWM frequency.

Regards the ADuM3223 - it is a good device and has a rise/fall time less than 100 ns!

Use opto isolator on the PWM pin of the IR2184

Note that this answer was made prior to the OP showing a picture of the magnetic coupler he proposes (this is not an opto coupler despite the question explicitly stating "opto-isolated" in the title).

PWM operating frequency is going to kick you in the teeth unless you are running sub 10 kHz or you can find a really good opto-isolator that has fast rise and falls times.

A hand waving typical rise and fall time for an opto might be about 5 us and this will be into a quite low load resistance of about 100 ohm. Into 1 kohm it could rise to 20 us or more. Now, if it takes 20 us to deliver a PWM "edge" then you don't want to be switching too often or the rise/fall time inefficiences will be warming things up. So maybe 20 us represents 5% of the period of your PWM.

If you work that out you'd choose a PWM frequency of 2.5 kHz because 2.5 kHz has a period of 400 us and 5% of that is 20 us. But it's a little worse because there are two edges to consider in each PWM cycle so, you'd choose 1.25 kHz as your PWM frequency.

Use opto isolator on the PWM pin of the IR2184

Note that this answer was made prior to the OP showing a picture of the magnetic coupler proposed (note this is not an opto coupler despite the question explicitly stating "opto-isolated" in the title).

PWM operating frequency is going to kick you in the teeth unless you are running sub 10 kHz or you can find a really good opto-isolator that has fast rise and falls times.

A hand waving typical rise and fall time for an opto might be about 5 us and this will be into a quite low load resistance of about 100 ohm. Into 1 kohm it could rise to 20 us or more. Now, if it takes 20 us to deliver a PWM "edge" then you don't want to be switching too often or the rise/fall time inefficiences will be warming things up. So maybe 20 us represents 5% of the period of your PWM.

If you work that out you'd choose a PWM frequency of 2.5 kHz because 2.5 kHz has a period of 400 us and 5% of that is 20 us. But it's a little worse because there are two edges to consider in each PWM cycle so, you'd choose 1.25 kHz as your PWM frequency.

Regards the ADuM3223 - it is a good device and has a rise/fall time less than 100 ns!

1
source | link

Use opto isolator on the PWM pin of the IR2184

Note that this answer was made prior to the OP showing a picture of the magnetic coupler he proposes (this is not an opto coupler despite the question explicitly stating "opto-isolated" in the title).

PWM operating frequency is going to kick you in the teeth unless you are running sub 10 kHz or you can find a really good opto-isolator that has fast rise and falls times.

A hand waving typical rise and fall time for an opto might be about 5 us and this will be into a quite low load resistance of about 100 ohm. Into 1 kohm it could rise to 20 us or more. Now, if it takes 20 us to deliver a PWM "edge" then you don't want to be switching too often or the rise/fall time inefficiences will be warming things up. So maybe 20 us represents 5% of the period of your PWM.

If you work that out you'd choose a PWM frequency of 2.5 kHz because 2.5 kHz has a period of 400 us and 5% of that is 20 us. But it's a little worse because there are two edges to consider in each PWM cycle so, you'd choose 1.25 kHz as your PWM frequency.