Compact 300V 300mA high side switch with PNP

[chipeater]

Hi guys,
 
Quick question. I have to imagine a “high side switch” to switch a kind of big power relay.
The particularity of this relay is that you have to apply a very high voltage in the order of 250-300V for 20ms and then maintain it with a voltage of 24V ( I don't have the ref ).
 
I need something as compact as possible, and I don't have a 10-20VDC rail for the bootstrap power supply of a Mosfet driver.
It's a microcontroller that activates the coil excitation, so I need isolation between the MCU and the high voltage.
 
The load is symbolized by resistor R1 ( 300mA ), resistors R2 and R3 are 400V resistors and R3 is a 3W resistor:



I need to do more or less the same thing for the 24V rail, but without the high-voltage issue, and couple the two collectors together via diodes.

Question:

Do you have any criticism on this schematic ?
Can you think of a more compact, more elegant, or better solution?

Thank you very much 

[magic]

A more compact solution would use an NPN or N-MOSFET to switch R3. But failure of such transistor may expose the MCU to high voltage, while an optocoupler provides better protection. The optocoupler must be rated to withstand 300V on the secondary, not all of them can do it, I hope you chose a suitable one.

R2 is probably not really needed, and it doesn't need to withstand high voltage - it is paralleled by the BE diode of the PNP, so it will see less than 1V unless the PNP is disconnected or blown.

[moffy]

Don't forget to put a flyback/spike suppression diode across the relay coil.
If you use a P-MOSFET you can use a zener to limit the gate voltage and reduce the current needed to switch the high side switch.

[Doctorandus_P]

With 300V power supply you have to be careful with all the little details. PS2535 can withstand up to 350V on the Collector - Emmiter side, but how can you be sure your power supply never has peaks over that? FZT857 is also only rated to 300V, which is likely too low for your application.

R3 is going to dissipate 3W and will surely overheat if you use a small resistor. Apart from power dissipation, resistors also have a maximum voltage rating and that is something you have to watch out for too in a 300V application.



https://www.renesas.com/us/en/document/dst/ps2535-1-ps2535l-1-ds
https://www.diodes.com/assets/Datasheets/FZT857.pdf

[mawyatt]

Why not use the Opto-Coupler with a HV NPN for and Opto-Coupled Darlington.

Best,

[chipeater]

With 300V power supply you have to be careful with all the little details. PS2535 can withstand up to 350V on the Collector - Emmiter side, but how can you be sure your power supply never has peaks over that? FZT857 is also only rated to 300V, which is likely too low for your application.

R3 is going to dissipate 3W and will surely overheat if you use a small resistor. Apart from power dissipation, resistors also have a maximum voltage rating and that is something you have to watch out for too in a 300V application.



https://www.renesas.com/us/en/document/dst/ps2535-1-ps2535l-1-ds
https://www.diodes.com/assets/Datasheets/FZT857.pdf

The PNP is a FZT758 so its 400V.
R3 is a 400V resistor and switching the coil is just a 20ms operation so I'm not too worried.
I'm actually much more worried about the transiant voltage, but I haven't found an optocoupler capable of supporting more than 350V, and TVS have a breakdown voltage that's far too wide to be used with a reasonable margin. I'm not sure what to do.

[chipeater]

Why not use the Opto-Coupler with a HV NPN for and Opto-Coupled Darlington.

Best,

That is very nice 
What is the purpose of R2? Is it to have a known potential on the base of Q1 when the phototransistor is not conducting?
This avoids the need for a large HV and high power resistor.

[mawyatt]

Why not use the Opto-Coupler with a HV NPN for and Opto-Coupled Darlington.

Best,

That is very nice 
What is the purpose of R2? Is it to have a known potential on the base of Q1 when the phototransistor is not conducting?
This avoids the need for a large HV and high power resistor.

R2 just provides a path for leakage currents from the Opto-Coupler NPN transistor and the Collector-Base leakage of Q1 when OFF, not critical 10K to 100K should be fine.

Yes, R3 is eliminated.

Best

[voltsandjolts]

^ Note that when OFF there could be 300V across the phototransistor, so be sure to use a part rated for 350 or 400V.

https://www.mouser.co.uk/c/optoelectronics/optocouplers-photocouplers/transistor-output-optocouplers/?maximum%20collector%20emitter%20voltage=350%20V~~400%20V&rp=optoelectronics%2Foptocouplers-photocouplers%2Ftransistor-output-optocouplers%7C~Maximum%20Collector%20Emitter%20Voltage

[profdc9]

You might want to consider a Darlington configuration so the optoisolator doesn't have to switch as much current, and the off and on states are in cutoff/saturated as possible.

[MasterT]

I'd prefer initial circuits, as more reliable in case of capacitive load or overcurrent . Darlingtons usually are slow, and consequently load current likely to pass via optocoupler, that was not design to dissipate much power

[mawyatt]

^ Note that when OFF there could be 300V across the phototransistor, so be sure to use a part rated for 350 or 400V.

https://www.mouser.co.uk/c/optoelectronics/optocouplers-photocouplers/transistor-output-optocouplers/?maximum%20collector%20emitter%20voltage=350%20V~~400%20V&rp=optoelectronics%2Foptocouplers-photocouplers%2Ftransistor-output-optocouplers%7C~Maximum%20Collector%20Emitter%20Voltage

That's no different than the original PNP based circuit shown by OP, both must sustain the supply voltage when off.

Best,

[mawyatt]

You might want to consider a Darlington configuration so the optoisolator doesn't have to switch as much current, and the off and on states are in cutoff/saturated as possible.

See #4. The NPN version the Opto-Isolator only switches the base current of Q1 plus the small amount of Vbe/R2.

Best,

[David Hess]

I'm actually much more worried about the transiant voltage, but I haven't found an optocoupler capable of supporting more than 350V, and TVS have a breakdown voltage that's far too wide to be used with a reasonable margin.

For the PNP design, a high voltage TVS or zener could be placed in series with the optocoupler to reduce the voltage across the optocoupler.  A high value resistor then placed across the optocoupler sinks the leakage through the TVS so the voltage across the optocoupler is now the supply voltage minus the voltage across the TVS.

Several TVSs in series could be placed across the optocoupler to protect it from voltage surges.

A high voltage NPN or N-channel MOSFET cascode transistor could be added in series with the optocoupler.  This would allow a low voltage optocoupler to be used.  It also allows the optocoupler output to be current limited.

I like the last option however biasing of the cascode transistor could be tricky.

Alternatively, a photovoltaic optocoupler like the TLP591B can drive the base-emitter or gate-source and never see the high voltage on the collector or drain.  It will supply 10 volts or 50 microamps.  Switching time for a huge MOSFET is still just 1 or 2 milliseconds, so much faster in your application.

[mawyatt]

I'd prefer initial circuits, as more reliable in case of capacitive load or overcurrent . Darlingtons usually are slow, and consequently load current likely to pass via optocoupler, that was not design to dissipate much power

The Darlington Transistor configuration can be quite fast, Avantek (later acquire by HP) introduced RF Darlingtions ~4 decades ago, later Mini-Circuits also offered these.

Here's an old article discussing RF Darlington use.
https://www.arrl.org/files/file/Technology/microwave/8702023.pdf

The Opto-Coupler Darlington shown in #4 the Opto-Coupler's output transistor's collector current is limited to Q1's base current, so Q1 will likely give up the magic smoke before the Opto-Coupler goes, in which case they'll both be toast!!

If one is concerned about a capacitive load charge current then some form of current limit needs to be added, be it a emitter and/or collector resistor, or active current limit with emitter sense resistor and NPN. Here's an example.   

Best,

[PCB.Wiz]

Hi guys,
 
Quick question. I have to imagine a “high side switch” to switch a kind of big power relay.
The particularity of this relay is that you have to apply a very high voltage in the order of 250-300V for 20ms and then maintain it with a voltage of 24V ( I don't have the ref ).
 
I need something as compact as possible, and I don't have a 10-20VDC rail for the bootstrap power supply of a Mosfet driver.
It's a microcontroller that activates the coil excitation, so I need isolation between the MCU and the high voltage.

How important is idle power ?

If the 250-300V is somewhat stable, you can reduce the voltage stress on the opto with a zener, either series or a shunt regulator.

You can get 300+V PFET, which have lower gate energy needs.

You did not show any 20ms timing parts, is that done in the MCU with two pins ? 

If not, you might want to add a RC for the 20ms and maybe a schmitt, to keep things fast on the edges.
With no schmitt, it can be simpler but you need to check the slow turn off SOAR of the HV switching device.

If this has wires to the 'big relay' you might consider some current limit on the HV switching part too, so it does not instantly fail on fault.

[MasterT]

I'd prefer initial circuits, as more reliable in case of capacitive load or overcurrent . Darlingtons usually are slow, and consequently load current likely to pass via optocoupler, that was not design to dissipate much power

The Opto-Coupler Darlington shown in #4 the Opto-Coupler's output transistor's collector current is limited to Q1's base current, so Q1 will likely give up the magic smoke before the Opto-Coupler goes, in which case they'll both be toast!!

If one is concerned about a capacitive load charge current then some form of current limit needs to be added, be it a emitter and/or collector resistor, or active current limit with emitter sense resistor and NPN. Here's an example.   

Best,

Not really, BJT can pass ~1A base current before it's turn on, optocoupler in this time frame would dissipate 300 Wt.
Your second circuits is a good example of what should not be done, current limiting transistor would pass all transition current via optocoupler, guaranteed failure on first try.

[mawyatt]

I'd prefer initial circuits, as more reliable in case of capacitive load or overcurrent . Darlingtons usually are slow, and consequently load current likely to pass via optocoupler, that was not design to dissipate much power

The Opto-Coupler Darlington shown in #4 the Opto-Coupler's output transistor's collector current is limited to Q1's base current, so Q1 will likely give up the magic smoke before the Opto-Coupler goes, in which case they'll both be toast!!

If one is concerned about a capacitive load charge current then some form of current limit needs to be added, be it a emitter and/or collector resistor, or active current limit with emitter sense resistor and NPN. Here's an example.   

Best,

Not really, BJT can pass ~1A base current before it's turn on, optocoupler in this time frame would dissipate 300 Wt.
Your second circuits is a good example of what should not be done, current limiting transistor would pass all transition current via optocoupler, guaranteed failure on first try.

Care to make a serious $ wager on that

Best revisit your circuit analysis and semiconductor physics before making unqualified statements like that

Carefully analyze the circuit before jumping to unqualified statements, note the added two resistors, might save you some serious $ in case you accept the wager

Best,
 

[nctnico]

How about using a MOSFET output opto-coupler? These can be had with reasonable high current and high voltage handling abilities. Toshiba has quite a wide range of these.

[mawyatt]

How about using a MOSFET output opto-coupler? These can be had with reasonable high current and high voltage handling abilities. Toshiba has quite a wide range of these.

That would work and be simpler, any recommendations (are they expensive)?

Best,

[PCB.Wiz]

The PNP is a FZT758 so its 400V.

That has poor gain at 300mA, so you need a 'better' part.

Since this is a ~20ms pulse you can use a N-MOS fet with a bootstrap gate drive, that a > 20V opto coupler can manage.
You can drive that opto either from a timed MCU pulse, or a RC + NPN + emitter resistor.
If using a RC drive, the opto gain has some influence on the pulse, so you would pick a 'better' gain-span defined part.

[mawyatt]

If the OP chooses the NPN version shown in #4 and #14 there's a number of available low cost HV NPNs.

Here's a quickly done LTspice simulation of the NPN version mentioned with Current Limit. Since we had no HV NPN model for ~300ma, we just paralleled 4 2SCR346P LTspice models. For the Opto-Coupler we just used the LTspice model for the popular PC817A since it doesn't include a breakdown parameter for the output transistor.

Note the current from the Opto-Coupler, displacing the "myth" mentioned above in #16 that's it's going to "blow up" due to the excessive 1A base current required by the output pass NPN at Turn On

Not really, BJT can pass ~1A base current before it's turn on, optocoupler in this time frame would dissipate 300 Wt.
Your second circuits is a good example of what should not be done, current limiting transistor would pass all transition current via optocoupler, guaranteed failure on first try.

Plots are from no Load Capacitance, Rload = 1.5K, then Rload = 1K, then Rload = 1.5K & Cload =1uF, then 10uF then lastly 100uF.

Best,

[nctnico]

How about using a MOSFET output opto-coupler? These can be had with reasonable high current and high voltage handling abilities. Toshiba has quite a wide range of these.

That would work and be simpler, any recommendations (are they expensive)?

As I wrote, Toshiba has many different types. But these are kind of expensive. In modern oscilloscopes these opto-MOSFETs are often used to switch between DC and AC coupling.

[mawyatt]

How about using a MOSFET output opto-coupler? These can be had with reasonable high current and high voltage handling abilities. Toshiba has quite a wide range of these.

That would work and be simpler, any recommendations (are they expensive)?

As I wrote, Toshiba has many different types. But these are kind of expensive. In modern oscilloscopes these opto-MOSFETs are often used to switch between DC and AC coupling.

Interesting devices, however very slow (~1ms). Not surprising considering the semiconductor mechanisms employed, but might be highly useful where switching speed isn't required.

Best,

[chipeater]

Hello, thank you everyone and thank you mawyatt.

I will try to optimize with other NPN. I need to make compact but I also like to take some margin.
Are there any particular parameters to look at for the darlington configuration to work with the optocoupler?

I don't think it's possible to be as compact with a N Mosfet solution. You need a driver with a 10-20VDC power supply for the bootstrap power supply + HV bootstrap capacitor.

Just to switch the coil of a relay, I think that's excessive.