Over current protection devices question

[Chris Wilson]

With devices like this, limited to a max of 36V how do you apply the technology to say an RF amp running on 60v or even rectified 240V mains? And secondly, with the amp powered, when a signal is applied, let's say the digital amateur mode, WSPR commences, there is a high current for a short time. how do you allow for this yet provide effective over current protection for rapidly failing devices like FET's? Thanks

http://www.ti.com/product/INA300

[PartialDischarge]

With devices like this, limited to a max of 36V how do you apply the technology to say an RF amp running on 60v or even rectified 240V mains?

That device would be useful only in a low side application if your Vcc goes beyond 36V. You can easily build your own high side with the appropriate components using for example the circuit on page 11 from http://www.ti.com/lit/an/snoa620/snoa620.pdf , there are other circuits out there

And secondly, with the amp powered, when a signal is applied, let's say the digital amateur mode, WSPR commences, there is a high current for a short time. how do you allow for this yet provide effective over current protection for rapidly failing devices like FET's? Thanks

You can't have it both ways. Best you can do is filter the sensed current and allow for a peak in current during a certain amount of time. Anyway semiconductors are much faster in failing that you'd be in protecting them so a small delay would be ok anyway.

[dmills]

You use a high side current sensing arrangement, either a resistor and INA-Something or something from the likes of LEM, then a comparator, D flipflop and a small transistor to yank the RF sands bias supply or trip the power supply offline. 


I would note that there is a nasty trap if you yank the drain voltage on an amp built with significant amounts of RF feedback, the loss of feedback makes the gate voltage swing increase and risks puncturing the gates.

Personally I like a PIN diode switch in the drive circuitry in combination with killing the bias voltage, gets it shut down in a few microseconds and transient protection diodes can take the 10W or so of drive for that long.

Regards, Dan.

[Chris Wilson]

You use a high side current sensing arrangement, either a resistor and INA-Something or something from the likes of LEM, then a comparator, D flipflop and a small transistor to yank the RF sands bias supply or trip the power supply offline. 


I would note that there is a nasty trap if you yank the drain voltage on an amp built with significant amounts of RF feedback, the loss of feedback makes the gate voltage swing increase and risks puncturing the gates.

Personally I like a PIN diode switch in the drive circuitry in combination with killing the bias voltage, gets it shut down in a few microseconds and transient protection diodes can take the 10W or so of drive for that long.

Regards, Dan.

Thanks Dan, problems are it's Class D so no bias supply, and the huge caps in the supply mean I would have to trip after them, not a big problem save aesthetics modding the PCB. I imagine a relay is too slow, what device will switch off 60V or more at 40 Amps plus, very fast?

[dmills]

So just yank the gate drive (An and gate between the gate pulse generator and the mosfet driver for each fet)?

A mosfet makes a decent switch if you really wanted to, but you usually get into needing a little boost supply charge pump to keep the gate where it needs to be, or use a P Channel part.

Relays are actually usually fine providing your power supply has a current limited output, less so if you have a honking great cap waiting to vaporise the bond wires, they are of course way too slow to catch drain or gate over voltage.

Regards, Dan.

[Chris Wilson]

So just yank the gate drive (An and gate between the gate pulse generator and the mosfet driver for each fet)?

A mosfet makes a decent switch if you really wanted to, but you usually get into needing a little boost supply charge pump to keep the gate where it needs to be, or use a P Channel part.

Relays are actually usually fine providing your power supply has a current limited output, less so if you have a honking great cap waiting to vaporise the bond wires, they are of course way too slow to catch drain or gate over voltage.

Regards, Dan.

OK, understood, thanks again Dan for the great reply

[coppercone2]

How do you deal with the RF feedback path? Short it out to ground with a PIN diode that the pSU controls?

[dmills]

Clamp the gates (which is where the RF feedback can be a problem), not an issue obviously in class E.

For a standard sort of class A/B affair I like a turn on the DC injection bifiliar to sense the drain RF voltage, and a set of diodes and caps to sample the gate voltage, feed these signals to some comparators and have them latch a shutdown signal if drain voltage, drain current, temperature or gate voltage become excessive.

Shutdown switches off the bias, and switches on the current to a PIN diode that removes the drive power, you can do this within a ~10us easily.

For the time it takes the protection to react, that is what transient diodes and clamps are for, it does not take much to survive a kW for 10us or so.

Regards, Dan.

[Chris Wilson]

For completion this is the amp I would like to perhaps add over current protection to. Thanks again.

[dmills]

IR2110 has a perfectly good shutdown input, just drive it high from a flipflop when an error condition is detected, easy.
I would add a turn to the drain transformer to sense over voltage due to load mismatch of the inductive sort, but whatever works for you.

Regards, Dan.

[T3sl4co1l]

That's your old circuit with the 136kHz and the D flip-flop and gate drivers and IRFP260s (or whatever they were), right?

It's pretty basic as switching supplies go, to add current limiting to that circuit.  Example, limiting peak current by turning off the transistor (until it is next requested to turn on).  This adds a current sense resistor, comparator and latch.

This also neatly solves the problem of exploding finals when subject to bad load impedances (high SWR).  The current limit can be chosen to have no effect during normal (SWR within a certain limit) operation, or chosen to operate under certain conditions so that output power is limited (which may or may not introduce distortion, IM3 or such, depending on how exactly it's done).

Tim