Overload / Short Circuit Protection, Yet Again...

[Buckeye]

I'm working on a real basic BJT amp design, but want to incorporate short-circuit protection. So I've been doing some reading. In his article "Testing Amplifiers To Their Limits", Phil Allison gives an outstanding overview of VI limiting and some if its limitations (pun intended). (VI Limiters in Amplifiers)

The main problem is that to protect against shorts, you have to set the VI limit too low. Reason is from the SOA graph (see File:BDV66C limits.png - Wikipedia, the free encyclopedia) and the low current allowed at high Vce voltages. Note that a nominal Ic and a huge Vce is exactly what could happen during a short. And when you have a normal speaker load, high currents are allowed because Vce is small. So to be effective under short-circuit conditions, the conventional VI limiter limits current even when it shouldn't. It's a compromise solution, and maybe not a very good one.

I was searching the web and found this 1982 patent by Kaplan, Power Protection Circuit For Transistors. I think he's on the right track. His invention measures Vce to limit the current. It offers a better approximation to the SOA in the secondary breakdown region. But not many implement his approach because it's complex and costly.

So here's an idea... how's about we measure the Vce indirectly by measuring Vout and using that to gate the conventional VI limiter? I call this the Output-Gated VI Limiter (OGVIL). Here's the drawing of just the positive half of the typical BJT AB amp:

GatedLimiter.jpg

So you can see from the above, when a short is placed between output and ground, Q4 is off and the conventional VI limiter operates as usual. We are now free to set the current limit as low as needed to protect the transistor from its secondary breakdown. However if no short exists, Vout rises with the output signal, and Q4 turns on effectively reducing the gain of the limiting transistor Q3.

Some additional notes on the above circuit... Depending on your drive circuit, a diode may be needed in series with the collector of Q4 to prevent forward bias of its collector-base junction. Most likely the value of R4 will be zero.

I've done a bit of searching and haven't seen anyone try this approach for short-circuit protection. (There may be good reason for that .) Renardson did something like this in his post here: Overload and Short Circuit Protection.. He used a resistor (R1) from base of the limiter transistor, Q1, to ground. Q1 in his circuit is Q3 in my circuit. By using a transistor to ground, we're able to totally disable current limiting when Vce is low, i.e., Vout is high. We can even go wild and implement a log function in the current limiter to more closely match the SOA curve near the secondary breakdown region.

What do you think?

[audiohead]

Diode D1 in your circuit is drawn backward?

Art

[Buckeye]

oops, yes it is. thanks

[ontoaba]

Good idea..
Add capacitor too. With slow charge and fast discharge.
Then thermal sense, like NTC or transistors.

Actually I never use any SOA for myy amplifiers

[AndrewT]

Originally Posted by ontoaba

Actually I never use any SOA for myy amplifiers

your amplifiers should be using the SOA all the time they are operating.
It's when a device strays outside the SOA that you should be concerned about.

We don't care if you ignore SOA. WE know that SOA matters to the devices. They have to survive what we ask them to do.

[Buckeye]

Originally Posted by ontoaba

Actually I never use any SOA for myy amplifiers

If the amp is used for a powered speaker or subwoofer, speaker wiring is permanent and no protection is needed. But if there's a user involved in connecting the speakers...

The intended purpose of my amps is live sound and perfect reliability is more important then perfect fidelity. Users are not always (or more accurately are rarely) careful in making their speaker connections. Without SOA protection, the amp becomes and anchor or a door-stop.

What I'm suggesting here could possibly help fidelity by eliminating the limiter when Vout is high and the speaker load is normal. During normal operation, the power supply can limit the amp to its rated power. That's probably what's keeping your amps' operation inside the SOA. Just don't short those speaker wires!

[AndrewT]

Buckeye,
you are on the right lines. I have never seen q4 used like that.
Combinations of Zeners are more often used to obtain multiple slopes into the IV locus.

My version of a good IV limiter is that it must pass all valid audio signals to all valid audio loads. If it does this it can never have any effect on sound quality.
If abused the IV limiter should interfere with the signal and thus protect the devices from damage.

Capacitors have been mentioned.
I don't see any mechanism to allow high transient currents to pass uninterrupted while detecting and protecting from lower levels of excessive long term currents.

[MJL21193]

IMO, the best way to limit current to a short circuited output is via the power supply. Low side (0 volts) sensing and cut both rails when it is exceeded.
Leave the limiter out of the amp itself, where it may have an impact on performance.

[Buckeye]

Originally Posted by AndrewT

Capacitors have been mentioned.
I don't see any mechanism to allow high transient currents to pass uninterrupted while detecting and protecting from lower levels of excessive long term currents.

Agreed. Shorts in stage speaker wiring can happen any time and protection must be instantaneous. There doesn't seem to be any reason to keep it active longer than necessary.

Originally Posted by MJL21193

IMO, the best way to limit current to a short circuited output is via the power supply. Low side (0 volts) sensing and cut both rails when it is exceeded.
Leave the limiter out of the amp itself, where it may have an impact on performance.

That's a great approach. Are there manufacturers doing this? Seems like cost and complexity would suffer. I have some questions like: How is the output short sensed? Seems like you would have to cycle the power to see if it's still there. How is the rail power cut -- a couple big transistors along with their heat sinks? I haven't seen any cheap and simple circuits that do this can you point me at some?

Building you your idea... You know, it would be easier and cheaper to only cut the power to the input stage. Then you get the same benefit of nothing in the audio path, but protection by eliminating the drive signal.

That still needs a good method of sensing a short. You got anything?

[MJL21193]

I've seen a few that are high side sensing, but nothing that is low side. Low side is where I'd be doing it as it covers both rails simultaneously. I used low side sensing in my lab power supply HERE with outstanding results.
Cutting the rails would be the best way to go. This could be done a number of ways, but I think I would use power bjts, and incorporate a basic regulator in as well. Since the voltage drop across the devices is very low, power dissipation would be minimal.