Re: Re: Safe Operating Area Requirements
Hi Andrew,
I must admit that I am unfamiliar with Benson's spreadsheet and your version. Could you point me to it? It sounds very helpful.
Thanks,
Bob
AndrewT said:
Hi Andrew,
I must admit that I am unfamiliar with Benson's spreadsheet and your version. Could you point me to it? It sounds very helpful.
Thanks,
Bob
Hi Bob,
Yes, that's it. Instead of a triac across the outputs, use a triac with a resistor in series. Just provide good, constant gate drive and watch out for dv/dt effects. You can filter the gate (to ground) to avoid false triggering. In this way the amplifier will go to a slightly higher impedance very quickly once the shut down operates. Almost instantly.
This has the side benefit of loading the amplifier down should it go DC which may give the speakers a few more msec of life until the fuses can blow. This assumes your series resistance does not go open. I was never a fan of a direct short (triac bulk resistance only) as the currents are high enough to vaporize wide copper foil.
This should also appeal to the "no speaker relay" camp.
-Chris
Yes, that's it. Instead of a triac across the outputs, use a triac with a resistor in series. Just provide good, constant gate drive and watch out for dv/dt effects. You can filter the gate (to ground) to avoid false triggering. In this way the amplifier will go to a slightly higher impedance very quickly once the shut down operates. Almost instantly.
This has the side benefit of loading the amplifier down should it go DC which may give the speakers a few more msec of life until the fuses can blow. This assumes your series resistance does not go open. I was never a fan of a direct short (triac bulk resistance only) as the currents are high enough to vaporize wide copper foil.
This should also appeal to the "no speaker relay" camp.
-Chris
Perhaps I should not say this, but we are discussing protection solutions that we have worked with since the 1960's. The problem is: Obtrusiveness, in that protection is necessary, BUT it should reset itself and only happen with a serious overload condition.
For the record, V-I limiting was first brought forth to the design public in 'Wireless World' in the late 1960's. It was elaborated upon by Crown about 1970. It seemed, at the time, to be a good solution, but as stated in previous inputs recently on this thread, it IS obtrusive, and we attempted to design it out in the 1970's.
Alternative solutions, such has dual circuit breakers in the power supply, and Triac crowbar shorting have been used, BUT they have the problem of having to be reset.
Parasound's approach over the last 17 years, at least, is to sense with V-I, but to go into a timed out reset mode, by removing the output with a power relay. It is a useful solution, but not a perfect one, as the relay 'may' compromise the sound quality.
However, using a number of devices to spread the heat buildup either within the first 10ms or for a longer period can remove any need to set the relay sense so that there is any significant nuisance tripping, and still not easily break the amp. We secondarily protect with 4 series fuses in the power supply leg for each channel, and 1 primary fuse on the AC line. If these go, we have suffered a severe problem, but often just replacing the fuse makes the amp work again.
The ideal way would be to protect the output with a dual DC circuit breaker, but I have found that problematic with mosfets and V-fets. However, I have had pretty good luck with a number of amps using bipolar transistors.
For the record, V-I limiting was first brought forth to the design public in 'Wireless World' in the late 1960's. It was elaborated upon by Crown about 1970. It seemed, at the time, to be a good solution, but as stated in previous inputs recently on this thread, it IS obtrusive, and we attempted to design it out in the 1970's.
Alternative solutions, such has dual circuit breakers in the power supply, and Triac crowbar shorting have been used, BUT they have the problem of having to be reset.
Parasound's approach over the last 17 years, at least, is to sense with V-I, but to go into a timed out reset mode, by removing the output with a power relay. It is a useful solution, but not a perfect one, as the relay 'may' compromise the sound quality.
However, using a number of devices to spread the heat buildup either within the first 10ms or for a longer period can remove any need to set the relay sense so that there is any significant nuisance tripping, and still not easily break the amp. We secondarily protect with 4 series fuses in the power supply leg for each channel, and 1 primary fuse on the AC line. If these go, we have suffered a severe problem, but often just replacing the fuse makes the amp work again.
The ideal way would be to protect the output with a dual DC circuit breaker, but I have found that problematic with mosfets and V-fets. However, I have had pretty good luck with a number of amps using bipolar transistors.
Hi John,
I agree that protection can interfere with the music. I guess the other issue is that the higher your amplifiers rated power is, the more energy there is sitting around to go the wrong way. Often, this results in a destroyed woofer.
One could sense heat sink temperature and mute the signal until it cools somewhat. If you use a relay to ground for muting there should be no sonic penalty. In the event that you go DC, either the load should be disconnected and / or the output should be shorted. Since the amplifier is in a fault condition already, I'm not too concerned with damaging a few more parts. The amount of damage may be reduced by applying signal mute and bias shut off at the same time. If you could shut down the power supply (as in Carver - effective in this case) you can prevent further damage and also the speaker(s), or load.
In any event, the shut down should inconvenience the operators. If you let the equipment continue to run, it will be damaged. So even in a professional environment, such as an outdoor show, the equipment either dies, or shuts down. Either way there would have been no sound. You can not protect an idiot from themselves.
-Chris
I agree that protection can interfere with the music. I guess the other issue is that the higher your amplifiers rated power is, the more energy there is sitting around to go the wrong way. Often, this results in a destroyed woofer.
One could sense heat sink temperature and mute the signal until it cools somewhat. If you use a relay to ground for muting there should be no sonic penalty. In the event that you go DC, either the load should be disconnected and / or the output should be shorted. Since the amplifier is in a fault condition already, I'm not too concerned with damaging a few more parts. The amount of damage may be reduced by applying signal mute and bias shut off at the same time. If you could shut down the power supply (as in Carver - effective in this case) you can prevent further damage and also the speaker(s), or load.
In any event, the shut down should inconvenience the operators. If you let the equipment continue to run, it will be damaged. So even in a professional environment, such as an outdoor show, the equipment either dies, or shuts down. Either way there would have been no sound. You can not protect an idiot from themselves.
-Chris
john curl said:
Hi John,
I pretty much agree with what you have said here. I don't like SOA VI limiters, either. You're right, nothing new here; they have been around for ages.
As I pointed out, the main thing that is really important, at least with the vertical MOSFETs, is protection in the event of an output short circuit. If someone shorts the output of the amplifier, and the amplifier stops operating until the power is cycled, I don't see that as a problem. It is really no different than having a nearly-ideal circuit breaker with a 1 ms response time.
However, there are many ways that one could make such a scheme as I described self-resetting if they want to. The key thing would be to be able to sense when the short circuit condition has been removed. This can be as simple as occasionally applying a small current to the output during the period when the amplifier output stage is shut down, and testing to see how much voltage results.
Cheers,
Bob
Hi Bob,
-Chris
Consider using a 60 Hz signal (or 50 Hz) as it is available. It would give you a better idea if there was a short.
-Chris
Yeah, I hate that. I just had to drill out a batch of Fairchilds,
and dulled out 4 drill bits, just like that.
😎
and dulled out 4 drill bits, just like that.
😎
Nelson Pass said:
...you let me know next time you need at 2.5mm drill and a M3 tap , I may have a few to spare
Magura 😉
Hi Magura,
Hi Bob,
-Chris
I should buy some from you. I broke my only M3 tap (yep, in a heatsink ... last hole) a few years ago. I have not been able to locate one locally. The lack of a 2.5 mm drill bit may be the reason for the break. Yes, I was using oil with a drill press.
Hi Bob,
I've noticed that in On Semi's later devices. Too bad they discontinued the MJW0281A and MJW0302A, I was very impressed with them. Not every amp needs to be 200 watts. I've just ordered some of the others to check them out, so we shall see.
That's exactly something I have a use for. I'm looking forward to receiving some of the NJL devices as well. Things are looking up for On Semi and the audio amplifier business. We hobbyists may be in for some really nice parts to work with.
-Chris
I'd really like to see a new thread about biasing these devices and temperature compensation using the inbuilt diodes. The simple capacitor in the application note seems a bit simplistic.
Hi Pooge,
I think there is value in your idea. Please, cast the first stone!
-Chris 😉
The power is in your hands to make this idea come to life. Simply wish it and follow the steps to start a new thread.
I think there is value in your idea. Please, cast the first stone!
-Chris 😉
anatech said:
Hi Chris,
Yes, I'm really impressed at how these parts stack up against Japanese parts. Finally!
Now if they'd just get off their metric butt and open those holes just a smidge for those of us who still use the good old 6-32!
Cheers,
Bob
andy_c said:
Here's how I do it.
I fuse the rails and implement a simple safety backup I limiter which limits at a high current – say double the peak load current into the lowest nominal specified load impedance.
For instance, I’m currently building an amp rated at 1kW into 2R. Peak load current into 2R = 32A, so I limit at 64A. This is an adequate safety factor and will allow the amplifier to deliver its specified output voltage swing into half the specified load impedance.
I monitor the output voltage and output current using a PIC uC with an internal 10 bit ADC with multiple MUX inputs. The uC performs the SOA and DC faults protection. This allows a lot of flexibility beacuse you can program in time delays before the protection circuits activate for specified load current values.
The amplifier is shut down by activating a relay that shunts the audio input signal to ground and cuts the rails to the power output stage. The rails are switched with either MOSFET’s or IGBT’s in series with the rail fuses. This is a latched condition, requiring the amplifer to be either turned off/on or a reset button pressed.
Cheers,
Glen
Hi Bob,
What that means is that we are metric, sort of. Most of our fasteners are UNC and UNF still. Most machining equipment is imperial as well. However, our US built cars are ? Yep. Metric. 🙄
Some metric fasteners can be had in bulk, but not at our local warehouse setups. What gives with that anyhow??? 😡 They want a fortune for metric hardware at the electronic jobbers and the smallest tap I can find is M4.
I really ought to order some bulk to pass down to my children. Maybe some belleville washers while I'm at it.
-Chris
Well .......... I'm Canadian.
What that means is that we are metric, sort of. Most of our fasteners are UNC and UNF still. Most machining equipment is imperial as well. However, our US built cars are ? Yep. Metric. 🙄 Some metric fasteners can be had in bulk, but not at our local warehouse setups. What gives with that anyhow??? 😡 They want a fortune for metric hardware at the electronic jobbers and the smallest tap I can find is M4.
I really ought to order some bulk to pass down to my children. Maybe some belleville washers while I'm at it.
-Chris
Hi Glen,
Okay, yes I am being picky. It's just shades of Mr. Rube Goldberg (?).
You only need to detect a diode drop worth of DC, or about a volt if that makes you more comfortable. Easy to do since a diode drop is also an E-B drop with filtering to give you an average.
Most of us are not that good with uP's and programming. I'm not for one.
-Chris 😉
Why not keep it simple? I understand that you are attempting to combine functions. This makes sense to a degree. But what happens if your uP crashes or gets hung while in operation? No protection.
Okay, yes I am being picky. It's just shades of Mr. Rube Goldberg (?).
You only need to detect a diode drop worth of DC, or about a volt if that makes you more comfortable. Easy to do since a diode drop is also an E-B drop with filtering to give you an average.
Most of us are not that good with uP's and programming. I'm not for one.
-Chris 😉
What about the roller coaster F(t) of the ON semi devices. What will happen if you exceed 5A, even for a moment?
anatech said:
From a hardware point of view, it is the simplest approach. The uC is already there to monitor the output voltage as part of the SOA protection, and it controls the switching of the rails and other protection circuits. The protection circuits are latching, and will trigger either by immediate, direct command from the uC, or if a monostable times out, shutting down the entire amplifier. The monostable is continually reset by the running uC. If the uC hangs, the amplifier simply shuts itself down once the monostable times out.
The uC has it’s own internal watchdog timer too.
Cheers,
Glen
john curl said:
Hi John.
My 1kW class B stage has 16 pairs of ON Semi TO-264’s. Peak load current is limited to 64A, giving a maximum of 4A per device.
Although I haven’t updated the schematic on my website yet, the class A stage with tracking low-V rails is being built with 10 pairs of devices, giving just over 5A peak Ic per BJT - not a problem, IMO.
I also think that some people here are reading too much into those Ic Vs fT graphs. The implications for stability may not be as drastic as some assume.
For instance, your JC-1 with it’s advertised "18 matched 60MHz BJT’s" (only the NPN’s are 60MHz, so I’m assuming you have 18 pairs of NPN/PNP complements??) is rated at 135A peak load current.
Isn’t the JC-1 also a bridged design? That means that the load current is not shared between 18 devices, but two pairs of 9 in series.
135A / 9 – 15A per device. How do you cope with that at low vce, even for a moment?
Cheers,
Glen