?????
"The low output impedance of the amplifier will cause large currents to flow even when there is no signal present.... surely violate the SOA of the output transistors"
Bob Cordell - Designing Audio Power Amplifiers - Section 15.3 p323/4.
Brian.
Don't be confused. That is only happened in huge feedback error corrected amplifier, that has enough offset voltage at the output, that most low impedance amplifier has. Also not all amplifier has low impedance output.
I have mosfet amplifier with very low output impedance and huge feedback working in class B that not draw high current when output shorted with no signal. Because it has inductor at output and snubber before it and low offset.
I have mosfet amplifier with very low output impedance and huge feedback working in class B that not draw high current when output shorted with no signal. Because it has inductor at output and snubber before it and low offset.
Last edited:
?????
"The low output impedance of the amplifier will cause large currents to flow even when there is no signal present.... surely violate the SOA of the output transistors"
Bob Cordell - Designing Audio Power Amplifiers - Section 15.3 p323/4.
Brian.
Brian,
the law of physics states you need a potential difference for current to flow. I trust Bob's book does not imply the contrary. These laws of physics existed since the discovery of electricity. Don't believe everything you read.
That depends on who writes it.
I'll stick to adding VI limiting to everything I build.
Brian.
I have mosfet amplifier too,..can you share schematic for inductor protection circuit ?
Mr Bob Ohm, Mr Bob Coulomb, and all the other Bob guys that discovered electricity. Some time before the internet.
I like Nico Ras idea of comparing the input to the output. Unlike him I have blown a couple of sets of output transistor due to too much volume for too long out of a too small heat sink, probably, and the 'obsolete" output capacitor of the ST120 saved my speakers both times. I've blown two irreplaceable tweeters with a wimpy tube ST70, so I find protection circuits interesting. Neither the amp corner of my L.R., nor under the pulpit at the church, is well lit and the chance of a speaker whisker hitting ground is non-zero, as well as the fact that 1/4 phone plugs short every time they come out and tripping over them and yanking them out of the speaker is a common stage problem. The CS800S has passed my "direct coupled transistors are dangerous to speakers" test, with ratification from Mr Athens moderator Sakis, with Peavey's microprocessor controlled SOA measurement system. They use current transformers encircling the speaker output wire to produce data for the microprocessor, which can pull the speaker relays if it gets worried. I took the speakers out in the front yard 4th of July and played the Tchaikovski cannon shots (1812 overture) as loud as I could go on the CS800s, and never heard a clip or protection trip. This level of complication can't be duplicated by the amateur, in my opinion. That is why I am building a 3 channel amp with output capacitors, the djoffe circuit plus my mod, better heat sinks than the ST120, and crummier (lower short circuit current) transformers, right now.
Last edited:
It is not protector, the amplifier just act like that.
I am never using special VI limiter, just fuses for my early bjt, then current limited (by VA stage, it has V+I feedback) for classH, then the current is limited by regulator in classG. I never got burned output stages except once caused by fake black-green output bjt (after thatt I never use fake again). And most happened are broken VAS transistor during overheated for few years (happened at my early BJT amps).Howdy,
I have not abandoned my thread. I've been off learning how to use a circuit simulator, but I've continued to read this thread with great interest. There are some great ideas here. I am especially taken by the idea to compare Vin to Vout. Seems like you can protect a whole lot more than just the output transistors with that approach... even the speakers from a C-E shorted output transistor.
But alas my goal is not that lofty. I want to protect a dirt cheap power amp from an un-informed or careless user with a dirt cheap protection scheme. That kind of puts me in the VI limiter camp. But I want to improve on the concept a bit.
I realize that being the dirt cheap sort, I'm likely to end up with counterfeit transistors making me a much better candidate for high-end protection. But the most likely scenario is that when one transistor goes with a C-E short, it immediately takes out the other one which also fails with a C-E short and the speaker is safe. Then I figure there's usually a fuse somewhere that trips before a fire starts.
I'm seeing that the last line of defense has become the thermal fuse inside the transformers. They didn't used to put those in a while back, and the musicians who chose to keep rocking with a 20-Amp car fuse in their amp often got the smoke and fire. Now they get an amp that's dead to the world... and a hefty bill to replace the transformer because most techs aren't willing to swap out the internal thermal fuse. And I don't blame them.
The output shorts I typically see are the ones where the speaker cabinet end of the 1/4" cord gets unplugged and the tip touches ground. I'm guessing we might see around a half-Ohm load. So I cobbled up a simulation of my circuit. And am getting started looking at it, but I wanted to share and see if anyone had any opinions. I'm using LTspice. A free download is available from Linear Technology. You have to mess with the libraries to get the models you will typically use and that takes a bit of time. (BTW, the guys on the "Software Tools" board were a real big help getting me started with this. I even got some replies from Bob Cordell.)
Here's my circuit:
If you want to play with the simulation yourself here is the the net list. You will need to delete the .txt from the end of the file name.
View attachment SOCheap150-limiter.asc.txt
Here is the output into a 4-Ohm load:
This plot shows the +/- rail voltages. I modeled the PSU with a 0.75 Ohm internal resistance and a 10kuF capacitor. Also shown are the Vout and the current through the negative side emitter resistor. It's out of phase with the output voltage so it's easy to see everything on the plot.
Notice that the current limiter is not active and current peaks of over 8 amps are permitted.
Next is the plot with a 0.5-ohm load representing a short. notice that the current is limited to below 4 amps and Vout is around +/-2.5V.
So that's as far as I got. One thing I noticed is that my gated limiter is very sensitive to offset voltage. This may be what shreds my dream of a cheap protected amp. There is no easy way to adjust the output offset voltage with this design. Even for small offset voltages, the asymmetry of this VI protection scheme is huge.
Let me know your thoughts.
-Buckeye
I have not abandoned my thread. I've been off learning how to use a circuit simulator, but I've continued to read this thread with great interest. There are some great ideas here. I am especially taken by the idea to compare Vin to Vout. Seems like you can protect a whole lot more than just the output transistors with that approach... even the speakers from a C-E shorted output transistor.
But alas my goal is not that lofty. I want to protect a dirt cheap power amp from an un-informed or careless user with a dirt cheap protection scheme. That kind of puts me in the VI limiter camp. But I want to improve on the concept a bit.
I realize that being the dirt cheap sort, I'm likely to end up with counterfeit transistors making me a much better candidate for high-end protection. But the most likely scenario is that when one transistor goes with a C-E short, it immediately takes out the other one which also fails with a C-E short and the speaker is safe. Then I figure there's usually a fuse somewhere that trips before a fire starts.
I'm seeing that the last line of defense has become the thermal fuse inside the transformers. They didn't used to put those in a while back, and the musicians who chose to keep rocking with a 20-Amp car fuse in their amp often got the smoke and fire. Now they get an amp that's dead to the world... and a hefty bill to replace the transformer because most techs aren't willing to swap out the internal thermal fuse. And I don't blame them.
The output shorts I typically see are the ones where the speaker cabinet end of the 1/4" cord gets unplugged and the tip touches ground. I'm guessing we might see around a half-Ohm load. So I cobbled up a simulation of my circuit. And am getting started looking at it, but I wanted to share and see if anyone had any opinions. I'm using LTspice. A free download is available from Linear Technology. You have to mess with the libraries to get the models you will typically use and that takes a bit of time. (BTW, the guys on the "Software Tools" board were a real big help getting me started with this. I even got some replies from Bob Cordell.)
Here's my circuit:
If you want to play with the simulation yourself here is the the net list. You will need to delete the .txt from the end of the file name.
View attachment SOCheap150-limiter.asc.txt
Here is the output into a 4-Ohm load:
This plot shows the +/- rail voltages. I modeled the PSU with a 0.75 Ohm internal resistance and a 10kuF capacitor. Also shown are the Vout and the current through the negative side emitter resistor. It's out of phase with the output voltage so it's easy to see everything on the plot.
Notice that the current limiter is not active and current peaks of over 8 amps are permitted.
Next is the plot with a 0.5-ohm load representing a short. notice that the current is limited to below 4 amps and Vout is around +/-2.5V.
So that's as far as I got. One thing I noticed is that my gated limiter is very sensitive to offset voltage. This may be what shreds my dream of a cheap protected amp. There is no easy way to adjust the output offset voltage with this design. Even for small offset voltages, the asymmetry of this VI protection scheme is huge.
Let me know your thoughts.
-Buckeye
I own a ST120 which uses one pair of current sensing transistors to protect two channels, each of which has 2 npn quasi comp output transistors. I'm not good enough on SS circuits to tell in your schematic which is the protection transistor, but it sounds like you are using 1 or 2 transistors to measure power supply rail current. This doesn't work, because when an output transistor on one channel starts to avalanche, the current is within the allowable current for two output pairs. So the output pair self destructs. The ST120 is obsolete in having an output capacitor, so the speaker is not damaged. The heat sink is not big enough on the protection transistor either, so if you don't catch the problem and turn it off, the protection transistor avalanches too, then the fuse blows.
So, without checking your math, I think you need one protection transistor per output transistor.
The ST120's biggest problem is modern hi gain replacement transistors allow it to pump out maybe 100 W per channel, and the heat sink is big enough for 60W.
If your going to pump out 100 W/channel you need 2 output transistor pairs, or more.
That brings up emmiter resistors. 0.1 in each emmiter leg is standard to share current between the different output transistors.
I have recently acquired a PV-1.3K power amp which has as it's only output protection, a crowbar diac-SCR board that determines that the DC voltage is too close to the rail (IE a transistor is shorted) and crowbars the rail to ground. Sometimes this blows the fuse. People have observed, sometimes this melts PCB traces first. Any, it dies dead, maybe not taking out your speaker.
PV-1.3K schematic at peavey.com product support catagory, PV amps brochure.
So anyway, I have been thinking of replacing all the 0.1 ohm emmiter resistors on the 10 output transistor, with these $.52 PCT thermistor "resettable fuses". 0.1 ohm nominal resistance, able to take out 72v, 3 amp hold current. A 1.3K can pump out about 13 amps total with 5 output transistor pairs. RXEF300 Tyco Electronics / Raychem PTC Resettable Fuses People don't like real speaker fast blow fuses, they say you can hear them. I wonder if you can hear these? Only near the ragged edge?
So, without checking your math, I think you need one protection transistor per output transistor.
The ST120's biggest problem is modern hi gain replacement transistors allow it to pump out maybe 100 W per channel, and the heat sink is big enough for 60W.
If your going to pump out 100 W/channel you need 2 output transistor pairs, or more.
That brings up emmiter resistors. 0.1 in each emmiter leg is standard to share current between the different output transistors.
I have recently acquired a PV-1.3K power amp which has as it's only output protection, a crowbar diac-SCR board that determines that the DC voltage is too close to the rail (IE a transistor is shorted) and crowbars the rail to ground. Sometimes this blows the fuse. People have observed, sometimes this melts PCB traces first. Any, it dies dead, maybe not taking out your speaker.
PV-1.3K schematic at peavey.com product support catagory, PV amps brochure.
So anyway, I have been thinking of replacing all the 0.1 ohm emmiter resistors on the 10 output transistor, with these $.52 PCT thermistor "resettable fuses". 0.1 ohm nominal resistance, able to take out 72v, 3 amp hold current. A 1.3K can pump out about 13 amps total with 5 output transistor pairs. RXEF300 Tyco Electronics / Raychem PTC Resettable Fuses People don't like real speaker fast blow fuses, they say you can hear them. I wonder if you can hear these? Only near the ragged edge?
Last edited:
They take a looong time to trip - 20 seconds and a lot of mayhem can happen in that amount of time.
By the time a smoothing capacitor has dumped all its charge into the output transistors the damage is already done. I suppose you could try fuses before the rectifier and after the smoothing caps.
On the other hand how likely is a short on the output?
I run a mobile disco and the leads get tripped over and pulled so I need to protect my amps a bit more from shorted speaker leads.
On the other hand how likely is a short on the output?
I run a mobile disco and the leads get tripped over and pulled so I need to protect my amps a bit more from shorted speaker leads.
Last edited:
Indianajo,
My goal is to protect the amp from a short, not the speaker from a shorted transistor. Usually, the emitter resistor will do that because failure of one transistor is usually followed by a failure of the other side, and then the emitter resistors blow. This sequence usually happens pretty quickly. (Although I did lose a pair of really nice headphones once when my headphone amp output shorted to the rail.)
The circuit in my simulation does not attempt to sense the rail current. Instead it's a conventional VI limiter, plus the speaker load voltage is sensed. If Vout is too low (a condition indicating a shorted output) then the maximum current is limited to a very small value. A low Vout and a high Ic are not permitted when the output transistors are subjected to a high Vce because that's outside the SOA.
Obviously all of the output and driver transistors have to be functional for this to work. But that's what I'm mostly trying to protect with this limiter.
I think you can get in excess of 100W out of a pair of BJTs these days. But for sure, the size needed for the heatsink hasn't changed much.
My goal is to protect the amp from a short, not the speaker from a shorted transistor. Usually, the emitter resistor will do that because failure of one transistor is usually followed by a failure of the other side, and then the emitter resistors blow. This sequence usually happens pretty quickly. (Although I did lose a pair of really nice headphones once when my headphone amp output shorted to the rail.)
The circuit in my simulation does not attempt to sense the rail current. Instead it's a conventional VI limiter, plus the speaker load voltage is sensed. If Vout is too low (a condition indicating a shorted output) then the maximum current is limited to a very small value. A low Vout and a high Ic are not permitted when the output transistors are subjected to a high Vce because that's outside the SOA.
Obviously all of the output and driver transistors have to be functional for this to work. But that's what I'm mostly trying to protect with this limiter.
I think you can get in excess of 100W out of a pair of BJTs these days. But for sure, the size needed for the heatsink hasn't changed much.
Last edited:
How likely are blown output transistors?
Survey the used equipment at the musician's consignment shop. Most of the speakers have one or more new drivers. Most of the amplifiers have shiny new output transistors on the heat sink on one side. The root cause, if not avalanched output transistors is? Yeah, maybe some torn cones. I've avalanched the output transistors twice, probably by demanding 100 W average out of my ST120 to fill the church. My CS800S is a lot softer cranked all the way up, (overall gain) I find out, with 8 output transistors per side instead of 2. The CS800S protects itself, but it is $999 retail here. The PV8.5 without the current transformers and microprocessor is about $499. Interesting about the 30 sec trip time. I had the ST120 protection circuit set to 2.25 amps overall last time the it blew up (no schematic or internet when I repaired it, I didn't know about the 6.5 A setting until last year) and I still cooked an output transistor pair (NTE181)'s. So maybe 3 A resettable fuses are a little liberal.
Not protecting speakers? A good protection circuit should do both. Protect the amplifier from shorted speaker terminals. Protect the speakers from shorted output transistors. Maybe wimpy emmiter resistors are part of an answer. Buckeye, does your protection circuit sense each output transistor individually?
Survey the used equipment at the musician's consignment shop. Most of the speakers have one or more new drivers. Most of the amplifiers have shiny new output transistors on the heat sink on one side. The root cause, if not avalanched output transistors is? Yeah, maybe some torn cones. I've avalanched the output transistors twice, probably by demanding 100 W average out of my ST120 to fill the church. My CS800S is a lot softer cranked all the way up, (overall gain) I find out, with 8 output transistors per side instead of 2. The CS800S protects itself, but it is $999 retail here. The PV8.5 without the current transformers and microprocessor is about $499. Interesting about the 30 sec trip time. I had the ST120 protection circuit set to 2.25 amps overall last time the it blew up (no schematic or internet when I repaired it, I didn't know about the 6.5 A setting until last year) and I still cooked an output transistor pair (NTE181)'s. So maybe 3 A resettable fuses are a little liberal.
Not protecting speakers? A good protection circuit should do both. Protect the amplifier from shorted speaker terminals. Protect the speakers from shorted output transistors. Maybe wimpy emmiter resistors are part of an answer. Buckeye, does your protection circuit sense each output transistor individually?
Last edited:
Blown output transistors are a common problem, no doubt. I'm the last one who'll argue that point, seeing how I put myself through college mostly replacing blown output transistors and everything else they took with them.
I don't remember ever seeing a blown speaker along with the blown output stage. There may have been some, but not enough to make any kind of lasting impression.
The exact problem you're describing is why I opened this thread. Sometimes high output current is OK and transistors won't blow. Other times they'll pop like popcorn. I referred to a link a while earlier that talks about why the conventional VI limiter is a compromise. When the output is shorted, even 2 amps is too much for many power transistors and the device will fail because of secondary breakdown. Just look at the typical SOA graph. I'm trying to come up with an improved limiter. Still a compromise, but a slightly improved one.
In the circuit shown, there are only 2 output transistors and the emitter current is sensed for each one. If there is a need to parallel outputs, the concept could be extended to sense each additional transistor. But me being dirt cheap, I probably won't do it.
I don't remember ever seeing a blown speaker along with the blown output stage. There may have been some, but not enough to make any kind of lasting impression.
The exact problem you're describing is why I opened this thread. Sometimes high output current is OK and transistors won't blow. Other times they'll pop like popcorn. I referred to a link a while earlier that talks about why the conventional VI limiter is a compromise. When the output is shorted, even 2 amps is too much for many power transistors and the device will fail because of secondary breakdown. Just look at the typical SOA graph. I'm trying to come up with an improved limiter. Still a compromise, but a slightly improved one.
In the circuit shown, there are only 2 output transistors and the emitter current is sensed for each one. If there is a need to parallel outputs, the concept could be extended to sense each additional transistor. But me being dirt cheap, I probably won't do it.
I'm a cheapskate, that is why I worry about my speakers. A peavey driver is $100 +, a pair of output transistors is $4. And only peaveys have a support program, My LWE III and KLH 23's were both orphans when the tweeter blew up. (ST70 tube amp). Hammond speakers, you get 3 with each $200 organ and it takes usually $200 in truck rental & a helper to get it home. How about sensing the current with optoisolators on each output transistor, then paralleling the opto output transistors to suck drive away from one output transistor- or if too much current, one output relay? Not a real sharp knee, but opto's are fast. Or 4/package op amps to sense the emmiter resistor voltage, then use the overvoltage out to trip a shutoff transistor or relay. My PV1.3K would need a 13 A Ic transistor to shut off a rail, even in switch mode not very affordable.
As far as SOA goes, the guys that put a thermistor on the heatsink are whistling in the dark. Some guy on here quoted an ON semi MJL part that had a 4th pin and an on die temperature sensor. That is getting close with op amp sensing: but they were $6 each. My recent MJ15015 lot (not counterfit, newark) were $1.50 each.
As far as SOA goes, the guys that put a thermistor on the heatsink are whistling in the dark. Some guy on here quoted an ON semi MJL part that had a 4th pin and an on die temperature sensor. That is getting close with op amp sensing: but they were $6 each. My recent MJ15015 lot (not counterfit, newark) were $1.50 each.
Last edited:
All of those approaches can be made to work, but they have their challenges. You've got to be real careful when limiting drive, because audiophiles call that distortion. That's one of the problems of VI limiters. Op-Amps are a real nice way to do the signal processing, but depending on the total power, the output could swing well over 100 Volts. Your CS800 is probably in excess of 120V P-P. Those usually aren't the dime per dozen, garden variety of op amps. Yeah, you could use optos, but then you're adding to complexity.
Like I said, I don't want to switch the rails -- just use a fuse for that. So you're still stuck controlling the drive. Then you have half of my circuit already. I'm adding a couple more transistors and resistors. (Actually 2 of the resistors in my circuit are jumpers. They're set at 0.1Ohm because I was playing with the values in the simulator and then was too lazy to take them out. The simulator won't let me have a 0 Ohm resistor.)
Why do you dislike my gated VI limiter approach?
Like I said, I don't want to switch the rails -- just use a fuse for that. So you're still stuck controlling the drive. Then you have half of my circuit already. I'm adding a couple more transistors and resistors. (Actually 2 of the resistors in my circuit are jumpers. They're set at 0.1Ohm because I was playing with the values in the simulator and then was too lazy to take them out. The simulator won't let me have a 0 Ohm resistor.)
Why do you dislike my gated VI limiter approach?
Last edited:
When an output transistor dumps the rail cap, the speaker coil goes, not the basket.That part is over $100 including shipping or big markup for the local dealer. If you're willing to put whatever circuit was on the front page on every transistor individually, then you might have something. I'm not a circuit reading guru, but I think you had two extra transistors per drive transistor, or was it one? To do the math of the dual slope SOA curve, there should be some integrating capacitors somewhere, that is probably why peavey does the math with microprocessors. Then if they get a deal on some other transistor, they just change the math. For the amateur it might be cheaper to sense the die temperature of the MJL something transistor with 15V op amps and achieve the same effect. I calulate I've blown $500+ in 1975 dollars on speakers with blown coils that can't be bought as exact replacements, that is why I'm a bit concerned.
I've put a lot of time into Djoffe's ST120 circuit, that is now sounding pretty good. With its $3 output cap its level of speaker protection is nearly absolute. Output transistor protection less good. Looking at splitting the single 6.5 amp current limiter into a 3 amp current limiter per channel, or maybe 2. Have already put temp sensing on the O.T's and cutting the idle bias current per side when hot. Your circuit is pretty dedicated to split supply amps, which are a totally different ball game, more dangerous to speakers. I'm not real clear whether q3 & q4 are drive dumpers to rail, or drive transistors, or one of each. I don't see how the SOA is calculated without capacitors. And the circuit would have to be calculated differently for each type of output transistor. A lot of people use MJ15024's, I'm using NTE181 and MJ15015. Do Q3 & Q4 just protect the upper transistor, or also the lower? Showing my ignorance. The ST120 has something like Q3 & Q4 on the main + rail, which limits both voltage to +72 and current to +6.5A. SImulations are nice, but my computer (ubuntu) is too stupid to look at microsoft spreadsheets or whatever they run on, I have to see the math in text form, sorry.
I've put a lot of time into Djoffe's ST120 circuit, that is now sounding pretty good. With its $3 output cap its level of speaker protection is nearly absolute. Output transistor protection less good. Looking at splitting the single 6.5 amp current limiter into a 3 amp current limiter per channel, or maybe 2. Have already put temp sensing on the O.T's and cutting the idle bias current per side when hot. Your circuit is pretty dedicated to split supply amps, which are a totally different ball game, more dangerous to speakers. I'm not real clear whether q3 & q4 are drive dumpers to rail, or drive transistors, or one of each. I don't see how the SOA is calculated without capacitors. And the circuit would have to be calculated differently for each type of output transistor. A lot of people use MJ15024's, I'm using NTE181 and MJ15015. Do Q3 & Q4 just protect the upper transistor, or also the lower? Showing my ignorance. The ST120 has something like Q3 & Q4 on the main + rail, which limits both voltage to +72 and current to +6.5A. SImulations are nice, but my computer (ubuntu) is too stupid to look at microsoft spreadsheets or whatever they run on, I have to see the math in text form, sorry.
Last edited:
that is precisely what a reactive speaker demands regularly. If a peak transient happens at that combination of frequency/slew changes then the amplifier is expected to pass some or a lot of current when Vce is high. The SOA must allow this.
The limiting must allow this.
The music demands it.
The listener demands it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.