Hi everybody,
Everybody knows about clipping in amplifiers, which happens when input signal exceeds its maximum limits and output starts to flatten between the power supply rails...If input drive exceeds, then the mosfet might fail instantly due to gate rupture.
But one thing...what would be the best protection implemented to protect the mosfet amps form overvoltage drive at inputs.
1.Zener diode at gate of mosfets[everyone uses it]
2.back to Back Zeners at input of amp.[some of us might use it.]
3.You guys tell me.
Waiting for your nice suggesstions....
Kanwar
Everybody knows about clipping in amplifiers, which happens when input signal exceeds its maximum limits and output starts to flatten between the power supply rails...If input drive exceeds, then the mosfet might fail instantly due to gate rupture.
But one thing...what would be the best protection implemented to protect the mosfet amps form overvoltage drive at inputs.
1.Zener diode at gate of mosfets[everyone uses it]
2.back to Back Zeners at input of amp.[some of us might use it.]
3.You guys tell me.
Waiting for your nice suggesstions....
Kanwar
This could be an intersting thread for me, as I'm in the middle of designing an amp and and looking for protection circuits.
This next URL shows in point 7 a current limiting protection. But I wonder if this has maybe influence on the sound colour/quality?
http://www.ee.und.ac.za/coursemain/DNE4EC1/notes/PwrAmpNotes.htm
The ETI466 has also a protection circuit thas goes out from the voltage over the emittor resistors.
But using this kind of protection isn't perfect, because the maximum amount off powerdissipation isn't produced when the voltage over the Re's is max.
The powerdissipation is also affected by the phase angle at the output of the amp.
Example:
PSU voltage +-48V / 190Wrms/4Ohm
*0° --> max Pd=130W --> Vgs=25V / ID=4.9A
*50° --> max Pd=340W --> Vgs=49.12V / ID=6.9A
--> by ID=4.9A --> Pdmax=39W
So, with this kind of protection you can never get the full output power.
This next URL shows in point 7 a current limiting protection. But I wonder if this has maybe influence on the sound colour/quality?
http://www.ee.und.ac.za/coursemain/DNE4EC1/notes/PwrAmpNotes.htm
The ETI466 has also a protection circuit thas goes out from the voltage over the emittor resistors.
But using this kind of protection isn't perfect, because the maximum amount off powerdissipation isn't produced when the voltage over the Re's is max.
The powerdissipation is also affected by the phase angle at the output of the amp.
Example:
PSU voltage +-48V / 190Wrms/4Ohm
*0° --> max Pd=130W --> Vgs=25V / ID=4.9A
*50° --> max Pd=340W --> Vgs=49.12V / ID=6.9A
--> by ID=4.9A --> Pdmax=39W
So, with this kind of protection you can never get the full output power.
Sorry gentleman, I've made a big mistake.
I've mistaken the phase angle at one certain moment with the output power.
BUT, with that exaple: (3x2SK1530 /55°C)
50° --> Imax is 7A to not cross the SOA curve
0° -->Imax is 10A to not cross the SOA curve
So you need to know the inductance/capacitance (phase angle) of the speaker system to get maximum output power without exceding the max power dissipation.
Greetz
Ben
Kanwar, this actually is on topic as any overload of the output will cause gate overdrive due to NFB trying to force the amp to do something it cannot. Besides, zeners are often used as current limiters, and if present, will quite simply protect the MOSFET from gate overdrive.
That being said, amps WILL clip. Either by current limit, or voltage limit (trying to go 'over' the power rail voltages). In both cases there is potential for gate overdrive. As far as protecting the MOSFET, a zener is perfectly adequate. The choice of using one, using one + a diode, or two anti-series, has little to do with which one is better, as the putpose is voltage limiting, but is a consequence of the difference in Vgs_bias and Vgs_limit. For complementary SF configurations, if (2xVgs_bias + 0.6V) > Vgs_limit, you can use one zener diode. In other cases, you have to use a zener + regular diode. Reason is simple, you do not want the forward biassed zener of one polarity MOS to limit the other polarity MOS, or seriously upset Vgs_bias. For complementary SF configurations, negative Vgs on the N channel part is automatically limited by limiting the maximum negative (i.e. MOS turned on) Vgs of the P part and vice versa - the negative voltage can at most be Vgs_limit - 2*Vgs_bias. Because of this I have found anti-series zeners to be unnecessary, but of course, they will work just fine, except that one of each anti-series zeners will really be working as a normal diode.
All the above examples imply no source resistors on the output transistors. If there are some, Vgs_limit include the voltage drop on these at the desired maximum current. This also means that Vgs at maximum current + voltage drop on the series source resistors (=Vgs_limit ~~= zener voltage), must be less than Vgs_breakdown, as the same zener limits gate drive for cases when output current is 0, so there is no voltage drop on the source resistors.
The real question is, do you want to protect your ears from gate overdrive - in other words, when an amplifier clips, given there is likely a NFB in place, ehat happens to the front stages? Are there any overhang effects? Any saturation and associated carrier storage problems in the front end transistors? Etc, etc. Unless the amp front end is designed for this not to be a problem (an interesting example of some of that can be found in the Stochino non-slew-limited amplifier), the only real solution is clipping the input signal when the gate gets close to the treshold - i.e. 'pre-clipping'. I think vAD is on the right track, though the idea needs serious polishing (sharper and better defined threshold, speed issues on the optocoupler...)
That being said, amps WILL clip. Either by current limit, or voltage limit (trying to go 'over' the power rail voltages). In both cases there is potential for gate overdrive. As far as protecting the MOSFET, a zener is perfectly adequate. The choice of using one, using one + a diode, or two anti-series, has little to do with which one is better, as the putpose is voltage limiting, but is a consequence of the difference in Vgs_bias and Vgs_limit. For complementary SF configurations, if (2xVgs_bias + 0.6V) > Vgs_limit, you can use one zener diode. In other cases, you have to use a zener + regular diode. Reason is simple, you do not want the forward biassed zener of one polarity MOS to limit the other polarity MOS, or seriously upset Vgs_bias. For complementary SF configurations, negative Vgs on the N channel part is automatically limited by limiting the maximum negative (i.e. MOS turned on) Vgs of the P part and vice versa - the negative voltage can at most be Vgs_limit - 2*Vgs_bias. Because of this I have found anti-series zeners to be unnecessary, but of course, they will work just fine, except that one of each anti-series zeners will really be working as a normal diode.
All the above examples imply no source resistors on the output transistors. If there are some, Vgs_limit include the voltage drop on these at the desired maximum current. This also means that Vgs at maximum current + voltage drop on the series source resistors (=Vgs_limit ~~= zener voltage), must be less than Vgs_breakdown, as the same zener limits gate drive for cases when output current is 0, so there is no voltage drop on the source resistors.
The real question is, do you want to protect your ears from gate overdrive - in other words, when an amplifier clips, given there is likely a NFB in place, ehat happens to the front stages? Are there any overhang effects? Any saturation and associated carrier storage problems in the front end transistors? Etc, etc. Unless the amp front end is designed for this not to be a problem (an interesting example of some of that can be found in the Stochino non-slew-limited amplifier), the only real solution is clipping the input signal when the gate gets close to the treshold - i.e. 'pre-clipping'. I think vAD is on the right track, though the idea needs serious polishing (sharper and better defined threshold, speed issues on the optocoupler...)
Hi ilimzn,
I have already mentioned the zener protection for gate overdrive in my first post....i think you misunderstood my topic....
I already know what you have said in your earlier post....
I just want to say that, if an amp [output load capability of driving 2 ohms]is loaded by only 16 ohms and input signal limit which is 1 volt for full output exceeds to somewhat 15 volts in input than the output starts to hard clip.....just finding the right solution to this problem.......
One way to go is using VCA voltage control amplifier, or using an Opto coupled FET, or optocoupled LDR to solve this problem....other i dont know yet.
Though i have seen some pro-amps using back to back zeners at input to clamp the over voltage at inputs and thus protects the output also......
regards,
Kanwar
I have already mentioned the zener protection for gate overdrive in my first post....i think you misunderstood my topic....
I already know what you have said in your earlier post....
I just want to say that, if an amp [output load capability of driving 2 ohms]is loaded by only 16 ohms and input signal limit which is 1 volt for full output exceeds to somewhat 15 volts in input than the output starts to hard clip.....just finding the right solution to this problem.......
One way to go is using VCA voltage control amplifier, or using an Opto coupled FET, or optocoupled LDR to solve this problem....other i dont know yet.
Though i have seen some pro-amps using back to back zeners at input to clamp the over voltage at inputs and thus protects the output also......
regards,
Kanwar
I've used zeners in a headphone amp where I had concerns that inputs of a higher voltage than normal line-level might be applied. While a simiulation will tell you that they cause some distortion at levels below the hard limit, the amount is actually so small (.000x%!) that you won't hear and can't easily measure it. Simple and effective. I suggest you not worry and just use the zeners -- no need to come up with something novel.
For a power amp they are used used to protect input circuitry especially in pro-amps where being "idiot proof" is a real concern. If the concern is with protecting the output devices perhaps a more satifying approach would be to reduce the closed loop gain. In the pro-world there is a sort of custom that the gain is set so that rated power is achieved with a 1V input. This is often cited in a "sensitivity" spec. Consumer audio sometimes used lower sensativity/gain because rated power is aimed more at increased headroom rather that average volume.
For a power amp they are used used to protect input circuitry especially in pro-amps where being "idiot proof" is a real concern. If the concern is with protecting the output devices perhaps a more satifying approach would be to reduce the closed loop gain. In the pro-world there is a sort of custom that the gain is set so that rated power is achieved with a 1V input. This is often cited in a "sensitivity" spec. Consumer audio sometimes used lower sensativity/gain because rated power is aimed more at increased headroom rather that average volume.
kanwar, look at the last paragraph in my post. What you are referring to is called 'pre-clipping' - the signal is clipped in a defined manner BEFOPRE it gets into the amp, i.e. at the amp input. In this manner it is never alowed to make the amp clip.
There are numerous simple ways you can do this, which do not impart unwanted degradation of the input signal. One of the more interesting ones I have seen used a transistor + diode clamp driven by two comparators, with the threshold levels derived from the power rails. In this way it could always account for rail droop, which is different depending on load. There are a number of other solutions where antiseries diodes are placed in a local feedback loop around the amp front end in order to reduce it's CL gain to a very low value once a preset output is reached, but i don't like this very much. A more complex solution with transistor clamps seems better to me because the clamping thresholds can be better controlled. I myself used a FET clamp (P and N), the gates are at fixed threshold potentials defined by a voltage divider, the Vgs is normally such that the FET is in a pinch-off mode, since both S connections are driven by the output of an OP-amp front end, and normally below gate potential. Both D connections are connected to the feedback node. When the OPamp output reaches within Vp of the threshold voltage, the FET starts conducting and the CL gain falls rapidly as the NFB is now closing through the FETs. It worked quite well, overdrive clipping had a softish characteristic as the FETs do not turn on abruptly. Due to the topology used, overcurrent limiting had a similar effect as well since the OPamp would try to overdrive the output, by increasing it's own output, so the same limiting appeared again. In this case the limiting was harder as the normal OPamp output voltages were tailored to get close to the thresholds for voltage clipping anyway. The OPamp worked in an inverting configuration so no further overvoltage protection was needed on the input save for a pair of diodes to the local OPamp power supply.
There are numerous simple ways you can do this, which do not impart unwanted degradation of the input signal. One of the more interesting ones I have seen used a transistor + diode clamp driven by two comparators, with the threshold levels derived from the power rails. In this way it could always account for rail droop, which is different depending on load. There are a number of other solutions where antiseries diodes are placed in a local feedback loop around the amp front end in order to reduce it's CL gain to a very low value once a preset output is reached, but i don't like this very much. A more complex solution with transistor clamps seems better to me because the clamping thresholds can be better controlled. I myself used a FET clamp (P and N), the gates are at fixed threshold potentials defined by a voltage divider, the Vgs is normally such that the FET is in a pinch-off mode, since both S connections are driven by the output of an OP-amp front end, and normally below gate potential. Both D connections are connected to the feedback node. When the OPamp output reaches within Vp of the threshold voltage, the FET starts conducting and the CL gain falls rapidly as the NFB is now closing through the FETs. It worked quite well, overdrive clipping had a softish characteristic as the FETs do not turn on abruptly. Due to the topology used, overcurrent limiting had a similar effect as well since the OPamp would try to overdrive the output, by increasing it's own output, so the same limiting appeared again. In this case the limiting was harder as the normal OPamp output voltages were tailored to get close to the thresholds for voltage clipping anyway. The OPamp worked in an inverting configuration so no further overvoltage protection was needed on the input save for a pair of diodes to the local OPamp power supply.
hi
It seems so - as the questions are more or less a copy of the questions we at Hunter Audio asked Kanwar over phone
Kanwar is a friend of mine - I dont want to creat a tiff or show anyone up -
It is only one of principles - which has prompted me to write here that - I dont in beleave solving professional problems -
through solutions sought at Google / Net
Common interest intrigue and doubt not withstanding - these are for what this forum i guess represents to help Diy By people who like us are a bit better informed being manufacturers / experienced fanatic hobbyists etc.
Suranjan Das Gupta
Transducer Design Engineer
It seems so - as the questions are more or less a copy of the questions we at Hunter Audio asked Kanwar over phone
Kanwar is a friend of mine - I dont want to creat a tiff or show anyone up -
It is only one of principles - which has prompted me to write here that - I dont in beleave solving professional problems -
through solutions sought at Google / Net
Common interest intrigue and doubt not withstanding - these are for what this forum i guess represents to help Diy By people who like us are a bit better informed being manufacturers / experienced fanatic hobbyists etc.
Suranjan Das Gupta
Transducer Design Engineer
HI PA,
We are already using diode clampers even at input of opamp and input of amplifiers, along with gate to source zeners for each mosfet, without these implementations the amp wont stand still even for a minute because the pro-mixers are in habbit of giving very large output signal voltage upto 10VRMS or 14V peak and the input senstivity of an amp is 1 volt and upon recieving 10V at input it would certainly get damaged in no time.... I think you simply dont know what to talk about and wandering here just with illicit questions.....and that too which doesnot bear any significance in common conscience
Ilimzn,
Your suggesstion is better, but in every pro-amp it comes with a VCA or something like that to limit the clipping of output signal to minimum and as fast as possible and we also implement it in our own style....... using opto coupled fet, but VCA also have a limit, I just want to search for something better than that....
..............................................................................................
I just post my questions here on the forum to get some more "better than best solutions" which we have already using......
...............................................................................................
Hi Hunter,
Nice to see you BOSS,
Yaar punjab is very hot 50deg Celcius at afternoon...and VAT is just implemented here and causing too much trouble for setting up of various sales issues for bussiness guys here.....
Suranjan how is the climate in KOLKATA might be hot.....
I think these guys are bit impassionate about my amps thats why they ask me for fishing ideas...what I call it as always searching the world for better than best options for just making the products simply much better......
regards,
Kanwar
We are already using diode clampers even at input of opamp and input of amplifiers, along with gate to source zeners for each mosfet, without these implementations the amp wont stand still even for a minute because the pro-mixers are in habbit of giving very large output signal voltage upto 10VRMS or 14V peak and the input senstivity of an amp is 1 volt and upon recieving 10V at input it would certainly get damaged in no time.... I think you simply dont know what to talk about and wandering here just with illicit questions.....and that too which doesnot bear any significance in common conscience
Ilimzn,
Your suggesstion is better, but in every pro-amp it comes with a VCA or something like that to limit the clipping of output signal to minimum and as fast as possible and we also implement it in our own style....... using opto coupled fet, but VCA also have a limit, I just want to search for something better than that....
..............................................................................................
I just post my questions here on the forum to get some more "better than best solutions" which we have already using......
...............................................................................................
Hi Hunter,
Nice to see you BOSS,
Yaar punjab is very hot 50deg Celcius at afternoon...and VAT is just implemented here and causing too much trouble for setting up of various sales issues for bussiness guys here.....
Suranjan how is the climate in KOLKATA might be hot.....
I think these guys are bit impassionate about my amps thats why they ask me for fishing ideas...what I call it as always searching the world for better than best options for just making the products simply much better......
regards,
Kanwar
It strikes me then that what you really need is a cross between a limiter and compressor.
I was under the impression you wanted to protect outputs from being overdriven in order not to smoke the FETs there - any of the approaches given in this thread can do that with indefinite input overload, BUT of course it does not prevent clipping, just controls it. The amp could survive this indefinitely, speakers connected to it are a different problem.
If you want some sort of overload protection that still alowes the amp to work 'normally' under overload conditions, then some sort of gain control is needed.
On the two integrateds I've made for myself, a relay attenuator was employed for volume control, using a small microprocessor. This was fed a clipping detect signal and would automatically increase attenuation as long as clipping was present. The clipping signal itself was conditioned so that very short clips would not reduce volume. On the other hand, if the clipping continued for a longer time and would persist even when the attenuation was at max, the amp automatically goes into error shut-down. That would be my take on it - one cannot expect a power amp to correct basic mistakes on the side of the sources. Sooner or later, someone has to actually correct the too high voltage out of a mixer or whatever, or this person is not worthy to be operating the equipment anyway.
I was under the impression you wanted to protect outputs from being overdriven in order not to smoke the FETs there - any of the approaches given in this thread can do that with indefinite input overload, BUT of course it does not prevent clipping, just controls it. The amp could survive this indefinitely, speakers connected to it are a different problem.
If you want some sort of overload protection that still alowes the amp to work 'normally' under overload conditions, then some sort of gain control is needed.
On the two integrateds I've made for myself, a relay attenuator was employed for volume control, using a small microprocessor. This was fed a clipping detect signal and would automatically increase attenuation as long as clipping was present. The clipping signal itself was conditioned so that very short clips would not reduce volume. On the other hand, if the clipping continued for a longer time and would persist even when the attenuation was at max, the amp automatically goes into error shut-down. That would be my take on it - one cannot expect a power amp to correct basic mistakes on the side of the sources. Sooner or later, someone has to actually correct the too high voltage out of a mixer or whatever, or this person is not worthy to be operating the equipment anyway.
Hi
It was quite hot here it cooled a bit recently - The heat wave was real bad , Solid Global Warming I guess
while fishing here include this :-
go to schematics connection / peavey forum etc. get their DDT circuit discription , Get hold of a Myers circuit , Go to Jbl technical right down below on amps - the schems are there - download two 1 mpa series , 1 mpx series , get a Qsc schem too , you may get the Qsc schem from Jbl site - it was there before
all of the above have solved your problem in different innovative ways - once you have been over all - you will i think be able to design one your self
if you yet choose to do it the internet way
take care
by the way speak to prem about your amps - lot of movement here and he needs good amps - so do we for our speakers
suranjan
It was quite hot here it cooled a bit recently - The heat wave was real bad , Solid Global Warming I guess
while fishing here include this :-
go to schematics connection / peavey forum etc. get their DDT circuit discription , Get hold of a Myers circuit , Go to Jbl technical right down below on amps - the schems are there - download two 1 mpa series , 1 mpx series , get a Qsc schem too , you may get the Qsc schem from Jbl site - it was there before
all of the above have solved your problem in different innovative ways - once you have been over all - you will i think be able to design one your self
if you yet choose to do it the internet way
take care
by the way speak to prem about your amps - lot of movement here and he needs good amps - so do we for our speakers
suranjan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.