jacco vermeulen said:
The answer is always obvious after the solution has been found.
I did not see many posting the answer to my question - ?? Maybe the ones that already knew did not bother with a simple question like this…?
Regarding the excel file, I think I is diffecult to read, I prefer something a bit more like what mikeks wrote.
\Jens
JensRasmussen said:
Well, yes and no Jens.
I had an eye on this subject but was unnusally busy this week. Trouble is the algebra is pretty messy for complex loads, and there is no pratical way to write it down in text only form so I planned to check, put it as clean (and revised) as possible and write with Word's equation editor and then clip and post a .gif.
If you are still interested, I promise to tackle it next week, let me know.
Unless of course others come out with a better solution.
Rodolfo
Merging soft power-up circuit with output protection. Please review.
Basic design:
Separate power supply for driver stage. Includes time delay to relay for soft turn-on of output power supply. The driver circuit includes a servo to control DC offset, and a shelf circuit which triggers if the output has significant DC from a short circuit or shorted output transistor.
Output power supply has soft turn-on resistor and shunt relay that shorts out this resistor after the timer on the driver board triggers and the shelf circuit is off. This relay is DPDT, and has a 1 ohm resistor between the output terminal and ground in the "off" position.
Main power switch turns on driver supply and the soft turn-on resistor path to the output power supply. When the timer on the driver triggers AND the shelf circuit signals "no short", then the turn-on relay is activated and the 0.1 ohm shorting resistor removed from the output.
If an output short occurs, then the relay falls open, the 1 ohm shunt resistor in put in parallel with the speaker to protect it, and the output power supply is run on the limited current from the soft power-up resistor. If the short goes away, normal operation is returned.
Basic design:
Separate power supply for driver stage. Includes time delay to relay for soft turn-on of output power supply. The driver circuit includes a servo to control DC offset, and a shelf circuit which triggers if the output has significant DC from a short circuit or shorted output transistor.
Output power supply has soft turn-on resistor and shunt relay that shorts out this resistor after the timer on the driver board triggers and the shelf circuit is off. This relay is DPDT, and has a 1 ohm resistor between the output terminal and ground in the "off" position.
Main power switch turns on driver supply and the soft turn-on resistor path to the output power supply. When the timer on the driver triggers AND the shelf circuit signals "no short", then the turn-on relay is activated and the 0.1 ohm shorting resistor removed from the output.
If an output short occurs, then the relay falls open, the 1 ohm shunt resistor in put in parallel with the speaker to protect it, and the output power supply is run on the limited current from the soft power-up resistor. If the short goes away, normal operation is returned.
Load line
At last made some time to address the load line issue under reactive impedance. Attached is a PDF with the derivation, and in the next post a simple spreadsheet to plot results both for current mode drive and for voltage drive.
Feel free to mercilesly point mistakes and observations
.
Rodolfo
At last made some time to address the load line issue under reactive impedance. Attached is a PDF with the derivation, and in the next post a simple spreadsheet to plot results both for current mode drive and for voltage drive.
Feel free to mercilesly point mistakes and observations
.Rodolfo
Re: Load line
I think you'll find your Ic Vs Vce plots are in significant error....
Instanteneous device dissipation will have either two maxima, or one maximum and one point of inflection depending on phase shift between Ic and Vce....(BJTs assumed)...
With a reactive load, the device is expected to deliver substantial current at high Vce....Your characteristics contradict this....
ingrast said:Feel free to mercilesly point mistakes and observations
Rodolfo
I think you'll find your Ic Vs Vce plots are in significant error....
Instanteneous device dissipation will have either two maxima, or one maximum and one point of inflection depending on phase shift between Ic and Vce....(BJTs assumed)...
With a reactive load, the device is expected to deliver substantial current at high Vce....Your characteristics contradict this....
Re: Re: Load line
Thanks for taking the trouble to provide feedback Mikeks.
I was puzzled by your comments. Both in the current mode and voltage mode drive, the posted graphs show significant current at peak voltage. To make it clearer I included for the voltage mode case, power dissipation graphs for the DC and 2500 Hz situations, where the contrast is obvious as shown in attached clip.
On the other hand I had previously checked and found a mistake in equation [4], the sign for the sum within square brackets should be + instead of -, but it does not change significantly the overall graphs shape.
Rodolfo
mikeks said:
Thanks for taking the trouble to provide feedback Mikeks.
I was puzzled by your comments. Both in the current mode and voltage mode drive, the posted graphs show significant current at peak voltage. To make it clearer I included for the voltage mode case, power dissipation graphs for the DC and 2500 Hz situations, where the contrast is obvious as shown in attached clip.
On the other hand I had previously checked and found a mistake in equation [4], the sign for the sum within square brackets should be + instead of -, but it does not change significantly the overall graphs shape.
Rodolfo
Attachments
2 Areas of output protection
OK
I have waded through the entire thread and would like to sumarise..
We are talking about 3 types of protection circuits here to deal with 3 types of increasingly sevear output stage issues
a) VI limiting
b) Thermal compensation
c) Output protection - The subject of the original discussion
a) VI limiting is required to allow an indetermined load to be handled gracefully by the amplifier. ( this may include temporary shorts )
b) Thermal componsation- including thermal overload- is put in place to prevent the SOA of the output devices colapsing to the point that it is violated by normal operation. As an extreem there may be a limiter that disconnects the output to allow for the heatsink to cool this may be self resetting, or it may latch to make it clear to the user that he is going above and beyond.
c) Output protection - This is put in place for when it has all gone wrong and a condition occurs that appears to be a failure in the amplifier. At this time we are cutting our losses and running. Thus a latching methodology seems appropriate ( or at least one that retries in a manner that is non damaging)
Three methods have been sugested
i) disconect the amplifier from the speakers
ii) disconect the power from the output stage
iii) Short the output(s) together
i) this has the advantage of a single point disconnect, However the internals of the amplifier may still continue to disintergrate. Depending on the nature of the fault. There are possible sonic issues of having a switch element in the signal path.
ii) more complicated as there are at least 2 disconnect points. The internals of the amplifier are protected from further disintergration. There are possible power supply problems of having switch elements in the high current output and simultaneous transition of the switching elements. Unless the high current PSU is a SMPS then it may be as disabling it's operation.
iii) Protects the speakers, but encourages the insides to disintergrate
Of these (to me) i and ii are acceptable and iii may be if the action of storting the speaker terminals together causes something else to happen - like it switches off the PSU, blows a fuse etc.
Now it seems to me that the best method is to have a layred approach, but everyone is entitled to help a semiconductor company be more profitable in whatever way they see fit.
Brian
OK
I have waded through the entire thread and would like to sumarise..
We are talking about 3 types of protection circuits here to deal with 3 types of increasingly sevear output stage issues
a) VI limiting
b) Thermal compensation
c) Output protection - The subject of the original discussion
a) VI limiting is required to allow an indetermined load to be handled gracefully by the amplifier. ( this may include temporary shorts )
b) Thermal componsation- including thermal overload- is put in place to prevent the SOA of the output devices colapsing to the point that it is violated by normal operation. As an extreem there may be a limiter that disconnects the output to allow for the heatsink to cool this may be self resetting, or it may latch to make it clear to the user that he is going above and beyond.
c) Output protection - This is put in place for when it has all gone wrong and a condition occurs that appears to be a failure in the amplifier. At this time we are cutting our losses and running. Thus a latching methodology seems appropriate ( or at least one that retries in a manner that is non damaging)
Three methods have been sugested
i) disconect the amplifier from the speakers
ii) disconect the power from the output stage
iii) Short the output(s) together
i) this has the advantage of a single point disconnect, However the internals of the amplifier may still continue to disintergrate. Depending on the nature of the fault. There are possible sonic issues of having a switch element in the signal path.
ii) more complicated as there are at least 2 disconnect points. The internals of the amplifier are protected from further disintergration. There are possible power supply problems of having switch elements in the high current output and simultaneous transition of the switching elements. Unless the high current PSU is a SMPS then it may be as disabling it's operation.
iii) Protects the speakers, but encourages the insides to disintergrate
Of these (to me) i and ii are acceptable and iii may be if the action of storting the speaker terminals together causes something else to happen - like it switches off the PSU, blows a fuse etc.
Now it seems to me that the best method is to have a layred approach, but everyone is entitled to help a semiconductor company be more profitable in whatever way they see fit.
Brian
The topics:
a) VI limiting
b) Thermal compensation
c) Output protection
My solutions are:
a) I would certainly go for Overkill+large SOA mosfets, because nothing stands tall in front of Overkill for reactive load driving capability[whether you have double or even triple slope SOA protection].
b) N-channel vertical Mosfets offer positive temp. coefficient and along with absence of dreaded Second breakdown voltage.
c) simply latch the input signal to mute and dissconect the power supply [secondary side AC]with the traics and hence forth the threat to output is eliminated....
regards,
K a n w a r
a) VI limiting
b) Thermal compensation
c) Output protection
My solutions are:
a) I would certainly go for Overkill+large SOA mosfets, because nothing stands tall in front of Overkill for reactive load driving capability[whether you have double or even triple slope SOA protection].
b) N-channel vertical Mosfets offer positive temp. coefficient and along with absence of dreaded Second breakdown voltage.
c) simply latch the input signal to mute and dissconect the power supply [secondary side AC]with the traics and hence forth the threat to output is eliminated....
regards,
K a n w a r
This is all well and good. What about isolated device failure? It happens and no one buys absolutely zero defect parts. Just imagine if one of the diff pair fails, or vas. Cracked solder joint anyone?
Speakers in flames.
Not unless we are so arrogant that "our amps don't fail". Direct quote from someone who should know better. This company still sells amps but brought in some real engineers. (Hint: One series was the "ST" series.)
Let's be realistic about theworld we live in.
-Chris
Speakers in flames.
Not unless we are so arrogant that "our amps don't fail". Direct quote from someone who should know better. This company still sells amps but brought in some real engineers. (Hint: One series was the "ST" series.)
Let's be realistic about theworld we live in.
-Chris
Workhorse said:
Kanwar,
I don't agree on your first point. You can have very effective slope/ SOA protection. You are free to go for overkill. It is less effective than SOA protection: You can never be ABSOLUTELY sure to be able to handle ALL possible situations, but with SOA protection you can. So, go for overkill if you like it, but don't tell us it is somehow better than SOA protection. It is not as good, and it is more expensive.
On point 3: Do you mean disconnect the supplies with the triacs or short them? If you mean disconnect, that will be difficult and complex to do.
Jan Didden
janneman said:
Jan,
I don't agree with you . I have seen many amps with so called SOA protection that failed when encountered with real world loads[highly reactive loads], whereas OVERKILL really helps to kill the "incapability of output stage to handle such realworld loads". I need not to tell you that overkill is better than SOA protection in Professional arena because it is a champion in reality and traditional SOA protection simply fall miserable. We do Overkill of output stage and also connect gate to source Zeners to each individual mosfet to conduct the current in safe limits[you may call it a SOA protection also]. Expense doesnt count when you have to run the workhorse amp [our amps are not costly as other big guys amp's cost ]continuously for days and to keep the show going , without any interupt.....
regarding disconecting the secondary of transformer with triacs is fairly a simple and yet effective method, not a complex one..we have done it in a simpler manner...I dont know why you feel it so complex...maybe you have different picture in mind regarding the triacs....
kindly please dont take it as an advertisement sort of thing , its simply an explanation of why we use overkill.
regards,
K a n w a r
oh Guru, dont ask these silly type of questions, because you already know that our amps dont cross-conduct even at 250KHZ.
how about your health , are you alright now?
cheers to Greg Ball
Workhorse said:
Ahhh, disconnect the transformer BEFORE the reservoir caps, yes you can do that. Now I understand why you need the overkill, because this type of short circuit protection is mostly worthless because the energy in the caps will cause damage irrespective of the disconnection of the xformer. Unless, of course, you have SOA protection.
Jan Didden
janneman said:
Yeah sure, a competently designed SOA protection would safe guard the output, BUT it will SAG the output power to much extent as compared with overkill of output stage...when encountered with reactive loads...
janneman said:
Ahhh, disconnect the transformer BEFORE the reservoir caps, yes you can do that. Now I understand why you need the overkill, because this type of short circuit protection is mostly worthless because the energy in the caps will cause damage irrespective of the disconnection of the xformer. Unless, of course, you have SOA protection.
Jan Didden
2 things came into action when the out put stage is subjected to short circuit.
1. input audio-signal to amp is muted, therefore no drive signal hence no output current and also not dissipiation.
2. Supply is disconnected which is also an another precautive measure to protect the output stage.
JensRasmussen said:
I think i am Talking about Speakers fitted with Passive Cross-overs, so there's no question for active speakers.....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.