| beppe61 |
Dear Sirs,
I understand that most of the commercial amps have some sort of circuit/device that limits the delivery of current to the load (speakers).
I read that in order to safeguard the life of the output devices some designer overprotect them limiting the peak current too much sometimes.
I would like to know if anyone has ever eliminated this circuit and what has been the outcome.
I wonder how this circuit is built.
If I could find this circuit in the schematic I would try to by-pass it in some way.
Thank you very much.
Kind regards,
beppe61 |
|
|
| ashok |
1. You are making a very general statement.
2. Yes some amps limit current more than what some users want.
BUT maybe the designer wants it that way.
3. If you are a DIY guy or FIDDLER of equipment , the current limit might be a blessing in disguise.
4. From your statements it looks like you "might" not be a techie guy. In that case do not alter anything unless you want 'holy smoke'.
5. Specify the amp in question to determine if it has been "over" protected.
6. Current limit must have prevented many DIY'ers amplifiers from becoming toast!
So be VERY cautious about what you want to do. If you have a scope and other test equipment , you could possibly do it - assuming you are a techie.
I think not many people would like to help you do this by remote control on an amp they know nothing about. If you have a specific amp in mind , ask again!
Cheers. |
|
|
| pinkmouse |
| I agree with Ashok. Beppe, if you want an amp that can supply current, then you would be best off building or buying something specifically designed to do so. May I suggest a Krellclone? :) |
|
|
| beppe61 |
Dear Sir,
thank you very much indeed for your extremely kind and valuable reply.
Let me be more precise between your statements.
>| quote: | Originally posted by ashok
1. You are making a very general statement.
I understand the problem is quite complex, unfortunately.
But it is my opinion that amp manufacturers should help customers providing them with figures useful to assess the driving ability of their products.
From what I have gathered here and there this value of max peak current delivery is a very important one and very seldom is in the specifications.
They feel bad to declare let's say a 5 A of max peak current when other amps can deliver 20 A ?
They fill a page with data of no real interest.
Better to say nothing. Or let me try it with my speakers.
I have compared with my speakers two amps:
1) 90W/8ohm ; peak current = ?
2) 45 W/8ohm; peak current = 20A
There is no comparison in driving ability between the two, being 2) a world better than 1).
I bought of course 1).
2. Yes some amps limit current more than what some users want.
BUT maybe the designer wants it that way.
No discussion about this. But they must declare it to customer.
I want an amp powerful and reliable.
Maybe I am asking too much.
My friend gave a look inside amp 2).
A very minimalist one with off-the-shelf components.
And with an output stage made out of just two darlingtons/channel.
As I said really a world better, at least with these beasts of Dynaudio speakers.
> 3. If you are a DIY guy or FIDDLER of equipment , the current limit might be a blessing in disguise.
Let's say for a moment I am not a DIY guy (i.e. the reality).
May I know the current deliver of an amp I want to buy or it must remain a secret until I test it with my speakers?
Here in Italy I have to buy before listening.
I have the home full of bad buys made on the basis of specifications.
> 4. From your statements it looks like you "might" not be a techie guy. In that case do not alter anything unless you want 'holy smoke'.
I see the point. I do not want to destroy anything for sure.
> 5. Specify the amp in question to determine if it has been "over" protected.
The amp is a Samson Servo 260.
At full power the woofers of the Dynaudio just mumble.
Something like a velvet fist in a ... velvet glove.
The sound is not bad at all just the lower part of the audio spectrum (<100Hz) is just missing with my speakers.
> Current limit must have prevented many DIY'ers amplifiers from becoming toast!
The amp 2) could be a breeze to clone indeed.
It has about 10 transistor /channel and a very remarkable sound.
But this is not my point. My point is that the common specs are not useful to establish the driving ability of the amp.
Max peak current maybe it is more useful, from what I have read.
> So be VERY cautious about what you want to do.
If you have a scope and other test equipment , you could possibly do it - assuming you are a techie.
I am not a techie. I am just a disoriented audio hobbyst.
> I think not many people would like to help you do this by remote control on an amp they know nothing about.
If you have a specific amp in mind , ask again!
Cheers. |
To end my ramblings, I understand that modding an amp without knowledge could destroy it.
If only these value of peak current were stated more clearly in the specs these could be very useful IMHO.
Even if I decided for the DIY route, only the UcD modules datasheets report the max current available from a module.
But which is for instance the max peak current of an AKSA amp?
10A, 20A, 30A ?
I could not find it anywhere..
I know the sound is fine but what about its driving ability with demanding speakers?
Of course to be sure I should buy a Krell and stop any discussion.
They could drive even a door.
Unfortunately the original Krell is out of my reach and the clone is too difficult to build.
Thank you very much indeed.
Kind regards,
beppe61 |
|
|
| beppe61 |
| quote: | Originally posted by pinkmouse
I agree with Ashok.
Beppe, if you want an amp that can supply current, then you would be best off building or buying something specifically designed to do so.
May I suggest a Krellclone? :) |
Dear Sir,
your suggestion is very wise, but my skills as a DIYer are very limited.
I need an amp in the kit form like for instance an AKSA amp.
I am afraid that the driving ability is quite different though, and I happened to like very much a deep and powerful bass response.
Anyway I also think that the Krellclone is just too complex.
Are you aware of amp in kit form with a good delivery of current?
Thank you very much indeed.
Kind regards,
beppe61 |
|
|
| pinkmouse |
| quote: | Originally posted by beppe61
...They could drive even a door... |
:D| quote: | | ...and the clone is too difficult to build... |
Really, it's not. If I can build it, anyone can. It looks complicated, with lots of components, but all it takes is a bit of care and attention. ;) |
|
|
| beppe61 |
| quote: | Originally posted by pinkmouse
:D Really, it's not. If I can build it, anyone can.
It looks complicated, with lots of components, but all it takes is a bit of care and attention. ;) |
Dear Sir,
thank you so much for your extremely kind and valuable reply.
I see that you are a UK citizen so I feel the urge to state that some of the best example of amp design come from your land.
Congratulations.
Some years ago I had a Cyrus One, then a Rotel, a Cambridge.
Now I have fallen in love with another UK amp ( I would like to say the name but I am waiting to purchase one maybe in the ebay.co.uk).
They are monos, 45W/8ohm but with a great current delivery (22A max peak current ! each).
They made my current hungry Dynaudio sing, and for the very first time my system was enjoyable.
The friend who lent them to me, has opened it and found very few components. Again a very simple even minimalist approach.
Think of 10 transistor per channel. A single darlington output pair.
I really do not know about copyright law about audio design.
But that would be a design to clone.
I looked at the schematic of the Krell. It is really to much for me.
I am a very unexperienced DIYer.
I do not know of any other kit with a great current delivery.
I understand that the company has ceased its activity nowadays.
This is indeed a pity because their designs were very good indeed.
Thank you very much indeed for you very kind support and advice.
Kind regards,
beppe61 |
|
|
| ashok |
Hi Beppe61 ,
You needn't worry about copyright etc. As long as you copy the design for personal use it should be OK. The problem arises only if you use it for commercial purposes ( meaning you start building copies and selling them for gain ).
A very large number of DIY guys are making copies of amps they like but cannot afford to buy ! In any case at the cost they make it , they will never make it commercially viable .
So you can go ahead and copy the amp you like. No one can stop you. However study the Krell clone carefully and any other kits available before picking one.
In addition , email Hugh at AKSA to find out about the current capability . He might even suggest something if you need higher current capability ! Maybe that amp IS your solution.
Cheers,
Ashok. |
|
|
| beppe61 |
Dear Mr. Ashok,
thank you so much fro your extremely kind and precious reply.
Let me please reply to you hereunder.
>| quote: | Originally posted by ashok
Hi Beppe61 , You needn't worry about copyright etc.
As long as you copy the design for personal use it should be OK. The problem arises only if you use it for commercial purposes ( meaning you start building copies and selling them for gain ).
Thank you very much indeed. Now I am feeling better.
I have absolutely no intention to use them for commercial purposes.
> A very large number of DIY guys are making copies of amps they like but cannot afford to buy !
In any case at the cost they make it , they will never make it commercially viable.
I know this too well. One starts to save money and ends spending 2, 3 times the budget.
> So you can go ahead and copy the amp you like. No one can stop you. However study the Krell clone carefully and any other kits available before picking one.
I have the old Krell stereo amps (the KSA50, 80 and 100) in the greatest esteem indeed. The mono version even higher as well, of course.
They were powerful, with a prodigious bass and nice full-range sound. They drove anything and have been the basis of the Krell success worldwide.
I would like very much to be able to clone one but as I said looking at the schematic posted here of the KSA50 I find them beyond my ability, that is very poor.
> In addition, email Hugh at AKSA to find out about the current capability .
He might even suggest something if you need higher current capability !
Maybe that amp IS your solution.
Surely taking into account that is a complete kit with a very well done users' manual it should be easier to build.
I will try to assess its driving ability (my recurrent nightmare after my choice of Dynaudios as speakers).
Cheers,
Ashok. |
Dear Mr. Ashok,
thank you so much for your extremely kind and precious advice.
Kind regards,
beppe61 |
|
|
| phase_accurate |
Regarding overcurrent protection:
Limiting current to a specific value is not going to help much in terms of device protection.
The most effective protection circuits are limiting current as a function of output voltage (so called SOA protection) and are therefore often employed. Their downside is that they work perfetly into resistive loads but not so into real loudspeakers where current and voltage can be out of phase and cause the circuits to limit much too early.
There are only to solutions to this problems:
1.) Overdimensioned output stages for which the protection can be more generously dimensioned. These can also help to increase the output stage linearity.
3.) Using amp principles that don't need SOA protection but only fixed-value overcurrent protection: Switching amps !
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Regarding overcurrent protection:
Limiting current to a specific value is not going to help much in terms of device protection.
The most effective protection circuits are limiting current as a function of output voltage (so called SOA protection) and are therefore often employed. Their downside is that they work perfetly into resistive loads but not so into real loudspeakers where current and voltage can be out of phase and cause the circuits to limit much too early.
There are only to solutions to this problems:
1.) Overdimensioned output stages for which the protection can be more generously dimensioned. These can also help to increase the output stage linearity.
3.) Using amp principles that don't need SOA protection but only fixed-value overcurrent protection: Switching amps !
Regards
Charles |
Dear Sir Charles Lehmann,
Absolutely agree on your first point, but I think Mr.Jan Didden and Mikeks wouldn't..agree because they think overbuilding the output stage is wastage of resources and could be cured with SOA limiter[Their opinion not mine]
But I would favour Overkill in output stage in pro-audio amp..
Yes , the switching amps would take anything resistive to highly reactive without any threats, provided an overcurrent protection implemented along them....
sincere regards,
K a n w a r |
|
|
| janneman |
| quote: | Originally posted by Workhorse
Dear Sir Charles Lehmann,
Absolutely agree on your first point, but I think Mr.Jan Didden and Mikeks wouldn't..agree because they think overbuilding the output stage is wastage of resources and could be cured with SOA limiter[Their opinion not mine]
But I would favour Overkill in output stage in pro-audio amp..
Yes , the switching amps would take anything resistive to highly reactive without any threats, provided an overcurrent protection implemented along them....
sincere regards,
K a n w a r |
Well, the reason for the overkill is given as if SOA doesn't work with reactive loads. But it does, that's actually the REASON for SOA protection. Now, if you need more SOA area to drive your speakers than is available in your output stage, obviously you need a bigger output stage or multiple parallel devices. I don't see why anyone calls this overkill. It isn't, it is just common sense engineering.
If on the other hand someone would use multiple devices in an unfounded fear that SOA protection is not enough, either the SOA is incompetently designed, or the available SOA is not enough (see first paragraph) or the person is irrational. And yes, in this last case it IS a waste of resources.
I would be willing to change my view though if someone comes up with a situation where SOA protection, well designed, would be insufficient. Since you Kanwar seem to like to project an air of expertise, that should be easy for you.;)
Jan Didden |
|
|
| Arty123 |
HI Beppe,
First off, It is important to see that current is not the only factor: If the amplifier is clipping, the speaker cone is stationary at maximum current, which equates to DC flowing through the voice coil. This will destroy the voice coil, as no cooling is available.
The optimum is a soft clipping characteristic, combined with overvoltage and overcurrent protection.
Have a look at http://www.hypex.nl, they do a number of excellent Class D amplifiers, with all the stuff you'll need to build a complete power amp, with all the protection you'll need for minimal cost. A UcD180 module is only 60 Euros....
I built my amp with UcD180's and it sounds better than any Krell I have ever listened to, I have listened to those and Mark Levinson, MacIntosh and numerous others.
Have a look at the Class D forum as well, it is full of good advice. Cheers, Arthur. |
|
|
| phase_accurate |
Hallo Flachi !
Hoe gaat et ?
Jan
I don't mean to say that SOA protection wouldn't work. It is just that there are amps out there whose devices and protection are dimensioned for nominal power into a resistive load only. If both are dimensioned with real-world (i.e. non-real loads to make things complicated ;) ) loads in mind then there isn't a problem.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by janneman
Well, the reason for the overkill is given as if SOA doesn't work with reactive loads. But it does, that's actually the REASON for SOA protection. Now, if you need more SOA area to drive your speakers than is available in your output stage, obviously you need a bigger output stage or multiple parallel devices. I don't see why anyone calls this overkill. It isn't, it is just common sense engineering.
If on the other hand someone would use multiple devices in an unfounded fear that SOA protection is not enough, either the SOA is incompetently designed, or the available SOA is not enough (see first paragraph) or the person is irrational. And yes, in this last case it IS a waste of resources.
I would be willing to change my view though if someone comes up with a situation where SOA protection, well designed, would be insufficient. Since you Kanwar seem to like to project an air of expertise, that should be easy for you.;)
Jan Didden |
Dear Jan,
I think you misunderstood the overkill concept,
The reason for overkill is not to overcome the SOA limitation, but to extend the capability of the Output stage as whole to project the minimum stress level on the output devices whenever a situation of a specific stress is encountered and thereby doing so would eventually increases the life of the equipment in field operations.....everyone likes to have a minimum stress level whether its a FET/BJT or you & me...its reality....
If 4 pairs of output devices with SOA protection are sufficient for handling a given reactive load, then adding 2 more pairs[total six] wont be stated as wastage, rather it should be seen as an added advantage to lighten the stress level even more..... which would always have a positive effect on the operation of the amp.... which is designed to operate in professional arena rather than in home environment...
I would only consider it as a wastage if the output devices pairs exceeds the six pairs to something eight or even nine pairs of devices, because in that case it is clearly evident that the designer has no understanding of what SOA protection is meant for....
SOA protection..ofcourse a well designed one is always sufficient but if you offer a "LITTLE EXTRA" in terms of output devices, it must not be regarded as a resource wastage...
K a n w a r |
|
|
| janneman |
| quote: | Originally posted by phase_accurate
Hallo Flachi !
Hoe gaat et ?
Jan
I don't mean to say that SOA protection wouldn't work. It is just that there are amps out there whose devices and protection are dimensioned for nominal power into a resistive load only. If both are dimensioned with real-world (i.e. non-real loads to make things complicated ;) ) loads in mind then there isn't a problem.
Regards
Charles |
I think you misunderstood. SOA protection is designed to safeguard the output devices, WHATEVER loads you throw at them. So, if the SOA is set up correctly, no imaginable load will destroy them.
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by Workhorse
Dear Jan,
I think you misunderstood the overkill concept,
The reason for overkill is not to overcome the SOA limitation, but to extend the capability of the Output stage as whole to project the minimum stress level on the output devices whenever a situation of a specific stress is encountered and thereby doing so would eventually increases the life of the equipment in field operations.....everyone likes to have a minimum stress level whether its a FET/BJT or you & me...its reality....
If 4 pairs of output devices with SOA protection are sufficient for handling a given reactive load, then adding 2 more pairs[total six] wont be stated as wastage, rather it should be seen as an added advantage to lighten the stress level even more..... which would always have a positive effect on the operation of the amp.... which is designed to operate in professional arena rather than in home environment...
I would only consider it as a wastage if the output devices pairs exceeds the six pairs to something eight or even nine pairs of devices, because in that case it is clearly evident that the designer has no understanding of what SOA protection is meant for....
SOA protection..ofcourse a well designed one is always sufficient but if you offer a "LITTLE EXTRA" in terms of output devices, it must not be regarded as a resource wastage...
K a n w a r |
Well, it is all very strange to me. So we design an output stage for a design load regime. Then we design a protection system to make sure they will not fail. And then we throw in something extra? For what? Lower the stress? Does that mean you expect it to fail unless you throw in the extra? That means your design wasn't sufficient in the first place. Very, very strange...
Jan Didden |
|
|
| phase_accurate |
| quote: | | So, if the SOA is set up correctly, no imaginable load will destroy them. |
Ahhhhh, now I see whre we misunderstood each other !! I didn't want to say that they won't protect your output devices, far from that. But they sometimes do that too early i.e. they restrict output power into reactive loads if the output-stage is dimensioned on the cheap and the SOA protection dimensioned accordingly. I don't critisise SOA protection as such but cheap undersized output stages.
There are amps that offer the same nominal power into resistive loads but differ in perceived power into real-world loads. One of many reasons can be the above.
BTW: With switching amps, max current limitation is about the same as SOA limitation.
Regards
Charles |
|
|
| Workhorse |
Dear Jan,
You still Misunderstood....its extremely Strange
THE SOA Protection is designed to Safeguard the output devices when encountered with reactive, or simply say Unsafe Loads..which would otherwise damage the devices.....And the output will not be destroyed in case of well designed SOA protection....because SOA Limiter is always there to safeguard the devices...No destruction is possible with any type of load....until your SOA protection isn't damaged
The Fact is that when you add a little extra, this increases the Long period operation reliabilty in high temperature conditions and ease the output as a whole....Because the reliability of a BJT/FET device is always inversely proportional to its operating temperature...
Thereby when we Parallel More Pairs of output devices ..it simply Increases the Available DIE-AREA and complements the Heat Dissipation over Larger surface area of Heatsink and ensures cooler running of device through increased heat removal from the device die area attached to the heatsink and boost the reliability
K a n w a r |
|
|
| janneman |
| quote: | Originally posted by Workhorse
[snip]If 4 pairs of output devices with SOA protection are sufficient for handling a given reactive load, then adding 2 more pairs[total six] wont be stated as wastage, rather it should be seen as an added advantage to lighten the stress level even more..... which would always have a positive effect on the operation of the amp.... which is designed to operate in professional arena rather than in home environment...
I would only consider it as a wastage if the output devices pairs exceeds the six pairs to something eight or even nine pairs of devices, because in that case it is clearly evident that the designer has no understanding of what SOA protection is meant for.... [snip] |
(sigh) OK, so you say that adding a few extra pairs have a pos effect on the operation of the amp. What it it, please? Do you mean longer life time? How much longer? Do you have any information on how much longer? Or is the few pairs extra just something that 'feels good' ?
I also have no idea why adding 2 pairs is a Good Thing, while adding 4 extra pairs shows a misunderstanding of the SOA meaning.
Jan Didden |
|
|
| Workhorse |
Dear Jan,
I think I have already answered it in my previous post..please review it..... |
|
|
| beppe61 |
Dear Mr. Arthur,
thank you so much for your very kind and valuable reply.
First of all I have to declare my complete ignorance in electronics.
I am trying to learn something but without basic knowledge I have no hope.
And even the basics are really hard to understand for me.
>| quote: | Originally posted by Arty123
HI Beppe, First off, It is important to see that current is not the only factor: If the amplifier is clipping, the speaker cone is stationary at maximum current, which equates to DC flowing through the voice coil. This will destroy the voice coil, as no cooling is available.
The optimum is a soft clipping characteristic, combined with overvoltage and overcurrent protection.
Have a look at http://www.hypex.nl, they do a number of excellent Class D amplifiers, with all the stuff you'll need to build a complete power amp, with all the protection you'll need for minimal cost. A UcD180 module is only 60 Euros....
As a matter of fact I have already posted questions in the Class D forum about UcD modules as beppe61.
I read great things about these products.
They will be present in a lot of amps at the next CES in USA in January, I understand.
> I built my amp with UcD180's and it sounds better than any Krell I have ever listened to, I have listened to those and Mark Levinson, MacIntosh and numerous others.
What about their driving ability with demanding load ?
I read some reserves about this.
I have Dynaudio speakers. To wake-up the woofers and control them a very powerful amp is mandatory.
Which speakers do you use with UcD180 ?
Have a look at the Class D forum as well, it is full of good advice. Cheers, Arthur. |
Thank you very much Mr. Arthur
Kind regards,
beppe61 |
|
|
| beppe61 |
| quote: | Originally posted by Workhorse
Dear Sir Charles Lehmann,
Absolutely agree on your first point, but I think Mr.Jan Didden and Mikeks wouldn't..agree because they think overbuilding the output stage is wastage of resources and could be cured with SOA limiter[Their opinion not mine]
But I would favour Overkill in output stage in pro-audio amp..
Yes , the switching amps would take anything resistive to highly reactive without any threats, provided an overcurrent protection implemented along them....
sincere regards,
K a n w a r |
Dear Sir,
thank you so much for your extremely kind and valuable reply.
Unfortunately the technical discussion is very much beyond my reach.
Anyway I find extremely interesting this your statement:" the switching amps would take anything resistive to highly reactive without any threats ". Very interesting.
Thank you very much indeed.
Kind regards,
beppe61 |
|
|
| beppe61 |
| quote: | Originally posted by janneman
I think you misunderstood. SOA protection is designed to safeguard the output devices, WHATEVER loads you throw at them. So, if the SOA is set up correctly, no imaginable load will destroy them.
Jan Didden |
Thank you Mr Didden.
I am very sorry not to be able to take part in the discussion but I like this idea of SOA protection.
As I understand correctly, this kind of protection can guarantee that the output devices are used up to their maximum performance.
I think it is very much more refined protection system that a trivial current limiting device/circuit.
Am I right ? My knowledge is very limited.
Kind regards,
beppe61 |
|
|
| janneman |
| quote: | Originally posted by beppe61
[snip]I think it is very much more refined protection system that a trivial current limiting device/circuit.
Am I right ? My knowledge is very limited.
Kind regards,
beppe61 |
Yes and no. The concept is really not that complex. Normally you would feed the voltage across the emitter resistors of the output stage to a transistor base. That voltage is really a measure of the output transistor current Ie (Ie * Re = Vbe of the protection transistor). That transistor then starts to conduct if the current gets too high. You with me so far?;)
The protection transistor is connected to the driver stage, so when it starts to work it will steal drive current and thus limit the output current that will be available. So far so good.
The SOA implementation will take a resistor from the power supply voltage and use it to inject extra current into the protection transistor base. The protection transistor is connected with its emitter to the output, which is close to Ve of the output device. The output transistor Collector is normally connected to the power supply. So, the current injected by the extra resistor is really a measure of the output transistor Vce! (Injected current = Vce/Rinjection).
So, now the protection is activated not only by the output Ie but also the Vce. So, when Vce rises, smaller Ie will already trigger the protection because the protection transistor is 'helped' by the injected current which depends on Vce.
It is a little more involved because the SOA is not a nice linear combination of Ie and Vce, but that is the principle.
Jan Didden |
|
|
| AndrewT |
Hi Beppe,
a simple fixed current limit will not protect the amp from failure and it will also limit too early. If you want I can explain this statement.
A VI limiter that comes close to the long term and short term SOAR curves and takes account of device temperature will protect the output stage from failure and it will not limit too early.
The VI limiter achieves the exact opposite of the fixed current limiter.
The design of a temperature compensated VI limiter that permits both long term limiting and allows short term peaks will be a technical achievement that many would want to copy. From past threads and other research I fear we are a long way from achieving this.
In the meantime you are relying on the designer compromising as best he sees fit to get the VI limiter as close as he wants to prevent warranty failures and only a few understand the complexity enough to avoid early limiting.
Listen and hope. |
|
|
| Arty123 |
Dear Beppe 61,
My current (and have been for a very long time) speakers are Klipsch La Scala, and yes, I realise they are very efficient, and that is why I chose an amp that is very quiet, has no problems delivering an open soundstage.
For your Dynaudios, if you really want to shake walls, a UCD400 or UCD700 would do nicely. With the UcD's you will find that you are able to turn up the volume without the sound becoming aggressive or tiring.
The support of Hypex is unsurpassed, and for a DIY newbie that you say you are, the build of such an amp would be easy.
This is how I built my monoblocks(see picture) : They deliver more than 100Watts into 8 Ohms, and are stable with an 8 Ohm load with 2 Uf in parrallel, which is a simulation of a Quad electrostatic speaker. |
|
|
| beppe61 |
| quote: | Originally posted by janneman
Yes and no. The concept is really not that complex. Normally you would feed the voltage across the emitter resistors of the output stage to a transistor base. That voltage is really a measure of the output transistor current Ie (Ie * Re = Vbe of the protection transistor). That transistor then starts to conduct if the current gets too high. You with me so far?;)
The protection transistor is connected to the driver stage, so when it starts to work it will steal drive current and thus limit the output current that will be available. So far so good.
The SOA implementation will take a resistor from the power supply voltage and use it to inject extra current into the protection transistor base. The protection transistor is connected with its emitter to the output, which is close to Ve of the output device. The output transistor Collector is normally connected to the power supply. So, the current injected by the extra resistor is really a measure of the output transistor Vce! (Injected current = Vce/Rinjection).
So, now the protection is activated not only by the output Ie but also the Vce. So, when Vce rises, smaller Ie will already trigger the protection because the protection transistor is 'helped' by the injected current which depends on Vce.
It is a little more involved because the SOA is not a nice linear combination of Ie and Vce, but that is the principle.
Jan Didden |
Dear Mr. Didden,
thank you greatly for your extremely kind and valuable reply.
Unfortunately all this explanation is very well beyond my abilities of understanding.
Just a question: are you aware of commercial amps that employ this kind of very smart protection circuits?
Thank you very much indeed.
Kind regards,
beppe61 |
|
|
| beppe61 |
| quote: | Originally posted by AndrewT
Hi Beppe,
a simple fixed current limit will not protect the amp from failure and it will also limit too early. If you want I can explain this statement.
A VI limiter that comes close to the long term and short term SOAR curves and takes account of device temperature will protect the output stage from failure and it will not limit too early.
The VI limiter achieves the exact opposite of the fixed current limiter.
The design of a temperature compensated VI limiter that permits both long term limiting and allows short term peaks will be a technical achievement that many would want to copy. From past threads and other research I fear we are a long way from achieving this.
In the meantime you are relying on the designer compromising as best he sees fit to get the VI limiter as close as he wants to prevent warranty failures and only a few understand the complexity enough to avoid early limiting.
Listen and hope. |
Thank you very much Mr. Andrews for the very interesting explanation.
In your opinion, all amps have some kind of current limiting devices/circuits or there are also amp without them?
They coudl be indeed a bottle-neck with demanding loads.
Of course output stages and power supplies must be built in order to be up to the task.
Kind regards,
beppe61 |
|
|
| beppe61 |
Dear Sir,
thank you very much again for your very kind and valuable reply.
Let me be more precise hereunder.
>| quote: | Originally posted by Arty123
Dear Beppe 61, My current (and have been for a very long time) speakers are Klipsch La Scala, and yes, I realise they are very efficient, and that is why I chose an amp that is very quiet, has no problems delivering an open soundstage.
Actually the idea of changing speakers is another possible solution.
Dynaudio ask for powerful amps with an high damping factor.
If not the bass is weak, soft, slow and without control.
> For your Dynaudios, if you really want to shake walls, a UCD400 or UCD700 would do nicely.
After all my readings I think that at least a 400 is mandatory.
I would be very interested to hear from Dynaudio e UcD owners.
> With the UcD's you will find that you are able to turn up the volume without the sound becoming aggressive or tiring.
The support of Hypex is unsurpassed, and for a DIY newbie that you say you are, the build of such an amp would be easy.
Actually that is a fundamental point.
> This is how I built my monoblocks(see picture) : They deliver more than 100Watts into 8 Ohms, and are stable with an 8 Ohm load with 2 Uf in parrallel, which is a simulation of a Quad electrostatic speaker. |
My sincere congratulations ! Great work indeed from all points of view.
Again dear Friend, I thank you very much for your very kind and valuable support.
I will report on my final solution.
Kind regards,
beppe61 |
|
|
| Workhorse |
| quote: | Originally posted by janneman
It is a little more involved because the SOA is not a nice linear combination of Ie and Vce, but that is the principle.
Jan Didden |
Hi Jan,
Well Said,
But if you look at SOA curves in case of Mosfets , they are much linear than BJT's in terms of Vds and Id combination....
Correct me if I were wrong...
K a n w a r |
|
|
| janneman |
| quote: | Originally posted by Workhorse
Hi Jan,
Well Said,
But if you look at SOA curves in case of Mosfets , they are much linear than BJT's in terms of Vds and Id combination....
Correct me if I were wrong...
K a n w a r |
You are right, of course. Power FETS are not subject to secondary breakdown. This occurs in BJT as a result of high current with high Vce. That causes what is called "current hogging" where a small island on the die heats up more than another part, the hottest part will conduct more current, will become hotter, will conduct more etc. Because in effect only a very small part of the total chip or die will then conduct all the current, the allowed Pc gets much smaller. This current hogging occurs especially with higher Vce, so that is why the SOA shown much less allowed Pc at higher Vce then what you would expect from the DC Pc number.
Power FETs dont have this, so they can absorb higher powers at higher Vds compared with similar bipolars.
Jan Didden |
|
|
| Workhorse |
Thanks Jan....:)
K a n w a r |
|
|
| amplifierguru |
What you say is true enough, BUT!
Second breakdown derates at around 10% /100C c.f. thermal derating of bipolars so your 250W TO-3 can well hold it's own against 150W TO247 MOSFETs for quality audiophile art. Maybe nothing in it.
In practice the MOSFET seems more rugged and forgiving of lack of SOA protection.
Cheers,
Greg |
|
|
| AndrewT |
Hi Amp Guru,
10%/100C ?
Do you mean SOA falls by 10% when the temp rises by 100Cdeg?
That gives 90% at 125degC cf. 100% at 25degC.
I would think that power BJTs do a lot worse than this.
It is a bit difficult to compare because the manufacturer's data is generally on two different factors:- the Tc derating and the SOAR derating. Then you combine them for overall effect and resulting output maxima.
Can you explain or clarify? |
|
|
| janneman |
| quote: | Originally posted by amplifierguru
What you say is true enough, BUT!
Second breakdown derates at around 10% /100C c.f. thermal derating of bipolars so your 250W TO-3 can well hold it's own against 150W TO247 MOSFETs for quality audiophile art. Maybe nothing in it. [snip]
Cheers,
Greg |
No. The point is that because of the SOA limit, that 250W BJT can only dissipate say, 80 W at a Vce of 80V (1A max at 80V). Maybe you really should take a good look at a BJT power SOA curve at this point.
Jan Didden |
|
|
| mzzj |
| quote: | Originally posted by janneman
You are right, of course. Power FETS are not subject to secondary breakdown. This occurs in BJT as a result of high current with high Vce. That causes what is called "current hogging" where a small island on the die heats up more than another part, the hottest part will conduct more current, will become hotter, will conduct more etc. Because in effect only a very small part of the total chip or die will then conduct all the current, the allowed Pc gets much smaller. This current hogging occurs especially with higher Vce, so that is why the SOA shown much less allowed Pc at higher Vce then what you would expect from the DC Pc number.
Power FETs dont have this, so they can absorb higher powers at higher Vds compared with similar bipolars.
Jan Didden |
APT seem to have different idea about mosfet SOA and current hogging. Maybe i should have made separate topic on this...
http://www.advancedpower.com/TechSu...px?id=3855&fd=0 |
|
|
| EWorkshop1708 |
I just add as many pairs as possible.
Why not, if you are able to afford multiple transistors, (or got a bunch of samples ;) ) parallel a bunch and have a really rugged output stage.
So in the bad case of a speaker connected with too low impedance, or the load gets reactive, you will be sure the amp will handle it. |
|
|
| Workhorse |
Hi Jan & everyone,
In Traditional Output Device Protections designer usually consider mainly 2 parameters Vds & Ids across the output device and than implement a SOA protection...which is though a very good in case of Reactive Loads & Resistive Loads.....
But when a Pure Resistive Short Circuit is encountered, the SOA limiter Still limits the output but the Protection always goes on cycling again & again whenever the threshold of over-current detection is crossed or reached but this still leaves some finite output current which is always greater than Idle current through the output device in No signal conditions and thus results in temperature rise.....if not dealt with some means through which one could make a Difference between an Over-Current & Short-Circuit...
I have a rather simple concept in my mind which uses Three Parameters Vds, Ids , Vout to detect the Status of Device health and upon some bad load limits the Gate drive of output Mosfet....
In case of Short-circuit.....The Idss rises very abruptly and at the same time Vout equals to Zero or tends to zero....If this condition is considered well enough than the circuit could be made to limit the gate drive of mosfet at once and there would be no need of cycling of Idss to trigger again & again the overcurrent threshold... because we have the Third key parameter Vout which is there to tell that there is no output and limit could be engaged indefinately untill the short circuit is removed and hence no dissipation of output results.....
Attached is a simple block circuit
Lets consider Short circuit
In case of normal operation the Vout would be always present & optocoupler would be OFF hence no limit
In case of Short Circuit ..The Vout tends to Zero and optocoupler gets ON and limits the Gatedrive....hence Ids is limited to idling current due to the bias offset Zener....[the value of Vz zener is choosen to offset the bias voltage upto 4 volts]
Waiting to hear any helpful suggestions regarding Pros-Cons of this idea....
K a n w a r |
|
|
| Workhorse |
| :confused: :scratch: :tons: |
|
|
| janneman |
Kanwar,
Basically you are right that in case of dead short there may be cycling, although that depends on the exact implementation of the protection circuit. Even if there is no cycling, the prolonged overload will lead to temp rise.
There are several ways to deal with that. The simplest is a shut down in case of overtemp. You would need that anyway in professional amps, so this feature for the SOA protection comes for free ;)
Another one is to make the protection system latching, so that after an overload, you have to reset it. I don't think that you would want that in pro audio.
Your proposd solution is in my view not optimal. For one, how do you decide if there is a dead short? Load 1 ohms? Load 0.1 ohms? There will always be border cases, and a dead short never exactly is a dead short. Experience tells us that when you try to define this kind of things, along comes a case that slips through the cracks. Another real issue is that if you built an ingenious system, it is almost impossible to test it under all conditions and you can end up with an amp that fails in order to protect the protection circuit...
Better is to build something that unconditionally protects the amp. The SOA will protect the output devices. An offset detector can protect the speakers in case something goes wrong in the amp anyway. A temp detector can protect against high temp either from prolonged heavvy duty in high ambient temp and as a bonus against prolonged overloads.
Wasn't it Einstein who said, "things must be made as simple as possible, but not simpler".
Jan Didden |
|
|
| Workhorse |
Hi Jan,
Yes you are right, the Dead Short only exists when its just happened at the output terminals of an amp...buts there is always a trap there if the short is encountered at speaker terminals then thats not the dead short because the Length of wire connecting speakers to amp has some finite resistance in order of few milliohms..which might ceases short circuit protection to function....
Secondly, Its also very difficult to make a difference in a set of conditions when a Short -Circuit or a Reactive load is encountered, in both cases..At Zero crossing of voltage during reactive loads..there is still finite current flowing through the device and similarly in case of dead short circuit the output voltage is zero, but current through device is finite...
Thirdly, Another problem is that when a short is encountered, The amp usually turns into open-loop type[as output is shorted to ground]..and if the power suppply rails are stiff...than the device which is activated to conduct the current would simply function as a power resistor and would now be conducting very high current with full Vds across it, which results in Very high dissipation and even failure because its unable to drive the Dead-Short Circuit ....
The SOA limiter if encountered with overcurrent or short circuit, then it starts cycling , which could give rise to High Temperature conditions, if not dealt with appropriate Thermal Breaker type protection..would eventually destrys the devices in prolonged stressful conditions....
But if we , further look into that matter there is still a point exists which shows us a difference between both short circuit and reactive load activity...and thats the output voltage..
If by some means we process the data obtained from Ids and Vout and ADD a Time Base Delay in order to overcome the False triggering during Reactive loads[Zero-Crossing voltage- finite current conduction similarity] then the short circuit protection could be made more effective....
Lets consider an example....
In an amp..if we implement another limiter favourably optocoupled one along with already present SOA limiter....
At idle the opto limiter would be engaged fully ON but the limiting wouldnot be activated as long as the gate voltage increases to overcome the Zener voltage[Vz is set to over come the usual gate bias which is 3 to 4 volts], But when ever a voltage signal is detected at the output , the gate voltage also rises but at same time the opto coupler is made to disengage thus no limiting......But whenever a short or severe over-current condition occurs ...2 things would happen at sametime..1.Gate voltage rises 2.output voltage drops to near zero..Simuntaneously the opto is engaged at that time and then limit the increase gate voltage to around Vz of zener and thus limits the Id conduction through the device...and its another feature is that its self-resetting..no cycling. no high temp rise during prolonged short circuit..self latching because as long as short exists there would be no or very low voltage at output hence opto limiter would be engaged thus limits indefinately.....though a Time Base Delay must be implemented to over come the false triggering during reactive loads ... Also SOA limiter is there to protect in case of excursion during reactive activities...
Correct me If i were wrong.....
here's the link to QSC short circuit limiting patent for reference....
http://patft.uspto.gov/netacgi/nph-...N/QSC&RS=AN/QSC
K a n w a r |
|
|
| AndrewT |
Hi Workhorse,
no. the SOAR limiter is there to prevent the output devices being driven beyond their specifications.
If the output stage is well designed it will be capable of what the designer chose for it to do. i.e. the designer decides on the load conditions and selects components accordingly.
If both stages of this design process are carried out competently then the limiter will never trigger for a valid load condition. i.e. never trigger early on any conventional signal condition nor on any loading within the design requirements.
The SOAR limiter can operate for both the excess load driving and short circuit/bad load conditions. It can also be made latching although very unusual.
Continuous and excessive DC offset (speaker protection) is better met by a separate detector and latching isolator (fuses even?). |
|
|
| Workhorse |
| quote: | Originally posted by AndrewT
Hi Workhorse,
no. the SOAR limiter is there to prevent the output devices being driven beyond their specifications.
If the output stage is well designed it will be capable of what the designer chose for it to do. i.e. the designer decides on the load conditions and selects components accordingly.
If both stages of this design process are carried out competently then the limiter will never trigger for a valid load condition. i.e. never trigger early on any conventional signal condition nor on any loading within the design requirements.
The SOAR limiter can operate for both the excess load driving and short circuit/bad load conditions. It can also be made latching although very unusual.
Continuous and excessive DC offset (speaker protection) is better met by a separate detector and latching isolator (fuses even?). |
Well said AndrewT.....
But i am searching for some more Concrete solution , rather than relying on SOA Limiter & Thermal breaker to carry out the Task.........
Prolonged heating of Transistor Die....results in much less life expectancy.......IMHO
K a n w a r |
|
|
|
