richie00boy said:
If your output fails short-circuit, then you'll have a large DC offset by definition....This invokes your DC offset protection, which should turn off the MOSFETs in your supply rails in a hurry...(less than 10nS).
There are cheap rugged N-channel MOSFETs available whose individual power ratings far exceed those of the vast majority of domestic power amps.
http://www.irf.com/product-info/datasheets/data/irfp4232pbf.pdf
Note that the MOSFETs in your supply rails must rated to withstand at least twice the value of each supply rail...
Mike, my whole point was that I was not confident that the MOSFETs used in the rails (I called them interruption MOSFETs) would be able to break the circuit without exceeding their SOA and thus failing short circuit so not protecting at all. I can get a good relay for £5, so the solid state equivalent has to match or better this cost and also not be too greedy on the real estate.
richie00boy said:
http://www.geofex.com/Article_Folders/ampprot/dcprot.htm
I think you'll find SOA is not a significant issue for MOSFETS used as switches....
The cost of course is another matter...
http://www.geofex.com/Article_Folders/ampprot/dcprot.htm
Ok...should the solid state relays in each supply rail be 'normally-closed' or 'normally-open'?
Thoughts anyone?
Thoughts anyone?
Thanks for the links, Mike. I go with normally open relays because under adverse conditions they are more likely to be open and thus still protect.
To mikeks : If you take this circuity as DC protection, all is easy ( relatively 😱) ), but I mean that this circuit can take function of current limiter. How you can make current sensing with low serial impedance ? By Hall probe or have you some different solution ?
Upupa Epops said:
No point involving your output stage's SOA protection in the DC offset protection system......
Each system can be entirely independent of the other....no problemo...
richie00boy said:
'Normally-open' as a fail-safe default is 'good practice' for electro-mechanical relays in series with the amp's foward-path.
Any specific reason why this should apply to their solid-state counterparts located in the supply rails instead of the signal path?
Solid state parts typically fail short. Conditions where these would operate are already not normal so excessive currents are to be expected. My vote is with a relay properly sized.
-Chris
-Chris
anatech said:
I am not at all sure that this is true infact.....I have only encountered one (very large!) relay (P&B) with a contact rating of 50A.
On the other hand a pair of these http://www.irf.com/product-info/datasheets/data/irfp4232pbf.pdf will sustain 100A at Tc=50deg. with ease.
Still, you may consider using the electro-mechanical P&B as follows:
http://www.diyaudio.com/forums/showthread.php?postid=440786#post440786
Hi mikeks,
I was talking about interrupting fault current. This can be extremely high. The turnoff characteristics become very important. If your device ratings are exceeded, the device may (probably) short. Relay contact ratings are for repeated operation. If the relay is damaged, I don't care. It's cheaper than the speaker(s). Therefore, the contact rating is for normal expected currents.
Hmmm, we don't need a 50A relay any more do we? In fact, 10 ~ 25A models should do for all but the largest amplifiers. A 3 ~ 5 A rating is fine for less than 100W in practice. The biggest problem that relays have seems to be oxidation of the contacts, then they burn. So, change the relay or clean it before you run into that.
One thing we have not mentioned is that when the output stage is disconnected from the speaker, we break a loop. I've seen amplifiers blown and speakers fried while the amp was unplugged due to close lightning strikes. That's got to be worth something depending on where you live or have a cottage. With over 25 yrs of service experience, you see many things.
-Chris
I was talking about interrupting fault current. This can be extremely high. The turnoff characteristics become very important. If your device ratings are exceeded, the device may (probably) short. Relay contact ratings are for repeated operation. If the relay is damaged, I don't care. It's cheaper than the speaker(s). Therefore, the contact rating is for normal expected currents.
Hmmm, we don't need a 50A relay any more do we? In fact, 10 ~ 25A models should do for all but the largest amplifiers. A 3 ~ 5 A rating is fine for less than 100W in practice. The biggest problem that relays have seems to be oxidation of the contacts, then they burn. So, change the relay or clean it before you run into that.
One thing we have not mentioned is that when the output stage is disconnected from the speaker, we break a loop. I've seen amplifiers blown and speakers fried while the amp was unplugged due to close lightning strikes. That's got to be worth something depending on where you live or have a cottage. With over 25 yrs of service experience, you see many things.
-Chris
anatech said:
Yes...if fault current is very high, then this is one good reason to avoid EM relays i would have thought, as these would probably have their contacts welded together as they tried to break the circuit.....In which case bye-bye 'speakers!!
MOSFETs are immune to this.
They can turn off faster than any EM relay known to man...no contact bounce....no arcing....no welded contacts.....no fuss.
OK Mikeks, I f...k on your theory, show us your concrete solution, or .... Is easy to present some theory without results.
Well, that depends on the circuit turning off the fets. In practice I have only seen a few welded contacts, and those were plainly too small a relay for the unit. I have seen more lightning damaged gear with no relay.
Each method can be valid and they each have their own problems. I tend to trust a part more than a circuit design these days. Another approach comes from Carver (defunct). A fault turns off the switching supply. It works very well and limits damage in the amplifier. The housekeeping circuits keep the unit off once this occurs. I like this better than trying to turn the rails off. What do you think?
-Chris
Each method can be valid and they each have their own problems. I tend to trust a part more than a circuit design these days. Another approach comes from Carver (defunct). A fault turns off the switching supply. It works very well and limits damage in the amplifier. The housekeeping circuits keep the unit off once this occurs. I like this better than trying to turn the rails off. What do you think?
-Chris
Upupa Epops said:
I f...k on you too old chap... 🙂
It's not 'my theory' as a matter of fact...see post #1
anatech said:
Actually it's the same principal.....
Not really, the primary currents are shut down (before the xformer) with one or two (PM2.0) devices. There is no direct impact on the rectified DC. I guess you could say the overall priciple is the same, but that's as close as it gets.
-Chris
-Chris
Whatever works and limits the energy. This works best in the PM2.0 as it has a higher switching frequency and small caps. The protection is very good at determining the diff between heavy signal and fault.
I don't like the sound of the amp (that's not related) but the darn things really pound.
-Chris
I don't like the sound of the amp (that's not related) but the darn things really pound.
-Chris
Greeting all. I just want to go back to RG's original point.
Firstly, it is a good/new idea. Thanks. But if we stick to the problem that RG has solved (ie. protecting expensive loud speakers from raw DC without degrading the signal) there is another option. I take it that most of us would like to avoid relays in the signal path to the speaker for sound quality/reliabilty reasons. Also I guess most of us would like to minimise the number of components between the PSU and the amp' for the same reasons. But if we do insist on a solid, uninterupted wire for these two functions there is still another option that will protect speakers.
I think this will work. I ran it in another thread last year and no one jumped on it. Instead of a relay in the signal path to the l/speaker you could put a relay across the output terminals of the amp'. (That is, in parrallel with the l/s) This of course would normally be open circuit during safe conditions but in the event of DC appearing at the output it would close and shunt the current to earth thereby protecting the speaker.
Things would need to be rated appropriately and the whole lot (amp/ PSU)can be shut down by having a slow blow fuse on the primary side of the PS transformer.
Firstly, it is a good/new idea. Thanks. But if we stick to the problem that RG has solved (ie. protecting expensive loud speakers from raw DC without degrading the signal) there is another option. I take it that most of us would like to avoid relays in the signal path to the speaker for sound quality/reliabilty reasons. Also I guess most of us would like to minimise the number of components between the PSU and the amp' for the same reasons. But if we do insist on a solid, uninterupted wire for these two functions there is still another option that will protect speakers.
I think this will work. I ran it in another thread last year and no one jumped on it. Instead of a relay in the signal path to the l/speaker you could put a relay across the output terminals of the amp'. (That is, in parrallel with the l/s) This of course would normally be open circuit during safe conditions but in the event of DC appearing at the output it would close and shunt the current to earth thereby protecting the speaker.
Things would need to be rated appropriately and the whole lot (amp/ PSU)can be shut down by having a slow blow fuse on the primary side of the PS transformer.
- Status
- Not open for further replies.