I don't know if this has been around before, so I thought I'd try it here.
I don't like the idea of protection with relays. Too prone to relay failure. So I comtemplated my navel for a while and decided to use MOSFETs in series with both the power supplies to the amp.
Big digital switching MOSFETs are cheap, and if you have them switched full on or off you're not worried about the effect on audio quality. I set up a pretty standard DC voltage sense and used that to turn the MOSFETs in the power supply lines off when a fault was found.
It worked pretty well on the somewhat modest amp (100W) I tried it on.
I don't like the idea of protection with relays. Too prone to relay failure. So I comtemplated my navel for a while and decided to use MOSFETs in series with both the power supplies to the amp.
Big digital switching MOSFETs are cheap, and if you have them switched full on or off you're not worried about the effect on audio quality. I set up a pretty standard DC voltage sense and used that to turn the MOSFETs in the power supply lines off when a fault was found.
It worked pretty well on the somewhat modest amp (100W) I tried it on.
It's not much of a schematic, and I'd have to go draw it.
Lemme see if I can describe it. For the negative rail:
one or more n-channel MOSFETs, source to the DC supply, drain to the amp. Resistor and gate-protection zener from gate to source; drive resistor to pull-up-to-ground driver. I used a PNP, but obviously any arrangement to switch the drive resistor up to ground would work. The PNP pulls the gate up, to turn on the MOSFET, gate voltage is limited by the zener, and turn off time is determined by the gate capacitance and the gate-source resistor. I used two MOSFETS in parallel, but as many as are needed can be parelleled there to keep DC drop low.
I used P-channels on the top side because I had a tube of IRF9640's on ebay cheap, but a high-side driver chip would let you use n-channels on the positive side. In my amp, the positive side was the mirror image of the negative side.
The MOSFETs are used purely as high speed, high current switches. You put enough current handling there by picking big MOSFETs and paralleling to handle the max current excursion on the power rail and most of the time they act like smallish resistances in series with the power rails. When you detect a fault, you drop the drive to the MOSFETs and they turn off before you smoke your speakers.
The MOSFETs can switch slowly because they only switch once per power cycle, and you only have to heatsink them to get rid of the saturated-on power losses.
By turning off both power rails on a fault, you keep the speakers from being toast. Selected properly, the MOSFETs cannot have their contacts welded...
Of course, anything that you want to detect can turn them off.
Lemme see if I can describe it. For the negative rail:
one or more n-channel MOSFETs, source to the DC supply, drain to the amp. Resistor and gate-protection zener from gate to source; drive resistor to pull-up-to-ground driver. I used a PNP, but obviously any arrangement to switch the drive resistor up to ground would work. The PNP pulls the gate up, to turn on the MOSFET, gate voltage is limited by the zener, and turn off time is determined by the gate capacitance and the gate-source resistor. I used two MOSFETS in parallel, but as many as are needed can be parelleled there to keep DC drop low.
I used P-channels on the top side because I had a tube of IRF9640's on ebay cheap, but a high-side driver chip would let you use n-channels on the positive side. In my amp, the positive side was the mirror image of the negative side.
The MOSFETs are used purely as high speed, high current switches. You put enough current handling there by picking big MOSFETs and paralleling to handle the max current excursion on the power rail and most of the time they act like smallish resistances in series with the power rails. When you detect a fault, you drop the drive to the MOSFETs and they turn off before you smoke your speakers.
The MOSFETs can switch slowly because they only switch once per power cycle, and you only have to heatsink them to get rid of the saturated-on power losses.
By turning off both power rails on a fault, you keep the speakers from being toast. Selected properly, the MOSFETs cannot have their contacts welded...
Of course, anything that you want to detect can turn them off.
The original idea is that the outputs are already dead, at least one of them. If you get a sudden DC offset, that's the presumption anyway, and the reason for protection circuits that open relays.
With that rationale, I just let them have maybe a millisecond to turn off.
I suppose you might want to try something like monitoring other junk to turn the amp off, but I was just after eliminating the output relay.
Hate relays. ptuui...
With that rationale, I just let them have maybe a millisecond to turn off.
I suppose you might want to try something like monitoring other junk to turn the amp off, but I was just after eliminating the output relay.
Hate relays. ptuui...
I was wondering if VI protection was better than protection in a output stage voltage regulator. I think FETs can hack more current for a very short duration than BJTs.
Maybe for BJTs VI protection is better than having regulator protection?
Whereas with FETs DC protection in reg's maybe adequate ?
Maybe it's a response related question(s)
------
Live on the Power Mosfet......
Maybe for BJTs VI protection is better than having regulator protection?
Whereas with FETs DC protection in reg's maybe adequate ?
Maybe it's a response related question(s)
------
Live on the Power Mosfet......
I had no idea that the MOSFETs were part of any regulator. They're switches, pure and simple. When they're on, current flows to the outputs. When they're off, no current goes to the outputs. Any regulator would be separate.
It is possible to use the MOSFETs as part of a regulator, I suppose, but that would be a sophisication I didn't envision, and might compromise the regulator.
This is not intended to replace VI protection. The amplifier remains exactly as it was, internal protections, and everything as it was. All I added was a fast, high current switch in both V+ and V- lines. No regulator or protection for the outputs is intended - this is to keep my very expensive speaker drivers from going up in smoke.
If your protection circuit is set off by an 'over current' sensing circuit, you could have only the drivers and biasing current sources be switched off and use a smaller device to do the switching, in the case of a big amp. As long as the output devices don't fail, the result would be the same in that no current would flow. The only problem though with setting a max current is when driving an inductive load, where the phase margin between V & I, might create a condition that will exceed the SOA. Of course in that case... (pooooofffff!)
Only problem I can see with all this is once the DC has been detected and rails switched off, the DC is no longer there - what stops the circuit then re-activating the rails, detecting and de-activating, re-activating ... continuously pumping the speaker with DC pulses? If a latching circuit is used, then some means of resetting is neccessary, but resetting will still give the speaker another blast if the fault is still present....
Cheers
Graeme
Cheers
Graeme
I looked into this idea a few months ago. I decided to fall back to the simple, reliable and cheap passive relay in the speaker output solution. My reason being that if an output device is short circuit (thus the DC on the output), then possibly most of the rail voltage is across the interruption MOSFET, yet also a very large current is flowing. This exceeds the SOA of the device and I could not be certain to myself that it would not fail short circuit and thus blow my speakers. I didn't want to have to parallel like 3 MOSFETs per rail at 10x the cost of the passive protection.
?
richie00boy said:
A contradiction in terms...methinks
Upupa Epops said:
Yes....particularly if used purely as switches in supply rails..:
http://www.irf.com/product-info/datasheets/data/irfba90n20d.pdf
I've repaired hundreds of 70's era amps and receivers that use relays for output protection, and the relay is about the most reliable part on those things. The ONLY trouble I see is that sometimes the contacts get a bit oxidized, and I have to pop the cover off and drag a piece of #600 paper through and a flush with DeOxit to get 'em to behave.
Once I got an old Marantz that blew up so badly the contacts were horribly pitted. Another time, a Sansui amp welded the contacts together. These are the exceptions.... Seeing as how the vast majority of amps/receivers I fix have blown outputs, if relays were that bad then I ought to be buying huge amounts of them to swap out. Instead, it's a rare occasion when I do.
This is not to say that the idea of a solid-state solution such as you all are discussing is not a cool idea and shouldn't be explored. I just wanted to smack down the notion that relays are unrelaible. If the relay is properly selected for the job at hand, it'll likely be the LAST thing on that amp to fail.
I really have to agree with EchoWars on this. My experience has been the same except I've seen a few more welded contacts. I have also seen many fuses in the speaker output (and power rail) with cigarette foil around the [blown] fuse. The big amps use industrial contactors, one per channel.
-Chris
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.