Mike, don't make me an idiot 😉 . If you can use mosfet as fuse breaker at rail line, you must drive him by higher voltage than is rail voltage. This driving voltage you can get from supply, which is connected in series with main rail voltage and which you can switch for example by some fast optocopler, which is driven from protector circuit. That's all 😉 .
Ahh, I was looking for photovoltaic drivers but a fairly thorough search on the IR website revealed nothing!
OK, time for some explanation.
I believe Upupa was referring to the technique of using a transformer with a couple of independent low voltage windings. A 5VA transformer with two independent windings of say 9Vac to 12Vac can be used to make two completely floating 10Vdc power supplies by attaching a full wave bridge (4x1N400x or a 6-pin DIP bridge), and a filter capacitor (say 470uF/16V) to each one. That makes two power supplies that have no ground reference and may freely be attached anywhere in an amp.
For instance, one good place would be to have the negative of one of these supplies attached to the most positive power rail. Although it's not completely necessary, you can attach a second one to the negative power rail. Now have two floating power supplies that ride wherever the amplifier power rails take them. You can then run your now-floating MOSFET gate drivers from that by opto coupling in the on/off signal with something esoteric - like maybe a transistor-output optocoupler. This can all be completely independent of the amp, save for a connection to the AC power at the wall.
Lest we get the semi-standard "But that's complicated and expensive..." we'll do the math.
Parts:
two secondary transformer - $8.76; alternatively, two single secondary 12Vac 0.06A transformers for $2.76 each, $5.52 (Mouser Electronics)
four power diodes - $0.16 (Digikey, any of 1N400x)
two 470F 16V caps - $0.72 (Digikey)
two 4N35 Optos - $0.64
Total for two opto isolated floating power supply channels = $7.04. Or if you choose to do one channel floating on the + rail and work the - rail inside the existing power supply, $3.52.
It's not complicated, and again it all depends on what your definition of "expensive" is.
I believe Upupa was referring to the technique of using a transformer with a couple of independent low voltage windings. A 5VA transformer with two independent windings of say 9Vac to 12Vac can be used to make two completely floating 10Vdc power supplies by attaching a full wave bridge (4x1N400x or a 6-pin DIP bridge), and a filter capacitor (say 470uF/16V) to each one. That makes two power supplies that have no ground reference and may freely be attached anywhere in an amp.
For instance, one good place would be to have the negative of one of these supplies attached to the most positive power rail. Although it's not completely necessary, you can attach a second one to the negative power rail. Now have two floating power supplies that ride wherever the amplifier power rails take them. You can then run your now-floating MOSFET gate drivers from that by opto coupling in the on/off signal with something esoteric - like maybe a transistor-output optocoupler. This can all be completely independent of the amp, save for a connection to the AC power at the wall.
Lest we get the semi-standard "But that's complicated and expensive..." we'll do the math.
Parts:
two secondary transformer - $8.76; alternatively, two single secondary 12Vac 0.06A transformers for $2.76 each, $5.52 (Mouser Electronics)
four power diodes - $0.16 (Digikey, any of 1N400x)
two 470F 16V caps - $0.72 (Digikey)
two 4N35 Optos - $0.64
Total for two opto isolated floating power supply channels = $7.04. Or if you choose to do one channel floating on the + rail and work the - rail inside the existing power supply, $3.52.
It's not complicated, and again it all depends on what your definition of "expensive" is.
Try the URL I posted:
http://www.irf.com/product-info/dat...s/data/pvin.pdf
or put the term "pvi" into the IRF.com search window. I just did that, and it turned up all the PVI.... devices, and when you click on "datasheet" it gives you the URL I posted. They do not make it easy.
Also see
http://www.toshiba.com/taec/cgi-bin/display.cgi?table=Category&CategoryID=7070
for their photovoltaic couplers: TLP0190B, TLP0191B, TLP0590B, TLP0591B.
Hi RG..
Thanks for a most illuminating discussion....i knew you would come through in the end.... 😉
Thanks for a most illuminating discussion....i knew you would come through in the end.... 😉
Interesting optocouplers for this applications : CP Clare - FDA 215, IAA 110 P etc. 😉 - look at www.cpclare.com 😎
ingrast said:
Hi Rodolfo,
Yes...i misunderstood Upupa there....
But why woudn't his idea work if the transformers were 'subject to swinging'?
They are floating supplies after all....
mikeks said:
Be it purely analog amplification, worst still if switching waveforms, you have a supply (including transformer winding) swinging with the signal. Think of all that capacitance both to ground and to other sensitive circuit sections ... and EMI ...
Probably another good reason to use p-channel switches for the high side.
Rodolfo
ingrast said:
I think what RG had in mind is a simple capacitor input supply...no switching involved....
..even then i fail to see why this should be a problem to circuit operation...since your gate-drive supplies are floating....
If cheap small transformers can be obtained (not cheap in the UK!! £20!!
), then upupa's suggestion is a sound one.....as availability is far superior to that of IRF and Toshiba's photovoltaic PVIs...As for capacitive coupling, the power supply in a class-B amp should always be well away from the gain stages.....
P-channel mosfets are expensive...less readily available, and of generaly poorer performance in this application...
Don't cry so loudly, Mike, these transformers cost by Farnell up to five pounds, not twenty 😀 . Are you poor student ? 😎
Upupa Epops said:
😀
And yes you pick up a suitable transformer for £5-8 from farnell. Why shop at digikey in the US???
Just for a bit of clarity, gentlemen:
I believe the discussion of "swinging" was predicated on the thought that the low voltage power supplies would be following the signal.
They will not. They will be held (reasonably) rock solid on the power rails, at plus-and-minus whatever as Mike as pointed out. True, the rails have ripple on the, but that's a movement of a few volts even under strenuous exertion. There are no rail to rail or ground to rail swings for the low voltage power supplies to follow. In fact, the whole point is that the MOSFET switches do nothing but sit at the power supplies until they are called on to do something, at which point audio quality has ceased to be an important point, as you're trying to save the speakers.
As a second item, for the few volts they do have to swing, they are driven by the lowest-real-impedance, highest current capability source in the amplifier - the DC power supply. Since the low voltage supplies are essentially a single node (compared to the power rails), the capacitance that has to be driven is the primary-to-secondary capacitance of the small power transformer. While this is not zero, it is still fairly small - I suspect sub-nanofarad - and there is a lot of current available to drive it.
When the high side driver is called on to stop current in the upper MOSFET, it shunts the MOSFET gate to its source. At this point, the low voltage supply goes along for the ride, tied to the source of the MOSFET, as the source heads for ground and the drain remains up at the power supply. The swinging is now driven by the MOSFET shutting off, and again is trivial compared to the presumably many amperes of fault current the MOSFET is trying to stop.
So I don't think swinging issues and problems driving the upper MOSFET are either issues of normal operation or of activation under faults. Could be with poor gate driver design, I guess, but no one here would do that, right?
I believe the discussion of "swinging" was predicated on the thought that the low voltage power supplies would be following the signal.
They will not. They will be held (reasonably) rock solid on the power rails, at plus-and-minus whatever as Mike as pointed out. True, the rails have ripple on the, but that's a movement of a few volts even under strenuous exertion. There are no rail to rail or ground to rail swings for the low voltage power supplies to follow. In fact, the whole point is that the MOSFET switches do nothing but sit at the power supplies until they are called on to do something, at which point audio quality has ceased to be an important point, as you're trying to save the speakers.
As a second item, for the few volts they do have to swing, they are driven by the lowest-real-impedance, highest current capability source in the amplifier - the DC power supply. Since the low voltage supplies are essentially a single node (compared to the power rails), the capacitance that has to be driven is the primary-to-secondary capacitance of the small power transformer. While this is not zero, it is still fairly small - I suspect sub-nanofarad - and there is a lot of current available to drive it.
When the high side driver is called on to stop current in the upper MOSFET, it shunts the MOSFET gate to its source. At this point, the low voltage supply goes along for the ride, tied to the source of the MOSFET, as the source heads for ground and the drain remains up at the power supply. The swinging is now driven by the MOSFET shutting off, and again is trivial compared to the presumably many amperes of fault current the MOSFET is trying to stop.
So I don't think swinging issues and problems driving the upper MOSFET are either issues of normal operation or of activation under faults. Could be with poor gate driver design, I guess, but no one here would do that, right?
Oh, I forgot, Mike - I only used Digikey and Mouser as illustrations as they are US mail order places that have real stock that can actually be had. I wanted to produce an illustration that did not have parts made of unobtainium.
Of course you'd look in your local supplier catalogs and pick a low price supplier of a low power, low voltage transformer. 240:10:10V should work nicely, and under 5VA would be dandy.
Of course you'd look in your local supplier catalogs and pick a low price supplier of a low power, low voltage transformer. 240:10:10V should work nicely, and under 5VA would be dandy.
R.G. said:
Right, and then I was the one understanding it wrong.
At all times I was thinking about the working devices, not separate dedicated rail switches. In the latter case there is no problem whatsoever.
But then, I believe it is more efficient componentwise to drive the working power devices with protection signals. The only limitation is catastrophic failure in any of these devices compromises load protection, while dedicated switches should encapsulate the failure.
For thousand dollars speakers, the cost and complexity of added circuitry is moot.
Rodolfo
- Status
- Not open for further replies.