Sorry, but this is the world of electronics and physics, not fantasy design land, you look at any component and you get a data sheets, that gives you information regarding that component so you can engineer it into your system.
life is not to short to provide data regarding a component, that is how this electronic design thingy works, by rational engineering, not by waving a wand and invoking magic fairy dust.
Esoteric Audio is full of suspect components that work by magic and have no data to back them up...to design with such components is comical, nay farcical, and it detracts from real issues, real education and reality.
I would presume you consider yourself to be wary of things such as email offering many millions of foreign capital for little outlay, yet will happily promote/defend or use a component with NO actual data to back it up other than anecdotal tales and the obligatory reference to military secrets...
life is not to short to provide data regarding a component, that is how this electronic design thingy works, by rational engineering, not by waving a wand and invoking magic fairy dust.
Esoteric Audio is full of suspect components that work by magic and have no data to back them up...to design with such components is comical, nay farcical, and it detracts from real issues, real education and reality.
I would presume you consider yourself to be wary of things such as email offering many millions of foreign capital for little outlay, yet will happily promote/defend or use a component with NO actual data to back it up other than anecdotal tales and the obligatory reference to military secrets...
Yeah, remember those 70's/80's integrated amps with low cut switches ?....cutting the VLF bottom end improved the sound but also wrecked the bass sound by introducing phase funnies.
Agreed, running very quiet 1/f amplification stages is a big help, but there is more fundamental stuff going on.
Dan.
Pretty much all of us listen to ferrite speaker magnets.
Alnico magnets sound different.
Ditto Neodymium magnets.
DC electromagnet speakers are another sound.
Dan.
Andrew, I well respect your abilities and your designs.
No supporting evidence? As expected.
I suppose you need special hearing abilities to detect the ferrite sound?
Please see above post.
Dan.
Ferrite speaker magnets in rescue of ferrite beads ?
Different material compositions, totally different working principles.
That said, hard magnetic materials differ from each other at tens of points but none of them has a sound attached to it.
Magnets (electromagnets or permanent magnets) on a speaker have to provide top plate gap’s magnetic flux.
For to do that in the most effective way, designers have to work with load lines and operating points, different for each magnetic material and magnet shape (for a given speaker).
Whatever they do, they do not ‘voice’ the magnets.
George
I have experienced this. Stable means the device will survive. Unstable means that if nothing is done very quickly, the device will fail.
In testing 4 inch wafers of a small signal NPN using an IMPACT II test system, I came across a very interesting problem. Every die on the wafer was passing wafer probe test limits, but when a small section of the wafer was cut off and chips sawn out, mounted, wire-bonded and tested for actual device parameters, every 4th device was failing. During visual inspection under a metallurgical scope using polarized light showed that there was a small melt spot on the exact same location on the exact same emitter finger of every 4th die.
I found the specific breakdown test which was causing the problem, BVce(sus). The Impact II test system will test a die by applying only the test limit to the device, and simply determines pass/fail that way. When you wish to know the actual value, the system will use an SAR routine to locate the actual value. During the test, the system turns the device on by applying the test current to the emitter and watches the base while slewing the collector voltage up. If breakdown occurs, the base will be pulled up off ground, and the test "fails", the breakdown value is below the tested voltage. If the test does not fail, the system sets a higher voltage, and slews the collector up again. This SAR routine eventually finds the correct value of the device.
The bad chips were passing the 60 volt test, and then failing every subsequent test, this because the next higher voltage tried blew the device, and the system eventually decided that the last passed value of 60 volts, was the device parameter. Note that the system never rechecks a passed value. So I was holding a test datasheet that had every 4th die as 60 volts exactly, the other 75% of the devices had a distribution of values consistent with the product, and the 60 volters were all DOA.
I investigated, determined that there was a primary mask artifact at the manu (I spoke to their diffusion engineers, learned about "primary mask" from them), and suspected the failure mechanism to be punch through (reach through). At the bad spot, the base width was reducing to zero at a high voltage, so the gain around the edge of that spot was high, and there was punch through conduction at the spot (not gain forced). I also determined that the high slew rate of the IMPACT system was preventing the horizontal spread of the hot spot area. (the high gain of the edges will spread the conduction hotspot if given sufficient time to use up the heat capacity of the silicon; if you slew too fast, you exceed that horizontal spread rate)). When I tested each device on an 576 curve tracer, the devices would not blow (it uses a 60 hz half sine waveform). Subsequent testing on the Impact would indeed destroy the devices even after passing on the 576. So, for low speed slew rates, the hot spot was stable, but for high slew rises during active ops, the hotspot is unstable.
A simple check for this phenomena is the comparison of Icbo Vs Ices, or BVcbo vs BVces, BVces being lower. If the device has the problem, these two will be different. This is easily seen on the 576 as well.
This is discussed in A. S. Grove, "Physics and Technology of Semiconductor Devices", page 230, John Wiley and Sons, 1967.
Note that the manufacturer was not required to recall or fix anything. This is because the devices were 40 volt devices, we bought them screened on wafer to 60 volts. The vendor never performed datalogging on the devices, so would never have found the problem.
An odd set of conclusions.
I have blown power mosfets in SMPS applications as a result of exceeding the slew rate limits of the devices.
The gate structure of these puppies is that of a honeycomb polysilicon web buried across the chip under the source metalization. As a consequence of the gate electrode horizontal resistivity and the capacitance to source and drain, there is a propagation velocity for gate signals across the die. If there is a large drain voltage and the device is turned on too fast (extreme gate drive and very low drain loop impedance), the device will destroy itself at the gate bonding pad because the cells there will attempt to carry the full drain current before enough cells are turned on.
Note they provide a picture showing the device blowing up right at the gate pad on the left.
If the device is fully on carrying drain current, and then the gate turned off extremely fast, the wave of cells turning off will run across the chip, and the per-cell current density will increase until failure occurs. This failure will be the farthest away from the gate pad, they have provided a picture of that as well.
I used this failure pattern to adapt the snubbers in the circuit to drop the offending slew rates.
This type of current slew rate issue is widely known with hockey puck switching devices, as the problem is more severe when the silicon is 3 to 4 inches in diameter.
Pardon any typos, too much to spellcheck..
jn
Last edited:
Elves, I use passive PFC.
Marcee, go make your own successful audio business based on anything, anything at all, and come back to complain that everyone is a duffous.
Jn, I enjoyed reading your dealing with a device. I much enjoy troubleshooting in that way, the intricate nature of fine details for which we take for granted at times.
Marcee, go make your own successful audio business based on anything, anything at all, and come back to complain that everyone is a duffous.
Jn, I enjoyed reading your dealing with a device. I much enjoy troubleshooting in that way, the intricate nature of fine details for which we take for granted at times.
It was group delay funnies. GD is most easily heard in the low freqs rather than the highs. GD can be EQ'ed flat or below audible threshold and then get benefit of no 1/f noise mucking up the over-all sound. You would just factor in the natural roll-off at the bass speaker with the cut-off filter to get low GD.
End result is Very clean, pure, accurate over-all improvement in sound.
Good chance of this occuring more and more ---- There is a trend towards more integration in system design..... those products can use DSP in them and produce excelllent audio reproduction. Maybe the new Hi Res systems coming out (and being shown at CES right now) will take us further towards better sound --- at least for the masses.
THx-RNMarsh
Last edited:
Okay, did some more looking - and the stable hot spot situation doesn't ensure self-destruction, rather the phrase is that often failure occurs because of "apparent avalanche of the collector-base junction, often at a voltage substantially below the rated VCEO of the transistor", at turn-off.
Hmmm ... a big "maybe" - as far as I can see no-one has a decent handle on guaranteeing good sound, it's more fluke that anything else - and change one element in the environment, something that doesn't make sense, and that high quality evaporates.
From my perspective, most are deniers; and the believers are floundering, trying to make sense of it all ..
- Status
- Not open for further replies.