get off the stump
seems to be an awful lot of huffing and puffing, and not much doing. If it's importanat enough to you, then do the experiment yourself. We're all ears.
seems to be an awful lot of huffing and puffing, and not much doing. If it's importanat enough to you, then do the experiment yourself. We're all ears.
3 days Sin Bin for repeated OT and now personal attack. This is an experimental thread. Not the old armchair thread. The ones who do the experiment set the tests.
We'd see plenty of content in frequency bins not present in the input signal at integer multiples of the input frequency. That's called harmonic distortion. The levels in those bins would be relatively independent of the input signal level, so as the signal level fell their energy relative to the signal would increase. Below a certain signal level corresponding to the width of the deadband we'd see nothing at all.
Grasped the concept now?
Hi abraxalito,
I wanted to say that. Shows you what breaking for supper will do.
SY, if I'm not very much mistaken, a ceramic core as used in electronic components is both non-conductive and chemically inert. Good when using ceramic to support the stuff that does what you are looking for - like a resistance for example. It isn't formed in a manner that might suggest it's use as a dielectric (like a ceramic capacitor).
My point?
Electrons do not easily travel through an insulator unless extreme electric fields are present. Measuring the ceramic form with any old ohmmeter (two wire measurement preferred) should suffice, and would also show any conductivity to the resistive element, should that be suspected. If you want to have a great deal of fun, use a hi-pot tester.
Anything a normal half-decent meter will not show will not be of sufficient magnitude to contribute to what can be heard.
-Chris
DARN!!
I wanted to say that. Shows you what breaking for supper will do.
SY, if I'm not very much mistaken, a ceramic core as used in electronic components is both non-conductive and chemically inert. Good when using ceramic to support the stuff that does what you are looking for - like a resistance for example. It isn't formed in a manner that might suggest it's use as a dielectric (like a ceramic capacitor).
My point?
Electrons do not easily travel through an insulator unless extreme electric fields are present. Measuring the ceramic form with any old ohmmeter (two wire measurement preferred) should suffice, and would also show any conductivity to the resistive element, should that be suspected. If you want to have a great deal of fun, use a hi-pot tester.
Anything a normal half-decent meter will not show will not be of sufficient magnitude to contribute to what can be heard.-Chris
I have a question.
Can we get or make a very high resolution digital recorder and ship it around to a bunch of different test systems, along with some BQPs and get some recordings from all the systems at the speakers running with and without the Bybees, however they may be used? I thought I read a while back some discussion about analysis software but you need the data first no matter what happens to it later. This way it could be possible to find in such a broad database a function or range of functions that account for the effect they're going to try to describe in the other thread. And possibly we could have the recordings to go along with their descriptions. Might even lead to better physical test ideas on the device itself. It may not wind up necessary to worry about matching clocks and samples blah blah. Maybe something will show up in standard audio test equipment, analog or digital. I mean we can ask them to run some tones too, or any signal. I can pitch in 50 bucks.
Can we get or make a very high resolution digital recorder and ship it around to a bunch of different test systems, along with some BQPs and get some recordings from all the systems at the speakers running with and without the Bybees, however they may be used? I thought I read a while back some discussion about analysis software but you need the data first no matter what happens to it later. This way it could be possible to find in such a broad database a function or range of functions that account for the effect they're going to try to describe in the other thread. And possibly we could have the recordings to go along with their descriptions. Might even lead to better physical test ideas on the device itself. It may not wind up necessary to worry about matching clocks and samples blah blah. Maybe something will show up in standard audio test equipment, analog or digital. I mean we can ask them to run some tones too, or any signal. I can pitch in 50 bucks.
Last edited:
Aaaaah - an out-of-control test process...
How would you EVER reliably control for the different test environments? And isn't it a little run-beorre-walk? I mean, it would surely make sense to see if just ONE system exhibited an identifiable repeatable effect....
The way things have gone so far with this I see no cart-horse relationship anyway. Many of the things so far planned can be done independently in any order. I'm not even suggesting that it's important to have a copy of the realtime drive signal. If there is the kind of difference in the "sound" some are claiming, it is possible (that's all I can say) that a difference could be trapped between Bybee/no bybee runs through the same source material. And we might have descriptions to go along with it, which Could wind up being as interesting when looking at the data as a very tighly controlled signal/analysis combination when looking for something in particular that may not be something the Bybee affects. I see that this is somewhat contrary to ideas about the limits of running tests with the intention to analyze a particular aspect of performance. I'd like to think that can be done and then know it all, but the people who think they might have an idea how these things work, or even make them, can't show a meaningful measurement from going the classic route. I'm just trying not to lose a lead or time. I agree with your last statement, but nothing about this thing makes sense. Finally, I think it would be interesting even if the ears in the group hear something but nobody is ever able to find a difference, even in the future. One problem is we can't have as impressive of a test on one system with one group of people unless camp is pitched somewhere and everyone shows up.
Last edited:
I hears ya. There may be a sound card that can cut it by now, I haven't been into high end PC stuff, but I know it's there. I figure you mean something coming back off the coil could contaminate the signal, but I Believe that if a difference is heard then that difference will be on the coil too. Maybe we can get a manufacturer to loan or rent their flagship product. I was thinking it might require something outboard the PC to be "trustable" enough.
Last edited:
Chris,
If you are doing that on a Windows box and you are not already using it, you might want to check out ASIO.
Note especially the part about how the Windows sound mixing kernel is not "bit perfect".
If you already know all that, then nevermind
Hi Andrew,
Most sound cards would be okay in SY's hands here. All we need to do is capture the signal at the loudspeaker end - say three or four runs. Then install the Bybee device and run another three or four runs. That will give you a number of sound files that can be compared by everyone with a 'puter across the world. A picture of each setup would confirm the procedure for everyone, and they can also run like (with or without) Bybee devices to make sure there aren't any uncontrollable factors that would make the test completely invalid.
I'd rather not see a dummy load used, but using a real loudspeaker in a box provides a source of contamination. If that can be solved, you have files everyone can examine. That should make everyone happy I would think.
-Chris
Most sound cards would be okay in SY's hands here. All we need to do is capture the signal at the loudspeaker end - say three or four runs. Then install the Bybee device and run another three or four runs. That will give you a number of sound files that can be compared by everyone with a 'puter across the world. A picture of each setup would confirm the procedure for everyone, and they can also run like (with or without) Bybee devices to make sure there aren't any uncontrollable factors that would make the test completely invalid.
I'd rather not see a dummy load used, but using a real loudspeaker in a box provides a source of contamination. If that can be solved, you have files everyone can examine. That should make everyone happy I would think.
-Chris
Hi Jacques,
Still, these errors should be pretty similar for each recording, and that's one reason why there are three to four recordings of the same setup,and for each setup. You did bring up a valuable point though.
-Chris
I certainly do. I've posted many times on the fallacy of "bit perfect" recordings.
Still, these errors should be pretty similar for each recording, and that's one reason why there are three to four recordings of the same setup,and for each setup. You did bring up a valuable point though.
-Chris
Theoretically, this could be a useful setup. In practice however, inductive reactance of the individual pieces will confound the measurement. It would be very difficult to assure that each device sees the exact same current through the audio bandwidth, and that the voltage taps required to view either the current or the voltage across the devices do not pickup time varying magnetic fields from the units.
Indeed, even the resistors themselves will measure differently as a result of the manufacturing process:::
BeO thin film resistors are made by deposition of the resistive material onto a cylinder of the ceramic. When a specific value of resistance is wanted, the manufacturer will use a laser to cut a spiral track into the film, and this raises the resistance of the device. This process is continued until the device meets the tolerance requirements for resistance. If you specify a 5% device, the trimming will stop when the device is just less than 5% below the nominal. If the manu programs the laser to burst cut the film when it approaches the tolerance band, they may have an algorithm which will approximate the amount of material to cut to get closer to the nominal value.
Needless to say, it is very possible for resistors from the same batch to have different thermal, magnetic, and hf characteristics piece to piece, and definitely from batch to batch.
The "infintesimal voltage drop bidirectional diode" statement is consistent with the tests done by VDH et al, and the later trials by JC. Measurement that low are error prone by nature, and extreme efforts are required in the test design to eliminate confounders. EMC issues are a prime confounder as this involves ground loop current "anarchy".. AP designed much of this out via isolated drive, JC and VDH did however, do an exceptionally fine job considering the equipment they had to work with..
BTW, axial lead resistors are typically dissipation rated with a specific lead length tied to a specific PC trace pad size. Covering the device does not impact the dissipation maximum by a lot, as radiation/convection do not normally play a big role in the rating.
Cheers, John
Mr John,
as in lead size determined ?
So what about sleeving the leads all the way up/down to the solder pad ?
(i'm known to emit X-rated rays, according to Langmuir that labels me as pathological)
The sleeve can only impact the lead radiation and convection, which is rather small given the dimensions.
The long term thermal rating of any device is dependent on the thermal conductivity of the path which takes the heat from the device to the "sink".
For a solid copper path as an example, the equation relating the thermal drop of the path is this: (you had to expect I'd use an equation, no??
..)R = t/K*A
R is thermal resistance in degrees C per watt
t is thickness in inches
A is area in square inches
K is the material constant in watts/degree C-inch.
For a copper cube of 1 inch dimension:
K = 10.2
R = 1/ (1 times 10.2)
A cube of copper will have a thermal resistance of 1/10.2, or .098 degrees C rise for every watt dissipated.
For a 40 mil diameter lead made of copper and 3/8 inch long..(3/8 will be the thickness in the equation as that is the heat path.)
A= pi (.02) squared.... 3.1415 x .02 x .02 = .00125 square inches
3/8 = .375 inches
R = .375/ (10.2 * .00125) = 29.4
A 40 mil copper wire 3/8 inch long will have a thermal resistance of 29.4 degrees C per watt. Using two of these to sink a resistor will produce a derating of 29.4/2, or 14.7 degrees C per watt. edit: note that this is only the lead temperature rise, it does not explain the temperature rise of a PC pad which the lead solders to, that is an additional entity to be considered.
Typical K's:
Cu: 10.2
Alu: 5.3
Ag: 10.84
Au: 8.06
Iron: .9006
Nickel: 2.14
BeO: (340-T)/61.7 (6.3 at 25 C)
Cheers, John
ps..I have to apologize, all of this is from memory, but I believe the numbers are accurate...except for the BeO formula, something there is troubling me...I'll figure it out...I think Brush wellman has the correct formula..
Last edited:
Hi John,
The only point I disagree with is the impact of a blocked airflow on a resistor body. Good air ventilation will remove a noticeable amount of heat from the part. It sets up it's own moving air column, so it's like a little tiny fan blowing away. Convection in action.
Look at the condition of PCB traces when large holes are located under a power resistor and compare that to PCBs that do not include a vent hole. Convective cooling is not the major heat path, but it can be substantial. As for the condition of the pads and traces, yeah. This doesn't seem to be well understood by many who are in a position where it matters. Pretty irrefutable evidence that the leads drain most of the heat.
I have seen power resistors covered with tubing (to guard against shock). The tubing is useless after it melts, and the resistor is useless after it opens up. Forget reading the value off the body.
-Chris
The only point I disagree with is the impact of a blocked airflow on a resistor body. Good air ventilation will remove a noticeable amount of heat from the part. It sets up it's own moving air column, so it's like a little tiny fan blowing away.
Convection in action.Look at the condition of PCB traces when large holes are located under a power resistor and compare that to PCBs that do not include a vent hole. Convective cooling is not the major heat path, but it can be substantial. As for the condition of the pads and traces, yeah. This doesn't seem to be well understood by many who are in a position where it matters. Pretty irrefutable evidence that the leads drain most of the heat.
I have seen power resistors covered with tubing (to guard against shock). The tubing is useless after it melts, and the resistor is useless after it opens up. Forget reading the value off the body.
-Chris
Hi John,
The only point I disagree with is the impact of a blocked airflow on a resistor body. Good air ventilation will remove a noticeable amount of heat from the part. It sets up it's own moving air column, so it's like a little tiny fan blowing away.
-Chris
Chris
You are not disagreeing with me, you are simply embellishing on a point that I did not discuss..
All you state is accurate... My discussion is merely about the resistor as per mil spec 19500..where there is really no concern over forced flow...
Cheers, John
- Status
- Not open for further replies.