It's my Cascode Bride Of Zen circuit, which I finally boxed up. I just posted about the problem here:
http://www.diyaudio.com/forums/pass-labs/157097-cascode-boz-4.html#post2444781
I think the circuit should be quiet, but probably just the way it's wired. Funny thing is I did a cascode Bride of Son of Zen circuit which is really quiet in much the same way. Oh well, tiime to move on I guess.
http://www.diyaudio.com/forums/pass-labs/157097-cascode-boz-4.html#post2444781
I think the circuit should be quiet, but probably just the way it's wired. Funny thing is I did a cascode Bride of Son of Zen circuit which is really quiet in much the same way. Oh well, tiime to move on I guess.
)When you first put in the 9510's you probably damaged something else that has not outright failed but has become noisy as a result.
Many years back when I was visiting Bose, I showed some of their engineers a problem with their model 402 loudspeaker. It is a small sound column with the four drivers tilted to maintain a constant coverage angle even as high frequency beaming occurs. It also has a plastic diverter the spreads the very highs and provides a place to attach the grille. To my ear the loudspeaker had a sibilance problem which I reduced by filling a structural void in the diverter.
While playing at Bose we set up 2 of the same model 402s one with stuffing one without. Some of the engineers could hear the difference others could not. It was suggested we were only listening to a typical production difference not the result of the stuffing, so we swapped speakers around. The sibilance change actually was less but still there.
So as a bit of motivation I bet that they could not measure the difference between stuffed or not. They had built their own FFT based measurement system. They quickly set up a nice B&K style microphone and power supply to their in house A/D and computer system and ran both curves.
The curves did show a bit of difference around 4,000hz. But what was interesting was the phase plot. At the point of interest it went off the top of the scale (180 degrees) and then came back in from the bottom (-180). One of the engineers apologized as he "knew" it was an error in their software that they had not yet tracked down and fixed.
So does anyone want to guess what caused the sibilance and why stuffing fiberglass into a void in front of the drivers reduced it? Why did the phase plot show a 360?
While playing at Bose we set up 2 of the same model 402s one with stuffing one without. Some of the engineers could hear the difference others could not. It was suggested we were only listening to a typical production difference not the result of the stuffing, so we swapped speakers around. The sibilance change actually was less but still there.
So as a bit of motivation I bet that they could not measure the difference between stuffed or not. They had built their own FFT based measurement system. They quickly set up a nice B&K style microphone and power supply to their in house A/D and computer system and ran both curves.
The curves did show a bit of difference around 4,000hz. But what was interesting was the phase plot. At the point of interest it went off the top of the scale (180 degrees) and then came back in from the bottom (-180). One of the engineers apologized as he "knew" it was an error in their software that they had not yet tracked down and fixed.
So does anyone want to guess what caused the sibilance and why stuffing fiberglass into a void in front of the drivers reduced it? Why did the phase plot show a 360?
Last edited:
Probably no displayable scale above 180°, so it suddenly jumps to -180°. This is normal for speaker measurement software, phase plot.
Yeah, that sounds reasonable. I don't think this circuit should be as noisy as it is at -81dBV. Though, I don't know if I want to go back and match a bunch of JFETs and start over, at least not for now, as I'm working on something else.
I put some sheilding around the trasnformer, and grounded it. It's covered in black tape so it's not visible.
Yeah, I agree. Though, the clip leads are just used to connect the shunt regulator to the circuit. And suprisingly it still does a real good job on the regulator side. I turned the regulator on the PCB into a capacitance multiplier and that feeds the shunt regulator.
The JFETs are kind of far from the MOSFETs though and that's probably not so great, nor is having a long feedback wire. So, I'm reluctant to try to make it work as is, and I don't plan on starting over.
Will see how it sounds though, noise and all.
The JFETs are kind of far from the MOSFETs though and that's probably not so great, nor is having a long feedback wire. So, I'm reluctant to try to make it work as is, and I don't plan on starting over.
Will see how it sounds though, noise and all.
I hope that 'down home' circuit design does not go away. Sometimes, you just have to try things, in order to get a 'feel' for the devices. In this world of computer simulation, it would seem that a computer would get you there first, and it often does, but not giving you the learning curve necessary, sometimes.
The biggest problem is the MODELS of the devices. I am sure that many have improved, BUT will they tell you which manufacturer they came from? Many devices deviate from the ideal in the textbook. For example, I have about 100 ea of probably the quietest complementary transistors ever made. Hi beta too! Yet, they were found to be useless in my patented Levinson JC-1 circuit. 10 times more distortion than any other devices. Why? I could show you with a transistor curve tracer, but it is NOT on the spec sheet, and I very much doubt in the SPICE model of the devices.
The biggest problem is the MODELS of the devices. I am sure that many have improved, BUT will they tell you which manufacturer they came from? Many devices deviate from the ideal in the textbook. For example, I have about 100 ea of probably the quietest complementary transistors ever made. Hi beta too! Yet, they were found to be useless in my patented Levinson JC-1 circuit. 10 times more distortion than any other devices. Why? I could show you with a transistor curve tracer, but it is NOT on the spec sheet, and I very much doubt in the SPICE model of the devices.
Why not, I'll take a stab at it. Stuffing fiberglass in front of a source of sound absorbs sound selectively, more effective at absorbing high frequencies. My guess is that the sibilance is due to a high frequency peak directly on the speaker's axis. This is a typical problem for loudspeakers that rely on relatively large cones for high frequencies unless they have a significant hf rolloff. Even the relatively small KLH Model 6 driver later used in other speakers like models 5, 12, and 17 had poor off axis response. It is virtually impossible for a single driver to have both flat on axis FR and flat total power output unless all of its energy is funneled through a tube in a single direction....or it is a pulsating sphere....or it has zero dimension.
The strange and interesing FR distortion produced by Snell type A, at least that's the way I hear it, is IMO due to a small piece of foam glued to the front of the tweeter. It has a similar effect I managed to duplicate the effect by other means. Lateral reflected energy in the vacinity of around 7 ti 8 khz greater than the the directly incident wave creates the effect. Once I was able to duplicate it at will and spent a day listening to it, I completely lost interest in that speaker.
- Status
- Not open for further replies.