The ironic thing is that we had to purchase a Quan-Tech in order to sort the Analog Devices AD844's. They switched the process to a larger wafer size and never spent the time to dial in all of the parameters. Now we have about a 20% fallout rate. At first the people at AD would at least give us credit for the bad parts, but that only lasted about a year. Now the party line is "They meet the spec."
But it could be worse -- at least they still make the danged thing. I had high hopes when Mark Brasfield was at National and they made a clone of it. The only problem was they forgot to connect pin 5. Silly, silly people. It would have raised the price $0.001 each. Oh well...
Charles Hansen
Ayre Acoustics, Inc.
Charles,
You can always buy the bare die and get it packaged with pin 5 to the outside. I once looked into having two '844 dies packaged in a DIL16 and for a few 1000 parts it came to about $ 6 each, including the dies.
This was about 10 years ago so may now be different but there are small-scale packaging outfits that will do this for you. Your choice of packages will be limited but that should not be an issue.
Jan
You can always buy the bare die and get it packaged with pin 5 to the outside. I once looked into having two '844 dies packaged in a DIL16 and for a few 1000 parts it came to about $ 6 each, including the dies.
This was about 10 years ago so may now be different but there are small-scale packaging outfits that will do this for you. Your choice of packages will be limited but that should not be an issue.
Jan
There was a super nice unit like that published in Linear Audio- I don't have them at hand right now, but I think it was by Samuel Groner.
I had a math professor who threw chalkboard brushes at those fallen asleep. Guy could hit a forehead at 30 yards, even tried 50 once.
Surprise part was that he gave no warning, to not get caught in a cloud of chalk, one also had to pay attention neither one of the neighbors was dozing off.
To reach a larger audience, he threw objects on the floor, large books, chairs, monday morning specialty around nine thirty.
And could totally humiliate someone for 10 minutes on end, in front of several hundred students.
(one attendant was royalty, but had a security companian to keep him awake at all times. few dozen years later, he got hit by an avalanche)
Last edited:
What affect did varying Z from a source have on the measured or listened results?
I had such a Quan-Tech but gave it away to a local college about 2 years ago. Why? Because once you have tested everything of interest currently made, it just sits around oxidizing. Then, after 5-10 years, you get it out and run the same tests on all the new devices available - as many of the old ones become no longer made. Other than that, it is really useful for R&D and for QA/QC sampling/tests in production.
-RNM
Last edited:
40 years ago, we did not think that MC loading Z was important. Later, maybe 35 years ago, we found input loading can be optimized for a particular cartridge. We also found that the common base connection created a subtle, sonic change, that appeared to sound like overdamping the cartridge. In 1975, I modified a JC-1AC to be either common base or common emitter input (100 ohms), and I found that the 100 ohm loading sounded better.
Its all in the implimentation.... As you know, I had bought the JC-1DC. We talked about what my tests on batteries showed etc. Others, like Peter Moncrieff thought lower than 100 was better sound thru lowering the cartridge distortion. So, I made my own MC preamp to see what affect it had for myself.
Some MC topology designs are more sensitive to the source Z than others (not just capacitance-freq response loading) and have a broad range of parameters affected in the amp.
Have you experimented with loading and how it changes the distortion coming from the cartridge itself? Particularly if you did measurements to determine any distortion changes. Or any ideas on how such a test would be done.
Thx-RNMarsh
Some MC topology designs are more sensitive to the source Z than others (not just capacitance-freq response loading) and have a broad range of parameters affected in the amp.
Have you experimented with loading and how it changes the distortion coming from the cartridge itself? Particularly if you did measurements to determine any distortion changes. Or any ideas on how such a test would be done.
Thx-RNMarsh
Last edited:
I have shown how a test instrument like the QuanTech, can quickly and easily sort single solid state devices, or IC's regarding noise. Since it is 'ready to go' at all times, it is easier to make accurate comparative noise measurements, of an individual device, at any reasonable current, voltage, or input impedance. This makes it useful to find noise differences in different batches of the same device, or different manufacturers equivalent devices, or similar devices from the same manufacturer.
Now, I do not expect audio amateurs to go out and get a QuanTech. But we manufacturers who do have one, know its value even today.
Now, I do not expect audio amateurs to go out and get a QuanTech. But we manufacturers who do have one, know its value even today.
Today, I do not use sliding feedback like the Levinson JC-1 or the COMMON BASE input configuration like the JC-1AC or DC. I found these later devices to be a failure, sonically, even though they had more dynamic range than the original Levinson JC-1. I believe this to be because of the COMMON BASE input, and its very low input impedance.
I do find that input loading is important and the best phono units should have variable resistive loading and it should be optimized by ear, while playing a record.
I have never measured any significant change in frequency response with loading, so I don't know why loading should make a difference, but it does. Usually a default value of 100 ohms is a good compromise.
I do find that input loading is important and the best phono units should have variable resistive loading and it should be optimized by ear, while playing a record.
I have never measured any significant change in frequency response with loading, so I don't know why loading should make a difference, but it does. Usually a default value of 100 ohms is a good compromise.
I think that the low noise bipolar transistor has been adequately covered, so I might show what we did about 40years ago toward using low noise jfets, by using a QuanTech noise analyzer.
Back in the 1960's jfets were only partially developed, had relatively low Gm, generally, and high 1/f noise. It had been shown in the middle 1960's that it was POSSIBLE to get very low noise from jfets, but that manufacturing defects usually made them more noisy than necessary. It would appear that in the late 60's and the early '70's, significant developments were made, and some jfets could be almost as quiet as any jfet today, but they were expensive.
A company named Crystalonics made a very nice, military quality jfet that could be very quiet but it was relatively expensive, and some batches were noisy. I found this out with the QuanTech. They sold these devices at $30 ea, but they did NOT test them before sale. I have a whole set of these devices that are relatively noisy, and some that are very quiet. This device was Nchannel only, and was called the C413N. We had hoped to use them for microphone input stages, but they were too undependable as far as noise, AND the company did not care.
This is where I started testing a whole range of jfets from Siliconix. They were making complementary devices in some cases and they made a lot of different ones.
So, I started testing one part, called the j113. They had a range of Idss that could be 10:1, so selection was necessary. There was a similar complementary Pchannel part called the j175. Here again, the QuanTech quickly shows the N material to be superior to the P material, so if you wanted the LOWEST NOISE, back 40 years ago, you had to stick to Nchannel jfets. I then expanded my search with the QuanTech and found the J110 Nchannel jfet as a darn quiet fet, that normally was as quiet as the C413N, at 1/10th the price. This is the jfet that we used for the input for the stage for the Levinson JC-2, designed in 1973. Without the QuanTech, I would never have found this part. Unfortunately, Siliconix changed its process in about 1976, and another part had to be found.
Once the Japanese got into making quality low noise jfets, much of the struggle of buying a decent complementary jfet pair was eliminated, and we started using Toshiba jfets in the Vendetta Research phono stage.
However, in 1990, Toshiba ALSO changed their processing, AND I had to send back 100's of them, that were excessively noisy. Once again the QuanTech came to the rescue.
For those who may never have seen or used a QuanTech, let me denote some of its features, that would be hard for an amateur to imitate.
The QuanTech DC servos the device under test to whatever current you want to test at. It also has a variable test voltage, although that is not as important.
Finally, it servos an AC signal through the device to adjust for changes in Gm, both with current and device type. This might be useful for bipolars, but for jfets it is absolutely necessary to get an accurate reading.
The IC noise tester is a simpler design that amateurs could emulate fairly easy.
Back in the 1960's jfets were only partially developed, had relatively low Gm, generally, and high 1/f noise. It had been shown in the middle 1960's that it was POSSIBLE to get very low noise from jfets, but that manufacturing defects usually made them more noisy than necessary. It would appear that in the late 60's and the early '70's, significant developments were made, and some jfets could be almost as quiet as any jfet today, but they were expensive.
A company named Crystalonics made a very nice, military quality jfet that could be very quiet but it was relatively expensive, and some batches were noisy. I found this out with the QuanTech. They sold these devices at $30 ea, but they did NOT test them before sale. I have a whole set of these devices that are relatively noisy, and some that are very quiet. This device was Nchannel only, and was called the C413N. We had hoped to use them for microphone input stages, but they were too undependable as far as noise, AND the company did not care.
This is where I started testing a whole range of jfets from Siliconix. They were making complementary devices in some cases and they made a lot of different ones.
So, I started testing one part, called the j113. They had a range of Idss that could be 10:1, so selection was necessary. There was a similar complementary Pchannel part called the j175. Here again, the QuanTech quickly shows the N material to be superior to the P material, so if you wanted the LOWEST NOISE, back 40 years ago, you had to stick to Nchannel jfets. I then expanded my search with the QuanTech and found the J110 Nchannel jfet as a darn quiet fet, that normally was as quiet as the C413N, at 1/10th the price. This is the jfet that we used for the input for the stage for the Levinson JC-2, designed in 1973. Without the QuanTech, I would never have found this part. Unfortunately, Siliconix changed its process in about 1976, and another part had to be found.
Once the Japanese got into making quality low noise jfets, much of the struggle of buying a decent complementary jfet pair was eliminated, and we started using Toshiba jfets in the Vendetta Research phono stage.
However, in 1990, Toshiba ALSO changed their processing, AND I had to send back 100's of them, that were excessively noisy. Once again the QuanTech came to the rescue.
For those who may never have seen or used a QuanTech, let me denote some of its features, that would be hard for an amateur to imitate.
The QuanTech DC servos the device under test to whatever current you want to test at. It also has a variable test voltage, although that is not as important.
Finally, it servos an AC signal through the device to adjust for changes in Gm, both with current and device type. This might be useful for bipolars, but for jfets it is absolutely necessary to get an accurate reading.
The IC noise tester is a simpler design that amateurs could emulate fairly easy.
"Had to", you guys must be smarter than that. The QuanTech was designed when FFT's were an intellectual curiousity. It is now an obsolete anachronism, nothing it does is not now reproducible easily. Samuel's circuit should do nicely, banks of filters with noise bandwidths carefully calculated is so 70's.
A QuanTech gives hard numbers, naturally averages, and parts can be changed relatively quickly. Too bad ADI does not screen their devices so that Charles has to.
But ADI is not the only culprit in allowing noisy devices. More than 20 years ago, I switched from the National LF411 to the AD711, because the AD711's measured noise was more similar to the spec.. However, IF I used AD711's or their rough equivalent today, I would probably test them, just in case, ADI forgot how to make them properly.
IF you think that Charles or I WANT to test EVERY PART because the manufacturer has gotten sloppy, you are mistaken. We just HAVE TO, to meet our personal standards, so that we do not sometimes release noisy products.
But ADI is not the only culprit in allowing noisy devices. More than 20 years ago, I switched from the National LF411 to the AD711, because the AD711's measured noise was more similar to the spec.. However, IF I used AD711's or their rough equivalent today, I would probably test them, just in case, ADI forgot how to make them properly.
IF you think that Charles or I WANT to test EVERY PART because the manufacturer has gotten sloppy, you are mistaken. We just HAVE TO, to meet our personal standards, so that we do not sometimes release noisy products.
What utter nonsense. If you design a product that relies on a device parameter that is not specified by the manufacturer you should not be criticizing the manufacturer for your design failure.
If you want to "select on test" for a particular parameter then so be it. You are free to do that yourself (it is called DIY) or you can pay the manufacturer to do it for you - they may or may not oblige.
- Status
- Not open for further replies.