Steve, why don't you buy me new test equipment? I have 'paid my dues' already.
This is a strange sort of 'snobbery'. It is like saying that my Porsche 944 is a piece of junk, and not a 'real' sports car, and because it is more than 20 years old, both in design and build, that it is useless.
My test equipment, works well enough to make these measurements, whatever the distortions are derived from, and the results have been consistent for the last 15 years, and with 4 different analyzers. The 1700B that I am using TODAY, is owned by Demian Martin, and has been worked on by Demian's technician and me to get an optimum null, less than .002%, even at 30mV operating voltage. However, similar results were gotten from 3 other 1700B's. It is not interference, and I have established a baseline of -118dB or so, with greater than 100 signal average, with the 'best' measuring cables.
Generally speaking, cables that have been sitting around for months or years, and non soldered or welded connections are the problems. Sometimes, very cheap cables actually measure quite well in the same test, and some very expensive cables fail.
This is a strange sort of 'snobbery'. It is like saying that my Porsche 944 is a piece of junk, and not a 'real' sports car, and because it is more than 20 years old, both in design and build, that it is useless.
My test equipment, works well enough to make these measurements, whatever the distortions are derived from, and the results have been consistent for the last 15 years, and with 4 different analyzers. The 1700B that I am using TODAY, is owned by Demian Martin, and has been worked on by Demian's technician and me to get an optimum null, less than .002%, even at 30mV operating voltage. However, similar results were gotten from 3 other 1700B's. It is not interference, and I have established a baseline of -118dB or so, with greater than 100 signal average, with the 'best' measuring cables.
Generally speaking, cables that have been sitting around for months or years, and non soldered or welded connections are the problems. Sometimes, very cheap cables actually measure quite well in the same test, and some very expensive cables fail.
Then I guess you're never going to get beyond "30 years ago." It's 2011, John. The world has moved on since "30 years ago."
It's not anything like that at all.
It's about your going on and on about distortions you're getting with 30+ year old test gear that aren't evident on more modern, more capable equipment.
Here's what it's like, John. It's like if Percival Lowell were still alive today and going on and on about the canals on Mars that he's seeing when he looks through his old telescope.
se
There is a 'grounding difference' between the ST1700 and the Audio Precision, however no one has shown my WHY this grounding difference would bring out differences between an Audio Precision and the ST1700, and furthermore, most audio circuitry used today, would have grounding more similar to the ST1700 than the Audio Precision which is transformer isolated. Does this matter, if so, why?
I will stick with my results, as they are consistent, and yet show differences with identical lengths of interconnects, with similar characteristic impedance (75 ohms normally) and capacitance.
I will stick with my results, as they are consistent, and yet show differences with identical lengths of interconnects, with similar characteristic impedance (75 ohms normally) and capacitance.
Model the path of the signal current.
All of the return current MUST go through the cable shield. If this is not the case, then what is being measured??
The transformer eliminates the secondary path. The 1700 does not.
Cheers, John
Consider building a piece of test equipment in a 3ru tall rack enclosure. It outputs say, 1V rms into a cable via a non isolated rca jack mounted to the front panel. Assume all this electronic stuff is on the right side of the chassis, and an 1/8th inch aluminum panel runs down the middle of the chassis.
Now, on the left side of the chassis, have a non isolated rca jack on the front panel, with some fancy circuitry inside to receive the signal from the cable. Within this side of the box, use a 10K load resistor to terminate the input.
Within the box, way at the back, have a wire that sends the exact same 1Vrms signal from the transmitter side to the receiver side, but do it using a differential signal scheme..
Now, at the receiver, subtract what came in the rca jack from what is received from the diff pair. And do so very well, such that you can get 100 to 120 dB of null.
Ok.. Now, let's look at signal path.
For the balanced pair, all current in the + wire is equal to and opposite to the current in the - wire.
What about the IC connected to the front panel? The test signal current going into the IC goes to the 10k load in the receiver, then where? All of this test current should go on the outer braid of the IC via the RCA jack. A coax does NOT shield properly if the net current in the cable is not zero. But, where is the return current going??
At lower frequencies, most of the current will travel along the front panel, from RCA jack ground to RCA jack ground, bypassing the IC shield. This is the path of least impedance at DC, and up into the Khz range.
This creates a loop. It makes the system sensitive to ALL stray em fields. ALSO, the current path becomes sensitive to both the IC shield resistances and the RCA jack insertion resistances. Remember, this is a parallel path composed of the IC shield, and a 1/8th inch thick aluminum panel. To get 50% of the current flowing in the IC shield, the shield and jack resistances will have to be very low..
The AP unit completely eliminates this second path.
Cheers, John
Ohh, this "1700" doesnt have isolated RCA jacks? Not separate and galvanically isolated psu's for signal generation and measurement? And no differential in/out? In that case its safe to assume that at audio frequencies there is almost zero going through the cable shield. Does that induce signal dependent disturbance?
Thank you John for taking the trouble to respond to my question in detail.
I keep the following sentences as the essential ones:
So, I guess the problem would be solved using isolated (from the chassis) ICs.
Now, I wonder why the oscilloscope manufacturers keep on using non isolated BNC ICs. Haven’t they investigated this problem?
I am
Regards
George
I keep the following sentences as the essential ones:
So, I guess the problem would be solved using isolated (from the chassis) ICs.
Now, I wonder why the oscilloscope manufacturers keep on using non isolated BNC ICs. Haven’t they investigated this problem?
I am
Regards
George
The 1700A & B have a common supply for input and output. The ground is isolated from the chassis with a switch. The input is differential with I think 50K to ground on each side and a pretty high CMRR.
I don't understand how the cable shield will pass current from the chassis to chassis points in the way described on a 1700. In a typical audio system there are a lot of "leakage" currents going from chassis to chassis in ways that are prone to degrade the audio and hard to fix, but in a test setup like this there should be no problems.
I don't understand how the cable shield will pass current from the chassis to chassis points in the way described on a 1700. In a typical audio system there are a lot of "leakage" currents going from chassis to chassis in ways that are prone to degrade the audio and hard to fix, but in a test setup like this there should be no problems.
Thank you John for taking the trouble to respond to my question in detail.
I keep the following sentences as the essential ones:
So, I guess the problem would be solved using isolated (from the chassis) ICs.
Now, I wonder why the oscilloscope manufacturers keep on using non isolated BNC ICs. Haven’t they investigated this problem?
I am
Regards
George
Scopes usually have 8-12 bit resolution.
The return current will take the lowest impedance path. Since the return path is two parallel ones, each with resistance and inductance in series, there will be frequency breakpoints. If the connector resistance changes such that the breakpoint moves, the meter can read differently. Recall, JC speaks about cleaning solutions on the rca's to effect changes? Bingo.
In all cases where low level effects are being measured, absolute control over all current paths is critical. The AP does this by design. Loss of current path control, vis a vis, non zero net current through a coax, causes susceptibility to external fields. JC also notes his measurements are inundated with TV horiz scan interference from neighbors.
edit: Within the last half decade, repeated requests on my part for schematic and physical details of the modified 1700 being used by JC after "extensive electronics upgrades" by him have gone unanswered. As have repeated requests for baselining of the equipment to determine the measurement floor and measurement accuracy.
JC reports results of his tests in the -100 to -120db range IIRC, and that requires exquisite attention to current path details. Without such, the results are not necessarily accurate.
Cheers, John
Last edited:
Now, I wonder why the oscilloscope manufacturers keep on using non isolated BNC ICs. Haven’t they investigated this problem?
I am
Regards
George
NEC.
Cheers, John
Last night, I ran this test for an engineer who came to visit after the hi fi show at the SF Airport. We ran several example cables. This is what I found: Some cables, some very expensive, looked sort of bad with higher order harmonics 5th, 6th, 7th, and 8th, primarily. We had some trouble, at first with CLEAN AND TIGHT CONTACTS. This is number one in causing problems. Anyone who thinks that you can just plug something together that is not all gold on gold, and not get measurable distortion, is a fool. But I cleaned the contacts with isopropyl alcohol and sometimes DeOxit, and EXERCISED the contacts by repeated insertions and I finally came to a baseline. I could never totally get rid of 2nd and 3'rd harmonic, BUT they would change in relative amplitude to each others with some cables. Most of the cables were one meter long, but 1/2 meter and 2 meter cables were also tested, with not much effect. Once we got the contacts clean and tight, some cables virtually 'flatlined' between 4th and 10 harmonics. Other cables showed a lot of higher order harmonics. Floating the source from the chassis did little to change the results.
It is still an interesting test. I have not tried it with an HP339 as I do not presently have access to one, but it might present the same results.
For the record, the cables that Bruno and I tried, selected by SE, did NOT show any significant distortion in this test. I don't know why, but they did not.
It is still an interesting test. I have not tried it with an HP339 as I do not presently have access to one, but it might present the same results.
For the record, the cables that Bruno and I tried, selected by SE, did NOT show any significant distortion in this test. I don't know why, but they did not.
Surface conductivity of gold plating measures all over the map. There are innumerable high carat gold alloys (99.7 - 99.9 pure, at best, typically) used commercially, depending on the source (typical alloying components include nickel, cobalt, arsenic, etc. to produce those brite shiney coatings everyone loves). When building schottky point contact diodes, for instance, for radio astronomy detectors, using GaAs patterned degenerate silicon, one needs to use dull 24 carat gold for optimal function. This, 40 years ago or so, nothing new or exotic. never seen an audio connector with pure gold on it (it would have to be dull)
John L.
John L.
Gold is a known high quality contact. One problem that I have is that most of the contacts in my test equipment are NOT gold. Therefore, I have to work hard to make sure that the surface film is pushed aside when making the connection, and that the surfaces are as clean as possible from grease or fingerprints before making the connection. It is OBVIOUS from the measurements that this is mandatory at these distortion levels with this test set-up.
For everyone else, heavy gold plated contacts have been the world standard and still are. Gold 'flashed' is NOT heavy gold, and while they 'may' be better than raw nickel or chromium, they are nothing special.
For everyone else, heavy gold plated contacts have been the world standard and still are. Gold 'flashed' is NOT heavy gold, and while they 'may' be better than raw nickel or chromium, they are nothing special.
- Status
- Not open for further replies.