Jamiemunz,
I agree with you, Bell Labs, and their phone lines.
Almost everything electronic that has ever been invented, was invented by Bell Labs, Armstrong, Raytheon, RCA, Tektronix Labs, and countless others; all in the past.
Everything else is just a modification of the above inventors.
One example is the differential delay line amplifier. Multiple tubes in cascade, with delay lines from grid to grid, and delay lines from plate to plate. Just one differential stage had to have 4 delay lines (2 grids, and 2 plates)
Perhaps first used in Tektronix scopes, this used Vacuum Tubes.
Next, there were those who did this with Bipolar Transistors.
Later, there were those who did this with GasFets (Triquint, a Tektronix Labs, later a Tektronix spin-off).
The most famous scope that used DC connection of the Vertical input of the scope, to the CRT vertical plates . . .
It had a bandwidth of 1 GHz (1,000 Mega Cycles / second in the terminology of the day, not Hertz).
It was very expen$ive, u$ed for nuclear bomb research, and the government as usual $pared no expen$e.
Actually, the input was a floating GR connector, that went to a coiled metal pipe that surrounded the differentail delay line, that connected to the CRT differential vertical plates.
What gave the 1GHz bandwidth was not just the the GR connector, and not just the delay line to the CRT.
Bandwidth was limited by the CRT itself. The Vertical plates were cut into multiple short plates, spaced one after the other (think capacitance from one plate to the plate that was on the other side), and there were air core coils that connected from each plate to the next plate (think inductance).
Wow, a differential delay line inside the CRT. Why?
The velocity (propagation speed) of the signal wave along the CRT vertical delay line plates, was the same as the velocity of the electron beam.
That meant each electron was moved by each pair of vertical plates. Moving the electron a little bit at each pair of plates, added up to moving each electron by more than just the little bit of one pair of plates; it added all those little movements to something that could be seen.
By the way, the linear sweep ramp was sent to the CRT horizontal plates, the horizontal plates were not connected to any connector outside of the scope.
Because of the delay lines in the vertical channel, the trigger for the sweep ramp that took time, was not a problem, because the 'late' start of the sweep ramp was earlier than the time delay of the vertical system, so the leading edge of the nuclear event was able to be viewed.
Necessity is the mother of invention.
So much for other scopes engineers who were not even born when the first 1GHz scope was created.
I am not degrading the genius of current scope engineers, I am just stating historical fact.
Those who do not know history, might make things even worse than they are now.
I agree with you, Bell Labs, and their phone lines.
Almost everything electronic that has ever been invented, was invented by Bell Labs, Armstrong, Raytheon, RCA, Tektronix Labs, and countless others; all in the past.
Everything else is just a modification of the above inventors.
One example is the differential delay line amplifier. Multiple tubes in cascade, with delay lines from grid to grid, and delay lines from plate to plate. Just one differential stage had to have 4 delay lines (2 grids, and 2 plates)
Perhaps first used in Tektronix scopes, this used Vacuum Tubes.
Next, there were those who did this with Bipolar Transistors.
Later, there were those who did this with GasFets (Triquint, a Tektronix Labs, later a Tektronix spin-off).
The most famous scope that used DC connection of the Vertical input of the scope, to the CRT vertical plates . . .
It had a bandwidth of 1 GHz (1,000 Mega Cycles / second in the terminology of the day, not Hertz).
It was very expen$ive, u$ed for nuclear bomb research, and the government as usual $pared no expen$e.
Actually, the input was a floating GR connector, that went to a coiled metal pipe that surrounded the differentail delay line, that connected to the CRT differential vertical plates.
What gave the 1GHz bandwidth was not just the the GR connector, and not just the delay line to the CRT.
Bandwidth was limited by the CRT itself. The Vertical plates were cut into multiple short plates, spaced one after the other (think capacitance from one plate to the plate that was on the other side), and there were air core coils that connected from each plate to the next plate (think inductance).
Wow, a differential delay line inside the CRT. Why?
The velocity (propagation speed) of the signal wave along the CRT vertical delay line plates, was the same as the velocity of the electron beam.
That meant each electron was moved by each pair of vertical plates. Moving the electron a little bit at each pair of plates, added up to moving each electron by more than just the little bit of one pair of plates; it added all those little movements to something that could be seen.
By the way, the linear sweep ramp was sent to the CRT horizontal plates, the horizontal plates were not connected to any connector outside of the scope.
Because of the delay lines in the vertical channel, the trigger for the sweep ramp that took time, was not a problem, because the 'late' start of the sweep ramp was earlier than the time delay of the vertical system, so the leading edge of the nuclear event was able to be viewed.
Necessity is the mother of invention.
So much for other scopes engineers who were not even born when the first 1GHz scope was created.
I am not degrading the genius of current scope engineers, I am just stating historical fact.
Those who do not know history, might make things even worse than they are now.
Last edited:
Yes, very poor S/N ratio in this thread. Your reference at least isn't wrong, which is a good start. I would quibble that sources/outputs with both lines driven shouldn't normally be called "differential" (although, there is such a thing, but very rarely used). Instead, they'd usually be called something like "actively-driven balanced", and outputs with one line driven are called "impedance balanced".
All good fortune,
Chris
Rather than leave this thread unanswered, maybe we could approach the topic of interconnecting (in stereo - two signals) line level signals in home installations. The classic case of two channel sources with two channel loads and both channels with common "grounds", signal returns/colds (the ubiquitous RCA/phono connection) has two flaws:
First is that signal return/cold is most often carried on the same conductor as shielding, so is corrupted by local fields. This can be easily fixed by using three conductors, two for signal and a separate shield, which can be connected at only one end, to isolate it from signal.
Second is that noise (from local fields) immunity in both coax and twisted pair wiring comes from having equal amplitude currents in both signal conductors of each channel. This condition is violated with return/cold conductors of both channels connected in common on both ends. Signal current will (must!) divide equally between the two identical return/cold paths, leaving almost nothing for a differential input* to work with - the signal current does not appear equally on both signal conductors, so interference cannot be "differenced" away.
These flaws are of course not fatal - we've lived with them our whole lives and we're not dead yet. For most folk these and PE grounding issues (which introduce yet another path for return/colds) are below the radar if everything below speaker level is kept local, everybody plugged into the same power strip, and we don't live underneath WWV.
If we wanted to improve on this though, but must keep two conductor RCA/phono ends, we could introduce three conductor interconnections. A first step would be to separate the wiring's shield from signal carrying conductors - and connect it at one end only.
An interesting second step possibility is to lift the receiving end's return/cold end off chassis by a few Ohms (call it 4R7). This input stage "ground" would include the first stage's grid leak, cathode bias, input stage B+ bypass capacitor bottom end, and any other first stage stuff referenced to the cathode. This restricts return/cold signal currents in the interconnecting wires to their own channels, restoring their local-field noise immunity. This can be done with RCA/phono connectors or of course with XLR 3 pins.
* All inputs are "differential" in the sense that they respond to the difference between two points in the circuit. An input calling itself a "differential input" is one with a better ability to reject signal/interference/noise/whatever common to both points, and the topic of "balanced" is only concerned with relative abilities to reject common-mode whatever. No other magic applies.
All good fortune,
Chris
First is that signal return/cold is most often carried on the same conductor as shielding, so is corrupted by local fields. This can be easily fixed by using three conductors, two for signal and a separate shield, which can be connected at only one end, to isolate it from signal.
Second is that noise (from local fields) immunity in both coax and twisted pair wiring comes from having equal amplitude currents in both signal conductors of each channel. This condition is violated with return/cold conductors of both channels connected in common on both ends. Signal current will (must!) divide equally between the two identical return/cold paths, leaving almost nothing for a differential input* to work with - the signal current does not appear equally on both signal conductors, so interference cannot be "differenced" away.
These flaws are of course not fatal - we've lived with them our whole lives and we're not dead yet. For most folk these and PE grounding issues (which introduce yet another path for return/colds) are below the radar if everything below speaker level is kept local, everybody plugged into the same power strip, and we don't live underneath WWV.
If we wanted to improve on this though, but must keep two conductor RCA/phono ends, we could introduce three conductor interconnections. A first step would be to separate the wiring's shield from signal carrying conductors - and connect it at one end only.
An interesting second step possibility is to lift the receiving end's return/cold end off chassis by a few Ohms (call it 4R7). This input stage "ground" would include the first stage's grid leak, cathode bias, input stage B+ bypass capacitor bottom end, and any other first stage stuff referenced to the cathode. This restricts return/cold signal currents in the interconnecting wires to their own channels, restoring their local-field noise immunity. This can be done with RCA/phono connectors or of course with XLR 3 pins.
* All inputs are "differential" in the sense that they respond to the difference between two points in the circuit. An input calling itself a "differential input" is one with a better ability to reject signal/interference/noise/whatever common to both points, and the topic of "balanced" is only concerned with relative abilities to reject common-mode whatever. No other magic applies.
All good fortune,
Chris
Last edited:
Yes, and how do we best balance this against the diminishment of the receiving end's (defined as minimal) differential ability? Stephe has raised a very deep but poorly explored topic.
Cordell (probably a lot of others too) has proposed the 4R7 resistor separation of input signal return/cold from the rest of signal common, but in the simpler context of a truly differential (modern semi-con) input stage. I'm thinking that it can be massaged to include a small DC offset in a trendy Now's-a-Gogo design.
About to find out, because my building partner itishifi.com (who does all the work) and I are making a new amp with this and other special attention to "grounds". Because, compactrons. And he has way too much energy.
A way to ameliorate bad RF issues would be to bring the otherwise unconnected ends of shield to chassis with small (ceramic?) caps. Easy for DIY but a pain with antiques. Just another reason to include XLR/Cannon connectors in parallel on all DIY builds - less drama.
Much thanks, as always,
Chris
Cordell (probably a lot of others too) has proposed the 4R7 resistor separation of input signal return/cold from the rest of signal common, but in the simpler context of a truly differential (modern semi-con) input stage. I'm thinking that it can be massaged to include a small DC offset in a trendy Now's-a-Gogo design.
About to find out, because my building partner itishifi.com (who does all the work) and I are making a new amp with this and other special attention to "grounds". Because, compactrons. And he has way too much energy.
A way to ameliorate bad RF issues would be to bring the otherwise unconnected ends of shield to chassis with small (ceramic?) caps. Easy for DIY but a pain with antiques. Just another reason to include XLR/Cannon connectors in parallel on all DIY builds - less drama.
Much thanks, as always,
Chris
Bill Whitlock of Jensen Transformers recommends that you connect the shield at the receiving end to XLR pin one with a 10 nF ceramic disc. That should take care of RF. I go one step further. I connect the 10 nF cap to the solder tab on the metal body of the cable XLR at the receiving end. This gives the shortest possible connection of shield to chassis, provided of course that chassis XLR connector has a good connection to the metal chassis. Serrated washers are your friends here. This is not my idea. I read about it somewhere on some forum. Unfortunately I can’t remember where, so I can’t give credit where it’s due. I’ve converted all my interconnects to this configuration. I suppose this may contribute to the complete and utter silence of my special Altec VOTTs driven by an all tube amplifying chain.
Something functionally very similar can be done with RCA/phono connectors, but it requires folk to think outside the box. For DIY, maybe just include a nearby chassis ground connection, as is done for turntable phono inputs. Mostly probably doesn't matter - but a lot more significant than magic coupling capacitors, and free.
All good fortune,
Chris
All good fortune,
Chris
Could you explain the values for this network? Wouldn't they be cable dependent?
Just watched a new YouTube video expressing some questions about "balanced" outputs and maybe I can offer something to help clarify the situation.
CMRR, the ability to ignore or reject interfering signals, comes from two things: signal carrying conductors in close proximity so that they receive the same level and phase of interference (twisting helps them in this respect) and a differential receiving end. This is one that responds preferentially to only the difference between the two signal carrying conductors and, as much as possible, rejects (does not produce output from) anything common to both conductors. Note that this must happen on the receiving end of the wires.
It is not necessary for both conductors to have driven signal on them; it's only necessary for them to be presented with the same source impedance (so that interfering fields generate the same voltages in both) for the differential receiving end to reject the interference. In the very old days both conductors would be driven, or sometimes just floated, but modern semi-pro stuff is usually "impedance balanced" meaning pin 2 is driven by an opamp through a 100R-ish resistor and pin 3 goes to signal ground through the same value resistor. This is also completely compatible with ordinary RCA/phono inputs, so win/win. No output transformers are needed for the potential noise immunity advantage with a differential input.
Side note: XLR/Cannon jacks don't have to be used for "balanced" interconnections, but it's expected on commercial equiptment. For us DIYers they work just fine for single-ended connections - even better - so could (I say should!) be included in parallel with RCA/phono jacks for easy compatibility.
All good fortune,
Chris
CMRR, the ability to ignore or reject interfering signals, comes from two things: signal carrying conductors in close proximity so that they receive the same level and phase of interference (twisting helps them in this respect) and a differential receiving end. This is one that responds preferentially to only the difference between the two signal carrying conductors and, as much as possible, rejects (does not produce output from) anything common to both conductors. Note that this must happen on the receiving end of the wires.
It is not necessary for both conductors to have driven signal on them; it's only necessary for them to be presented with the same source impedance (so that interfering fields generate the same voltages in both) for the differential receiving end to reject the interference. In the very old days both conductors would be driven, or sometimes just floated, but modern semi-pro stuff is usually "impedance balanced" meaning pin 2 is driven by an opamp through a 100R-ish resistor and pin 3 goes to signal ground through the same value resistor. This is also completely compatible with ordinary RCA/phono inputs, so win/win. No output transformers are needed for the potential noise immunity advantage with a differential input.
Side note: XLR/Cannon jacks don't have to be used for "balanced" interconnections, but it's expected on commercial equiptment. For us DIYers they work just fine for single-ended connections - even better - so could (I say should!) be included in parallel with RCA/phono jacks for easy compatibility.
All good fortune,
Chris
They are cable dependent. Typical coax is 50 or 75 ohm, and twisted pair can be close to 100. For my purposes, I don't feel the need to match perfectly, just provide something a bit better than 0 ohms. The capacitor only serves the purpose of addressing the RF range, so characteristic impedance is pertinent (and you will never find an interconnect with Z0 of 0).