chw@hpctdlb.HP.COM (Charlie Whiteside) (06/26/90)
I have used HP 54112D Digital Scope with great success in this area. The 54112 has a 64K deep capture buffer and samples up to 400 Msam/s. This allows the user to see the amplitude and transmission effects of a frame. It is especially useful when the trigger is set for the preamble of a frame. If you place the scope at the end of a segment you can see how low the amplitude of particular frames get ( At that end). It is definitely a very low level application (you need to know 802.3 and related info such as transmission line theory) but can point out some very tough problems that no other instrument can detect.
pat@hprnd.HP.COM (Pat Thaler) (06/27/90)
> > The NQA also checks the coax for DC bias voltage, and will look at the > DC component of a packet "as recovered by a 2 usec (approx) low pass > filter. This value is measured as a voltage but presented as a > current, since co-ax impedance is known. The IEEE specification for ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > the parameter is quoted as a current." "The collision detect > threshold is a value built into every transceiver on the network and > hence is subject to a manufacturing spread. The DC component launched > by a transceiver is also subject to a manufacturing spread and so > there is some scope for conflict in which some transceivers see > spurious collisions from certain stations but not other." > The DC impedance into which the node is transmitting is only _approximately_ known. The terminating resistor at each end should be 50 ohms +/- 1%, but the series resistance of the cable can be from 0 to 10 ohms. (Back to how long can the coax be and how many stations/connectors; series resistance and its affect on collision detect is one of the things which limit the length.) If the transmitting MAU and the tester were at the middle of a 10 ohm segment, the MAU is transmitting into a 30 ohm load rather than the 25 ohms it would see with no cable resistance. This introduces a measurement error of 20%. If the transmitting MAU is at one end of a 10 ohm segment and the tester is at the other end, the MAU is transmitting into a 27 ohm load. The tester is seeing 5/6 of the voltage produced because of the resistive divider of the cable and terminating resistor. A 10% measurement error (in the other direction). This ignores smaller sources of error such as the 1% resistor tolerance, leakage current from other MAUs, input resistance of other MAUs etc. A tester on the cable can tell you whether the MAU is grossly out of spec, but it cannot conclusively determine whether it is in spec. (The possible exception is if the tester was connected right next to the MAU and the tester measured the DC impedance to calibrate itself.) Pat Thaler
pat@hprnd.HP.COM (Pat Thaler) (06/27/90)
/ hprnd:comp.dcom.lans / pat@hprnd.HP.COM (Pat Thaler) / 10:03 am Jun 26, 1990 / > > The NQA also checks the coax for DC bias voltage, and will look at the > DC component of a packet "as recovered by a 2 usec (approx) low pass > filter. This value is measured as a voltage but presented as a > current, since co-ax impedance is known. The IEEE specification for ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > the parameter is quoted as a current." "The collision detect > threshold is a value built into every transceiver on the network and > hence is subject to a manufacturing spread. The DC component launched > by a transceiver is also subject to a manufacturing spread and so > there is some scope for conflict in which some transceivers see > spurious collisions from certain stations but not other." > The DC impedance into which the node is transmitting is only _approximately_ known. The terminating resistor at each end should be 50 ohms +/- 1%, but the series resistance of the cable can be from 0 to 10 ohms. (Back to how long can the coax be and how many stations/connectors; series resistance and its affect on collision detect is one of the things which limit the length.) If the transmitting MAU and the tester were at the middle of a 10 ohm segment, the MAU is transmitting into a 30 ohm load rather than the 25 ohms it would see with no cable resistance. This introduces a measurement error of 20%. If the transmitting MAU is at one end of a 10 ohm segment and the tester is at the other end, the MAU is transmitting into a 27 ohm load. The tester is seeing 5/6 of the voltage produced because of the resistive divider of the cable and terminating resistor. A 10% measurement error (in the other direction). This ignores smaller sources of error such as the 1% resistor tolerance, leakage current from other MAUs, input resistance of other MAUs etc. A tester on the cable can tell you whether the MAU is grossly out of spec, but it cannot conclusively determine whether it is in spec. (The possible exception is if the tester was connected right next to the MAU and the tester measured the DC impedance to calibrate itself.) The variation of transmit levels and collision detect levels is why people post that they ran their 10BASE2 segment 250 m with no problems. If you are lucky in the MAUs you choose, this can be true, but as you add or move MAUs, you may find some don't work or don't work at certain positions on the cable. Pat Thaler
spurgeon@sirius.cc.utexas.edu (Charles Spurgeon) (06/27/90)
In article <2230093@hprnd.HP.COM> pat@hprnd.HP.COM (Pat Thaler) writes: >> >> The NQA also checks the coax for DC bias voltage, and will look at the >> DC component of a packet "as recovered by a 2 usec (approx) low pass >> filter. This value is measured as a voltage but presented as a >> current, since co-ax impedance is known. The IEEE specification ... >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >The DC impedance into which the node is transmitting is only >_approximately_ known. The terminating resistor at each end should >be 50 ohms +/- 1%, but the series resistance of the cable can >be from 0 to 10 ohms. (Back to how long can the coax be and how >many stations/connectors; series resistance and its affect on >collision detect is one of the things which limit the length.) > ... > >A tester on the cable can tell you whether the MAU is grossly >out of spec, but it cannot conclusively determine whether it is >in spec. (The possible exception is if the tester was connected >right next to the MAU and the tester measured the DC impedance to >calibrate itself.) > Thanks for the information on coax transmission characteristics and the problems of measuring same. It's always interesting to hear more about how the analog electronics really function while delivering the "digital" bits between computers. Having written a fair amount of documentation I know that this sort of information can be complex and hard to describe, but I always find your postings quite clear. The amount of engineering that goes into making a high speed bus connection between computers work reliably under nearly all circumstances is really impressive. I take my hat off to the IEEE 802.3 committee members who have worked so hard all these years to make the Ethernet/802.3 system as reliable and trouble free as it is. In a perverse way, I suppose it's a testimony to the efforts of these engineers that people manage to get signals down "Ethernets" that egregiously violate the specifications. :-) Here's some more specs from photocopies of the NQA manual that I made while it was on evaluation: The "implant" they use for sensing signals is supposed to be installed in the center of the coax, although they concede that it can be installed at the end of the cable "as a last resort." The manual verbiage at this point implies that the reason for this is to improve the DTDR traces. The NQA also has a regular transceiver connection to the coax, which should not be any closer than 2.5 meters to the implant and no further away than four times that distance. The combination of a transceiver and the implant gives the NQA the ability to both sense the signals and identify the MAC address of the controller emitting the signals. This is something I haven't seen in any other low level signal analyzer approach. In their specifications they cite: "Packet Jitter +/- 3 ns accuracy. 1.5 ns resolution at the rear panel." "Packet AC Component +/- 10% accuracy. 4% resolution at implant." "Packet DC Component +/- 5% accuracy. 4% resolution at implant." "Packet Bit Rate +/- 200Hz accuracy, 400-4KHz deviation from 10MHz." "Packet Fall Time +/- 20% accuracy. 1.5 ns resolution." "Coax Bias Voltage: +/- 5ma accuracy. Resolution 5mV." Charles Spurgeon | | UTnet Network Information Center | spurgeon@emx.utexas.edu | University of Texas at Austin | ...!cs.utexas.edu!ut-emx!spurgeon| -------------------------------------------------------------------------