Comments to Fast-20 revision 3

Kevin Gingerich S=Kevin_Gingerich%S=Gingerich%G=Kevin%TI at mcimail.com
Fri Mar 3 11:10:00 PST 1995


 
 
  To  SCSI Reflector   X400
 
>From  Kevin Gingerich  GING
 
Subj  Comments to Fast-20 revision 3
 
1.      Technical Comment. Page no. 3, clause 5.1, line no. 3
Delete: It is recommended that the loads be evenly spaced, especially in 3
meter applications.
 
Reason: This statement is more appropriate for an annex and is not technically
correct. The device spacing is only constrained by a minimum distance (see
annex C) and uniformity is not required.
 
2.      Technical Comment. Page no. 3, clause 5.1, line no. 8
Change: ... shall be at least three times the stub length to avoid stub
clustering.
 
  To: ... shall be such that the added capacitance per unit length is no more
than 2 times the unloaded bus media's capacitance per unit length. (see annex C
for guidelines)
 
Reason: "Rules of thumb" do not belong in a standard. This rule can be followed
and result in a non-functioning bus from too much capacitance too close
together. The proposed requirement, if followed, will avoid node or stub
clustering regardless of the media and node capacitance.
 
3. Editorial Comment. Page no. 3, clause 5.1, line 10
Delete paragraph (line 10 through 13)
 
Reason: This paragraph is a repeat of the first paragraph in this clause.
 
4. Technical Comment. Page no. 3, clause 5.1, line no. 14
Change: The maximum ground offset voltage shall be maintained at or below 50
mV.
 
To: The magnitude of ground offset voltage between any to devices shall be
maintained less than 50 mV.
 
Reason: Voltage potentials occur between two points and can be either positive
or negative (depending upon the zero reference). The existing sentence
specifies neither.
 
5. Technical Comment. Page no. 3, clause 5.2, line no. 5
Change: ... shall be at least three times the stub length to avoid stub
clustering.
 
  To: ... shall be such that the added capacitance per unit length is no more
than 4.4 times the unloaded bus media's capacitance per unit length.
 
Reason: "Rules of thumb" do not belong in a standard. This rule can be followed
and result in a non-functioning bus from too much capacitance too close
together. The proposed requirement, if followed, will avoid node or stub
clustering regardless of the media and node capacitance.
 
6. Technical Comment. Page no. 3, clause 5.2
Add: The magnitude of ground offset voltage between any to devices shall be
maintained less than 3.6 V with up to 8 devices and 2.5 V with up to 32 devices
on the bus.
 
Reason: It can be shown that sufficient common-mode voltage potentials on a
loaded bus with minimum receiver input impedance can exceed the +/-60 mA output
current assured by EIA RS-485 compliant drivers.
 
7. Technical Comment. Page no. 4, clause 6.1.2, line no. 5
Change: b) VOH (high-level output voltage) = 2,5 to 3,7 V d.c. (signal
negated);
 
To: b) VOH (high-level output voltage) < 3,7 V d.c. at IOH = 0 mA (signal
negated);
 
Reason: I assume that this requirement is to set an upper bound on the level to
which an active negation driver can go. (Requirements c), d) and e) define the
operating points at other output currents.) The dc operating point of a circuit
must define both voltage and current.
 
8. Technical Comment. Page no. 4, clause 6.1.2, line no. 6 and 7
Change: ... at IOH = 7 mA (signal negated);
 
To: ... at IOH = -7 mA (signal negated);
 
Change: ... at IOH = 20 mA (signal negated);
 
To: ... at IOH = -20 mA (signal negated);
 
Reason: The direction the current IO is defined as positive in the IOL
requirement and therefore, must be negative for IOH.
 
9. Technical Comment. Page no. 4, clause 6.1.2, line no. 8
Change: IOH (high-level output current) > 22 mA (signal negated) @ VOH < 2,0 V
d.c.
 
To: IOH (high-level output current) > -22 mA (signal negated) at VOH < 0,2 V
d.c.
 
Reason: See 8 for the sign change.
        The symbol "@" is not used in the previous requirements.
        The basis for the -22 mA minimum active-negation current is that this
current plus the -48 mA from the terminators will achieve a valid high level
with a fully loaded bus. This current requirement occurs when the bus is
discharged or near zero volts. The 48 mA requirement for the terminators is
specified at 0,2 V and is repeated here for consistency.
 
10. Technical Comment. Page no. 4, clause 6.1.2, line no. 9
Replace Note 2.
 
With: Note 2. The negative sign here refers to the direction of current flow
and is not an algebraic convention.
 
Reason: The current note 2 is attempting to rephrase the quantitative
requirements and is not needed for interpretation. The proposed Note 2 is
needed for interpretation.
 
11. Technical Comment. Page no. 4, clause 6.1.2, line no. 16
Change: ... circuit shown in figure 1 with a load capacitor (CL) of 15 pF
+/-5%.
 
To: ... circuit shown in figure 1 with a load capacitance (CL) of 10 pF to 15
pF.
 
Reason: CL should be the capacitance introduced by the measurement circuit.
Most 10x scope probes do not hold such a tight tolerance for capacitance.
 
12. Technical Comment. Page no. 4, clause 6.1.2, line no. 17
Change: a) trise (rise rate)= 520 mv per ns maximum (0,7 V d.c. to 2,3 V d.c.);
        b) tfall (fall rate)= 520 mv per ns maximum (2,3 V d.c. to 0,7 V d.c.).
 
To: The magnitude of the rate of change of the output voltage shall not exceed
520 mV/ns between 20% to 80% of the steady-state output levels.
 
Reason: In 15+ years in the engineering profession, this is the first occurence
of the terms "rise rate" and "fall rate" that I can remember. Besides, trise or
tfall are usually associated with time and not voltage slew rates; there is a
polarity associated with slew rates; and there are any number of ways to get to
the same goal. This is my preference.
 
13. Technical Comment. Page no. 4, clause 6.1.2, line no. 19
Change: ... in figure 1 with a load capacitor (CL) of 200 pF +/-5%.
 
To: ... in figure 1 with a load capacitor (CL) of 10 pF to 15 pF.
 
Reason: The SCSI bus is a distributed parameter circuit and is closely modeled
as a transmission line and cannot be simulated with a simple lumped circuit.
The 47 ohm resistor in figure 1 represents the distributed load. If you accept
this fact, you cannot justify a load of 200 pF as representing a transmission
line. The standard requirement should be defined for the application and not
the test engineers' convenience. With the distributed parameter model there is
no lumped capacitance so, CL must be from the measurement circuit. A 10x scope
probe has a tip capacitance of 10 pF to 15 pF.
 
14. Technical Comment. Page no. 5, clause 6.1.2, Figure 1
Delete the "bubble" from the logic representation of the driver.
 
Reason: This is an electrical test circuit and logical requirements should not
be implied.
 
15. Editorial Comment. Page no. 6, clause 6.2.1, line 6
Change: ... signals described in except ...
 
To: ... signals described, except ...
 
16. Technical Comment. Page no. 6, clause 6.2.1, line 8
Change: The characteristic impedance ...
 
To: The differential impedance ...
 
Reason: Characteristic impedance is a parameter of transmission lines and not
of resistor networks.
 
17. Technical Comment. Page no. 7, clause 6.2.2
Add: All timing requirements shall be verified with the test circuit of figure
3. Timing instances shall be defined at the point where the difference voltage
is zero.
 
Reason: There is no test circuit defined for compliance testing. There needs to
be one and the second sentence is needed for timing definition.
 
19. Technical Comment. Page no. 7, clause 6.2.3, line 1
Change: The input characteristics of each differential signal pair, ...
 
To: The differential input characteristics of each signal pair, ...
 
Reason: The requirements are for the differential inputs, not the pair.
 
18. Editorial Comment. Page no. 7, clause 6.2.3, line 3
Change: a) conform to EIA RS-485 ),
 
To: a) conform to EIA RS-485,
 
19. Technical Comment. Page no. 7, clause 6.2.3, line 5
Change: c) measure no more than 25 pF of capacitance.
 
To: c) measure no more than 15 pF of capacitance.
 
Reason: The capacitance to ground of each signal of a differential pair,
including interconnection and interface silicon, is about 30 pF.
Differentially, this is 15 pF. This is one of the advantages of differential
signaling.
 
20. Technical Comment. Page no. 7, clause 6.2.3
Add figure 3 and renumber subsequent figures. Repeat figure 10 "Differential
Test Circuit" of SPI, lower VT to 4.5 V, and add 10 pF to 15 pF from each
output to ground.
 
Reason: See 17.
        VT is a schottky diode drop from 5 V.
The resistors simulate the bus and the capacitance covers the measurement
circuit (see 13).
 
21. Editorial Comment. Page no. 7, clause 6.2.4
Delete this clause.
 
Reason: Although the value of DIFFSENS can be debated, this clause is a repeat
|from SPI and imposes no new requirements for Fast-20. It is covered by
reference.
 
MSG continued in next MSG





More information about the T10 mailing list