SCSI Terminators: Determining Passive vs Active

[jmacz]

Everything I've read about the filter says that it provides termination capacitance. The service manual mentions it's needed for Apple drives shipped before March 19, 1990 and third-party drives. There are a bunch of notes about it in this doc too. I'm guessing that piece would be harder to disassemble...

Sounds like the answer to the question I had about internal termination in the other thread is that filter. But interesting that the doc @dougg3 linked keeps stating the IIfx takes only one internal scsi device and that the filter goes between the scsi cable and the drive. I thought you could have more than one internal device (I am running two) and if so, where would the filter go? between the cable and the first device or between the cable and the last device (Ie the one with termination enabled)?

 

[David Cook]

Maybe it just adds the 2.2uF and 0.01uF caps just like what you pointed out in your other post?

Ding! Ding! Ding! Give this man a prize.

I just broke out my LCR meter and took another shot at measuring the 'filter' portion of the parts. Previously, I had been testing signal pins between terminator power and ground -- assuming a change in inductance, capacitance, or resistance would show up there. This time, I just sort of randomly tested pins.

Between pin 26 (terminator power) and pin 24 (n/c) there is an embedded 2.2 uF capacitor. And between pin 26 (terminator power) and pin 28 (n/c) there is an embedded 0.01 uF capacitor. It is weird that they chose the no-connect pins rather than between terminator power and ground. But, it explains why they wired ground to the n/c pins on the 'terminator' portion of the parts -- to connect the capacitors.

These capacitor values are the same values as Apple uses on their IIfx black terminator and newer gray passive terminator.

Based on figures 11-5, 11-7, 11-8, and 11-9 from Apple's hardware guide, plugging the 'filter' onto an otherwise empty internal SCSI port on the:
Macintosh II and SE: Will connect the capacitors to ground and enable them.
Macintosh SE/30, IIx, IIcx, IIci, IIfx: Will not connect the capacitors to ground
Adding an internal drive or internal terminator which contains traces connecting ground to pins 24 and 28 will also enable the capacitors. This is the most common case.

Apple's recommendation to plug the filter nearest to the internal drive is appropriate. You want the capacitors nearest to the switching signals to reduce noise. Farther away will provide some benefit, just not as good.

I assume newer drives, terminators, and Macs include adequate capacitors, thus diminishing returns for the added capacitors of the filter.

 

[jmacz]

Apple's recommendation to plug the filter nearest to the internal drive is appropriate. You want the capacitors nearest to the switching signals to reduce noise. Farther away will provide some benefit, just not as good.

@David Cook what if you have multiple internal drives:

Internal Connector <-> Drive A (middle connector on the ribbon) <-> Drive B (last connector on the ribbon) with Termination jumper enabled

In this case does the filter go between the middle connector and Drive A or between the last connector and Drive B?

 

[David Cook]

@David Cook what if you have multiple internal drives:

Internal Connector <-> Drive A (middle connector on the ribbon) <-> Drive B (last connector on the ribbon) with Termination jumper enabled

In this case does the filter go between the middle connector and Drive A or between the last connector and Drive B?

The filter's purpose is to help the terminator. The capacitors provide little fast-acting reserves of terminator power.

In your example, attach the filter to the last connector (Drive B) with the termination jumper enabled -- because you want the filter nearest to the terminator because it helps the terminator.

In Apple's example without an internal drive, a little terminator cap is placed on top of the filter. There again, the filter is nearest to the internal terminator.

Newer hard drives already have the capacitors and/or use active termination. The filter is not needed in that case, but won't hurt.

 

[jmacz]

Great, thanks for confirming.

 

[dougg3]

Ding! Ding! Ding! Give this man a prize.

Yay! Thanks for confirming. It all makes sense.

For what it's worth, pins 20, 22, 24, 28, and 30 are all GND on a random Quantum hard drive I just tested. Seems like there is a bit of varying info out there about some of those pins. They're also wired to GND on SCSI2SD, ZuluSCSI, etc.

 

[David Cook]

Here is a Mac Memories DataMate DM800-09-R active SCSI terminator making the rounds on eBay.



This is an external terminator with a Centronics 50-pin connector.



The circuit is the usual 2.85V voltage regulator with 110 ohm pull-up resistors (laser trimmed in this case). The mix of capacitor values is slightly different than normal, but perfectly valid.



The regulator is a Burr Brown REG11172 -- specially targeted for this purpose.

 

[Zuulander]

Here is an unusual terminator, in regards to the quantity of discrete components used.

View attachment 54958

Top of board. Looks like capacitors to locally store terminator power. Voltage regulators on left and right??

View attachment 54959

Bottom of board. Transistors?

View attachment 54961

Angled to better show the connection to pins.

View attachment 54962

I think this is a Forced Perfect Terminator. Of the three kinds of terminators this is required for differential SCSI buses. Quite a bit of overkill for SCSI I, but seeing that the device is marked SCSI II it's probably intended for a server application. A FPT will function like an active terminator in non-differential applications.

Edit to add additional details about device in question: From an archived document: SunExpert

Wide SCSI Terminators
A series of wide SCSI terminators has
been introduced by Aeronics. The
FPT-27 Platinum Series terminators
terminate all 27 lines of the wide SCSI
bus using the company's own forced
perfect termination (FPT) technology.
The company says this gives users the
ability to run the SCSI at its maxi-
mum lengths, and support the maxi-
m um number of devices, without
common glitches. The product cor-
rects, for example, the inherent imped-
ance mismatch problems that are the
frequent bane of wide SCSI. Prices
range from $12 to $26.

 

[Iesca]

I have a couple of 25-to-25 pin pass-through terminators. One is HP-branded and is marked "SCSI TERMINATOR", I'm guessing it's passive, though it has a bit of a bulge to it, suggesting something inside has possibly gone awry...

Meanwhile, the second one is unbranded, and was sold online as an active scsi terminator, but it has no markings on it, though it does have a narrow barrel plug receptacle, suggesting that it is meant to be externally powered.