3 ms·

The ISA bus interface and NE2000 programming are not modern--although I'm sure there's some modern industrial motherboards that still have an ISA slot. But the

The ISA bus interface and NE2000 programming are not modern--although I'm sure there's some modern industrial motherboards that still have an ISA slot. But the design of this card is modern--everything is down to two ICs and a few other tiny surface mount parts that look more at home on a cellphone board than a PC ISA card from the 80's. Compare with pictures of actual old ISA NE2000-compatible cards (eBay a good source) and I'm sure the BOM is at least 2x of this NIC. Plus this is open source.

cosmotic 6y ago
Usage of surface mount and fewer but more integrated ICs are isn't what I would consider modern. Modern designs use those things, sure, but the technology and practice is not modern. It's actually pretty dated. Maybe not as-dated as a through-hole and discrete component design, but certainly not exclusive to recent (read: modern) designs.
metaphor 6y ago
Err...the only thing "modern" about this is you can purchase/fabricate one using new material and components in active production. If a modern cell phone integrated those SMT packages, it'd be the size of a brick, and the PCB itself is vanilla 2-layer Tg130 FR4...couldn't even be bothered with a controlled impedance layout, let alone length-match bus traces. Recently, bunnie talked about one "relatively modern" PCB design element---microvias---and explains why you just don't see them much in OSHW designs[1]...an excellent read if you get the chance. JLC is a well known PCB fab house because their low-volume prices are dirt cheap, not because their process is modern. [1] https://www.bunniestudios.com/blog/?p=6011 https://www.bunniestudios.com/blog/?p=6011
- LIV2 6y ago
  Why would anyone bother about controlled impedance or matched trace length on an ISA card of all things? At the speeds ISA runs at you could connect everything with wire coat hangers
  - metaphor 6y ago
    You've summarily conflated a signal's fundamental frequency with the significant harmonic contributions of its edges. Take a moment to review ISA v2.01 spec[1] § 10.4: > To minimize crosstalk between signals and bus reflections, the driver rise/fall rates must be greater than 3ns; only LS or ALS devices are allowed. Faster parts such as F and AS cannot be used to drive the bus. If you thought it odd that normative language like this explicitly calls out tech by name, then you might have some sense of just how concerned Intel was about interoperability at the time...no thanks to wire coat hanger designers. Also consider the language of PC/104 v2.6 spec[2] § 4.7, which is probably one of the last remaining modern platforms with native ISA bus integration: > Whether [ac] termination is needed and where it should be located, is dependent on the specific system configuration and must be determined by the system designer. Now consider for a moment why this language and the paragraphs that precede it exist in this independent standard. This design is a commidity peripheral that will be integrated into arbitrary systems. Length matching literally costs nothing...but even this is apparently a "stretch goal" given the obvious fact that this design completely dismissed the normative max 2.5" stub length requirement. So no, I disagree. For the sake of sensible SI that's too often conveniently dismissed by hobbyists, keep your wire coat hangers in the closet. [1] https://archive.org/details/bitsavers_intelbusSpep89_3342148/page/n65/mode/2up https://archive.org/details/bitsavers_intelbusSpep89_3342148... [2] https://resources.winsystems.com/specs/PC104Spec.pdf https://resources.winsystems.com/specs/PC104Spec.pdf
    - LIV2 6y ago
      > Length matching literally costs nothing Well arguably it does cost some time yet is utterly pointless on an ISA card (unless your traces are 40 meters long I guess) there's plenty of pedantic things you could waste time on designing an ISA card

cosmotic 6y ago

Usage of surface mount and fewer but more integrated ICs are isn't what I would consider modern. Modern designs use those things, sure, but the technology and practice is not modern. It's actually pretty dated. Maybe not as-dated as a through-hole and discrete component design, but certainly not exclusive to recent (read: modern) designs.

metaphor 6y ago

Err...the only thing "modern" about this is you can purchase/fabricate one using new material and components in active production. If a modern cell phone integrated those SMT packages, it'd be the size of a brick, and the PCB itself is vanilla 2-layer Tg130 FR4...couldn't even be bothered with a controlled impedance layout, let alone length-match bus traces. Recently, bunnie talked about one "relatively modern" PCB design element---microvias---and explains why you just don't see them much in OSHW designs[1]...an excellent read if you get the chance. JLC is a well known PCB fab house because their low-volume prices are dirt cheap, not because their process is modern. [1] https://www.bunniestudios.com/blog/?p=6011 https://www.bunniestudios.com/blog/?p=6011

LIV2 6y ago
Why would anyone bother about controlled impedance or matched trace length on an ISA card of all things? At the speeds ISA runs at you could connect everything with wire coat hangers
- metaphor 6y ago
  You've summarily conflated a signal's fundamental frequency with the significant harmonic contributions of its edges. Take a moment to review ISA v2.01 spec[1] § 10.4: > To minimize crosstalk between signals and bus reflections, the driver rise/fall rates must be greater than 3ns; only LS or ALS devices are allowed. Faster parts such as F and AS cannot be used to drive the bus. If you thought it odd that normative language like this explicitly calls out tech by name, then you might have some sense of just how concerned Intel was about interoperability at the time...no thanks to wire coat hanger designers. Also consider the language of PC/104 v2.6 spec[2] § 4.7, which is probably one of the last remaining modern platforms with native ISA bus integration: > Whether [ac] termination is needed and where it should be located, is dependent on the specific system configuration and must be determined by the system designer. Now consider for a moment why this language and the paragraphs that precede it exist in this independent standard. This design is a commidity peripheral that will be integrated into arbitrary systems. Length matching literally costs nothing...but even this is apparently a "stretch goal" given the obvious fact that this design completely dismissed the normative max 2.5" stub length requirement. So no, I disagree. For the sake of sensible SI that's too often conveniently dismissed by hobbyists, keep your wire coat hangers in the closet. [1] https://archive.org/details/bitsavers_intelbusSpep89_3342148/page/n65/mode/2up https://archive.org/details/bitsavers_intelbusSpep89_3342148... [2] https://resources.winsystems.com/specs/PC104Spec.pdf https://resources.winsystems.com/specs/PC104Spec.pdf
  - LIV2 6y ago
    > Length matching literally costs nothing Well arguably it does cost some time yet is utterly pointless on an ISA card (unless your traces are 40 meters long I guess) there's plenty of pedantic things you could waste time on designing an ISA card