Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Anything QL Software or Programming Related.
User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Peter »

Dave wrote:Its behavior is correct when the Q68 is powered, but if there is traffic on QLNET when the Q68 is unpowered, signal will still be passed to the FPGA regardless
The FPGA is protected by the Schottky diode, so no danger for it.
Dave wrote:plus try to power the Q68 through the diode.
An active network line would see 47 Ohms to about 0.5V, parallel to the normal 165 Ohms to 0V.
No risk of damage for other connected machines, but indeed a non-functional (or not reliably functional) state for the network.
Dave wrote:My solution would be to use a pair of BSS138 transistors to create a level shifter that would be high impedence if unpowered.
How exactly would you wire them?


FrancoisLanciault
Trump Card
Posts: 167
Joined: Mon Aug 08, 2011 11:08 pm

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by FrancoisLanciault »

Peter wrote:
FrancoisLanciault wrote:About the electronic add-on needed, it is not clear to me what the red lines in the PDF means i.e. what needs to be removed from the original schematic.
All three resistors with direct connection to the transistor, and the transistor itself, are removed.
The idea was to show the difference between Q68 and QL circuitry. Sorry if that was misleading.
Ok thanks Peter, it is clear now. Looks easy to put together.

François


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Dave »

Dave wrote:My solution would be to use a pair of BSS138 transistors to create a level shifter that would be high impedence if unpowered.
How exactly would you wire them?[/quote]
bidirectional BSS138 voltage level shifter<br />(image courtesy sparkfun)
bidirectional BSS138 voltage level shifter
(image courtesy sparkfun)
BSS138.png (48.23 KiB) Viewed 5378 times
This is pretty much my standard level shifter. The only ugliness is grabbing 5V, but by removing the HV side power source and resistor, it will convert 5V to 3.3V to the FPGA, while allowing the 3.3V straight through in the other direction. If anything, they're really good shock protection.

I love the BSS138. I compulsively buy them whenever I see them at a good price.

https://www.onsemi.com/pub/Collateral/BSS138-D.PDF


User avatar
Peter
QL Wafer Drive
Posts: 1953
Joined: Sat Jan 22, 2011 8:47 am

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Peter »

Unfortunately, this does not seem to work. When the low voltage side (Q68 FPGA) device transmits a high level (3.3V), the MOSFET is tied high (off). The 10K pullup on the high voltage side is then not suffient to drive the network line. The 0.8 V (typ.) to 1.4 V (max.) loss over the Drain–Source Diode is also too much to drive the network line well.


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Dave »

In light of Peter's reassurances, I am releasing the following files under the standard open hardware license.

This board implements two QLNET sockets, and has an extension bus through connector. I already have all the parts in stock, and will order the PCBs in the morning. Given the good number of Q68 sales, I will make a batch of 50. The left socket fits in the current "power button" hole and the right socket 10.5mm to the right of that, but not quite above or blocking access to the left SD socket. The power switch can be relocated inside the case, in a new hole on the side, or omitted if a wire jumper is included.

I'll just let this float here for a day before ordering PCBs so I can fix errors, incorporate any enhancements or feedback comments. If there are errors, I will replace the files in this post.

Enjoy.

Edited to fix small error.
Schematic
Schematic
PCB
PCB
BOM
BOM
Gerber_Q68 top deck_20190613f1.zip
Gerber
(27.71 KiB) Downloaded 174 times
Last edited by Dave on Tue Jun 18, 2019 4:39 pm, edited 1 time in total.


martyn_hill
Aurora
Posts: 909
Joined: Sat Oct 25, 2014 9:53 am

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by martyn_hill »

Terrific, Dave!

I couldn't spot your 'mistake', but take your word for it!

One thought - those 2mil pitch headers are more awkward to find connecting cables for than the more common 2.54mil pitch - could it work to include the 2mil (to connect to the Q68) and a 2.54mil for the pass-through connector?


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Dave »

All the files are there if you'd like to do that. If a couple of other people request it also, I could add a second header on .1" pitch, and either could be fitted.

The 2mm header was quite problematic. It is positioned too tight up against the RJ45 jack, so shrouded headers cannot be used. There's about 2mm too little clearance. Just be careful when mounting a card onto the connector that pins are aligned properly. The board stops just shy of the upright supercapacitor that provides clock power. Longer boards do interfere if they sit low enough so a cut-out for it is recommended.

I am defining a standard, as follows: The connector on the Q68 side will always be female. This makes it much less likely to fail due to bend pins. The locating pin on the header must always have the locating lug on the right (so, facing the SD slots and JTAG port.) If all others following me follow this scheme, all cards will fit with all others.

The PCBs will be blue, to match.


Derek_Stewart
Font of All Knowledge
Posts: 3928
Joined: Mon Dec 20, 2010 11:40 am
Location: Sunny Runcorn, Cheshire, UK

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Derek_Stewart »

Hi Dave,

Nice PCB, but a couple things I do not like.

Front mounting Network sockets. The Front will look very cluttered, I would prefer these on the left hsnd side (looking from the front).

New hole for the power switch, there are two Aluminium guide rails to drill through, which there is a risk of drill jamming.

I am fitting male 2x12 DIL Pin Headers.

Can not agree to this unofficial standard, you should discuss a proposed standard before trying to force it through.


Regards,

Derek
bixio60
Brittle Membrane
Posts: 110
Joined: Sun May 04, 2014 7:05 am

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by bixio60 »

Peter, Derek,
QLnet it is very interesting and valuable and .....May be it is time for the "Q68 Plus" : with Cache and QLnet ?
I am afraid I am not able to do any hw mod on my Q68.

Fabrizio


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Announcing availability of a QLNET driver for the Q68 (ND-Q68)

Post by Dave »

Martyn Hill, Esq. wrote:One thought - those 2mil pitch headers are more awkward to find connecting cables for than the more common 2.54mil pitch - could it work to include the 2mil (to connect to the Q68) and a 2.54mil for the pass-through connector?
Like this?
Screen Shot 2019-06-18 at 1.03.49 PM.png
Screen Shot 2019-06-18 at 1.10.43 PM.png
Derek_Stewart wrote:Hi Dave,

Nice PCB, but a couple things I do not like.

Front mounting Network sockets. The Front will look very cluttered, I would prefer these on the left hsnd side (looking from the front).

New hole for the power switch, there are two Aluminium guide rails to drill through, which there is a risk of drill jamming.

I am fitting male 2x12 DIL Pin Headers.

Can not agree to this unofficial standard, you use discuss a proposed standard before trying to force it through.
The net sockets can not fit to the sides. They have to mount slightly beyond flush with the metal surface. Side mounted, it would become impossible to slide the Q68 board into the existing cases as the jacks have a fixed and rigid relationship to the PCB, whereas the power switch does not because it is on a flying lead. You suggest putting the net sockets to the sides, but then in the same breath suggest it would be too hard to put the power socket on the sides because it is hard to drill. The power socket embeds quite nicely into the outside upper slot, just by sanding the top and bottom edges of its lip off. ;)

The front may look cluttered, but unfortunately the back is fully occupied. Since there is a power cord in the front, it made sense to have other similar sized cables as close to it as possible. I tried to put the second net port over the power socket, but there isn't clearance. This is the trade-off with Peter's incredibly compact and efficient design. I am probably going to have to shave the left edge of the board very accurately, so it tucks in very flush against the power socket. I did consider trying to use a pair of corded sockets, but then the sockets are not switched and termination would not work correctly. I might even have to move R2 and R3 up behind the jacks to make the board slimmer. I know I am good on the supercap side but on the power jack side it is *very* tight.

I'm speccing the female header to be on the Q68 because it's the correct choice for this application. The address and data lines are directly linked to the FPGA and RAM. The FPGA has clamping diodes on most pins, but the RAM does not. The pluggable connector supplies power so it would be less than ideal because clearances require unshrouded pins.
The experts wrote:Electronic designers often select female jack connectors for fixed mounting on electronic equipment they design. This is usually done because female connectors are more resistant to damage or contamination, by virtue of their concealed or recessed electrical contacts. A damaged motherboard connector can result in the scrapping of an expensive piece of electronic equipment. The risk of expensive damage is reduced by relegating the more exposed male contacts to connecting cables, which can be repaired or replaced at lower cost.
I will supply both headers loose, so if someone opens their Q68 and finds a male header they will be able to proceed. It's a soldering required job anyway. That way, they can do what they want.

Follow-up to this is a floppy interface. It uses the same IC used on the SGC, so drivers already exist in SMSQ/E. They just need the target address and spacing of registers editing. I've done the hardware, but the driver is beyond my skills.

Again, much appreciation to Martyn Hill for his excellent work with this driver.


Post Reply