As I mentioned a few posts back, the Brain Board/Wozanium is a complex system and I’ve been bitten by a couple of small issues with the initial version 4.0 firmware.

The first few Brain Boards shipped (version 4.0 Wozanium firmware) have a couple of bugs. Because of the first issue, I’ll be shipping updated PROMs to those few individuals who were lucky enough to get the first units. Shipments going out after today will have version 5.0 firmware with these issues fixed. Here is the errata and workaround.

Errata 1:
Depending upon how the memory in your system powers up, the wozanium drivers could be clobbering the first few bytes of page zero memory. This will prevent the simple dump text example on the back cover of manual from working. It doesn’t look like BASIC uses these locations, so BASIC programs run fine.

Workaround:
Using the A1 monitor, set location $BC25 to 0.
BC25:0

Errata 2:
Upon power up, if reset is typed before the screen is cleared, there is a less than optimal display seen.

Workaround:
To avoid this, after power up, always clear screen, before typing reset. If you type reset first, you can always clear screen afterward with the right arrow key, with no harmful effects.

I was having a third issue, which was a picked pixel on the screen when running Wendell’s Star Trek on my system during the display of certain menus. The picked pixel was consistently the same, but the occurrence of the problem was very erratic. It would usually appear soon after powering up and would sometimes stop happening after reloading the program. After many attempts to try to get this to consistently repeat and detailed code review, I had strong suspicions that it was some kind of odd hardware issue. Running 3 different memory test utilities failed to uncover any memory problem. However, since re-seating the second bank of memory in my system, it has not reappeared. I’ll continue to try testing over the next week or so.

Before sending replacement PROMs to Brain Board version 4.0 firmware owners, I’d like to get mileage on version 5.0 and also make sure the picked pixel problem is gone for good. This will probably take a few weeks. If you are a version 4.0 owner and want updated PROMs sooner, rather than later, let me know and I’ll send them out now.

They look great – just like the first batch with the silk screen printed right on the copper in the power supply section. – I’m working through the wait list now, notifying people that I have units ready to ship. By next week I hope to be completely caught up.

I’m a little short of two part types. I only have a few 74161s which is needed for location D-8 instead of 74161As and I may have to ship a few kits with the A parts, but I’ll ship the 74161s for location D-8 to anyone who ends up in that category, once the backordered parts arrive. See this post for info on the problem with the 74161 parts.

http://www.willegal.net/blog/?p=299

The other part is the P-8667 transformer, but I think I have enough of those in stock to satisfy immediate demand. In any case, the transformer is a current production part, so if I run short, I can refill stocks quickly.

The keyboard adapters are sweet looking boards if you ask me. It took me about an hour to build up one keyboard adapters, that is after the three hours it took to remember how to program the AVR correctly.. So far it looks fine, but I have a little more testing to do. Here is the front and back of the PCB and the board I just built.

The second Mimeo 1 boards are also in. One little hitch though. This time, the fab house decided to mask the silk screen with the solder mask, against my explicit instructions. The only place where you can tell the difference is in the power supply section as shown here. I didn’t notice it at all until the third time I was reviewing the board. The rest of the board looks exactly like the first batch, as it should, since the same CAD files were used.

I’m going to complain to the fab guys tomorrow and see what they will do about it.

The cause of this problem is rather simple, just like the circuit. The base to emitter voltage drop of the MPS3704 transistor is .6 volts. In order to light the LED, you need an input voltage high enough to cover the base-emitter voltage drop of 2 transistors. This means the circuit is designed to light the led when an input signal of about 1.2 volts is seen. This is around the voltage of a normal line level signal, but the earphone output of my vintage cassette deck is nowhere near this, which is why the LED does not light. The shape of this signal is more important than the magnitude, which is why I can read data fine with my cassette deck, even at such low levels.

Well the fuel gauge story is not quite over. My fuel line became disconnected after a couple of drives, spilling gas on the driveway when I went to take the car for a short errand today. I think I just didn’t get it completely seated before putting in the clips. Just in case, I’m buying new clips tomorrow. At least the fuel lines are easily accessible.

Spent the afternoon replacing the fuel gauge sending unit on my ’04 Chevrolet Malibu. Had to drop the gas tank and two of the four bolts holding it in place were very hard to reach. What a pain – at least I didn’t have to deal with stuck or broken bolts. Last sending unit I replaced was in a ’91 Ford Escort and that car had an access panel under the back seat. So much easier to do that it is hard to imagine why having an access panel isn’t a standard design feature.

Bad wipers on the offending fuel gauge sender

Had to buy an overpriced sending unit from Chevy, since aftermarket companies only sell the entire fuel pump assembly and that costs about $500. The wipers on the original unit were just plain worn out. I hope that this one lasts a bit longer than the original.

Spam and the general unfriendlyness of phpBB has caused me to port my phpBB post over to this wordpress Blog – hopefully I’ll have better luck with this. I think blog software is better suited to what I am doing, anyway.

Check out this web page.
http://www.willegal.net/preamp/preamp.htm