Cloning the ET-3404 6809 adapter for the Heathkit ET-3400

In this video, I make a clone of the ET-3404 6809 adapter:

Purpose

I’ve become somewhat of a Heathkit ET-3400 enthusiast lately, after building my ETA-3400 memory/io clone. I decided to also clone the ET-3404, which is a 6809 adapter board for the ET-3400. Why? Simply for completeness sake.

The official “experiments” for the ET-3404 aren’t all that exciting, they’re mostly focused around understanding the differences in the microprocessor. Differences such as addressing modes, register changes, etc. There’s not much flashy interfacing like you would find in the other labs. Nevertheless, the ET-3404 fit an important niche in history. Recall that back in the day all of this stuff was new, computers were expensive, and the ability to upgrade your trainer and learn about the 6809 would have been valuable.

Design

I pretty much used the original design of the ET-3404 and made a few relatively straightforward changes to it, to pick a different ROM and to allow some additional flexibility. The schematic is shown below:

ET3404 Clone, Schematic

As you can see, I chose to use an AT28C64 EEPROM instead of whatever the original board used. The AT28C64 is popular, easy to program and erase, and I use them in other projects.

The AT28C64 is a larger ROM than the original, so I included some solder jumpers to change around the address lines so that additional space could be addressed. I have not tested this capability — I set the jumpers as described in the schematic to address only 2K of the ROM.

I also included a jumper to select whether the E line is unbuffered. I found two different variants of the original ET3404 schematic, one buffered and one buffered. I’m not sure if one variant is more “right” than the other, but I allowed the option to be chosen by the builder.

Implementation

The completed board is shown below:

ET3404 Clone, Top View
ET3404 Clone, Bottom View

The board design is straightforward. Please note the positions of the five solder jumpers, which are all the way to the left. This is the standard configuration to address 2K of the EEPROM.

The two terminal blocks came from adafruit. If you can’t find them at Adafruit (they were out of stock when I needed them), you might be able to find a supplier such as Mouser or Digikey that sells Adafruit products.

The pins sticking out the bottom of the board are a little hard to source — I found them on ebay. They are *NOT* standard 0.100 header pins, which would be too large to plug into a socket. They are special machine turned pins.

If you’re modifying an ET-3400 (not ET-3400-A)

Heathkit produced two generations of the trainer, the ET-3400 and the ET-3400-A. The 6809 board is a drop-in replacement for the CPU on the newer version, the ET-3400-A trainer. However it is not so simple on the older version the ET-3400.

The ET-3400 requires that you replace the clock driver IC with a small header that includes a resistor and capacitor. This unfortunately requires you to disassemble the ET-3400, including removing the main pcboard from the top shell of the case, as the clock driver chip is not accessible externally. It’s a bit of a pain, doubly-so as once you tire of the 6809 and want to return your trainer back to the 6800, you have to disassemble it again and re-insert the original clock chip.

Bill of Materials

IdQtyDescriptionSource
IC11MC6809P MicroprocessoreBay
(see note)
IC21set of machine-turned pinseBay
IC3174HCT367Ndigikey
IC4174HCT20Ndigikey
IC51AT28C64 EEPROMdigikey
C1,C2227pF ceramic capacitordigikey
C3-C640.1uF ceramic capacitordigikey
R1-R3310K Resistordigikey
U$1,U$222×4 breadboard socketadafruit
Q114Mhz CrystaleBay
S1140-pin IC socketdigikey
S3114-pin IC socketdigikey
S4116-pin IC socketdigikey
S5128-pin wide IC socketdigikey

Note: It is important that you get the 6809P, and not the 6809 or 6809E. The “P” includes the built-in clock generator logic. I had numerous problems with either counterfeit or failed 6809P processors. See the following section.

Bad 6809P ICs

One US supplier 2/2 ICs that I received were bad. One China supplier, 2/8 ICs that I bought were bad. This is a significant prevalence of bad ICs and it existed across suppliers from different geographic regions. I’ve spoken with other ET-3400 enthusiasts and they reported similar experiences of receiving many bad 6809P.

The symptoms of the “bad” IC were that it did not generate E or Q clock pulses. There doesn’t seem to be a better way to evaluate them other than to buy several and try them in-circuit.

Interestingly, I did try to force clock signals into these “bad” ICs, similar to how the 6809E (the variant without an internal clock generator) is supposed to be used. I found that with an external signal generator, it was possible to get these ICs to function. The weirdness then is that 6809P are behaving like 6809E. Given that the problem happened from multiple suppliers with different lots of ICs, I’m skeptical that it’s a simple counterfeit. My pet theory is that there is an internal fuse or other feature that differentiates a 6809P (with clock generator) from a 6809E (without clock generator) and that fuse has degraded over time. That’s just speculation though.

The bottom line is that obtaining a functional 6809P may be a challenge for this project, and if you don’t have a source of known-good 6809P then it may be difficult to ascertain whether a failed board is due to a bad IC or a different problem. The telltail check is to put a scope on the E and Q outputs and see if they’re dead or alive.

Leave a Reply

Your email address will not be published. Required fields are marked *