ECS Passive Component Question

[JohnPCAE]

In the ECS daughterboard that plugs into the cartridge slot, the MCLK signal gets connected to two small green capacitors C16 and C17, and a small blue inductor L6. Does anyone know what their values are? I've been searching the web for information on how to interpret the value bands and I keep seeing conflicting information. Has anyone measured them or seen a schematic?

 

Edited October 13, 2021 by JohnPCAE

[carlsson]

Perhaps something on this discussion can help you find a definitive decoding:

https://www.eevblog.com/forum/beginners/capacitor-looks-like-a-resistor/

[JohnPCAE]

It seems like the capacitors might be 56pF and the inductor might be 56µH, but that seems odd. I figure they're using a tank circuit to recover MCLK from the rather severely attenuated version on the cartridge pin (because the Master Component has a 5.6k resistor on that pin that attenuates the heck out of it). The thing is, an L-C circuit with those values resonates at 2.84205 MHz, whereas MCLK is at 3.579545 MHz. Something is very odd here.

Edited October 13, 2021 by JohnPCAE

[+Lathe26]

The 56pF and 56uH values are the same ones I see on other ECSes (at least that are the values I decoded them as).  These are the values I put into the ECS schematic.

 

Looks like you're working on reverse engineering the ECS.  Do you have a project in mind?  I created a schematic of the ECS a while ago but have been hanging onto it with the intention of releasing it later with other ECS reverse-engineering.  Unfortunately, general life issues and other projects have gotten ahead of completing this work (ECS still on the back burner).

[JohnPCAE]

I'm working on my video overlay project but I'm getting occasional glitches that I think are related to how I'm reconstructing a square MCLK signal from what's on the cart pin. I need a decent MCLK signal that is in phase with what the Inty uses internally because I need to be able to synthesize one that has four times the frequency for NTSC color output (and I need its phase to also track the Inty's internal MCLK). So I wanted to see how the ECS does it to see if there's a better way. My method is rather complex: I'm using an AD8031 amplifier to amplify the signal and then a MAX913 comparator to generate a square wave with a duty cycle that's close to 50%.

 

This morning I ran some simulations in LTSpice and it seems that changing the capacitors to 68pF would give a better result, though the resulting sine wave would be exactly out of phase of the original. Leaving them at 56pF seems to generate a wave that's about 90 degrees out of phase. I also found that the phase of the resulting MCLK using Mattel's method is ***highly*** sensitive to capacitance, so much so that their method might not work for me at all anyway. A variable capacitor would be a must.

 

The most important thing is that I need the phase to be tightly controlled from build to build, or the color won't be accurate.

 

 

Edited October 14, 2021 by JohnPCAE

[JohnPCAE]

This is bizarre. I removed the 330k resistor that was providing hysteresis as the datasheet said to not add hysteresis. Unfortunately, that had no effect. Then, I added a 1N715 Zener diode between my ~14MHz MCLK4 signal and ground, and now the clock is rock-stable. It was just a wild guess, but I don't understand why that made all the difference. Before I added the diode, I noticed that merely probing MCLK4 with the oscilloscope made it stable. Anyone have any insight as to why this might be the case?

Edited October 14, 2021 by JohnPCAE

[+Lathe26]

You might already be aware of this, but there are 2 MCLKs on the ECS:

1. One MCLK comes out of the Intellivision, is altered by the LC circuit mentioned in your prior post, and goes into the chip-on-board (aka the blob-top chip).  This is the only place that MCLK goes.
2. The other MCLK comes out of the chip-on-board, passes through L5 (likely a ferrite bead), goes through the ribbon cable, is exposed on the J9 expansion connector (connector not always populated), passes through resistor R36 (5.6KOhm), and finally exits the game cartridge connector J6.

I am not sure what the chip-on-board does to the MCLK input versus its MCLK output.

[JohnPCAE]

I know that the 5.6k resistor is there because cartridges typically ground that pin. I also knew that the reconstructed MCLK entered the blob chip, and surmised that it would also exit because the sound chip would need it. What had me scratching my head was why the blob chip would even need MCLK. Another interesting thing is that the board looks at certain pins that are grounded in the Master Component but go to special signals in the Intellivoice. It makes me wonder if the blob chip's purpose is some sort of Intellivoice enhancement. The fact that the board is looking at ~REN from the Intellivoice makes me wonder if the blob chip is an external speech ROM, but there is no reason why such a ROM would need MCLK (not to mention the rest of the system bus). So it would have to be something else. The board is looking at all of the control signals that such a ROM would need, though: C1, C2, C3, S IN, S OUT, RCL, and ~REN. Maybe it's just the bankswitching chip?

Edited October 16, 2021 by JohnPCAE

[+Lathe26]

The blob chip needs MCLK to for the serial port's baud rate (used by both the cassette interface and the printer / AUX).  It even has a pull-down / pull-up input pin that controls the MCLK divisor for PAL vs NTSC markets (PAL MCLK needs a higher divisor to handle the 4MHz clock).  The bank-switching is handled by the ROM chip, not the blob chip.  I don't recall the chip handling any Intellivoice stuff but I might have overlooked that possibility.  I think the STIC shadow registers are suppressed by the MMI PAL chip and not the blob chip.