Lately I have been seeing many Amiga 600 and Amiga 1200 with bad Band Pass Filter (Z221) and Delay Line (Z222). This generally causing bad picture quality on composite and the RF-modulator.

There are many suggested solutions out there but none that really replaces the Z221 and Z222, until now. While attending to Datastorm 2018 I talked some with TBTorro and he said that he had done some tests with CXA2075 which was the high end version of the original CXA1145M which is the one used in Amiga 600 and Amiga 1200.

The modification is rather simple if you have some SMD soldering experience and you only need to components. So, why wait!

Bill Of Material

1CXA2075 RGB EncoderSOT
12.61 kOhm 1% ResistorThrough hole

CXA1145M vs. CXA2075M

The pinout is more or less 100% compatible between CXA1145M and CXA2075M. The main concern pins are Pin 8 and 9 on CXA2075M that is Audio In/Out on the CXA1145M and Pin 17 that is where we can enable the internal Trap which acts as the replacement for Band Pass Filter and Delay Line.


In my example I will show you how it is done on an Amiga 1200.

Start preparing the CXA2075M by cutting pins 8 and 9 as the picture show.

Remove the old CXA1145M from the Amiga 600 or Amiga 1200 mainboard. You can also remove Z221 and Z222, but we will disable them so it is really not necessary.

Flip the mainboard on the Amiga 1200 and locate the following area near the IDE 44 connector.

You should remove any of the following SMD components at this area. Note that some of the SMD components might already be removed from factory.

The de-soldering of SMD capacitors and resistors could be looking like this (pre-cleaning). Please be careful when cleaning the pads with solder wick. If you are using to much heat and pressure, you might damage the pads.


Clean the pads and add some flux of your choice and then put the CXA2075M in place. Attach one corner with solder, adjust the chip and put some solder on in the opposite corner of the CXA2075M.

When you are happy with the positioning of the CXA2075M, solder the rest of the pins to the mainboard, I these cases I prefer to perform a so called drag-soldering.

As you can see, I have already removed Pin 8 and 9 before soldering the chip onto the mainboard.

When the CXA2075M has been soldered to the mainboard, do a solder bridge between Pad 8 and 9. In my case I used a very short piece of the pin from the resistor.

This can also be done before the CXA2075 is soldered to the mainboard of course.

Prepare the 2.61 kOhm 1% resistor as following. Be aware so you don’t cut the pins too short.

Solder the 2.61k Ohm resistor between Pin 12 and 17 on the CXA2075M. This will add a Pull-Up so the internal TRAP switch is enabled and the CXA2075M will use the internal Band Pass Filter and Delay Line.

The final result should look something like the picture below. Be sure to connect between the mentioned pins, else you can damage your Amiga.


Alright, it is time for some testing. I hooked up my Amiga 1200 to a LCD TV with an RCA-cable between the composite port on Amiga 1200 and the LCD TV.

Power on the Amiga… Waiting… Waiting…

Holy shit! The result was probably one of the best quality pictures with composite I have ever seen. See for yourself!

And yes, the Mumin coffee cup is kinda of a thing I do 🙂


Well, I am super happy with this solution. The picture quality on the composite is now better than ever on my Amiga and I have more or less throw away any bad Z221 and Z222 that I find in the future and simply replace them with this solution.

If you have a working composite, please I would say it is not worth it unless you are a perfectionist 🙂

Over and out!

21 thoughts on “CXA2075M

  1. Jostein Kallevik

    Ohh this is nice stuff! I will look into this when I get the time, so this mod is bether than the 1645 then? Did the rgb out change or is it the same? I also have muminn coffe cup;)

    • It is a very nice mod. But the RGB out on A1200 does not use the CX1145 at all since it uses the BT101 video DAC. Let me know if do you do the mod and what you think of the result.

      • Jostein Kallevik

        I will test this mod for shure 🙂 yeah the RGB just pass-through i guess from pin 2-3-4 to 23-22-21. The C460 cap is it needed anymore? Just need to confirm that it is a 2.61K ohms resistor and not 2.61ohm, in your last drawing there may be a typo. Did you test the s-video out?

        Im hoping there will be a FPGA solution soon for the Amiga, tiny board and hdmi out 🙂

        Again Nice Work!

        • I think you could remove the whole circuit around vref and iref for the old CXA1145M. I need to verify. And yes, it is a typo and you write. The resistor should be 2.61 kOhm 1% so I will correct that as soon as I can.

          I have not tested S-Video out but will do it later and write a blog about that too. But what I could find was that it has very good quality and you only need to add 4 components.

  2. Do you know a reliable source of CXA2075M chips? They only seem available from Chinese sellers.

    • I followed the specification and that says 2.61k so I have not tried any other resistor. Can do the next time I do this mod but I do not know when that will be 🙁

  3. I cant get images from composite and/ or RF , but yes by RGB , i purchase CXA 2075 for this mod, is possible to restore composite output with this change? The resistors from bottom pcb MUST be take off, or can i leave them.?


  4. I did this mod and it was working fine initially, then something broke and I began having a weird video issue, both from the RGB and composite port, watch this video to understand:

    Note that every few seconds, an “almost correct” image shows up for a microsecond. You should be able to recognize Systest.

    It is NOT a problem with the new encoder, as I have proven that the bad signal is coming out straight from Lisa.

    I have already checked all the lines from Lisa and they look fine to me.

    Caps are fine too.

    The 28 MHz signal looks quite bad on a scope, but it looks the same as on another A1200 of mine, so I guess that’s OK.

    Any hints? Faulty Lisa?

    • Well, I have checked all signals coming from/going to Lisa against a working board and I cannot see any obvious differences. I guess I’ll try replacing Lisa, sigh. Thanks for your help.

Leave a Reply

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