ACA500plus odd mouse activity at clock > 7MHz

Hello,

I've been slowly upgrading my A500 Amiga's and just received my ACA500plus yesterday.

Installed Workbench 3.1, "acaflashrom"ed my KS 3.1.4 ROM into slot 1, rebooted and installed WB 3.1.4 on top, worked flawlessly

Thought I was in the clear, but...

When I set the CPU clock rate to 14MHz or higher and when the machine is given a task, like playing mod files,
the following video shows the 'idle' mouse pointer activity I see

https://youtu.be/LXp5_6HT56s

If I switch the clock frequency back down to 7MHz or remove the ACA500plus, the mouse pointer has 'normal idle' activity..it just sits there, not moving on it's own...

I see this with either my A500 tank mouse, or my USB bluetooth mouse; plugged into port 1.

Machine details:

The machine is an A500 rev5, one of the early ones with the factory transistor bodge to compensate for a defective Toshiba Gary.

https://forum.amiga.org/index.php?topic=72993.0

I've replaced the early Gary with a later revision Gary

I've replaced the 8371 Agnus with an 8372 Agnus (bought here)

I've re-capped the A500 with Würth Elektronik Aluminum Polymer caps

I've cut JP2, to get the second 512KB of ChipRAM from Ranger

I've installed a 2MB trapdoor Chip\SlowRAM board + RTCC and Gary adapter from (shop link removed)

I've swapped the 68000 for a Motorola 68010

I burned my own KS 3.1.4 EPROM and placed it in an A500 rev5 socket adapter which swizzles some pins and provides some needed pull-ups.

I'm using SwissBit 512MB and 4GB SLC with the C-500 controller chip, Industrial temp rated, controlled BOM, Compact Flash cards in the ACA500plus.

I rebuilt my A500 PSU which was dying, using a Mean Well RT-50B, 5V@4A, 12V@2A, -12V@1A module and adjusted 5V trim based on measurement taken at the A500 input filter.

3D printed an ABS UL94 V-0 cooling baffle to support an 80mm fan within the PSU housing, to further stabilize the PSU.

Hopefully this mouse pointer issue has been seen before and has a known fix,

or at the very least I've provided enough detail to prompt some ideas for next steps.

Handy too...we have a full electronics lab here.. just shy of a spectrum analyzer really.

Oh... and a complete set of A500 known good chips in ESD foam.

So, we can do a little measuring or swap some kit if the mood takes you..

I should warn, I've only just get back into Amiga, so if you need something software'ish done, I'll need some hand holding.

Electrical engineering, and technician work, no problem, still hand-soldering TQFP for fun

3AM post, figures I missed something

Machine details:(append)

Denise chip replaced with Indivision ECS V2 (bought here)

And this mouse pointer 'idle' action is not limited to a specific piece of software,

per the Indiana Jones Fate of Atlantis example.

At 14MHz I have to move the mouse slowly, like I'm moving in to click a tick box,

and then take my hand off the mouse when the pointer starts to jump.

It's 'like' some positions on the screen are more prone to initiate this behaviour.

I've also seen it just slowly drag the pointer to the right side of the screen...

As the CPU clock rate goes up, the behaviour becomes easier to trigger,

and in fact at 21MHz, you really don't have to try, even just running WorkBench.
The pointer still moves under mouse control, but you just have no fine aim, with it jumping around like that.

I'll see about updating the Indivision ECS V2 core...thank you.

12 years automotive design engineering, managing high amperage over 10 meter long cables..

Now in semiconductor industry.

I've replaced my board side bulk capacitors with enough mF and low enough ESR that I'm not concerned with the Mean Well's ripple.

If it was an issue...getting from 80mVp-p to 50mVp-p, I'd just throw a few inductors in the PSU case and low-pass away some of the Vp-p.

To be clear,

The original Commodore PSU does not manage cable drop or use remote sensing either.

If I was 'promoting' a PSU replacement solution, and I'm not,

I think I'd be very careful to point out that your focus is on providing something superior to the original,

something to support the higher loads that our machines must deal with in the 'now'.

Otherwise, it sounds like, and I'm sure you don't intend this,

it sounds like engineers who replace their existing failing PSUs with a Mean Well RT-50B or Mean Well RT-65B,

are somehow producing a PSU that is inferior to the original, 30 years old, failing, PSU.

Keeping in mind, there are no new ones to compare too...you have to compare to the ones that are dying...

I have scope traces...and this is a whole lot better than what came in the door with my Amiga

We should not let 'perfection' be the enemy of 'better than broken'

Tricky line to walk... Gotta stay on the positive side of that...

As for links, certainly..I can avoid providing links and instead describe make and model of things installed in the future.
Links are just so much more lazy and engineer like

Mouse issue seems to be fully resolved by applying Indivision ECS V2 firmware patch

I'll play so more, but we can close this one out I think

Thank you kindly,

Okay, so just in case I forget to trim my supply manually...

Since I use a PCB adapter board already to retrofit my A500 PSU's with Mean Well RT-50B's, which has heavy traces to carry the low-voltage away from the high-voltage side, I can use the existing Vss trace as a 'free' current shunt, and feed that signal into an $0.80 Atmel 328P, using the built-in A2D input OpAMP and 1.1V reference along with the bandgap temperature diode to make pretty reasonable estimation of cable current, and thus voltage drop.
Then add 6 resistors to make a 5 bit D2A, which trims the Mean Well RT-50B to compensate for cable drop automatically.
Then add a $0.40 NUD3124 MOSFET to switch my FAN for constant temp operation...I wanted a temp sensing fan anyway, so this is great.

20mV - 30mV step compensation is perfectly fine for me.

If I really wanted to isolate the 5V current, then I could use an INA138 on the 5V shunt line, but that adds a second $1.00, and is likely just overkill.
Coming from an automotive background I need tombstones to spend a dollar

As I mentioned, I could use an INA138 to isolate the 5V current, but since half the voltage loss on the cable is on the Vss side, I don't think it's going to be that impressive a difference in the end result,,,

Microcontrollers, control the relative position of your valves and pistons preventing collision at 5000 cycles per second for 5000 power-on-hours.

They are sufficiently trustable in this much lower risk application.

The MCU permits a one time auto calibrate cycle in the field, eliminating any trim elements.

As for over voltage, the Atmel would be fine to a full 6V, but a diode or two can provide the voltage drop to stay well under 5.5V

The mean well is ~$30 CDN over here and comes fully CSA\UL\FCC

The proposed compensation circuit is $1 CDN

The PCB adapter board is about $4

Our little group found an adapter PCB layout from a UK fellow and made a batch.
They needed a bit of re-shaping to fit our Amiga PSU enclosures.

I've already started modifying the adapter PCB layout to accommodate 'more' of the original Amiga PSU enclosures as we've found there were quite a few mold variations over the production life.

I'll just add the needed footprints for the compensation circuit, for those who want to spend the extra buck.

I think our group will give the MCU programmed for free to everyone in Canada that needs one.
We tend to give away boards to people on fixed income and the like, so an MCU, is no big.

But if it was a product I had to offer support on...yeah...I'd have to think about analog vs digital.

I do like that with the MCU we just hook a jumper wire between the A500 and the PSU, initiate auto calibration and it's finished...the MCU has correct compensation scale and you can loose the jumper wire.

Excellent idea
And the furthest reasonable point from power entry