Elethiomel builds a Phoenix H5C1

[Elethiomel]

Although this post looks like it comes just a minute or two after the last, I forgot to hit submit, so it's about 30 minutes later...

OK, I decided to pull the shifter and install it off the board in a breadboard and jumper the 4 aforementioned connections.

20220414_002222.jpg (421.34 KiB) Viewed 2072 times

Still nothing. OK, it can't be a short on the board. We're not connected to the board now.

I checked pin 1 and it reads 10k to ground and that's fine as per the schematic.
Pin 2 still reads 32MHz
Pins 20 and 40 are GND and 5V respectively.

We KNOW the shifter is good. I've swapped it out a few times and ran the shifter in another ST.

Maybe the clock is out of spec?

20220414_002151.jpg (314.5 KiB) Viewed 2072 times

The clock peak is a bit high at 5.21v, but look at the minimum voltage on the clock. It's nearly 2 volts! Could that be it? It's just not low enough to register as a logic 0!

OK, I placed a 10k between pin 2 and GND to pull it down a bit. Not much difference. 1K a bit more.
Then at 470R I got 16MHz on pin 39

I've noticed that the oscillator I received from the store is not the exact same one as the one pictured in the store. Here's mine:

20220414_003350.jpg (217.77 KiB) Viewed 2072 times

Here's the store one :

osc32084.jpg (14.25 KiB) Viewed 2072 times

Maybe they require slightly different support passives? Perhaps a higher value on R913?

Either way, I've got 16MHz! Some advice as where to go from here would be appreciated. Should I just bodge in a 470R on the board?

[Elethiomel]

I'm assuming you have 32Mhz on pin 2 as you have 3.5V. But you really need to scope to diagnose switching signals.

AFAIK, 32Mhz into the shifter, plus 0v,power, you should be getting 16mhz output.

You can try carefully bending pin39 shifter up enough to miss the socket, then try 16mhz then..

Have you removed the MMU ?

Just seeing this now. Thanks for the late night reply

[exxos]

Maybe they require slightly different support passives? Perhaps a higher value on R913?

Either way, I've got 16MHz! Some advice as where to go from here would be appreciated. Should I just bodge in a 470R on the board?

Great find! The oscs are a different brand now. The ones in the store were custom made. I've used them and not had any issues.

There is I think a 33R resistor from the osc to the shifter.. Try linking over see if that makes any odds.

Also sometimes loading the osc with like 470R to 0v can help.

I'll measure the 32mhz voltages on my board when I get home tomorrow.

[Elethiomel]

It was easier to bodge in the 470R than it would have been to bridge the 33R resistor (R913), so bodge I did.

20220414_020653.jpg (498.77 KiB) Viewed 2066 times

And boot!

20220414_020611.jpg (406.63 KiB) Viewed 2066 times

I still need to add the other peripheral sockets, plug in the Gotek, the UltraSatan, maybe move the ST536 to this machine (although I do have an unpopulated TF5362b...), do some stability testing and shop for peripherals, but I think things are looking a lot better

[Icky]

Another Phoenix H5 lives

[JezC]

Great result!

Reminds me that I still need to fix the problem with the floppy accesses on my H5C1...

[Badwolf]

Well spotted @Elethiomel!

Good work on the board too. Looks nicely done.

I wonder what the issue with the oscillator is? Can't sink enough current for the line capacitance? Would a buffer or schmitt trigger help going forwards, I wonder?

Interesting stuff & reminiscent of Exxos' recent Falcon SDMA pull-down, albeit much more aggressive.

BW

[exxos]

I wonder what the issue with the oscillator is? Can't sink enough current for the line capacitance? Would a buffer or schmitt trigger help going forwards, I wonder?

This is one of those "rabbit hole" type problems I have had for a long time. The oscillators seems to always "over volt". Even by a couple of volts. I think it causes the chips to latch-up internally. But I have not really investigated or spent any time on diagnosing it.

On the flip side of that, the STE booster oscillator is switching between 3-4volts. There is no possible way really that should even work. So not even sure the scope is telling the truth.

I think it probably boils down to the oscillator is oscillating Probably with all the inductance and capacitance of things at that frequency is probably resonating all over the place. So adding some resistors such as 470R is more on par with a termination resistor. There could be some other screwy rings going on with the oscillators but who the hell knows..

Interesting stuff & reminiscent of Exxos' recent Falcon SDMA pull-down, albeit much more aggressive.

Completely different problem. That is a biasing resistor for the clock patch is not really a "loading resistor" like done normally. I am writing updated pages for the wiki to explain all this ,but it all takes time.

[exxos]

Quickly tested my H5 32Mhz clock on shifter pin 2.

IMG_0473.JPG (52.83 KiB) Viewed 2021 times

Under and over volting by 1 V. Possible undervoltage doing the voltage locking up..

I guess if @Elethiomel Can test the clock now that resistor is there as he already posted the before image we can see what difference it made.... I assume it would just shave off about 0.5v high and low.

[Elethiomel]

Thanks all, I'm really chuffed to see this boot and to be honest, having a problem and being able to solve it was great fun!

Here's the clock on shifter pin 2 with the system running. All ICs are in and a Gotek is in use.

20220414_122059.jpg (274.52 KiB) Viewed 2013 times