Proof of concept : a 15W USB Power Delivery charger for an Atari STe

[sporniket]

I finally got to make this experiment, that was a success : it did power my Atari STe (no add-on, a GOEX floppy emulator instead of a legacy floppy drive), and it did not instantly kill it. I don't have a scope to look at voltage spikes and noises, so I am bound to trust that the USB-PD requirements are enough to not fry my STe little by little :?: (I was quite confident, as I also dit the same with my thomson MO5, even if I had to buy a bigger charger to have an output voltage of 15V)

Anyway, let's get started :

Here is the charger under test : a small wattage (15 W, 5V at 3A) USB power delivery charger, bought 10€ at a hardware and tools big store.

2022-11-11--00-usb-charger-5V-3A.jpg

And here is the schematics and the contraption powered on. The USB trigger is this one https://fr.aliexpress.com/item/1005003336833794.html (NOT an affiliated link ;) ) ; the 5V->12V converter is from exxos shop, item #223 (NOT an affiliated link too :D ) ; I put a very small load (2KOhm resistor + Led) for the sake of testing the voltages. I was too lazy to plug the power resistors I just bought at exxos shop.

2022-11-11--01-schematics-and-light-on.jpg

Testing the 5V rail and the 12V, seems ok.

2022-11-11--02-test-5V-rail.jpg

2022-11-11--03-test-12V-rail.jpg

Now it's time to test (plus I also tested a self made gamepad, that experiment is a fail).

2022-11-11--04-ready-to-test.jpg

It's alive ! I let the computer run for a few minutes, no problem. (once again, I don't have a scope, so I cannot check for voltages spikes and noise)

2022-11-11--05-alive-0.jpg

2022-11-11--05-alive-1.jpg

2022-11-11--05-alive-2.jpg

That's all for today, it's time to sleep...

[exxos]

I've been meaning to look what enables the 9v fast charging feature on devices.. They spike from 5v to 9v. But not sure what a triggers the fast charging.Might just spike by default. Might be worth looking into how all that works ;)

[sporniket]

Power Delivery as well as "Quick/fast charge" variations relies on a protocol between the charger and the device to get voltages higher than 5V, and thus need specialized chips. I looked at power delivery as it is a spec from USB, so not tied to a particular vendor, and it allows up to 20V at 5A for now, and recent update allows to go higher, way above what I need for a ST or a H5.

As a side note, IIRC with a USB-C charger (not power delivery), one can get a current of up to 3A by pulling down the CC1 and CC2 (Channel Config 1 and 2) with specific values, the charger sense the voltage drop on those lines to enable higher current (the USB cable should of course support that, meaning that the voltage drop caused by the cable must be within specs to enable high current).

For more power hungry modded configuration (e.g. like a ST with a Raspberry Pi as RGB2HDMI or Pi Storm) I plan to use one of those specialized ICs to get the power, add some DC step downs, a soft latch button for switching on/off and voilà. (Yes it's not that simple, but you get the idea :) ). My plan is essentially motivated by not having the computer directly plugged to main.

EDIT : some readings... (external links because pdf not allowed here)

• TI primer : https://www.ti.com/lit/wp/slyy109b/slyy ... 8072870706
• STM presentation (heavily oriented toward their portfolio to make a power adaptor) : https://www.st.com/content/ccc/fragment ... USB_PD.pdf

[sporniket]

After a few sessions with my contraption, I have noticed that while switching on I was expecting some noise... that did not come. No humming, compared to the original PSU.

[exxos]

Would be interesting to see how spikey it all is.

[sporniket]

I believe that next month I will finally be able to get myself an oscilloscope.

[sporniket]

Today I finally switched on my brand new oscilloscope, calibrated the probes, and measured the voltage going out the USB trigger (the little thingy that is plugged to the power adapter and convince it that it can deliver 3Amp at 5V)

To do the measure, I used the pin of the probe and the little spring to ground (I remember having read a post from @exxos explaining that).

2023-04-16--probe.jpg

The probe is in 10x mode, as well as the oscilloscope.

2023-04-16--voltage-measure.png

So, if I did things correctly, the voltage is inside 4.7-5.1V range, with no noticeable spike. In other words, seems safe enough to me.

[exxos]

So, if I did things correctly, the voltage is inside 4.7-5.1V range, with no noticeable spike. In other words, seems safe enough to me.

That looks rather nice. If you really wanted to get "into it". Just put your scope on AC on like 50mV / div and it will center on the screen and you can measure the actual ripple. You'll have to fiddle with the timebase to see anything meaningful though.

For example.

filter-min.png

[sporniket]

Ok, thanks for the tip, I will try that the next time I unshelf all my stuff :)

[sporniket]

That looks rather nice. If you really wanted to get "into it". Just put your scope on AC on like 50mV / div and it will center on the screen and you can measure the actual ripple. You'll have to fiddle with the timebase to see anything meaningful though.

Today I was feeling like making other measurements, and I succeeded to put the probes on the adapter output so that my hands were free (thanks to a via to ground left untouched, the springy thing to ground the probe could then maintain the tip against the Vcc wire.

AC mode, 50ns per horizontal division, 50mV per vertical division (edit: now I wonder whether I zoomed too much comparing to your reference pictures). The trigger level setup to catch the biggest oscillation that comes up regularly. I did 3 capture, it seems fairly stable to me :

dso_01_01_00_33_01.png

dso_01_01_00_33_05.png

dso_01_01_00_33_10.png