Future PSU designs

[exxos]

My new switch mode chips came today

Some more hacking to my demo board... I worked out that start circuit would need 1uF capacitor in order to give about a couple amps soft start charge into about 10,000uF capacitance. So to be on the safe side I just used 2.2uF. This gives about 300ms start up time. I guesstimated that this will give about 1 A soft start charge.

In any case, thankfully this is solve the problem and I have added on a 4,700uf low ESR capacitor on the power supply had no problem starting up with this value. Before he could not start up with even just 1,500uf.

Oddly even know this new chip should have slightly better efficiency than the old one, it is actually drawing a little more than expected so I will investigate this and see if I can find out why...

IMG_1908.JPG (127.91 KiB) Viewed 7162 times

EDIT:

Code: Select all

9V	1.17A	10.53W	94.21%
10V	1.05A	10.50W	94.48%
11V	0.96A	10.56W	93.94%
12V	0.87A	10.44W	95.02%
13V	0.81A	10.53W	94.21%
14V	0.75A	10.50W	94.48%
15V	0.70A	10.50W	94.48%
16V	0.66A	10.56W	93.94%
17V	0.62A	10.54W	94.12%
18V	0.59A	10.62W	93.67%

Okay, so, my previous best efficiency was 94.70 @ 15V input. So the efficiency loss was not as bad as I initially thought. In actual fact the best efficiency is now at 12 V at 95.02% efficiency! So this new IC is capable of marginally better efficiency than the previous one overall efficiency figures are averaging out around 94.50% anyway so there really is not much difference with any input voltage really..

So now am now confident that this circuit can ultimately be used to replace my previous power supply design. Also as stated before, this design does not need any electrolytic's on the low-voltage side. It can just use ceramics. This will also save a lot of PCB space and will also save on the assembly time. This circuit does need a few more extra components, but at least these are still surface mount, it will be a lot quicker and easier to assemble anyway. Of course the bottom line figure is I was looking for better efficiency.

So now I am happy with this design, I now need to look at 12V switch mode mode circuit's and trying find one which has the best efficiency at around 15 V input. As mentioned before, likely 24 V would give the best efficiency with a lot of chips, but I'm sure I found one somewhere the best efficiency at 16 V. So I just need to wade through millions of datasheets again until I can narrow down my candidates list to just a handful of chips to try

[exxos]

IMG_1910.JPG (125.78 KiB) Viewed 7149 times

New demo board came today

This one is wide up now to output 12 V.

Regulation as below..

121.png (3.91 KiB) Viewed 7149 times

So basically 20mV ripple peak to peak.

Efficiency figures as below..

Code: Select all

11.97V 22R 0.545mA 6.523W

11V	NA
12V	NA
13V	0.51A	6.63W	98.39%
14V	0.47A	6.58W	99.13%
15V	0.44A	6.60W	98.83%
16V	0.41A	6.56W	99.44%
17V	0.39A	6.63W	98.39%
18V	0.37A	6.66W	97.94%

So here we have 545mA load which simulates a couple of hard drives on the 12 V rail. Regulation and efficiency is extremely good. I'm a little sceptical about these figures, but there will always be some small error in the tests anyway. But it does give a very good indication that this IC is very usable.

I also tried 125mA load as this simulates the stock Falcon power requirement on the 12 V rail. Efficiency and regulation figures was still extremely good.

Next up is to make sure soft start circuit will work with higher output capacitance values...

Stock circuit started up with the 125mA load and 4,700uF, but failed at the higher 500mA test current. So another tweak to the soft start as I did the 5 V IC and it started up fine.

The demo board uses a 15uH inductor, I do not have any values near that, but I will probably try a 22uH inductor in the future to see if I can get a little bit better ripple regulation...

So I added to 10uH in series for 20uH.. It does not look pretty but it works...

IMG_1911.JPG (103.87 KiB) Viewed 7143 times

And the regulation..

1222.png (3.86 KiB) Viewed 7143 times

And I tried a original 47uH..

IMG_1912.JPG (99.62 KiB) Viewed 7143 times

And the regulation..

1247.png (3.66 KiB) Viewed 7143 times

So the 20uH did not really change anything. The 47uH basically half the ripple to about 10mV. Will probably stick with 22uH for this, so I can keep the resistance of the inductor low, and keep the physical size low as well.

So when I get some free time (probably a long long long time away) I will start creating a new power supply design based on these two regulator chips

[exxos]

Couldn't sleep last night, so did a basic layout of the next gen PSU..

ffcc3e67829fd5362e9aca1e8ad88518.png (76.62 KiB) Viewed 7102 times

[exxos]

Chopped... and prototype pcb ordered

c774c59a90627f87c78358d0941a33d1.png (26.03 KiB) Viewed 7093 times

I am going to try and add a voltage selector on the final, for 110V/230V.

[guus.assmann]

Hello Exxos,

Congrats on this new product.
Regarding the voltage selection, I have a suggestion.
I would make it such that it needs soldering to set the correct voltage.
That way, it's not possible to select a wrong voltage by accident.
And I guess this will effectively prevent claims and incidents.

BR/
Guus

[exxos]

Regarding the voltage selection, I have a suggestion.
I would make it such that it needs soldering to set the correct voltage.
That way, it's not possible to select a wrong voltage by accident.
And I guess this will effectively prevent claims and incidents.

I was trying to add a voltage selector switch, but struggling for space So might have to be some jumper links, though bare links is a bit of a no no..

[guus.assmann]

Hello Exxos,

A selector switch is exactly what I wouldn't recommend.
Yes, it's convenient. But also very easilly put in the wrong position.
And that may be hard to prove.

Why not have a couple of holes in the PCB.
Depending on how the wires are soldered in those holes, it is a voltage setting.
And holes in a PCB are relatively cheap. But also hard to do wrong.

BR/
Guus

[exxos]

Yes, it's convenient. But also very easily put in the wrong position.
And that may be hard to prove.

Why not have a couple of holes in the PCB.
Depending on how the wires are soldered in those holes, it is a voltage setting.
And holes in a PCB are relatively cheap. But also hard to do wrong.

The idea is that I don't want to keep 2 voltage versions of the PSU again. One always goes out of stock faster than the other, no matter what voltage I do more of. So having a selector means I can just have one PSU in stock.

It could be done with jumper links or variations with soldering on the PCB, but really no point, as currently I am tapping the transformer for either 110V or 230V directly and supplying them that way depending on what is the orders. If I did this on the PCB, it would just result in more soldering. Technically soldering the switch and soldering the is more work, but it is the trade-off for just keeping one power supply type in stock.

I would supply them all as 230V selected due to testing anyway. Running on 110V would mean the power supply would likely malfunction, but without damage. If someone ran 230V into 110V, the transformer would get extremely hot and there would be a burning smell after about 30 seconds. If the power supply was left for several minutes in this state, the transformer would start burning and melting and smoking and it would be pretty obvious that someone had selected the wrong voltage. Though if someone cannot even select the correct voltage, then they should not even be attempting to upgrade the power supply in their machine in the first place.

The only alternative that I can think of would be to do as you say, but I do not solder any of the transformer primary windings at all, and let the customer do this. Though I think going by total confusion of the colour wiring it causes people, I do not think that is a good direction to go down. People at the moment already struggling to even wire up the transformer up how I supply them now. If there was just 2 wires to solder and they were the same colour are both voltages, then this would prevent his constant confusion with people. This way they just simply select the voltage on the selector and they are done.

[Bikerbob]

Well I would like to say, that I would be interested in this.

I am in Canada, so a 110v is my need, and a switching or one that can be switched is better, because it opens up a world market for the machine if I ever want to sell it. I would also agree that for you .. much easier going forward.

What is the projected price point of this one Exxos?

Thanks

James

[exxos]

I am in Canada, so a 110v is my need, and a switching or one that can be switched is better, because it opens up a world market for the machine if I ever want to sell it. I would also agree that for you .. much easier going forward.

A small number of people requested a switchable option over the past year. Some people travel around and take their Atari with them, so having a switchable is useful for them as well.

What is the projected price point of this one Exxos?

I honestly cannot say at the moment as I have not price anything up yet. The parts on the board are generally more expensive than my previous designs, but there is also half the amount of parts. So really I'm hoping the price will stay basically the same as it is now.

The only possible exception here, is that I am going to have to get these assembled as I just do not have time to do them myself. I do not think it would add huge costs anyway. I doubt it will be more than £10 price increase, if the price did actually go up. But it is way too early to think about pricing yet, as I am still manufacturing my previous design currently.