Future PSU designs

[exxos]

So does this mean that I have to throw my old Blue one out that I got from you last year?

Of course, its now old and obsolete and red looks cooler than blue

but will need to buy a new one soon to rebuild yet another toy... great work as always...

No progress yet with the new ones.. busy with STE boosters and have 3 Falcons to repair which been stacking up for a while to do next

[dhedberg]

No progress yet with the new ones.. busy with STE boosters and have 3 Falcons to repair which been stacking up for a while to do next

Would you mind documenting the process of repairing the Falcons, by creating 3 blog posts or similar? Would be interested to know their initial state (what is not working), what you look for, why you do what you do, what components you replace, etc. I'd be interested for sure and we'd all be able to learn something from it. Why not record the repairs and put it on youtube?

Sorry for hijacking the thread.

[exxos]

Would you mind documenting the process of repairing the Falcons, by creating 3 blog posts or similar? Would be interested to know their initial state (what is not working), what you look for, why you do what you do, what components you replace, etc. I'd be interested for sure and we'd all be able to learn something from it. Why not record the repairs and put it on youtube?

Problem is time.. I am literally spending 14 hours a day on this stuff and failing badly in keeping up... doing videos and documenting it all would mean I would have to stop working on the PSU's, STE boosters etc to make time for it.. Theres been no progress in the STF boosters for over a month.. I did post a round up of most stuff.. but no time to update it viewtopic.php?f=20&t=696&p=4793 I spent 3 hole days last week just replying to emails and messages

I did post one flacon repair up already though...viewtopic.php?f=23&t=80&p=206#p206

[exxos]

Very small update..

Spent about 2 hours trying to find all the bits for the new board... then realised I don't have a stencil for the new layout yet going to order one this week and try and build up a final board for testing soon.. The previous board is 99% the same viewtopic.php?f=9&t=255&start=40#p6800

Just I added more via's for heat transfer and release mostly. I also changed the 5V resistor divider to normal through-hole parts. They were getting hot being SMT near the IC.. probably wouldn't do any harm, but now I can lift them off the board a fraction to get them away from the hotter pcb areas.

Ultimately I think I will get the PCBs made up for me (again lack of time) .. I did factor this in partly when I designed this new psu, in that the final selling cost shouldn't be any higher than the current ones in my store.. Will just see how the next months pan out with everything

[exxos]

Finally got my new stencil and started building up 1 final board for testing out....

I need to order some more bits to finish it off as either forgot to order them or lost them

I expect these will be on sale towards the end of the year. I'm not building these up myself anymore , they will get shipped off to be built. Though I will do final checks and testing before they make it to my store.

1.jpg (36.89 KiB) Viewed 503 times

[Axnfell]

Really cannot wait, in need of 4...

[exxos]

Looking for cheapest source for parts, I mean just 4 parts and its pushing £1,000

1.jpg (41.85 KiB) Viewed 483 times

2.jpg (20.15 KiB) Viewed 483 times

I was looking towards getting 100 PCBs made.. but I think I will have to order the coils in lower numbers and buy them as the PSU's sell. I mean the transformers cost about £12 each, 100 of those would be £1,200..

So will probably have to do smaller batches for my store, 4 variations (ST/Falc & 230V/110V) and do 5 of each. Then when they sell, buy more transformers and get those assembled.

What concerns me though, is parts go out of stock, and when they do , it can be several months before they come back in stock again. Which is mostly why I started buying stuff in bulk..

This stuff isn't for the faint hearted thats for sure

[exxos]

Finally some testing!

Typically 5V on the ST is around 1.2 - 1.8amps.. Here I am loading 4.5amps.

12V almost nothing uses it anymore, but I am loading with 2amps.

Efficiency isn't great at these high loads on both rails, but approx 88% which is a lot better than my previous design. 88% is still very good anyway.

5V only load on 2.5amps is about 92% efficiency. Lower loads around 1.5A are probably even better efficiency. Maybe around 95%. So its about as good as things get!

( as a side note, my PSU may say 15.1V (well it did when I re-tested) , but actually on the PCB its 14.1V, so if anyone uses the meter readings from below, will get worse figures than I calculated. But these are only quick tests anyway.)

IMG_3284.JPG (75.62 KiB) Viewed 432 times

EDIT:

..and the ripple & noise on the 5V rail under 4.5amps load..

5v______.png (80.84 KiB) Viewed 430 times

Pretty much 20mV p-p ( 0.02V) total ripple and noise (mostly noise) So really pleased about that figure. Actuall ripple is almost nothing, about 2mV (0.002V).

Note - x10 probes used in the above test.


EDIT2:

Efficiency figures are done powered from my bench power supply, so this does not include efficiency loss in the main toroidal transformer (which are generally very high efficiency anyway) and does not include efficiency loss across the bridge rectifier. Though I have used schottky diodes for lowest voltage drop and highest efficiency possible. So the entire design really cannot be made any more efficient than it currently is now anyway.


EDIT3:

The crowbar protection on my old PSU was on the 5V and 12V rails.. On this new PSU it is also on the main DC input from the rectifier (about 15VDC) with a trip of about 20VDC. This was talked about before, but should there be any nasty mains surges which may last a while, or enough to overcharge the main reservoir capacitor, the fuse will blow to protect the 5V and 12V regulator circuits and of course the computer attached.

The trip voltages for 5V and 12V are approximately half a volt higher. So should the 5 V rail go higher than approximately 5.5V, will blow the fuse and cut power. Of course in reality there should be no circumstances where this could happen unless there was huge catastrophic failure like someone drops a screwdriver across the PCB etc!

EDIT 4:

Small change from previous designs was to use normal through-hole resistors for the voltage feedback network. I did this as the PCB can get warm over time and I did not want to risk "cooking" smaller SMT resistors over time. So I used general through-hole resistors on the 5V regulator and will lift them the PCB a couple of millimetres to allow some airflow around them. The problem with resistors and indeed the parts, is as the temperature profile changes, so does the output voltage. So wanted to minimise this drift by keeping the resistors out of the heat path as much as possible.

[mikro]

12V almost nothing uses it anymore, but I am loading with 2amps.

Fun fact: CT2 uses 12V. Why it's fun? Because I could compare both yours and Centuriontech's and well... Centuriontech PSU didn't make it. After an hour of usage or so it simply refuses to work anymore while yours work perfectly.

We're looking into what is causing it (happened twice in a row) but anyway, just wanted to let you know that your former PSU shines even in 12V area.

[exxos]

Fun fact: CT2 uses 12V. Why it's fun? Because I could compare both yours and Centuriontech's and well... Centuriontech PSU didn't make it. After an hour of usage or so it simply refuses to work anymore while yours work perfectly.

We're looking into what is causing it (happened twice in a row) but anyway, just wanted to let you know that your former PSU shines even in 12V area.

Thanks

I wasn't aware (or remember) anything uses 12V like the CT2. I wonder what current it pulls from the 12V rail ?

I don't know what output the Centurio psu has current wise, I half remember seeing it was 2amps on the 5V rail ? I don't remember seeing what the 12V rating was. It is likely simply hasn't got enough amps to power the CT2. The transformer he has on there looks small, does it have any numbers so can identify it ?,does it get hot ? I think the stock Falcon was 1.8amps alone, I can't exactly remember.

My previous power supply design could actually push 6amps on 5V, but the limiting factor really was the heat output. Generally 2-3amps I considered normal in a lot of cases. The first design the 12 regulator couldn't could be screwed down to the metalwork to aid with cooling, so generally was thermally limited to about 1amp. But this was fine in a lot of cases anyway. The later revisions I moved it around so it could be screwed down. I guess technically this could work at a few amps as well. Design is more than capable of running at a few amps on each rail, but the efficiency figures dramatically drop around 3Amps load.I would be surprised if anything used more than that anyway.

My new design has had a total revamp of course, much more efficient design and can handle 5amps on the 5V rail easily. Technically the maximum is 6amps , this is limited by the chip itself. 12V can easily push 2amps as shown in my images above, and is limited by the chip itself to 3amps. Like my previous one, it can deliver a serious amount of amperage. Only the new one is a lot more efficient so they wouldn't get anywhere near as hot at higher amperage levels.