Future PSU designs

Repairs & upgrades.
frost
Posts: 6
Joined: Wed Oct 25, 2017 12:53 pm

Re: Future PSU designs

Post by frost » Thu Jun 07, 2018 3:28 pm

Can't wait to have some news :)

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Fri Jun 22, 2018 7:45 pm

I have been wondering about the possibility of main spikes damaging power supply circuits.. So I attempted to do some simulations..

The crappy simulation model has V1 as a simple 15VAC since wave gen (simulate transformer secondary). The actual resistance measured and the transformer is 1 ohms, so added that in, but in reality it is probably higher, but it does not really seem to affect simulation much anyway..

D1 is just a simulate a single diode rectifier, C1 is the main reservoir capacitor with a relatively huge 500mR series resistance. R2 is a dummy load resistor will draw around one amps..

V2 is a spike generator which will over volt the 15VAC to 30VAC. Basically what this means is a 230VAC input to the transformer would have to spike to double that voltage..

cct1.jpg
cct1.jpg (26.16 KiB) Viewed 288 times

The first simulation is as follows..

sim1.jpg
sim1.jpg (56.67 KiB) Viewed 288 times

Blue is the 15VAC sinewave input to the single diode rectifier. The green line the capacitor voltage and the redline spike generator.. Which lasts about 220uS.

Oddly with this setup the spike does not seem to be affected by the huge capacitance at all... But I have actually seen this happen before a normal Atari power supplies where spikes can rise a few bolts over the actual RMS output voltages. Simulation might not be a good one but I think it probably works well enough for investigation.

In the next test I reduce the series resistance of the capacitor to 100mR...
sim2.jpg
sim2.jpg (57.73 KiB) Viewed 288 times
Now the voltage spike is a lot less and only actually around 1V higher now. I made the spike with double , and the voltage only went up a fraction. The next test was to remove the load resistor, but this did not seem to really change anything. I was half expecting the spike to shoot up again but it did not seem to.

With the fact that there is a transformer involved in the real power supply anyway, I would assume a huge spike on the primary would be greatly reduced by the amount of inductance in the core.

I cannot think of any real way to reliably test this, but I was more just wondering if I should add overvoltage protection to the actual input voltage to the circuit.. But I think with quality components anyway, and looking at my crappy simulation, that I think it would be unlikely that a mains spike could damage anything anyway. Of course my simulation is literally driving the transformer at 200% voltage spike, realistically don't think would happen in general use. Even so, sensitive equipment should always be plugged into a filtered mains adapter anyway.

The only other possible reason to add this protection as if someone used a 120V tap transformer and ran it at 230V.. Obviously this would send double voltage through the secondary and likely damage to circuit.. Though so far nobody seems to have been that silly :)
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

stephen_usher
Posts: 96
Joined: Mon Nov 13, 2017 7:19 pm

Re: Future PSU designs

Post by stephen_usher » Fri Jun 22, 2018 10:11 pm

The UPSs at work (Oxford, UK) have seen over-voltages up in the 290v range and brown-outs of about 170v. Oh, and then there are those sub-second almost zero volt to over-volt transients. I'm sure the latter are the killers (especially for UPSs!).

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Fri Jun 22, 2018 10:35 pm

stephen_usher wrote:
Fri Jun 22, 2018 10:11 pm
The UPSs at work (Oxford, UK) have seen over-voltages up in the 290v range and brown-outs of about 170v. Oh, and then there are those sub-second almost zero volt to over-volt transients. I'm sure the latter are the killers (especially for UPSs!).
The voltage here seems to be more like 250V and can jump to 270 around 5PM.. Though its mains spikes more than RMS voltages. Things like lightning strikes on power lines which can spikes mains.. blow up sub-stations and such :twisted:
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Mon Jun 25, 2018 2:48 pm

Third revision pasted up and in the oven...

Mostly I need to work out the overvoltage protection values next...

There may well be 2 more these made up if anyone wants to call dibs on them..

psu.jpg
psu.jpg (96.3 KiB) Viewed 240 times


EDIT:

Onto some testing :)

IMG_2863.JPG
IMG_2863.JPG (148.73 KiB) Viewed 237 times
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Mon Jun 25, 2018 7:47 pm

Decided to build a proper dummy load.. This way I can leave it running for a long time.. I need to either switch on it somewhere as I linked 2x 1R resistors in series for the 5 V rail.. So 2.5amps.. If I bridge over one resistor then I can get 5amps load. Of course this is serious overkill amperage is way more than any ST will ever need.. But really it is just to simulate the power supply running for a long time on the worst-case conditions.. It will also allow me to check the thermal aspects easier..

IMG_2869.JPG
IMG_2869.JPG (119.67 KiB) Viewed 222 times
IMG_2868.JPG
IMG_2868.JPG (93.4 KiB) Viewed 222 times
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Forgottenmyname
Posts: 74
Joined: Fri Mar 23, 2018 8:22 pm

Re: Future PSU designs

Post by Forgottenmyname » Tue Jun 26, 2018 7:56 pm

You just can't beat a good old analogy meter

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Thu Jun 28, 2018 2:57 pm

I tried with a transformer input today to check the O/C voltage.. I don't actually have a 12VAC transformer to hand ,previous transformers were 15VAC, so I used my Variac to down the voltage to test that way.. I have ordered transformers just to try properly..

Though my concern is that the 5V circuit has a maximum of 18VDC input. Under load this would be around 15-16V.. which is fine of course. Though my current test today is a transformer going up to 19.19VDC without a load on the PSU. Basically because the main reservoir capacitor is charging up to the peak AC voltage, which is rather high the maximum input voltage of the 5V driver.

So I will test again with a real transformer, and of course mains voltages and transformer tolerances come into play also.. I will have a overvoltage input protection, this will blow the fuse if the voltage goes to high, but of course this could actually happen if the power supply was not actually powering anything :roll: It's not really a huge issue, as switch modes should generally not be operated without a load anyway.. But it is more of a annoyance than anything.

The only solution here is to find some really rubbish diodes to drop the voltage to the 5V circuit. While this would ordinarily be (and actually is) a efficiency loss due to the voltage drop.. The 5 V circuit will actually run more efficient the lower the voltage goes. I'm not going to get into huge calculations with all this again, but it will likely even out pretty much efficiency wise anyway.

One other thing was bugging me was the track on the bottom of the board which links the power to the 5V circuit.. It actually goes a fair way down the board and actually eats away into the thermal path of the 12V circuit (only on bottom layer, to player has good thermal path anyway) . Of course not much uses 12 volts these days, and it will run easily at a couple of amps anyway. It may actually be beneficial to use some diodes to drop the voltage as it adds a little bit of headroom to the 5 V circuit, and it also basically becomes some jumper links so I can do away with the power rail to the 5V circuit which is a PCB track... One thing I do not really like about using diodes as they will get rather hot.. It's something I was trying to avoid this design..

There are other options to consider yet, I may try a MOV across the reservoir capacitor.. Basically these are voltage ratings where the resistance is very high say 15VDC.. So it does not really do anything about voltage.. But say at 18VDC, the resistance will rapidly decrease and it will act as a loading resistor on the input.. These will also be useful I think for clamping any potential voltage spikes which may come in from the mains or transformer etc. so likely I will towards that as a solution.
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Fri Jun 29, 2018 3:36 pm

I looked into MOV's but the voltage curve simply to slow for what I wanted. I even tried a simple Zener as a clamp, and if the voltage goes to high, Zener overheats and shorts out. It was still a very slow curve anyway . So the only way to do this properly is to have a active voltage clamp, which basically uses a transistor to load the power rail. This way I can accurately set the voltage limit to 20volts. Anything higher, the amps drawn across a transistor rapidly rise and clamp the voltage hard.

In reality the only time the circuit would trigger would be during huge voltage spikes, which I think would be rare, and mostly snubbed by the large reservoir capacitor anyway. Even so, I have added in a clamp to blow the fuse on overvoltage conditions anyway. But as said before, it is possible voltage will float up high enough without a load that it could well blow the fuse. During normal operation when the power supply is powering the ST, this is not a problem at all as the voltage rapidly drops down to around 16volts.

So I really do not know if it is worth adding this circuit on or not ? It is just increasing the cost the final thing, for the sake of not blowing the fuse if someone ran it without a load on it..

I guess a better method would be to use a second rectifier on the board and use half wave rectification of the transformer. In effect to drop the voltage to around 10VDC for the 5V circuit... Though the problem there is space on the PCB and the need for a second reservoir capacitor :roll:

So half wave I forgot it literally is half the wave.. Kind of like how split power rails work.. Not too fond of that idea as it depends on the reservoir capacitor even more..

So I am thinking if I got a 6VAC transformer instead of a 12VAC one, then I could link the secondaries in series and use 2 full wave rectifiers also series. Basically then there is 2 supply rails in series.. I tap of the first one for around 10VDC for the 5V rial, and across both of them for about 20VDC for the 12V rail.. It would need another rectifier, and reservoir capacitor, but these could be smaller ratings the 12 V rail generally does not need many amps.. I need to look into the best solution there.. Mainly the second reservoir capacitor is the issue due to lack of space on the PCB :roll:
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

User avatar
exxos
Site Admin
Posts: 2575
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: Future PSU designs

Post by exxos » Fri Jun 29, 2018 6:07 pm

So near and yet so far... Actually thought I need a second fuse and second crowbar as well :roll:

Technically the 5V rail should be a couple more % efficient running from a lower voltage. I also realised I can re-use the 4700uF's I buy already.. so will actually help me out in not having to buy larger caps!

p1.JPG
p1.JPG (99.51 KiB) Viewed 85 times

EDIT:

I think the only way it's all going to fit is to use a smaller transformer :roll: I just realised I cannot use the top left part of the PCB as the mains wiring is very close to that point and the diodes will probably be in the way as well.

Though I also just realised that 6VAC work out about 8VDC which is a little lower than I would have liked. The next step is 9VAC which works out about 12.5VDC, it is ideal, though with 2 of these in series it gives 25VDC.. Which is actually over the voltage limit of the 12 V regulator. So I wouldn't actually be any better off going with that method :roll: really I need to go with 25VDC caps the physical diameter small enough to fit a couple of them on the board.

The only compromise here is to go with a full wave rectifier for the 5V rail, then go with a halfway for the 12V rail, to trying keep the voltage below its maximum. I will have to wire this up on breadboard first to see how it pans out.. Though I don't really like the idea of half wave either.. So still thinking of the best solution here.. I might have a look around for some other voltages of transformer see if I can find something more fitting..

Overall, I think it would be easier just to stick to the crowbar for overvoltage protection and just let the fuse blow. Though it does run some risk that if someones mains voltage is a lot higher than it should be, it runs the risk of blowing the fuse.. in that case such people should really be using a UPSU to regulate the mains voltage.. Of course this is still only without a load on the PSU.. So I think I may just leave it be, as its going to ramp up costs a fair bit just to prevent the fuse blowing without the PSU being loaded... I mean everyone should know by now swithcmodes don't like being run without a dummy load ;)
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Post Reply