tofro wrote: ↑Sat Feb 24, 2024 10:29 am
The switching regulators apparently tend to create initial spikes on their power output that seems to disturb the QL's power-on-reset circuitry.
Current spikes at the input are typical, unless there is a soft start feature.
But a severe voltage spike at the output of a switching regulator means a mistake in the circuitry. Could be insufficient of badly placed capacitors.
With proper capacitors according to data sheet, I never saw any problems replacing 7805 with switching regulators.
My Murata 150115 regulator has been in my main QL (German model) for 5 years and has been fine, including for the use-case of dumping hundreds of original mdv carts. I haven't actually used it much recently with GC or Trump cards, but it certainly had no issue with those in the past.
I wonder if there are fake regulators that simply aren't as good as real ones from reputable manufacturers.
https://QXL.WIN
Collector of QL related computers, accessories and QL games/software.
Ask me about felt pads - I can cut them to size and they have proved excellent for mdv data recovery.
The only issue I could see is that some of the older switching regulator circuits operate in the lower frequencies and might produce radiated interference that could affect the microdrives, but more modern designs use very high KHz or even MHz range and so this is less likely going to be an issue.
As Peter has already pointed out - the output should be free of glitches and current spikes, as long as the design notes are followed, although I would say that plugging a regulator module in place of the 78S05 / 7805 regulator chip - you need to be aware you have about 2-4 CM's of lead which are effectively very small value inductances - and it's not a bad idea to put small value suppression capacitors across input to ground and output to ground adjacent to the regulator itself - somewhere between 10n-47n should be enough.
Some of the replacement modulator boards that have been made had quite a significant inductor on board, which can radiate - and they are not screened or canned, there have been reports of interference from those.
Having said that - I've used Tetroid regulator and the original Qpower (with a screen over MDV2 ULA) without issue, and the new Tetroid design PCB for QL has 2 switch mode regulator modules on board - and had no issues with these at all.
tofro wrote: ↑Mon Feb 26, 2024 8:44 am
Well, I think the answer to the OP's question is as simple as pragmatic: If it works, use it, if it doesn't, don't
7805 is a very inefficient analog regulator and that inefficiency comes out as heat. Often 60-70% efficient, depending on the input/output voltage.
Buck regulators are switching 'digital' regulators [1] that make a lot of switching noise but are very efficient - in the order of 92%+
The QL has adequate supply side capacitors for a 7805 but not for a buck supply. It isn't a matter of amount of capacitance, but more the frequencies that capacitor can filter and those frequencies it passes through back to the QL. That 9V pre-regulated rail runs along the back of the QL motherboard to the expansion connector, and from there into any expansion, which will have its own voltage regulator to make the locally needed 5V. So Gold cards do not run off the 7805 at all - that 1.5A rated part in the original QLs was just for the mainboard itself.
Now that has been explained, using a 92% efficient regulator instead of a 60% regulator means that what was a *for example* 800mA constant load has been reduced to a 550mA constant load at the voltage regulator. This means the power supply is more lightly loaded, and more of that 9V feed is available for expansions. People would be surprised how little the current draw of a running unexpanded QL is! Even the microdrives have their own 7805s. Replacing those with a more efficient modern counterpart with excellent filtering makes a huge difference to the system load.
In terms of interference to the microdrives, this has always come primarily from the video system. The area around the MC1337P is a particular problem for the microdrives. If you're not using composite or UHF, the 1377 can simply be removed. Removing R48, R49, R50 and R51 de-powers the entire composite/UHF subsystem and reduces total QL EMI by over 1/3rd.
How can this be practically applied? If you're using a modern buck regulator "switching" supply....
Solder a 4.7uF 25V tantalum capacitor across pins 1-2 with the positive end on pin 1.
Solder a 10uF 10V or 15V tantalum capacitor across pins 2-3 with the positive end on pin 3.
The positive end is usually indicated with a line or a +
A large electrolytic capacitor is just too fat and slow to help with filtering these types of noise. The smaller the cap, the higher the frequency it will filter.
