A while ago I decided to upgrade and tune my rather old (30MHz Version 2.9) because I wanted to have a higher precision counter function.
After reading this enormous number of postings, with some very good information (thank you guys), I decided on three items.
The power supply, the oscillator and the driver opamps for the waveforms.
The latter two are easy and have been described various times.
The power supply has been talked about many times, with several options, but there are only a few examples for a replacement.
I decided to design a completely new supply, that would let me adjust the +/- 12V outputs to +/-15V as well.
Going through my parts collection, I picked a 24VA block transformer that I could mount on a protoboard. The transformer is a little heavy with 2 seperate windings of 12VA each, and an output of 15VAC. Both secondary windings are fused with 0.8AT PTC fuses. The primary has two seperate 115V windings.
The secondary winding voltages are specified for full load at the minimum mains voltage input limit so tend to be a little bit higher than you might expect when running on a light load at normal mains voltage. That "15v rms" may be more like 17v rms no load in practice which, after allowing for a generous 2 volt drop in the bridge rectifier and the 1.4 crest factor of an rms voltage for sine waves, would see the smoothing capacitor(s) storing a peak voltage close to the 22 volt mark which will likely drop to an average voltage around the 20 volt mark with an average current loading of 150mA when driving 50 ohm loads with a square wave peak to peak voltage swing of 24 with a 30mA 'vampire load' allowance.
When the regulators are set for 15 volts under this condition, they'll have to dissipate 750mW each. At a 12v setting this increases to 1200mW each which is a significant amount of heat energy to dissipate via the unsinked tab of a TO220 device (free air dissipation for most TO220 packaged devices is typically 1.5 to 2 watts maximum @25 deg C ambient).
That heatsinking you're going to use will be vital in (literally!) this case on account the ambient temperature within the case will be considerably higher than a mere 25 deg C (more like 50 deg but I still don't have a suitable temperature sensor to actually check this - that's just an 'educated' guess right now. It could be even higher in the absence of any active cooling - I have a small 50mm sq by 10mm deep 5 volted 12v fan fitted into mine which has made the two original hot spots on the top of the case disappear without trace.
These dissipation estimates are rather conservative based on the information you gave and the actual, measured, values are unlikely to be less without the effects of drop out from supply ripple making its presence known (at least as far as the 15 volt option goes).
It makes sense to use a switching type drop in substitute for the 7805 regulator (better than 90% efficient units with <30mV ripple at 1MHz or higher being cheaply available). As far as I can see, none of the logic ICs are powered from this 5 volt rail directly which is only used to feed the three on board LDO ICs (3.3, 2.5 and 1.8 volt rails iirc) so any switching ripple here is of no consequence. The other 5 volt rails (both plus and minus) are derived from LDO regulators fed from the +/-12 rails (via 8 volt LDO regulators istr) so are also immune to the ripple on this 5 volt psu rail.
Given a modest airflow around the regulator heatsink (best to use a single large heatsink for both 12/15 volt regulators), this analogue PSU design should be reliable enough. Without the addition of a cooling fan, reliability becomes somewhat questionable (it was already questionable to start with without any modifications being applied).
I decided to use simple LM318 and LM338 voltage regulators for both analog supplies. The only specialty is the circuit around the trimmers, because they are China quality, and therefore cannot really be relied upon. The worst case is when the runner looses contact, creating a much higher output voltage then you intended, and this could potentially blow up the FY6600.
For the digital voltages, I selected a simple DC-DC Buck convertor because doing that with normal regulators would create too much heat. I specifically did not want a fan, with all it's generated noise inside the box. I also used some parts from the old supply, specifically the line filter and the 5V choke.
As I've already hinted (and also in previous postings), I think it's a mistake to discount the cooling fan option. Although removal of the original PSU along with its own 2 watts contribution to the heat loading will compensate for the additional heat produced by a replacement analogue supply, you'll still see a net increase of waste heat raising the temperature within the box which not only isn't good news for the original components it's also counterproductive to keeping those analogue regulators at a safe working temperature even when well heatsinked.
Unless everything else on your bench is passively cooled and totally silent to protect unusually sensitive hearing, I doubt you would find any objection to the barely perceptible whisper quiet noise of a small 5 volted 12v cooling fan which has proved in my case to be more than sufficient to drop the internal temperatures by a good 20 deg C. For all the cooling effect of those vent slots, particularly when propped up on its bail stand, you might as well seal them all up to prevent dust ingress and make it proof against liquid spillage accidents.
My unit is intentionally still floating. This can be fixed easily with a BNC/USB connection to another instrument.
Which begs the question, "Why not fit a low impedance (1K to 10K resistor rather than the conventional 1 or 2M) "Static Drain resistor" between signal ground and the PE tag of a C14 socket?". You can buy filtered, switched and fused C14 connectors which will fit the back panel (along with some strengthening in that area of the flimsy fibreglass board used for the back panel) to replace the original C8 two pole connector (this is an FY6600, right?) which will save you recycling parts from the original psu (keeping it intact for use as a spare or to allow reversion to the generator's original state).
This, for example:-
https://www.ebay.co.uk/itm/Male-C14-IEC-RFI-Filter-Flange-Mount-5-x-20mm-faston-Rated-At-2-4A-250V-ac/223637128377?epid=1905379730&hash=item3411cfc4b9:g:aeAAAOSwE5Vc6tiR
To reduce some heat hot-spots from the three on-board regulators, I added some sticky heat-sinks to them.
Enjoy!
I'm not quite sure what "sticky heat-sinks" are. Hopefully, if they're relying solely on a 'sticky glue' to secure them, the glue used won't be failing at temperatures less than 150 deg C. Also, three regulator heatsinks? The 5v dc-dc converter surely won't be in need of a heatsink (not at half an amp or less) and, in view of the possibility of outputting a 24Vpp square wave at mHz frequencies into a 50 ohm load, it would be better, as I've already mentioned, to mount both 12 or 15 volt regulators on a common larger heatsink to keep each individual regulator cool during the resulting protracted periods (seconds to tens or hundreds of seconds) that they could end up handling each half of such waveforms in turn.
Incidentally, the circuit you've attached seems overly complicated to me, at least compared to a more pragmatic dc-dc converter alternative to the analogue regulator solution you're proposing. I rather think you're "making a rod for your back" with this upgrade project. Assuming it provides the hoped for benefit of eliminated switching noise (with the lid off, of course!), you may yet be forced to change your mind over the wisdom of keeping it "fanless".
In view of the fact that only 5% or less of the input energy appears as useful signal power on the BNC outputs, it's worth taking a note of the mains input energy consumption (wattage) beforehand so as to compare this wattage figure to the new figure after completing your PSU upgrade exercise. The difference will represent the additional heat being dissipated within the box, giving you some indication of the magnitude of the thermal issues this upgrade will present you with.
Most other contributors to this thread who've upgraded to an analogue PSU of their own design have recognised the need for a cooling fan but some have underestimated the cooling requirements, usually recognising just in time the need to fit a fan with one or two seeing the consequences where they've had to replace damaged parts when a regulator IC has failed, input to output short circuit, causing a massive overvolting event. You need to be very mindful of the potential thermal problems with such analogue supply upgrades so do take care.
JBG