For the C1 power tank and C2 Vref filter you may use whatever you like but for C3 there are rules else you will make it oscillate. Read the 100kHz ESR and capacitance range notes on the positive schematic. There is an impedance at 100kHz column in some high spec electrolytic types datasheets like the Pana FM etc. Those figures are usually reliable. I found them close when testing at 100kHz with the LCR Meter. So you don't strictly have to verify but simply choose from the datasheet. When the 100kHz permitted impedance range is met (0.12R-0.22R) choose the lower uF value model that falls in range. No more than 100uF though regarding its capacitance. There are also alternative lead spacing pads for C2 (7.5mm 5mm) & C3 (5mm 3.5mm). Unspecified for 100kHz impedance C3 capacitor types must be measured on 100kHz LCR for ESR first.

In the previous post's photo you can see it with EPCOS power tanks (4700uF/35V) and Panasonic FM C2 (470uF/25V) C3 (100uF 35V). Also in post's #14 second photo. C2 can be up to 16mm diameter and C3 up to 10mm diameter and there is some extra space around them also. Placed in a way that if you will mount the M2 MOSFET vertically they will not obstruct a screwdriver to fix it.

Nice for low profile in height caps combo. Robust for working hours spec also. I listened to it in the Soekris. Very nice. But when with Nichicon Gold Tune tanks and Nichicon ES (BP) non polars in later test it was simply better resolving and more melodic. Given the fact that the UltraBiB needs consume little spare current so it does not need (not even likes) Hot-Rod current setting, the M2's dissipation is low enough not to create high heat in C2 C3 vicinity. 85C "for audio" caps will also have a good life there. In my opinion the Nichicon ES (BP) Muse is a great affordable line of "audio" capacitors. Dead on for values and with very low DF & ESR in test. In first post's photo you can also see the Gold Tune but with C2 C3 Nichicon KZ Muse.

Yeap, I was one of the lucky too!
I had one symmetrical of pre-production pcb, that build it with no any special capacitors or resistors.
I kept exactly the ESR requirements for the C2, C3 capacitors but not boutique at these positions.

Attachment 679584

I done some measurements on oscilloscope with technical load, everything was OK.
Also I done some THD, IMD measurements by FFT and compare the results between UltraBiB 1.3 and BiB1.1 on one output stage of AK-4490dac.

I impressed with the lower 2nd harmonics that UltraBiB had vs BiB1.1 at the 1KHz THD testing.
After, an acoustical test with the same dac following. First the listening was done with BiB1.1 and after the same dac with UltraBiB.
From the first minutes, something has changed, the clarity and definition of music was more and acoustic stage was improved also.

I think, that Salas target the "sweet spot" and I am very sure that this reg will go fine!
Welcome UltraBiB 1.3 to diy community.

Lemon thanks for your Beta Testing report. The reg is actually good with any brand & type of capacitors when they fit spec as you found but I also owed to report what I found subjectively little better or at least I thought of as little better.

George thanks for your Beta report. The version in post one is final. But it will be done in high grade black & gold triplet. Oops there is only the SSLV1.1 in your photo :D

Excellent! When will PCBs be available? I'm not in a hurry, but looking forward to upgrade my Soekris DAC.

Regards

Tea said to me he will manage to squeeze it in the forthcoming GB (imminent). I had sent him the triple sections version* data as earlier on as possible to assemble and Beta test himself.

*Three regs on one board in breakable V scored sections. Positive, negative, positive, arranged. Exactly as in SSLV1.1 BiB. Same dimensions and mounting distances also (mechanical reference is the four green CCS Leds BiB 1.1 iteration board). Same transformer needs. Instant upgrade from BiB to UltraBiB. In case of installed sense leads if not bothered to remove them they can simply go along the main leads in the output connector.

Looks great Salas looking forward for the group buy and I currently use 1.1 on my Soekris, so as per your post just a drop in replacement with 1.3 :)

Regarding the Kelvin wiring on the outputs I see that all the beta testing is done without this, so we can go ahead without the Kelvin on the 1.3. I presume that we still need to use a load resistor of the actual source to set the DC output voltage as we used to do with the 1.1 version.

Thanks

You may use a dummy load resistor to emulate the actual load and the final work heat on M2 for sinking and warm up drifting. But the output power node is already locally closed in the 1.3 so you can pre-set the Vout regardless of dummy loading or not. It takes 15 minutes and about 1-1.5% negative drift to thermally settle. Then on its rock steady especially after half an hour. For 5V to 25V applications only, the trimmer can be 10K giving more precision in that range. When first assembled you may see as much Vout as Vin due to a trimmer addressing the full Vout range (or a large portion of it), is set somewhere in the middle position when delivered from the manufacturer but turn it down and it will soon listen. Don't connect to a real load before setting Vout. Due to its large range of Vout possibilities (more than most bench supplies) you may forget it might start high on first step. Set it alone first.

Thanks Salas, yes thats a good point of not putting a load resistor for the first time setup with the trimmer being at the highest resistance which might have come from the manufacturer and we forgetting to reduce before soldering to the board. Once the voltage is setup with the minimum and then using a load resistor to set the required voltage is perfect.

Oups tomorrow will attach the correct one! Getting old with the eye sight!