Adventures in low impedance power supplies

Here's my latest crazy project, intended to answer the question 'How low can you go?' in power supply impedance. No listening tests yet, I've just finished assembling this stack of hexagon capacitor arrays out of cheap electrolytics. Total capacitance of the top 4 tiers is comfortably above 1F. The top tier is designed for lowest ESR at HF using Nichicon 1500uFs, the next tier is a trade between ESR and uF. The last two are the highest capacitance density I could find (a total of 182 4700uFs). I've yet to connect the bottom two hexagons to the upper tiers but the top two are connected by wires running through the gaps between caps.

The whole is intended to push the limits of low frequency noise in powering a couple of TDA1545s which will sit on a small PCB supported on thick wires from the centre of the top tier. I am aiming to get a capacitive reactance (at 50Hz) at the DAC supply pins below 3mohm and ESR below 1.5mohm.

Update1 : connected the bottom two tiers together and wired into my existing DAC. Result not promising as the bench supply showed the caps charging slowly and then the process stopped abruptly at 1.9V while still drawing 0.5A (the current limit). I figured I must have connected at least one capacitor in backwards! Took the tiers apart again and inspected them all - nope, nothing wired in backwards.
So I took them over to another separate bench supply and applied volts. Sure enough, even with 3A (the max from that supply) I couldn't get the voltage to budge beyond 2.5V. I was then thinking 'OK, one of these caps is going to be faulty, I will just wait until it identifies itself by exploding'. So I waited, and waited, and waited. No explosion came and no cap was even bulging but in total they were soaking up 7.5W. Given that I figured they couldn't all be faulty I had to come up with another explanation.
I decided to check some of the caps I had left in the bag (of 200). They showed fairly high leakage current (10-20mA) at first which diminished quickly. So I figured there must be some kind of 'forming' process going on and that 3A wasn't enough juice to form all the caps in parallel. So I removed some of the link wires on the hexagon and charged the caps in banks. Bingo, this worked. Eventually I got each of the two whole arrays drawing less than 10mA with 8V bias and left it like that for several minutes to give them a good workout. I share this for anyone crazy enough to try what I'm doing here

I tried out just tacking on the bottom two tiers into my existing DAC arrangement but the impedance was dominated by the interconnecting wires (4mohms in total) so a bit of a waste of all those caps! More to follow when I have a dedicated DAC board to feed from the wedding cake....

Incidentally the bottom two tiers measure -j0.004ohms/50Hz on my LCR meter (which can only read up to 9999uF on its regular capacitance scale but can read XC). So calculating back from this impedance gets to around 800,000uF.

Update2 - pin7 filtering matters! I have a 'postage stamp' DAC sat atop the first two tiers now and figured that pin7 filtering would not matter with such a quiet supply. Wrong! It needs some attention - I tried one of the 4700uF caps that needed forming first. It worked for a few minutes very nicely but then suddenly the volume dropped almost to zero. Checking the cap I found its gone low impedance, so definitely not recommended. Next up - trying a few of the Nichicon 1500uFs.... Oh and yesterday I ordered some different form-factor 4700uFs, this time in 10*30mm cases so in theory I can pack them closer together. The ones I have currently are 12mm diameter. I'll be able to try a 7-on-a-side hexagon with these

Update3 - the new caps need some time to break in. This is shown on the current meter on my bench supply, during the first day or so there are low rate periodic bursts of current beyond what's needed for the two DACs (10mA or so). These have now subsided and the meter's rock steady now. Dynamics are awesome on the right recordings, (Decca in the main) piano has more weight and 'authority'....

Update4 - received the 4700uF/6.3V caps this afternoon. Very impressive measurements for the size and price. They're looking very good in terms of capacitance (many overstate the value by up to 20%) and ESR. A new tier is coming together which might just obviate the need for the Nichicons....

Update5 - built up a 5-a-side tier with the new caps. Measurements indicate XC=10.8mohm @50Hz, ESR 1.2mohm. Pic attached - worth listening to alone I feel as its a much more practical size now Incidentally, the same 'forming' effect is apparent with these caps as the last 'wide-body' ones - unless you have a very beefy power supply its best to form them up a few at a time. I formed this array in two stages - the centre three 'loops' and then the outer two.

Update6 - could not resist enlarging the hexagon with the new caps to 7-a-side. Now it measures -j5mohm @ 50Hz. At first the sound's a little bit 'ragged', cap arrays take a few dozen hours to bed-in. But the Nichicon top tier has been surpassed by the measured performance of these caps in ESR, I've just the one layer now, working out around 600,000uF. We shall see how it develops over the next few days.

Update7 - 10,000uF/10V caps work well (ChengX brand) although don't actually have 10V rating as the leakage current shoots up quickly above 8V. A six-a-side hexagon measures -j4mohm with around 1mohm ESR @ 50Hz. They didn't show any 'sickyness' on powering up the whole array (91 caps in parallel). At the current Taobao price of 0.46 each, with these caps a 1F array can be built for around 55RMB (under $10).

Update8 - I just found a kindred spirit in power supply thinking, here : https://www.ultraaudio.com/features/2008_07_01.htm

Excerpt : As of today, I have not found the limit of the capacitors I can install in a power amp or preamp without hearing additional improvement.

Attached Thumbnails