Building an Unregulated Linear Power Supply

[oneguy]

I am going to avoid shielding testing. Just posting data for the curious. I already have enough audio based neuroticisms. 

[Gavin1977]

15 hours ago, oneguy said:

For those curious about the amount of noise put into the line by well made audio components, take this as a sample point:

 

Everything unplugged from wall in bedroom (measured at wall):

 

All network gear plugged in (measured at power strip). UniFi USG, UniFi Switch 8, modem, Stealth 8 speakers:

Everything above plugged in but all audio gear plugged into isolation transformer turned off (measured at isolation transformer power strip):

 

All audio gear turned on (measured at isolation transformer power strip). Yggy LiM, 2x Hypsos, Mercury Streamer, ULPS V3:

 

Sam condition as above except with a 200VA load on the ULPS (fluctuating between 9 and 10mV):

 

I plan to post some measurements later related to the ULPS from my EMF meter in case the group is curious. Just need to figure out all the sceneries I want to test. Will also test after the 220,000uF cap and after the Toroidy transformer. 

 

…Don’t sleep on your server 

Great job on this - apparently 5mV is the level to aim for to achieve a black background for amplifiers, so looks like you’re right on the money.  No point chasing less than 5mV apparently (obviously source components, preamps etc… need to be really quiet because their signal gets multiplied by the amp, so quieter the better).

[Gavin1977]

It's live:  

 

No fancy pictures on this one, as I tried to right it up quickly before I go on holiday!

[Exocer]

13 hours ago, Gavin1977 said:

It's live:  

 

No fancy pictures on this one, as I tried to right it up quickly before I go on holiday!

Cool write up! Thanks

 

Brief update: 

 

My amp ended up in the repair shop after all, but, even with my temporary amp the improvements going from Saligny standard to Taiko rectifier were worthwhile. My immediate impression is that the sound is bigger and more transparent. Anyone else get to install it and listen yet?

 

Cheers!

-Rob

[MarcelNL]

game on; the Taiko rectifier and M-AG and HC caps arrived :-)

 

I'm considering aborting my vacation so I'm able to keep the server on 24/7 to get through burn in asap.  ;-)

[Gavin1977]

Price reduced if anyone is interested:

 

 

[MarcelNL]

update; it's far too hot to do anything requiring brain power so I postponed installation until after the vacation...even shut down the server in an attempt to not heat up the house any further, sigh.....

[bit01]

I agree. The Power Saligny is a better fit for the ULPS feeding the Taiko DC DC-ATX than the Standard Saligny when powering my server i9-10900K/Aorus Master MB. The result is a bigger fuller yet relaxed music experience with seemingly better frequency balance. The Standard seemed to tilt toward bass heavy and thinner up top. A worthy upgrade thanks to @MarcelNLfor the shout out.

 

I caution though that the screw terminal standoffs are SMD, simply soldered to the surface and not anchored into the board. One needs to go lightly on the torque.  One of them  (-ve) snapped off on mine when I tightened it  and cracked  the solder joints. I repaired it by re-soldering. One could say I overtorqued or that the solder joints were not adequately done - Either way there is some room for design improvement there IMO. Otherwise thumbs up.

[MarcelNL]

Agree, the power connectors could be better!

[oneguy]

After several iterations I think I’m solid on my next ULPS plans. 

This is inside the standard modushop case that most people buy. The thicker lines are silver plated copper bus bar. Only about 11.25” required. The thinner lines are 2x2mm Angelique. About 6” total required.

 

I think this is just about as short of runs as you can manage thus lowering resistance. Also only 4 ring terminal connectors are need (besides for the bypass caps) lowering the resistance of having a multitude of connections. 
 

Bonus:

I found out that the DCR listed on the Mouser website for the 195J10  is incorrect. Mouser references 0.070ohms. After looking at the spec sheets there were transcription errors with the C30, C50, E100, G30, G75, J10, J30, M20, M30, P20, R10, T10, D25, and K6 here (https://www.mouser.com/datasheet/2/177/195_196-980958.pdf). I notified Hammond and they fixed the table (corrected version: https://www.hammfg.com/electronics/transformers/choke/195-196). 
 

Long story short, the 195J10 has a slightly higher DCR than the 195J12 (0.087ohms vs 0.072ohms) and not the other way around as previously thought. Because of that I ordered 195J12 for the other ULPS I plan on building. For reference, the 159ZJ is most commonly used has a DCR of 0.160ohms. 

 

Same basic layout just slightly different choke dimensions for a J12 based ULPS. 
 

 

I plan on having the J12 version built next month and if anyone wants a J10 version let me know. I have everything on order except the transformer and whatever output connectors you may want. 

[drjimwillie]

I purchased a bunch of parts for the ULPS from an audio friend including the Mundorf HC 22,000 uF cap. 
 

Someone, I think it was Exocer, mentioned that they received the Mundorf AG 22,000 uF cap and they liked the sound of it better than the HC.

 

this had me check the cap I received from my friend to see if it was the HC or the AG. It was the HC but I noticed that the output is 80VDC. 

When I look at the Taiko 220,000 cap I see that it is the HC, but the output is 40VDC. 

The 22,000 cap that comes with the rectifier is the AG but it is also 40VDC. 

My guess is that the two 22,000 caps Needed for the ULPS should match. Both 40VDC And both AG. What do you think?

 

now I see when I go to build the UL PS that I will have another issue, the 220k cap has screwtop and the two 22k Caps have wires. 
How do I put this together?

I know that one of the caps attaches to the rectifier board, the board has two circles on it, can I connect both 22k caps to the board?

[Exocer]

Yes, it was me. One other person here on the forum, @BCRich swapped the first cap to AG. I’ll let him speak for himself 😎. 
 

There is no “voltage must match” rule here. Stick with the 80V HC. I have a 63V 22,000uf AG cap ill try in the near future.

 

The board doesnt appear to be designed to fit two Mundorf AG sized caps. I connect one AG cap to the rectifier PCB, then send the positive output from the rectifier pcb to the 22,000uf HC cap, the negative out to star grounding point (same as all negative poles, they all connect at one location).

 

If you have two HC caps, you will not mount those directly to the pcb.

[oneguy]

If you are referring to the Taiko rectifier board, the C2 spot is for the 22,000uF capacitor. C1 is for the bypass capacitor. 
 

The 80v/40v caps cap be used in the same system. It doesn’t matter. Those ratings aren’t what they output. They are the maximum rated voltage potential that they can handle before breaking down. 

[MarcelNL]

Suggest to read up on PSU components before potentialy creating juicy explosions when inadvertedly getting things wrong! No offense intended!

 

 

[oneguy]

@ExocerWith only 3 nodes (two caps and a rectifier output) plus the power output is star grounding even worth it? To me is seems that adding an extra ring terminal interfaces of a star point not on a capacitor terminal would kind of negate any benefits. 
 

What I view as optimal is taping negative for the power output off of the 22,000 uF which effectively makes the star to node length zero there and the star to rectifier and star to 220,000 cap is just the length of wire their respective wires. 

[MarcelNL]

IME the first cap and bypass is more important for sound/has more effect on 'voicing' than the other ones.

[Exocer]

7 minutes ago, oneguy said:

@ExocerWith only 3 nodes (two caps and a rectifier output) plus the power output is star grounding even worth it? To me is seems that adding an extra ring terminal interfaces of a star point not on a capacitor terminal would kind of negate any benefits. 
 

What I view as optimal is taping negative for the power output off of the 22,000 uF which effectively makes the star to node length zero there and the star to rectifier and star to 220,000 cap is just the length of wire their respective wires. 

I do not know if it's "worth it", but I do know that star grounding where possible has been a worthwhile route to take (I have tried with and without, and preferred with).

 

I also removed a number of lugs going with the PCB mounted AG vs all HC. I never added a ring terminal for star grounding. If I had a choice, I would rather star ground at a sturdy ring terminal than at a PCB (which is inherently wimpier, though, removing terminals has always been a step up in SQ). A balancing act.

 

More experimentation needed to determine the best approach but I am very much thrilled with what I mentioned above.

[oneguy]

This is what I was trying to convey:

 

Both are effectively star grounding but the lower options reduces the amount of rings required by 2. 

[Exocer]

8 minutes ago, oneguy said:

This is what I was trying to convey:

 

Both are effectively star grounding but the lower options reduces the amount of rings required by 2. 

The lower option is what I'm currently doing. I support that approach. I never had an additional grounding point on the chassis for this.

[oneguy]

Dry fit looks good gonna have to jack up the chokes and caps considerably to get them level with the 220,000. 

[oneguy]

Anyone have a source where I can order Panzerholz? I am looking at either that or Permali to go underneath the chokes (36mm thick for the J10) and 22,000 HC cap (29mm thick).

[SK8]

4 hours ago, oneguy said:

Anyone have a source where I can order Panzerholz? I am looking at either that or Permali to go underneath the chokes (36mm thick for the J10) and 22,000 HC cap (29mm thick).

I also tried hard to purchase small pieces of Panzerholz for my DIY chassis but at the end I gave up. I tried contacting a personal seller in the UK and an online shop in Germany. You can find the on-line shop in Germany if you Google "panzerholz zuschnitt". It's very promising, different thicknesses, different sizes available. I went ahead and placed the order. I received the confirmation email (all in German) but no delivery for a long time. I then sent multiple emails to ask for the status politely, in English and in German (translated by google) but never got any reply. At the end, I had to get the refund by escalating to PayPal for issues resolution. If you have friends in Germany who can speak German, perhaps you can try your luck here:

 

https://www.xn--huth-mbel-57a.de/Webshop/-c61535559

 

The tank wood I am using is from the US, the purchase experience was much much better:

https://www.richlite.com/

 

It's kind of wood similar to Panzerholz but is made of epoxy and papers instead of natural hard wood. The density is very high, slightly less than Panzerholz but the wood also sinks in water. It's very hard and heavy and is being used for the fretboard of some guitar brands. Do some research before placing your order. I purchased 2 pieces of 12mm thickness version of their wood in 12"x12" size for US$100:

 

 

https://www.richlite.com/collections/partial-sheets/products/partial-sheets-little-tahoma

 

Hope this helps.

 

[SK8]

This is how I plan to vertically mount my Taiko Noratel transformer to the chassis, 12mm tank wood on 5mm copper plate, no contact between the metals.

 

 

 

 

[MarcelNL]

Why not use Pertinax, or phenolic resin pressed cotton (similar to Pertinax)?

[lwr]

9 hours ago, SK8 said:

This is how I plan to vertically mount my Taiko Noratel transformer to the chassis, 12mm tank wood on 5mm copper plate, no contact between the metals.

 

 

 

 

Beautifully done.  Thanks for the link: https://www.richlite.com/collections/partial-sheets/products/partial-sheets-little-tahoma.  Order placed.  Would you please provide links for the 5mm copper plate, and for countersinkable screws and bolts that you used.

 

Thanks