Building an Unregulated Linear Power Supply

[MarcelNL]

57 minutes ago, oneguy said:

All VCap leads and cases show as open loads. 
 

Maybe an unnoticed pin hole in the cap heatshrink got to the party started then from there it cascaded? My mind is still boggled. I am going to put a layer of silicone conformal coating on the cap in the spot where the case is exposed then a bit of electrical tape on top of that when it dries. 

sounds likely IF it was a short between the cap case and the rectifier...hard to say without sacrificing a rectifyer and they're too expensive for that.

 

Think you should be good to go just sticking the rectifyer in the air or as you suggest by adding some insulation.

[oneguy]

The new rectifier came in last Tuesday and I got around to putting it in today. 0.08Amp with a 1k ohm 2w resistor for the load and 0.5 amp fuse for protection. 103.7 line voltage. 
 

Currently running it at idle for the next 24 hours to check rectifier temps then upsample if everything is good. Highest amp draw during boot was 2.11 amps. 2 amp fuse currently installed. At idle I’m seeing 0.65-0.79 amps. Rectified voltage measuring 27.86vdc. 

 

All amp measurements done with a kill-a-watt in the primary side. 

[elan120]

4 hours ago, oneguy said:

The new rectifier came in last Tuesday and I got around to putting it in today. 0.08Amp with a 1k ohm 2w resistor for the load and 0.5 amp fuse for protection. 103.7 line voltage. 
 

Currently running it at idle for the next 24 hours to check rectifier temps then upsample if everything is good. Highest amp draw during boot was 2.11 amps. 2 amp fuse currently installed. At idle I’m seeing 0.65-0.79 amps. Rectified voltage measuring 27.86vdc. 

Have you measure the secondary output voltage from your transformer?  You can use it to check what is the voltage drop across the rectifier.

 

As a reference, my system bootup peak power draw measured between the output of power supply and the server is around 175W, and with BIOS set at default CPU frequency with both Turbo and Hyperthreading enabled, idle is around 45W, running HQPlayer with no upsampling is around 65W, and running HQPlayer ASDM7ECv2, Poly-sinc-ext3 from 44.1K to DSD256 is around 115W ~ 120W.

 

[oneguy]

3 hours ago, elan120 said:

Have you measure the secondary output voltage from your transformer?  You can use it to check what is the voltage drop across the rectifier.

 

As a reference, my system bootup peak power draw measured between the output of power supply and the server is around 175W, and with BIOS set at default CPU frequency with both Turbo and Hyperthreading enabled, idle is around 45W, running HQPlayer with no upsampling is around 65W, and running HQPlayer ASDM7ECv2, Poly-sinc-ext3 from 44.1K to DSD256 is around 115W ~ 120W.

 

21.1vac rectified to 28.4vdc is what I just measured with the server still running. 29.8vdc would be the perfect rectified voltage. I’ll check to see what the numbers are with the server off after the 24 hour run at idle. The kill-a-watt seems to have settled out at 0.64 amps being pulled from the wall.
 

The rectifier was basically at room temperature when I touched it and it’s been running for 7 hours so far with the HDPLEX case open. 

[oneguy]

HQPlayer ASDM7ECv2, poly-sinc-xtr-no, 44.1K to DSD128 results in 1.94A from the wall and a 200va draw and 23.2-23.4vdc. That should mean about 8.5-8.6 amps going to the DC-ATX. 192va is the worst case rating for the Taiko Transformer. 

DSD256:

235va 2.28a 22-22.3vdc

 

DSD512:

310va 2.96A 21.25vdc

 

All VA and A measurements were done at the wall and all VDC measurements done at the capacitors.

 

After only maybe 20-30 sec the rectifier was hot enough for me to only be able to keep my knuckle on it for a moment. I think I left my non-contact in the US so I don’t have any actual numbers for temps. But know that fact I could see where last time I used it and started off on DSD512 it probably melted the heatshrink on the capacitor case. 
 

I fired it up for one more time at DSD512 and vdc dropped down to 18.8 which is nuts after about 10sec. I’m not sure what prompted that. The rectifier got hot quickly!

 

Thoughts?
 

I’m obviously exceeding the current rating of the chokes by a fair margin. When I get back to US voltages it should give me about 30% more voltages which will drop the amps going through the chokes. Replacing the Taiko transformer with something that can output a higher voltage will also help with the choke current. 

 

 

[MarcelNL]

seems to me you are exceeding the current rating of either the transformer or the rectifier, or both ...the rectifier should be able to cope with 20A...

I'd undo everything and start a fresh....measuring voltages at every step, primary transformer, secundary transformer, voltage from the rectifier etc etc....

 

What transformer are you using, not a center tap version by any chance?

Saligny Specifications:

• Operates from DC to 600Hz.
• AC operating voltage 6Vac to 51Vac.
• DC operating voltage 9Vdc to 72Vdc.
• Iq = 1,5mA.
• Continuous load current up to 20A or more depending on mosfet configuration.
• Over 200A pulsed current at Ta = 25Celsius ( Max Rθjc = 1.0C/W, pulse duration ≤100 μs, duty cycle ≤1%).
• Require a minimum output capacitor of 1000μF
• Do not support centre tapped transformer.

[oneguy]

Transformer amps is only 3 per the kill-a-watt

Rectifier can handle 20. Even when it dipped to 18v that should only have been 17A. 
 

All my voltages are posted above. Primary transformer is 103.7vac secondary at idle is about 21 vac, rectified about 28vdc. 
 

I am using the Taiko transformer. 

[oneguy]

Transformer is rated at 192va but can handle 400va according to Emile. Highest load I put on it was 310. 
 

I can’t see a single figure I’ve exceeded except for the chokes.

 

I have no idea if the Taiko is center tapped but others are using it with this rectifier so I would assume it’s good to go. 

[Rovo]

Hello oneguy, 

 

some weeks ago I tried to power my HQPlayer upsampling PC (i9-10900K plus RTX A4000 16GB) with a HDPlex 500W ATX LPS. It got hot in a matter of minutes and the power from the wall did rise from 200W to 400W. The "load" on the HDPlex was too high. 

 

I am now using a Toriody 400VA 230V/24V transformer, Saligny Power rectifier, Hammond 195J10 chokes and Mundorf Mylitic HC capacitors (20.000uF-20.000uF-210000uF). Power from the wall max. approximately 200W (upsampling to DSD512). After some running hours I can detect no temperature problems. 

 

I have the impression that you have to be careful not to overload electronic components, otherwise strange things might happen. 

[oneguy]

@RovoI have 195J10 to swap in but the other one I had arrived broken. I ordered them thinking I was going to have to swap them in and seeing as how that’s the only component that I can can identify as exceeding l think that’s the most likely way forward. 
 

I will also have to add some capacitance to throw them in. 

[MarcelNL]

my suspicion is that the inductor is saturating effectively making it a resistor, then overloading the rectifier...you are way over its power rating and the specs do not provide a saturation current, yet I suspect you are well over it's saturation current at high power computing.

(I never understood the choice for this specific TINY Hammond choke in this application to be honest)

 

[oneguy]

Makes sense. 
 

@elan120whats the VA rating on your transformer?

[bit01]

7 hours ago, oneguy said:

 

I’m obviously exceeding the current rating of the chokes by a fair margin. When I get back to US voltages it should give me about 30% more voltages which will drop the amps going through the chokes. Replacing the Taiko transformer with something that can output a higher voltage will also help with the choke current. 

 

 

Right!

[elan120]

5 hours ago, oneguy said:

@elan120whats the VA rating on your transformer?

I am using Toroidy Audio Supreme 400VA transformer.

 

 

8 hours ago, oneguy said:

DSD256:

235va 2.28a 22-22.3vdc

 

DSD512:

310va 2.96A 21.25vdc

 

All VA and A measurements were done at the wall and all VDC measurements done at the capacitors.

 

After only maybe 20-30 sec the rectifier was hot enough for me to only be able to keep my knuckle on it for a moment. I think I left my non-contact in the US so I don’t have any actual numbers for temps. But know that fact I could see where last time I used it and started off on DSD512 it probably melted the heatshrink on the capacitor case. 
 

I fired it up for one more time at DSD512 and vdc dropped down to 18.8 which is nuts after about 10sec. I’m not sure what prompted that. The rectifier got hot quickly!

 

8 hours ago, oneguy said:

I’m obviously exceeding the current rating of the chokes by a fair margin. When I get back to US voltages it should give me about 30% more voltages which will drop the amps going through the chokes. Replacing the Taiko transformer with something that can output a higher voltage will also help with the choke current. 

 

When I run this setup using ASDM7ECv2, Poly-sinc-ext3 from 44.1K to DSD256, the power draw is around 120W, and will have peak up to 125W.  The voltage output from power supply dropped to around 29.5VDC, with peak current around 4.2A.  If the same peak output power of 125W is applied to your voltage reading of 22VDC, the peak current will be 5.68A.  I did a short run at DSD512 using non-EC modulator, and the power draw from power supply is around 140W, using your output voltage reading at 21.25VDC and 18.8VDC, the current is 6.58A and 7.44A, and your point about exceeding the choke current rating of 5A is correct.

[elan120]

8 hours ago, Rovo said:

I am now using a Toriody 400VA 230V/24V transformer, Saligny Power rectifier, Hammond 195J10 chokes and Mundorf Mylitic HC capacitors (20.000uF-20.000uF-210000uF).

@Rovo, Could you detail your configuration (the xxx part below)?

 

Toroidy 400VA > Rectifier > Mundorf (xxx uF) > 195J10 > Mundorf (xxx uF) > 195J10 > Mundorf (xxx uF)

[MarcelNL]

@elan120 I see him write ; (20.000uF-20.000uF-210000uF

[Rovo]

small correction: 22000uF - 22000uF - 210000uF

The 210000uF is made up of 22000uF plus 4 times 47000uF

[Rovo]

With respect to temperatures for 200W from the wall: 

- the copper plate of the Taiko DC DC-ATX is getting warm to hot (you can hold you finger on the copper plate but after some 10 seconds you are glad you can pull your finger away)

- the Toriody is getting warm to hot to the touch (also here after holding your hand on the transformer after 10 seconds you want to pull your hand away although less than for the Taiko)

- for the other components there is no significant temperature increase

I would guess the temperature for the Taiko and tranformer are 40 to 50 degrees Celsius for a room temperature of 20 degrees Celsius

[AngeloVRA]

On 5/15/2022 at 8:29 AM, oneguy said:

Aftermath: 2 chips removed and 1 displaced. 
As you can see from the underside of the chip (they are all the same part number) there is one large pad connecting 4 of the pins and and the other 4 pins connect to their own individual pads. 
 

There was enough solder present to reflow one of the them back but I need to get some solder paste for the other 2. 

 

 

Hi @oneguy,

 

Since the rectifier board was face down, its quite possible that it was pressed against something which shorted the exposed/protruding ends of the SMD capacitor C1 on the board. That capacitor is connected to V- and V+. The output overload could then have quickly heated up the other components resulting in failure.

 

Unsolicited but I'd recommend you find some way to fix/secure the rectifier (e.g., on a PCB mounted on brass standoffs) even if it results in slightly longer wires, etc.

 

Good Luck!!

 

 

[oneguy]

1 hour ago, AngeloVRA said:

Hi @oneguy,

 

Since the rectifier board was face down, its quite possible that it was pressed against something which shorted the exposed/protruding ends of the SMD capacitor C1 on the board. That capacitor is connected to V- and V+. The output overload could then have quickly heated up the other components resulting in failure.

 

Unsolicited but I'd recommend you find some way to fix/secure the rectifier (e.g., on a PCB mounted on brass standoffs) even if it results in slightly longer wires, etc.

 

Good Luck!!

 

 

That may have happened after the superheating of the board but I don’t think it happened before it.  The metal of the capacitors is the only conductive material on that side of the rectifier and to get to that the heatsink would have had to be melted by the superheated rectifier first. 

[BCRich]

Hi everyone,

I am about to re-wire my V3 ULPS with the Mundorf Angelique Wire and the Noratel Transformer. I am aware that the Soft Start is not needed with the Noratel. My question involves the Neg Poles on the Caps. My first build everything flowed in a straight line on the Pos side; last cap to + output. For the Neg side I picked a Cap in the middle for the star connection; - Output to connector from there. My thinking was less wasted wire. Obviously the runs to the output connector are different lengths. Does it matter where the Star is formed? I noticed in Elan’s build that he has two paths to the star and both outputs to the connector are from the last cap. Does this factor into Sonic Character? If so to what extent? I honestly don’t have the time to try every scenario so feedback would be appreciated. Here is my current layout…..

Thanks…..Mike

[MarcelNL]

IME keeping the wire stretches from the C to the choke short is more important, so is using ONE negative point. WHich point you best use for negative is subject to some testing...it differs across schematics IME.

[BCRich]

13 minutes ago, MarcelNL said:

IME keeping the wire stretches from the C to the choke short is more important, so is using ONE negative point. WHich point you best use for negative is subject to some testing...it differs across schematics IME.

Understood….I actually misinterpreted a suggestion which resulted in the location of my chokes. I was advised to keep them perpendicular to each other and decided to just keep them away from each other. I intend on correcting that as well and position them similar to Elan’s layout.

Thanks….Mike

 

[Dev]

2 hours ago, BCRich said:

Hi everyone,

I am about to re-wire my V3 ULPS with the Mundorf Angelique Wire and the Noratel Transformer. I am aware that the Soft Start is not needed with the Noratel. My question involves the Neg Poles on the Caps. My first build everything flowed in a straight line on the Pos side; last cap to + output. For the Neg side I picked a Cap in the middle for the star connection; - Output to connector from there. My thinking was less wasted wire. Obviously the runs to the output connector are different lengths. Does it matter where the Star is formed? I noticed in Elan’s build that he has two paths to the star and both outputs to the connector are from the last cap. Does this factor into Sonic Character? If so to what extent? I honestly don’t have the time to try every scenario so feedback would be appreciated. Here is my current layout…..

Thanks…..Mike

 

 

are those foils pasted on the side purely for vibration damping of the chassis or for other reasons ?

 

[BCRich]

2 minutes ago, Dev said:

 

 

are those foils pasted on the side purely for vibration damping of the chassis or for other reasons ?

 

It is a Dynamat type material. Definitely has damping properties and I would imagine the foil is containing some of the irritants as well.