Popular Post elan120 Posted May 24, 2020 Popular Post Share Posted May 24, 2020 Disclaimer: This power supply build is one of many different possibilities and options available that can fulfil the need for a high performance linear power supply. Do search around to see what choice best fits your requirements and priorities before taking this route. Same also applies to the build method and process as there are many different options available, where this build is based on either my current knowledge or tools available. Everything you see in this thread is purchased at regular price as I am not sponsored or affiliated with any suppliers, where the intent is purely knowledge sharing, so hopefully you enjoy the read. The journey for this build started last year as the new HQPlayer EC modulators took SQ of my system up a good margin, but since my server at the time can only push up to DSD128 with this new modulator, I started to plan for the current server where linear power supplies was in scope of this server build. The challenge I soon faced was the higher power supply output current demand from the CPU, an i9-9900KS, which made the selection of capable power supplies on the market very limited. I went ahead and completed the server build using a standard ATX power supply to get the server running while continuing looking for power supply solutions. The first attempt to improve the power supply was trying Uptone JS-2 to power the CPU directly, unfortunately, it doesn’t have enough current output to satisfy the demand during bootup sequence. Next option is using a HDPlex 400W AC-DC adapter with 19V output connected to a HDPlex 800W DC-ATX converter to supply power to the CPU while keeping the ATX power supply on ATX connection. This combination worked and the SQ is noticeably better than the standard ATX power supply. With this setup running, I was able to estimate the current requirement for this CPU, which prompt the succeeding step by replacing the HDPlex 400W AC-DC adapter with a Keces P8 running at 20V, again, only to power the CPU, and the SQ continue to trend up, so finally, I added another Keces P8 to HDPlex 800W DC-ATX converter to supply power to ATX connector, and again, moved the SQ up another notch. After more time looking over limited options to power this server, I decided to source a few key components from Sean Jacobs (SJ going forward) with the plan for building two separate power supplies, hoping they will successfully provide quality power to this server. This first build is the single rail higher current EPS power supply, once this is completed, I will move on to build the ATX power supply that will have 4 rails, 3 of the 4 rails will be used for ATX and the 4th rail with a 5V output is intended for the JCAT Net Card Femto. In this EPS power supply build, due to the higher output current, there is going to be a high current path and a low current path, and point to point soldering will be used for the pre-filter section. To begin the build, there is some basic metal work to get all the components fitted in various locations. The first thing is getting the layout put together for all the key components and I used a basic CAD program to get the final layout decided before ordering the enclosure from Modushop. The heaviest component is a 400VA transformer from Toroidy, with weight close to 9 pounds, it is essential to have the enclosure bottom cover reinforced. In the pictures below, it shows a couple L-brackets were added to strengthen the bottom cover so it won’t bow when this transformer is installed. In addition, instead of using only two self-tapping metal screws per side of the bottom cover, one additional screw is added in the center, and replaced the self-tapping metal screw by tapping all screw holes to accept M4 button head screws and finished by using M4 nuts to help support the overall weight. Key component layout: Aluminum L-brackets added with standoffs to support transformer base plate: The 4 phillips screws were later replaced with the same button head cap screws used for the brackets: Test fit the transformer base plate: Once the transformer support was done, 4 additional holes were drilled to mount the two Mundorf MLytic HC capacitors. With the bottom cover reinforcement and pre-filter capacitor mounting done, time to get the back panel prepared for all the components. There will be a total of 3 openings, one for the switch/fuse/IEC module, one for the DC output, and one for the output fuse holder. In addition, two M3 screw holes were drilled and tapped to mount the switch module. Next is having the two heat sinks on the side prepared. The left heat sink panel has two M4 holes drilled and tapped to mount an IXYS VBE 55-06NO7 FRED bridge rectifier, two M3 holes to mount a SJ current booster. The right heat sink panel has two M3 holes to mount SJ regulator module and two more M3 holes to mount the other SJ current booster. In order to preserve the appearance of both heat sinks, these holes were drilled and tapped without protrude panels and simply using a heat shrink tubing helped to determine needed drill depth. Heat shrink tubing added to the drill bit: Drilled hole with proper depth: Test fit both bridge rectifier and SJ current booster module: The front panel is the last panel that needs work, where two M3 holes were drilled and tapped to mount the 25W resistor that will be used in the pre-filter section. All panels are done: More to come...stay tuned... Gavin1977, Nenon, Exocer and 13 others 3 11 2 Link to comment
elan120 Posted May 24, 2020 Author Share Posted May 24, 2020 11 hours ago, Gavin1977 said: would you be able to supply dimentions for me including the baseplate The transformer I use has the following dimension: Transformer diameter: ~154mm Height~65mm + 25mm (side-wiring) Base plate is 150mm x 150mm 11 hours ago, Gavin1977 said: I was wondering if feeding the JCAT card using bus-power, since this is from a SJ linear rail, will be pretty close to dedicated rail? Did you ever recieve any feedback from SJ on this, or did you always intend to provide a dedicated rail. I have always had better result using a dedicated rail for all the devices and JCAT card is no different, this is the reason I plan to have a dedicated rail just for the JCAT card. Building this separate rail will not add that much more work, once I am done with this build, I will post steps I take to build the next 4-rail supply, and you will see it is less involved than this build. Gavin1977 1 Link to comment
elan120 Posted May 24, 2020 Author Share Posted May 24, 2020 9 hours ago, RickyV said: Do you have a schematic drawing of the components layout? And what does the SJ current booster do? This might be of interest too, https://evotronix.eu/main/ I bought 4 LC modules for the two nuc power supplies I am building. Thank you for the link. Looks like they have good product offerings. I did put together a schematic and ran simulations for the pre-filter section before deciding what I will incorporate in this build, but since it may or may not fit other supply requirements, I will think about whether to post it here later. SJ current booster is a module Sean provide to help raise the output current from the base level at 1.5A to 5A. This supply will have two, wired in parallel to raise the output current up to 10A. The output Sean provide is quite conservative, actual maximum output can be higher, but other considerations will need to be planned more carefully, such as heat dissipation, transformer rating...etc. Link to comment
Popular Post elan120 Posted May 24, 2020 Author Popular Post Share Posted May 24, 2020 With all the metal work completed and components mounted, time to wire everything together. Due to the weight of the transformer, the transformer will be installed last and wiring will be done at the end. To start, a 6 gauge copper wire was used to ensure there is enough cross section for the ground path between the two pre-filter capacitors negative terminators. Next is to solder using Neotech 12 gauge solid core copper wire as high current wire from the bridge rectifier to the first pre-filter capacitor. The same Neotech wire was also used to connect between the voltage inputs from both SJ current booster modules and the second pre-filter capacitor. Solder wire between bridge rectifier negative output and first pre-filter negative terminal: Solder wire between bridge rectifier positive output and first pre-filter positive terminal: Solder wire between second pre-filter positive terminal and both SJ current booster voltage inputs: Next up is the wirings between SJ Voltage Regulator and SJ Current Boosters. RickyV, StreamFidelity, motberg and 4 others 2 5 Link to comment
Popular Post elan120 Posted May 25, 2020 Author Popular Post Share Posted May 25, 2020 36 minutes ago, Exocer said: what about a dedicated 19v rail into something like an HDPLEX 400W/800W DC-ATX converter directly to an expansion card? Based on my experience thus far, having dedicated rail work better, and this is the key reason why for this and next power supply build, so I can remove the two LPS's feeding the HDPlex 800W DC-ATX converter. ASRMichael and Exocer 2 Link to comment
elan120 Posted May 25, 2020 Author Share Posted May 25, 2020 21 minutes ago, Nenon said: Don't forget the isolation pads on the regulators. Very good eyes Nenon and thanks for the reminder. I only put them on the heat sink to see how they fit and plan for the wiring so didn't bother with the mica pads, but when I am ready to wire them, they will for sure have the mica pads added. Link to comment
elan120 Posted May 25, 2020 Author Share Posted May 25, 2020 6 hours ago, RickyV said: Ah you have the SJ regulators too, that why I was confused about the booster modul. Does SJ sell the regulators now to anyone? And may I ask what they cost? Yes, these are the key components I sourced from Sean, but the cost question is best answered by Sean. He has different options available for these modules, and I was quoted with a package total of 5 rails with specific options, so I really don't know what the individual price was. RickyV 1 Link to comment
elan120 Posted May 25, 2020 Author Share Posted May 25, 2020 46 minutes ago, basillus said: What is the reason that you use an 80v electrolyte that fills a lot They are indeed bulky, but they were designed specifically for high current application that fits this power supply requirement well. In addition, they have very low ESR rating which is also another important criteria I was looking for. Finally, as a side benefit, it comes with chassis mount configuration, which is easier to have them installed and wired. The need for an 80V rating is based on the power supply design topology where the transformer secondary output is rated at 18VAC, this output will rectify to about 25.5VDC, with +/- 5% of local power fluctuation, it could go as high as 26.7VDC, so higher voltage rating on these capacitors is essential for this power supply. beautiful music 1 Link to comment
Popular Post elan120 Posted May 25, 2020 Author Popular Post Share Posted May 25, 2020 All wires used in the SJ voltage regulator module are low current path, and Mundorf 18 AWG solid core silver are being used here. To connect between SJ voltage regulator and SJ current booster is very straightforward, as all the ports are clearly labeled to help make sure where the wires should be soldered. The ground wire from SJ voltage regulator is routed to the 6 AWG ground bus between the two Mundorf pre-filter capacitors. A KEMET 0.1uF C0G (NP0) ceramic capacitor was added right after the bridge rectifier to help suppress high frequency noise. At this point, all the front section wiring is getting close to be done. Time to move on to the back panel area. To start, the output fuse holder and output connector, a GX16-4 connector, is installed and Neotech 16 AWG solid silver wires are soldered between them. Once the back panel is bolted to the enclosure, output wiring from both SJ current booster is soldered to the fuse holder soldering pad located at the bottom of the fuse holder. Once the back panel is bolted to the enclosure, output wiring from both SJ current booster is soldered to the fuse holder soldering pad located at the bottom of the fuse holder. In addition, a Mundorf Metalized Oil-Impregnated Polypropylene bypass capacitor is also added to the output connector. As a safety precaution, when building a power supply, it is very critical to have the earth pin from the IEC connector connected to the power supply chassis. The picture below show the ring connector used with 18 AWG GREEN/YELLOW pigtail bolted to the chassis, and later it will be soldered to the earth pin of the IEC connector. More wiring work will come soon... Dev, HeeBroG, Gavin1977 and 4 others 1 5 1 Link to comment
elan120 Posted May 27, 2020 Author Share Posted May 27, 2020 On 5/25/2020 at 5:48 PM, Dev said: Looks beautiful. Thanks for sharing. I bet it would sound awesome when its done 👍 Thank you, Dev, for the kind word. Link to comment
Popular Post elan120 Posted May 27, 2020 Author Popular Post Share Posted May 27, 2020 Once the back panel output section wiring is done, it's time to move to the front panel. The only component on the front panel is the 25W power resistor between the two pre-filter capacitors. It will also have the high current Neotech 12 AWG solid core copper wire soldered to both terminals. Two bypass capacitors were also added to the second pre-filter capacitor, a 4.7uF Mundorf Aluminum Oil Series Metalized Oil-Impregnated Polypropylene film capacitor and a KEMET 0.1uF C0G (NP0) ceramic capacitor to help blocking the high frequency noise. Here is an overview of the progress thus far: Next up is the transformer installation...heavy stuff! RickyV, dctom, beautiful music and 2 others 5 Link to comment
Popular Post elan120 Posted May 28, 2020 Author Popular Post Share Posted May 28, 2020 Installing the transformer is fairly straightforward, other than being heavy, it basically drops in the space reserved nicely. Once the transformer is bolted down with nylon insert lock nuts, it is time to solder the primary wires to the IEC power entry module, and the reserved safety pigtail is soldered to the earth(ground) pin of the IEC together with the ground wire from the transformer primary wirings. After all the primary wires are soldered, I left the secondary wires unsoldered and did a quick functional test by plug in the power core and turn on the power switch to measure the voltage on the transformer secondary output. This transformer was ordered with 0-18VAC spec., and it was measuring about 19.1AC, which is about right without load. Once the transformer functional test is successful, secondary wires were cut to proper length, the final soldering work is to solder them to the bridge rectifier. With the final soldering done, the entire circuitry is now complete and the next functional test is to see how this power supply performs. To do this, I normally will start with conservative approach by using undersized fuses to make sure that no load voltage reading at different locations is as expected. For this power supply, there are four voltage readings I am interested in: Voltage at the positive terminal of the second capacitor. This will indicate the voltage after the rectifier. For this power supply, with no load, 18VAC secondary output should rectify to about 26+ VDC. Voltage at the input of the voltage regulator. This should be the same as the voltage reading above. Voltage at the output of the voltage regulator. This should read slightly above 12VDC as an indicator that the voltage regulator is working properly. Voltage at the DC output connector. This will likely be reading a bit lower than the voltage reading at the output of the voltage regulator due to some line loss. To do the initial voltage measurements with a new device, my preferred method, once again, the conservative approach, is using a variac to slowly increase the input power to the transformer primary and see how the first voltage reading listed above responds, and carefully observe the other three voltage readings once this voltage reading starting to go above 14~15VDC until the input power is at the line level. This approach is a safer way to do the initial power on test, anytime reading isn’t correct, simply shut down the variac to find out the root cause. If a variac isn’t available, the only way to test would be to take a deep breath, turn on the power, and hope nothing bad happened. When all four voltages measured are as expected, a light amount of output load can be added, and gradually increasing the load together with a proper size fuse to see how the power supply is performing with the same four voltage readings. The load I use is the low cost method, a resistive load, using different value power resistors to check what voltage readings I get as resistor value decreases. Another handy tool I use is an in-line watt meter, again, low cost, albeit not extremely accurate, but very effective to see how current and wattage respond to different loads. Final test is to increase the load all the way to the power supply maximum rating at 12A by using an 1R/200W power resistor. The test went well, and at this time, it is time to adjust voltage regulator output. To do this, voltage readings from (3) and (4) will be taken while adjusting the blue color potentiometer on the voltage regulator module. Voltage reading at (3) is the regulator output, which was pre-adjusted at the factory, but this is not going to be the voltage reading at the destination due to line loss. Once the power supply is connected to the load, adjust voltage output while monitoring voltage reading at (4) to be slightly above 12V. In my case, it turned out to be 12.02V. Once this is established, go back and check voltage reading at (3), it should not be higher than 12.4V, and if it is also satisfactory, the voltage output adjustment is done. With the final voltage output adjustment done, this project is now officially completed. Time for some pictures and install the power supply to the server. I will post more listening impressions once it has some burn-in time, but the first impression is very positive. This was a fun and rewarding project, and I hope you enjoy reading this build process along the way, once I have all the parts received for the ATX power supply, I will post the build process here as well. auricgoldfinger, Altec, beautiful music and 12 others 4 8 3 Link to comment
elan120 Posted May 29, 2020 Author Share Posted May 29, 2020 29 minutes ago, Nenon said: You may want to try to connect the last small Mundorf bypass capacitor directly to the output rather than having one lead going back to the star ground. It may help a little with the high frequency filtering. Or if you have space, move it to your computer chassis and connect it across the DC connector pins there (if you are using a connector). That sounds interesting and I will give it a try this weekend along with other things. Time to move in to the "tweak" mode. Link to comment
Popular Post elan120 Posted June 21, 2020 Author Popular Post Share Posted June 21, 2020 Finally received all the parts needed to build the ATX power supply couple weeks ago, and while building the power supply, I also divert to work on few cables to connect the power supplies to the server. Below is more of an update on what I did with the ATX power supply build rather than detailed build process since most of the build process is similar to the EPS power supply. This ATX power supply will have a total of 4 rails, where 3 of them will be used for ATX connection, the 4th rail will be used to power the JCAT Net Card Femto. As a side note, the 4 Sean Jacobs regulator modules and the 2 ReCap modules are upgraded models, both ReCap modules are installed with Mundorf capacitors, and all wires used in this build is Moundorf solid core silver. The first step is getting all the metal work done, and the process is exactly the same as before. Here is the end result right before assemble key components. Next is having assemble all key components, but instead of leaving the transformer until the end, due to wiring from transformer secondary to pre-filter boards are space limited, it is decided to install the transformer in the beginning. Picture below show all 4 Sean Jacobs regulator modules, 2 Sean Jacobs ReCap modules (4 channels total), and a 400VA Toroidy Supreme Audio Grade transformer installed and secondary wires soldered to the 2 ReCap modules. With the transformer secondary wiring done, it is time to solder outputs from ReCap modules to the four regulator modules. On this build, I decided to add a 0.01uF MLCC bypass capacitor to the output connector for the JCAT Net Card. Before connecting output wires from the regulator modules, I close up the rear panel and soldered transformer primary wires to the power entry module. In this step, it is essential to stress the importance of connecting IEC power input ground to power supply chassis as a safety precaution. This is a simple step to add in the build, but often seen omitted in commercial power supplies. Below are two pictures showing how this power supply is done, where the first picture is the pigtail connected to the chassis, and the second picture show this wire soldered along with transformer primary wires to the power entry module. With the power entry wiring completed, the remaining wiring would be soldering wires from regulator modules to the two output connectors. Here are pictures showing completed left side wiring. Here are pictures with all output wires completed. Now with the wiring job completed, time to install the front panel, power up the supply using the same procedure described previously, measure adjust output voltages under load, and more pictures showing completed ATX power supply. This is the final episode of both EPS and ATX power supply build. It was a very fun project and I hope it was helpful sharing the details here. If you decided to build a power supply, please share some pictures here. For anyone interested in cable building, I post one of the cables build for this server here. Cazzesman, LTG2010, Exocer and 8 others 1 8 2 Link to comment
Popular Post elan120 Posted June 21, 2020 Author Popular Post Share Posted June 21, 2020 1 hour ago, Nenon said: You have to do a follow up on listening impressions Sounds like a plan...will do that after some burn-in time. Nenon, RickyV and austinpop 1 2 Link to comment
elan120 Posted July 30, 2020 Author Share Posted July 30, 2020 5 hours ago, Tatomek7 said: Do you have some first listening impressions? Yes, the initial listening impression after about a month of burn-in is very positive. At this time, the initial burn-in for the EPS power supply is either done or very close being done as the improvement started to become difficult to detect. Overall, without going into too much details, this power supply is a very significant step up in SQ compare to the different setups mentioned in my first post that I have tried. The details, background noise level, sound stage, dynamic, impact, all took a very positive step forward, and the music presentation is so much more enjoyable with this power supply. 5 hours ago, Tatomek7 said: Did you check, as a bypass capacitor Miflex ones? Duelund are the best, but Miflex adds speed and its on the lower budget side I only have tried a limited number of tweaks thus far, and I do have few more I plan to experiment once everything is settled a bit more, and different bypass cap being one of tweaks I plan to explore. 5 hours ago, Tatomek7 said: Particularly I'm interested with the first project with only 2 big Mundorf capacitors, which I personally would be afraid to install. Just curious, what reason(s) preventing you from considering these caps? Link to comment
elan120 Posted July 30, 2020 Author Share Posted July 30, 2020 @Blackmorec, Thank you for the comment. I think you are right on regarding the lengthy burn-in needed for these components used in both power supplies. After a month of burn-in on the first one (EPS), although the improvement is slowing down, but I can still sense the change from time to time, and it is so much more fun and enjoyment having them in the system now. I will have more detailed impression updates later, but even at this point, I am very impressed with the result, and having inputs from Sean Jacobs on various questions really helps. 16 minutes ago, Blackmorec said: Quality power supplies do seem to have an especially powerful influence on improving sound quality in digital systems. Fully agreed! RickyV 1 Link to comment
elan120 Posted July 30, 2020 Author Share Posted July 30, 2020 26 minutes ago, k27R said: Why the 0.01uF MLCC bypass capacitor to the output connector for the JCAT Net Card? The main purpose for 0.1uF MLCC cap is for HF filtering. 27 minutes ago, k27R said: Why just for that output and not the others? They were installed on the ATX cable connector side but not on the power supply output connector due to physical space constraints. It should work better with caps on the receiving end in general, so it was decided to place them there. k27R 1 Link to comment
elan120 Posted August 13, 2020 Author Share Posted August 13, 2020 14 hours ago, Fourlegs said: I have been curious about just how much difference using the Toroidy Supreme Audio Grade transformers will make as opposed to sticking with the Toroidy Audio Grade transformers. I have two identical SJ power supplies built to power a Chord Dave (5v, +15v, -15v), one with and Audio Grade and the other with the Supreme Audio Grade and I'm doubtful that I can hear any difference. Unfortunately, I won't be able to provide more information on this, as I only used the Supreme Audio Grade transformers. As for the difference, the Supreme transformer is made using slightly higher grade steel, that is the main functional difference. Shielding is the same on both models, as is size of core. The Supreme has the steel can that should help with noise reduction, and be quieter. The cost of the Supreme is about 50% more, but the total cost compare to all other components, I thought is fairly insignificant. Link to comment
elan120 Posted August 14, 2020 Author Share Posted August 14, 2020 11 hours ago, Fourlegs said: Thanks. I will try to find time to do some more listening comparisons between the two supplies that are identical apart from the two types of transformers. It might be possible to involve Sean Jacobs in this as he had been booked in to come to me for a day to listen to as many versions of his power supplies that we could out together between us but it was postponed due to lockdown just the day before we were due to get together. Maybe we can reconvene soon. That will be an excellent data point. Looking forward to your listening impressions and hopefully Sean's as well when this pandemic is behind us. Fourlegs 1 Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now