I made this linear power supply based on LM 338 with pretracking regulator and the regulated output is 11.8 V. I use a 10 V zener to reg the output.
The trans is 13.5 V ac out and at least 400W capability and is source from an old medical laser printer.
I made it to power an atom based pc which utilize a 60 W smps.
The reg power supply will not power on the pc .
What are the special requirements to use a linear power supply for pc use ??
pointers are most welcomed.
thanks
kp93300
The trans is 13.5 V ac out and at least 400W capability and is source from an old medical laser printer.
I made it to power an atom based pc which utilize a 60 W smps.
The reg power supply will not power on the pc .
What are the special requirements to use a linear power supply for pc use ??
pointers are most welcomed.
thanks
kp93300
Attachments
Althoug linear PSU's are less complex and have less noise than switching ones, still they require a much bigger power transformer and filter caps, also the power devices have to cope with much more dissipation si heatsinks have to be much larger too, in other words they cost much more and occupy way more space in the case, and this is only to be verry short.
13.5Vac, with pre-tracking regulator and series regulator !!!!!
What is the output of the transformer at the smoothing capacitors?
You may find that the regulator drops out when demand goes a little bit higher than just ticking over.
Linear PSU's have one major drawback, for verry good stabilised outputs you have to take into account the minimum possible mains voltage and at that level the psu still has to deliver the needed power, that means higher voltage outputs for the transformer, high current capability and ofcourse much power lost on the series regulator, at hig mains level the PSU will experiance the lowest efficiency and the most power lost on the regulator, all that means you will end up with one big PSU ( real big sometimes ) and much more money spent on it, so that is why swithcers are prefered.
Hi
Thanks all.
The commends are all valid.
I start this exercise for experimentation. I just want to hear for myself whether there is any difference in sound quality from my atom based min atx sound system.
I realise that with the pretracking reg , the Vout do not have enough headroom . I think it requires about 5 to 6 V differnce between the input and output.
I remove the pretracking reg and manage to power the pc for 2 minutes before it went into shutdown because of overheated Lm338 !
cheers
kp93300
Thanks all.
The commends are all valid.
I start this exercise for experimentation. I just want to hear for myself whether there is any difference in sound quality from my atom based min atx sound system.
I realise that with the pretracking reg , the Vout do not have enough headroom . I think it requires about 5 to 6 V differnce between the input and output.
I remove the pretracking reg and manage to power the pc for 2 minutes before it went into shutdown because of overheated Lm338 !
cheers
kp93300
2nd attempt with LT 1084 --success
Hi
I rebuild another module of the power supply for my atom based pc with linux os.
Try it for a few hours and is stable.
Bridge rectifier is the hottest at about 55 C. There is minimal increase in temperature for the 2 LT 1084 .
Those who thinks that power supply for pc audio does not matter should build one to try for themselves.
I will spend the next few days trying to optimize it. I do not have a scope and i seek advice from more experienced diyer. The circuit is from LT 1084 datasheet . The original circuit do not have cap in the output.
I wonder why is there no 120R from V out pin to adjust pins of the top LT 1084.???
cheers
kp93300
Hi
I rebuild another module of the power supply for my atom based pc with linux os.
Try it for a few hours and is stable.
Bridge rectifier is the hottest at about 55 C. There is minimal increase in temperature for the 2 LT 1084 .
Those who thinks that power supply for pc audio does not matter should build one to try for themselves.
I will spend the next few days trying to optimize it. I do not have a scope and i seek advice from more experienced diyer. The circuit is from LT 1084 datasheet . The original circuit do not have cap in the output.
I wonder why is there no 120R from V out pin to adjust pins of the top LT 1084.???
cheers
kp93300
Attachments
Provide some measurements, it dosn't matter whether you use an SMPS or a Linear, and SMPS's have many advantages, not least being they operate well above audio frequecies.
In fact I would like some views on why people think a linear PSU wouyld be better than an SMPS for a digital based music server, especially a PC where you have numerous gates switching anyway, is it gonna make any difference, not realy. What will make a difference if you are realy worried is isolation qas near the final digital conversion as possible and seperate supplies for the analogue side of the DAC and the following analogue circuitry, optical or wireless from the PC, or SPDIF/I2S galvonicly isolated.
okay
I do not know their products
no worries I just dont like to see them and their voodoo linked on a mostly fairly rational site. add up one of their servers with all the upgrades and you can finish well over 50K, including very fast processors and interfaces etc which puts MORE stress on power supply and ground routing quality and causes MORE noise, for NO benefit for audio playback (despite their miraculous claims), which in truth is quite easily handled by old or low power computers.
its completely needless expense, just so they can be exclusive and add their cut to the many useless, even harmful upgrades. its a joke.
Last edited:
the point is, as marce as touched upon above, the audio subsystem should be build in such a way that the PC doesnt matter and that is quite doable, tuning the computer power supply is a pointless endeavour, because when working correctly the computer is still a noise source at bandwidths that even the best linear or shunt supply is powerless to deal with.
I was skeptical at first but I do hear a difference and the improvement in bass is pretty obvious.
The drop out from the Amanero combo usb to i2s converter is less. The amanero board is powered separately from another linear power supply.
i wonder is the sound difference due to cleaner ground plane.
My apologies, i do not have a scope to use.
this mourning i insert a inductor in the positive rail before the regulator. The pc will not start .
i hope to invite a few friends to do a-b testing between smps and linear ps for pc .
cheers and happy new year.
kp9330
The drop out from the Amanero combo usb to i2s converter is less. The amanero board is powered separately from another linear power supply.
i wonder is the sound difference due to cleaner ground plane.
My apologies, i do not have a scope to use.
this mourning i insert a inductor in the positive rail before the regulator. The pc will not start .
i hope to invite a few friends to do a-b testing between smps and linear ps for pc .
cheers and happy new year.
kp9330
what makes you think you have a cleaner ground plane? LDOs, which includes the 1084, generally have pretty crappy transient response and either way what has it got to do with the ground plane? the PC is still covered with little SMPS and the processors are still the same high bandwidth, fast switching, very transient load. all this still dumps into the ground plane, just like before.
Last edited:
Happy New Year
When you do impedance diagrams of PCB power and ground planes using some form of powe3r integrity software, the impedance is not even across the plane at any frequcies and the higher the operating frequency the worse it gets. When any digital system is working especially multi voltage (PC will have at least 5+ on board seperate supplies, DDR memory has 2 or 3), these are bouncing up and down all the time as the gates switch, often having an instantaneous current draw of 100A's. The only way to improve this is during layout, with correct de-coupling, layer stack, local supply location, power islands, shape of power plane islands. When doing a DDR memory interface it uis reccomended to simulate the layout using signal integrity software, this is because these interfaces run at the limit. So with all that going on and adding to the noise spectrum, your front end supply isn't gonna make a lot of difference (unless you have a bad supply and these can be either linear or SMPS). I have worked on systems with both linear and SMPS main supplies and the resulting noise superimposed on an analogue waveform after the DAC is the same (without nit picking with FFT), all high frequency related to system clocks and switching cycles. Isolation was the only cure (planar transformer, capacitive shield, optimised for bandwidth). I would be interested on information on how audible if at all this high frequency noise is, and also does it have any noticable effect further down the line, if it isn't filtered out or isolated.
A further thought is how immune our systems are from EMC interference, the ever increasing use of wireless devices has greatly increased not only the amount but frequency range of RF noise present in our envoironment...
When you do impedance diagrams of PCB power and ground planes using some form of powe3r integrity software, the impedance is not even across the plane at any frequcies and the higher the operating frequency the worse it gets. When any digital system is working especially multi voltage (PC will have at least 5+ on board seperate supplies, DDR memory has 2 or 3), these are bouncing up and down all the time as the gates switch, often having an instantaneous current draw of 100A's. The only way to improve this is during layout, with correct de-coupling, layer stack, local supply location, power islands, shape of power plane islands. When doing a DDR memory interface it uis reccomended to simulate the layout using signal integrity software, this is because these interfaces run at the limit. So with all that going on and adding to the noise spectrum, your front end supply isn't gonna make a lot of difference (unless you have a bad supply and these can be either linear or SMPS). I have worked on systems with both linear and SMPS main supplies and the resulting noise superimposed on an analogue waveform after the DAC is the same (without nit picking with FFT), all high frequency related to system clocks and switching cycles. Isolation was the only cure (planar transformer, capacitive shield, optimised for bandwidth). I would be interested on information on how audible if at all this high frequency noise is, and also does it have any noticable effect further down the line, if it isn't filtered out or isolated.
A further thought is how immune our systems are from EMC interference, the ever increasing use of wireless devices has greatly increased not only the amount but frequency range of RF noise present in our envoironment...
The only valid reason I can come up with for me wanting to justify it to myself (I've thought about building a linear psu for my pc for ages) is so I can keep computer psu borne noise out of the mains wiring which affects ham radio on shortwave, and at the same time charge a couple of large SLA batteries at 13.5v float.
I then make sure the pc is completely sealed with copper tape and then install aluminium fan filters.
Aside from that also making the computer as rugged and reliable as possible, I believe that Linear power supplies are far more rugged than smps units, feel free to tell me I'm wrong though.
Aside from that does the original poster and a few other posters it seems know about Project 77?
Here is a link: http://sound.westhost.com/project77.htm
I then make sure the pc is completely sealed with copper tape and then install aluminium fan filters.
Aside from that also making the computer as rugged and reliable as possible, I believe that Linear power supplies are far more rugged than smps units, feel free to tell me I'm wrong though.
Aside from that does the original poster and a few other posters it seems know about Project 77?
Here is a link: http://sound.westhost.com/project77.htm
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.