PDA

View Full Version : DIY Linear Power Supply MKII (lots of pics!)



crooner
2007-02-16, 00:00
Hi Fellas,

One year ago I built a linear power supply for my Squeezebox based on the popular Power One HB5-3/OVP-A unit. See this thread:

http://forums.slimdevices.com/showthread.php?t=21505

I've been very happy with the performance of this Power One unit. However, I wanted to upgrade the enclosure and the power supply itself to match my new Audio Research components.

I wanted something similar to fellow member Jabberwockie:

http://forums.slimdevices.com/showthread.php?t=23227&highlight=HB3

But with a larger extruded aluminum case. This would allow me to add an extra circuit board with a pair of high quality filter caps and bypass.

For this application I chose a pair of 8,200 uF Elna Cerafines and a 1uF high quality film bypass. Total capacitance is 16,400 uF compared to 10,000 uF of the original single cap.

The other small caps on the board were replaced with Elna Cerafine and film caps as well.

A single IEC/power switch and fuse assembly was added. For the DC connection I chose a 5 pin DIN jack and plug. Everything was wired with top quality Kimber Kable.

The IEC inlet has internal EMI filtering.

I like blue LEDs :-) so I added one to the front plate as well.

Take a look at the many pics. This power supply is rock solid at the 5V preset voltage.

Comments welcome!

crooner
2007-02-16, 00:01
More pics....

Skunk
2007-02-17, 09:25
Hi crooner,

Do the sense lines run to the DIN socket, or through it to the DC lead?

crooner
2007-02-17, 10:07
I'm not using the sense lines. Would seem like heresy, but I have it wired just like it was 1 year ago. Works fine for me. The sense lines are jumpered to the outputs.

Besides, I'm a using a short length of wire to the SB3, so there shouldn't be any apparent losses. It measures exactly 5 volts at the barrel plug.

Skunk
2007-02-17, 13:15
I'm not using the sense lines. Would seem like heresy, but I have it wired just like it was 1 year ago. Works fine for me. The sense lines are jumpered to the outputs.

Besides, I'm a using a short length of wire to the SB3, so there shouldn't be any apparent losses. It measures exactly 5 volts at the barrel plug.

It looks great.

Did you get the board off to add the caps? I wasn't sure how the transistor comes off.

crooner
2007-02-17, 13:39
Thanks Skunk!

The TO3 cased transistor comes off after desoldering the center pins and removing the two screws. It takes a bit of wiggling but the PCB came off relatively easily.

crooner
2007-02-17, 13:43
Oh, by the way...
I fired up the SB3 with the tweaked up power supply last night. The improvement was not subtle. Lots of low level information retrieval. Decay seemed much more natural and voices had a more realistic body to them.

Havoc
2007-02-17, 15:48
Putting those caps after that lenght of wire is silly. Might just have added some common kitchen-garden variety of caps.

crooner
2007-02-17, 15:51
Putting those caps after that lenght of wire is silly. Might just have added some common kitchen-garden variety of caps.

Remember the wire is low loss Kimber in a twisted pair configuration to reject noise. I doubt the thin copper traces on the PCB could be any better, even if they are short.

Greg Erskine
2007-02-18, 01:03
Remember the wire is low loss Kimber in a twisted pair configuration to reject noise. I doubt the thin copper traces on the PCB could be any better, even if they are short.

Hi crooner,

Nice job on the power supply. Do you have a schematic of the PSU? I'd be very interested in seeing one, but haven't seen one posted anywere do far.

As a side point, for filing away in the appropriate place (bin?)

When designing PCBs for power supplies or amps, it is considered "good" practice not to make T traces connecting your main filter caps to the power rail. Also, long leads on caps is considered "bad" practice because it adds inductance. Adding wire is like making long T traces and long leads. Having said that I have no idea if it really matters in this application as I haven't tried it.

regards

crooner
2007-02-18, 01:10
Thanks for your reply Greg.
Unfortunately, I don't have the schematic even though it appears to be a pretty standard design.

As for the long leads, I had two choices: try to "jam pack" the extra caps on the board or use an additional PCB. I decided to go for the latter, since it was much more elegant. As with all things, there are compromises. Added inductance could certainly be a problem. However, it appears the power supply works perfectly so I'll leave well enough alone.

Originally, I was just going to transfer my supply to a different case for asthetics. My old case was cheap thin aluminum and the holes I punched were pretty sloppy.

It then ocurred to me: why not do some small improvements as well? I had the Cerafine caps laying around in my parts bin so I decided to give them a go.

I didn't want to go overboard with filter capacitance since this would load the power transformer too much. So, I added just a few thousand extra uf :-)

crooner
2007-02-18, 01:15
Also, I've noticed that filter caps are usually soldered in PC boards in low voltage/wattage applications.

In the case of high current amplifier power supplies, for instance, they are almost universally hard wired to the chassis. The wire is usually a low gauge type to handle the extra current.

I know my feeble supply is only 3 amps max, but I feel better knowing that at least in some instances the caps are not necessarily on PCB traces :-)

crooner
2007-02-18, 01:35
I was checking the specs for the Elna Cerafine caps and found this:

**************************************

FEATURE

True Audio Reproduction by the suppression of electrical noise due to external vibration
Low Loss.
Good Current Capacity.
High Frequency and Temperature Stability.
Good Long Term Mechanical Reliability


ELECTRICAL PERFORMANCE

Low Dielectric Absortion Factor.
Low Equivalent Series Resistance
Low Self Inductance.
Low Dissipation Factor.
High Insulation Resistance.

********************************

It appears that, in addition to their well known low ESR qualities, they are also low in self inductance.

Hopefully, this should help compensate for the additional lead length...

crooner
2007-02-18, 01:49
This guy recommends adding a choke to improve filtering:

http://www.muzique.com/lab/filter.htm

Greg Erskine
2007-02-18, 01:49
Also, I've noticed that filter caps are usually soldered in PC boards in low voltage/wattage applications.

In the case of high current amplifier power supplies, for instance, they are almost universally hard wired to the chassis. The wire is usually a low gauge type to handle the extra current.

I know my feeble supply is only 3 amps max, but I feel better knowing that at least in some instances the caps are not necessarily on PCB traces :-)

Hi crooner,

You raise a good point. It doesn't matter if it is a PCB trace or a piece of wire, what is important is how it is connected.

I'll have to keep an eye out for the schematic, I'm sure it will turn up eventually.

regards

crooner
2007-02-18, 01:57
Greg:
What do you think of the guy using the choke?
I know chokes were common in the old tube days to smoothen out the B+ because filter capacitance was very expensive back then.
Regards,
crooner

crooner
2007-02-18, 02:03
If any fellow member has an schematic of the Power One HB5-3/OVP-A, please let me know!

Greg: It uses the ubiquitous LM723 regulator if this helps...

Greg Erskine
2007-02-18, 04:41
Greg:
What do you think of the guy using the choke?

Hi crooner,

I've never used a choke in a power supply, but reading the PSU design books, CLC filtering is better than CRC filtering, which is better than just straight C filtering. "Better" means less ripple. Does this matter in a SB PSU? I understand high current, high quality inductors are expensive and that's why they aren't used as often as they should. $4 does seem cheap though.

I've only tried CRC filters on my SB PSUs so far, so I haven't made comparisons.

Thanks for the info on the regulator. I haven't read the specs on that one before.

regards

Skunk
2007-02-18, 23:43
Thanks Skunk!

The TO3 cased transistor comes off after desoldering the center pins and removing the two screws. It takes a bit of wiggling but the PCB came off relatively easily.

Odd that mine wasn't screwed down whatsoever, just soldered. Also, there is a clear plastic spacer between the transistor and case. Does yours have the same?

I lifted one of the pads at the input, so cut the trace out and ran the wire there as shown, and would be interested in your thoughts. To me it looks like there is hardly any trace at all between the input and cap/diodes, when the cap is on the board. I thought that best practice was keep the cap close to the input, but could very well be wrong.

crooner
2007-02-19, 01:36
Hey Skunk!
Nice job removing the PCB!
There isn't much trace in there probably because the PCB itself is so small. If you can get a single cap with higher capacitance that would fit, go for it. Otherwise use a short length of wire as close to the chassis as possible, twisted pair of course!

crooner
2007-02-19, 01:37
Oh and yes, mine had the plastic spacer, important to prevent the center pins to touch each other or the chassis.

Skunk
2007-02-19, 02:21
Oh and yes, mine had the plastic spacer, important to prevent the center pins to touch each other or the chassis.

That makes sense, thanks.

Disassembly is the easy part, and where most of my projects end. Hopefully I'll have the sticktoitiveness to get it all back together.

crooner
2007-02-19, 07:24
Skunk: I'm investigating about power supply caps and external/additional wiring. Found an article from an old magazine.
Using a little extra wire may not be as bad after all!

Skunk
2007-02-19, 10:08
Skunk: I'm investigating about power supply caps and external/additional wiring. Found an article from an old magazine.
Using a little extra wire may not be as bad after all!

I've been reading at Rod Elliot's site and he seems to prefer the P2P wired PS as well.

There's an incredible amt. of credible info there, e.g. under the 'other components' subheading he talks about the choke inputs you referenced, and the required diode upgrade. http://sound.westhost.com/psu-wiring.htm

FWIW by 'not much of a trace', I meant the triangle is actually a very wide thick trace, almost a plane, so it shouldn't be assumed to be an inferior connection compared to wires.

crooner
2007-02-19, 14:16
The Dynaco Stereo 400 power amplifier from the mid to late 1970s used what they called their C-100 "Energy Storage Unit". An additional box contaning banks of huge capacitors. The caps were all wired in parallel with a bleeding resistor. I'm figuring there was quite a bit of wire in there, not to mention the "umbilical cord" that hooked up to the amp!

Greg Erskine
2007-02-20, 00:28
Hi crooner,

There is nothing wrong with using wire in PSUs, as you point out it has been done for years. Manufactues only use PCBs because it is cheaper, not because it is better. But...you shouldn't make a T junction for the caps with wires or PCB traces.

Wire between parallelled caps acts as a resistance, which is good, because it forms an CRC network, wire as an umbilical cord is also OK, it acts as a resistance forming a CR network but you should have decoupling caps at the very end of the umbilical cord.

I will draw a little picture if I haven't been clear.

regards

crooner
2007-02-20, 16:29
I made some changes to the wiring. I shortened the leads to the extra PCB with the caps. I also untwisted them.

Skunk
2007-02-20, 23:21
I will draw a little picture if I haven't been clear.

regards

I would like to see the little picture :-)

This paper goes into great detail on bypassing long serial supply traces, which I'm guessing is what you mean by t junction (isn't that a plumbing term?): http://www.designers-guide.org/Design/bypassing.pdf

crooner
2007-02-21, 00:58
LOL, I thought he meant a "T junction" was formed by twisting the two conductors which would create inductance, like winding a coil of sorts...

In any case, a shorter wire to the caps in my PS doesn't hurt anything at all :-)

crooner
2007-02-21, 01:03
Hey Skunk, that's a great paper, I'm reading it right now!

BTW, what do you think of my latest PSU cap wiring iteration?

I haven't listened to my system yet, but I think I won't be messing with it anymore!

And how's your PSU doing? did you put it back together?

Greg Erskine
2007-02-21, 02:43
I hope this little diagram shows what I mean. You can replace the traces with wire if you like. Yes, that is a good article Skunk.

Skunk
2007-02-21, 13:28
I hope this little diagram shows what I mean. You can replace the traces with wire if you like. Yes, that is a good article Skunk.

Yes Greg, that is helpful, and quite a nice rendering I might add.

Being green to DIY, it would also be helpful to me if you could explain how crooner's paralleled caps could be taken out of the t-trace scenario, in terms of wires being added/removed/shortened/bypassed.

That link is hard to get through. Actually it's impossible for me, if comprehension is a prerequisite. I was throwing it out there as a guess, mostly.

Thanks again for taking the time to make that drawing! The rendering not completing the picture for me is not its fault, I assure you :-)

crooner
2007-02-21, 16:05
Thanks to Greg and Skunk. You are true gentleman folks. Much appreciated.