Basic questions for the power supply

kepa1

Member
2004-04-07 2:59 pm
Paris
Hi all,

this is time to complete my UCD amp plate project but things get a little complicated :whazzat:

[IMGDEAD]https://img.photobucket.com/albums/v221/kepa1/DSCN6318.jpg[/IMGDEAD]

I measured the output of both bridge rectifiers and got 0 V DC instead of 30 - is that normal? the bridges are brand new and the transformer output 32 V AC.

Also, I wonder if the unused supply rails should be connected to the ground from the rectifiers or from the capacitors, as shown on the photo (just ignore that black cable hanging around).

And last, the switch I bought has three connectors, how are they supposed to work?

Thank you
 
Hi,

Start by taking a step back and a real deep breath.

Then, look up the data sheets for the parts you have in question, bridge rectifiers for example, that will tell you how to wire them properly if you're in doubt, pinouts etc, they're usually pretty clear though.

Then, once you know how all the parts work, make yourself a wiring diagram on paper, it should work on paper before you ever wire it, then show us that diagram and we can be alot more help. The picture is useless to me at least.

Alternatively, base it off of what you know will work, someone else's wiring diagram. Saaaaay perhaps, from http://www.zero-distortion.com/techno/powersupply/powersi_05.htm This should really be all you need.

Hope you didn't fry anything costly.

Chris

PS, I wired up a bridge rectifier backwards once (tired) it squeeled like crazy and the xformer got HOT but it all lived. Seems they can take a good beating though. It might be a good idea to place your rectifiers alot closer to your caps though, shorten the wires that is.
 

kepa1

Member
2004-04-07 2:59 pm
Paris
Ok, so the PS is just the same as this one, except it only has one capacitor per rail :

[IMGDEAD]http://www.plastichead.net/audioworks/content/amps/ucd180-1/ucd_psu.gif[/IMGDEAD]

I remeasured yesterday evening and the transformer outputs 32VAC :up: but i got this strange result 1. when measuring simultaneously the + / - pins of both bridge rectifiers.

Measuring the + pin and the ground gave 0 V.

Maybe there is another way to test?
 

classd4sure

Disabled Account
2004-02-13 3:35 pm
www.diyaudio.com
The drawing looks OK, that's got to work right?

Triple check your rectifiers yet? Test them on the diode scale just to make sure the parts are good to start with.

At this point take your drawing and trace down every wire one by one, highlight/ mark them off as you go, something in your circuit isn't going to match the drawing. Like for instance, why do you only have three wires going to the upper bridge rectifier in your picture?

Cheers,
Chris
 

soongsc

Member
2005-03-26 2:31 pm
Taiwan
It's hard to tell what's going where in you picture because the transformer is not seen and the wre color coding is not consistent. But if you look at the rectifiers, there is a cut corner, that corner corresponse with where the + side of the rectifier where two arrows point to.
 

kepa1

Member
2004-04-07 2:59 pm
Paris
Shame on me... the multimeter was calibrated at 20V and couldn't read the voltage coming out the bridges :eek: In the end, I read 28V DC out of them.

I have wired both bridges to the capacitors and everything looks good now. the caps deliver a mighty 44V...

... But I still have questions for you guys ;)

These big caps appear to discharge very slowly - half an hour after switch off, they still output 38V :bigeyes: after a night I could measure almost 5V.
Should I wear plastic gloves to work on the circuit?

Also, is it OK if the - terminal of cap#1 is connected to the + terminal of cap#2 at the star ground point? or should they be separates to avoid any kind of short circuit or electric shock on the chassis?

Then should the star ground point be isolated from chassis (mine is made out of aluminum and the star ground point is the screw in the middle of the plate)?

Soongsc, the transformer is screwed to the other side of the plate - It's so big I couldn't place it but there.


Cheers
 

classd4sure

Disabled Account
2004-02-13 3:35 pm
www.diyaudio.com
Heh heh.... been there! That's when I take a nap. Honestly though, aren't those the best kind of problems?


Gloves? Can't kill yourself at these levels, well you'd have to really try..

No, what you should do is discharge your caps alot faster to keep them happy, yes it loads the supply down a little but...call it insignificant I guess. That depends on how you to do it too :) I recommend you spend more time at the link I gave you a few posts ago. Click the back arrow on the page a few times to get you to the first and most basic supply and then read how they evolved and you'll get it.

You're putting your caps through a slow death by letting them self discharge slowly like that, they'll tend to want to reform to the lower voltage.

What you want between your caps is a good thick, short wire or bus bar that'll really let the ripple current flow nicely between them. Along this line flows rather turbulent currents, that means it's not a good place for star ground placement as it would pollute all your system grounds.

Connect them together with as short a wire as you can. Then "T"-off that wire and connect to your star ground. This leaves most of the turbulance flowing between the caps and out of your high quality ground point.


As far as floating the HQG off the chassis, that's more of a preventive / band aide cure for ground loops which you should be first trying to avoid.

Good way if you want to mass produce I'm thinking, but it doesn't seem like a hard addition after the fact so I'd try without first, and see what happens?

Cheers
 

kepa1

Member
2004-04-07 2:59 pm
Paris
Now How comes i get 10VDC on the Output?

Hi,

It's too bad, this thread has got to be re-open... :xeye:

Well, after replacing the small cables you see on the picture and setting them the good way, I measured 10VDC on the cable connected to the UCD speaker outputs (no AC voltage).

No speaker was connected (fortunately) and the DC is the same whether I connect a source or not.

Each capacitor outputs 43VDC and V+/V- on the UCD give 86V; I guess this is normal?

I'll re-check everything including the grounding which may be faulty, but in the meantime if you have any suggestions, please comment on this.

Alain
 
Why don't you wire the psu like this. You save one retcifier and the everything becomes more simple.
As long as the two secondaries are mached, theres nor problem.
 

Attachments

  • psu_wiring.gif
    psu_wiring.gif
    7.1 KB · Views: 411
A question of power

sovadk said:
Why don't you wire the psu like this. You save one retcifier and the everything becomes more simple.
As long as the two secondaries are mached, theres nor problem.



This is the traditional way of doing it. It does have the advantage of fewer parts and costs less to build. This is why most manufactures use it. This is a compromise as all things electronic are. In this case the trade off is less cost for greatly increased ripple current in the grounds. The only practical way for it to work is to have very short and very heavy ground connections. Using 2 bridge rectifiers eliminates the ground to transformer connection therefore the ripple current in the grounds is eliminated as well. The output from the bridges then can be routed to an input side of the cap bank and the output side becomes the only ground connection. It will still have current in it but it now will only be signal current.
Taking care of your grounds and ground connections is one true path to sonic nirvana! Bottom line; lower noise floor and better sound!
Roger
 
Re: Now How comes i get 10VDC on the Output?

kepa1 said:
Hi,

It's too bad, this thread has got to be re-open... :xeye:

Well, after replacing the small cables you see on the picture and setting them the good way, I measured 10VDC on the cable connected to the UCD speaker outputs (no AC voltage).

No speaker was connected (fortunately) and the DC is the same whether I connect a source or not.

Each capacitor outputs 43VDC and V+/V- on the UCD give 86V; I guess this is normal?

I'll re-check everything including the grounding which may be faulty, but in the meantime if you have any suggestions, please comment on this.

Alain

Please post another picture of how it is connected now and someone should be able to help, 10 volts is defiantly not right.
Roger
 

soongsc

Member
2005-03-26 2:31 pm
Taiwan
Re: A question of power

sx881663 said:




This is the traditional way of doing it. It does have the advantage of fewer parts and costs less to build. This is why most manufactures use it. This is a compromise as all things electronic are. In this case the trade off is less cost for greatly increased ripple current in the grounds. The only practical way for it to work is to have very short and very heavy ground connections. Using 2 bridge rectifiers eliminates the ground to transformer connection therefore the ripple current in the grounds is eliminated as well. The output from the bridges then can be routed to an input side of the cap bank and the output side becomes the only ground connection. It will still have current in it but it now will only be signal current.
Taking care of your grounds and ground connections is one true path to sonic nirvana! Bottom line; lower noise floor and better sound!
Roger

I remember doing a pre many years back using the "traditional way" and always saw a glitch when the centertap was connected to the ground.
 
Taking care of your grounds and ground connections is one true path to sonic nirvana! Bottom line; lower noise floor and better sound!

Is the supply ripple really that important when the caps are still large. The UCD and other amplifiers have a huge supply rejection especially at law frequencys. I've also build my own Class D amplifier using a Mueta like approach (cycle by cycle hysteresis control from after the output filter) and godt a 120dB+ dynamic range (my soundcard can't mesure the noise floor, it's to small) and the small 50Hz peak I saw in the spectrum (20dB above the noise floot of -120dB) was because noting was shielded. The 50Hz peak and the meaured noise floor was unaffected by whether or not the amplifier was turned one.
The distortion however was affected by the size of the supply caps, but this has nothing to do with the use of one or two retcifiers in the power supply.
Gounding internally in the amplifier affects also distortion, but I have improved this a lot by using a starpoint ground approach.
 

kepa1

Member
2004-04-07 2:59 pm
Paris
Here we go for the photos:

[IMGDEAD]https://img.photobucket.com/albums/v221/kepa1/DSCN6330.jpg[/IMGDEAD]



[IMGDEAD]https://img.photobucket.com/albums/v221/kepa1/DSCN6331.jpg[/IMGDEAD]

This is my little plate taking fresh air outside the window ;) (of course the transformer is screwed to the opposite side of the plate). The wires around the caps are arranged this way to let space for an optionnal MKP 0.22µF cap.

But still it's not working, it's even worst than before with -22VDC at output :dead: - I don't know why it 's higher, i just fixed the wire form the caps to ground a little tighter.

Something to tell you : not all terminals are crimped - I just pushed the wired inside them (to test the amp first) and it is stable.

Alain
 
Grounding issue?

Please remove the plastic jackets from the terminals and solder the wires to them if you don't have a crimper. Put shrink tubing over them for protection afterward. The ground wire from your common point is probably open or high resistance. It should also be heavy gauge and connected directly to the ground between the 2 caps. This wire has to handle the full current and needs to be away from the star point so as to not contaminate it.
Roger
 
To ground or not to ground that is the question!

sovadk said:
Is the supply ripple really that important when the caps are still large.

It is important only if you are interested in achieving the best possible performance from your unit. Having ultra large ultra low ESR caps means the ripple current peaks will also be ultra large. This makes it even more important to secure the grounds with very low resistance connections, short paths, twisted pair wires and avoiding having this ripple current circulating inside of signal grounds. Using 2 bridges you can totally separate the ripple current from the signal ground and your measured noise will improve. For me in some cases this has made the difference between having a workable or not workable amp design.
Roger