Advice First PSU build

[goldfinger]

Hi All,

New to this forum and great to be here as seems lots of great information and input from members. Ive only recently got into DIY electronics and Im fairly new to it all so please forgive my newbie questions, but as Im trying to build a power supply it’s something I would rather ask questions and be 100% clear on rather than risk doing the wrong thing with mains!!!!

Im looking to build a power supply for a chipamp LM4780 and I have a few questions and seeking advice an input into putting it together. Here is the info.

Ive recently bought a few bits and pieces to build a power supply.

1 x 240v 100VA Torodial transformer w/ 1 x secondary winding of 12VAC (for testing)
1 x 240v 400VA Torodial transformer w/ 2 x 22VAC secondary (for amp)
1 x PCB board – Bridged Rectifier PSU & a PSU cap board for power/filtering

I haven’t determined ecaps, filter caps, or rectifier yet but I have some caps/diodes lying around I could use for testing with the 12VAC transformer and breadboard if needed.

My main question (and confusion)surrounds 0v input.

Im sorry if my post sounds unclear but Ill try to point out an example

From the IEC connection I have live, neutral and ground. The live and neutral connect to the transformer primary, the ground/earth goes to the chassis (or whatever test earth is appropriate)

I then take 1 or 2 secondaries to a full bridge rectifier which would be either 2 leads going to one ~ bridge or 4 leads going to two separate bridge rectifiers

Then I have positive/negative +/- dc coming out of bridge. Now in the case Im going into the cap pcb board (see attached pic) I have a 0v input what do I use as input? Do I connect another wire from the ground wire chassis connector (where I connected the IEC ground) to the 0v input? Then any ground connector on output what reference that?

Also, I have a PCB that has a Rectifier/CAP PSU on one board on input it has 2 x AC in and a 0v again Im unsure on how this connects. On the output it has 2 x + 2x – and 2x gnd (DC out).

 

Probably a long and confusing post but basically my confusion lies in the 0v connections seeing I have only 2 x or 4x secondary wires which are carrying the VAC and Ive already grounded the IEC connection so Im not sure how I connect. I hope this makes sense.

 

Really appreciate some feedback and my apologies if my question is confusing or ill informed.

 

Thanks Guys

[Ian.M]

Provided the two secondaries are *EXACTLY* matched, you can wire them in series and treat them as a single centre-tapped secondary.  A center-tapped secondary can provide separate positive and negative rails from a single bridge rectifier connected to its ends, with the 0V rail taken from its center tap.  This is more efficient than separate secondaries with the DC side of their bridge rectifiers placed in series, and gives a slightly higher output voltage as the center tapped arrangement only has one diode Vf drop in the feed to each rail.

[mariush]

Some audio amplifier chips can work with single power supply (a positive voltage) or with a split power supply (a positive voltage and a negative voltage). There are pros and cons to both.

A bridge rectifier converts AC voltage to a DC voltage with lots of ups and down, but basically only positive voltage referenced to the minus pin on the rectifier. Capacitors after the bridge rectifier smooth out this DC wave and try to make it as flat as possible.
So the minus on the bridge rectifier is your 0v, your ground.

If you have a transformer with a center tap (the secondary side has three wires, the center wire beeing connected to the center of the secondary winding), or you have a transformer with two separate secondary windings allowing you to connect the two windings in the middle resulting in a three wire secondary, you can use a bridge rectifier to obtain a positive voltage and a negative voltage at the same time.  The center tap becomes your 0v , your ground.  The minus of the rectifier becomes -v and the positive becomes +v

Have a look at this page, where in the Full Bridge Rectifiers you have on the right side pictures showing you such a circuit, where you get +v and -v  using a center tap transformer: http://www.play-hookey.com/ac_theory/power_supply/ps_rectifiers.html


 
Also, here's some basic formulas.
The peak voltage after it's rectified by a bridge rectifier will be equal to Vac x 1.414 (square root of 2)  - 2 x Vdiode , where Vdiode is the voltage drop of one diode in the bridge rectifier. The Vdiode varies depending on what rectifier you buy and how much current flows through the rectifier, and how hot it is... usually it's anything between 0.8v and 1.1v
It's important to keep in mind this peak dc voltage when choosing capacitors, so that you pick a high enough voltage rating. Ex. for 22v AC you have 22 x 1.414 - 2v = 31.1 -2v = ~ 29v .
Imho this is too close for comfort to pick 35v rated capacitors and I'd use 50v rated capacitors.
The capacitance after the bridge rectifier can be determined depending on how low you're willing to let the voltage go. This formula approximates  C = Current  / [ 2 x Mains Frequency x ( V dc peak - V dc minimum) ]

So for example, going with your 22v AC 400 VA transformer, you have 22v AC and 200VA for each positive and negative voltage, about 200/22 = 9.1A AC so about 0.67 x Iac = 0.67 x 9.1 =  6.A 
Assuming you'll have peak DC of 29v and you want to keep your amplifier with at least 24v all the time and you're in US where you have 60 Hz mains frequency then let's figure capacitance up to a safe 6.5A:  C = 6.5 / ( 2 x 60 x (29-24))  = 6.5  / 120 x 5 = 6.5 / 600 = 0.01083 Farads or 10840 uF  minimum ... so I guess 3 x 4700uF would do, or 4 x 3300 uF, or 5-6 x 2200uF , whatever is convenient.

ps. On the PCBs in your post, the first board it seems instead of one single bridge rectifier (four diodes in a glass passivated package) they're planning to use 4 separate diodes in TO-220 (or similar) packages and have two diodes at a time on a heatsink, probably thinking it would be easier to cool the diodes this way. Still the same thing.

As for the second board, I don't know what the guy that did that was smoking. Seems to me it's some kind of board designed to overcome some limitations regarding height. The board is basically split in the middle, and you have capacitors for the -V on one side, and capacitors for +V on the other side.
It looks like they're making rows of 4 capacitors in series, which results in less capacitance but higher voltage rating. So for example you could put 4  1000uF capacitors rated for 16v and you'd have 250uF and capable of 64v or something like that (  in series, 1/Ctotal = 1/c1 + 1/c2 +... 1/Cn)
But, they screw it all up when they have in the center only 3 capacitors because they had to make room for those holes for V- and V+ on each side sides. Seems silly to me, you could just use a single capacitor rated for higher voltage and mount it horizontally instead of having 4 capacitors of lower height?
 

[goldfinger]

Provided the two secondaries are *EXACTLY* matched, you can wire them in series and treat them as a single centre-tapped secondary.  A center-tapped secondary can provide separate positive and negative rails from a single bridge rectifier connected to its ends, with the 0V rail taken from its center tap.  This is more efficient than separate secondaries with the DC side of their bridge rectifiers placed in series, and gives a slightly higher output voltage as the center tapped arrangement only has one diode Vf drop in the feed to each rail.

Hi Ian ,

Thanks heaps for your reply appreciated. So given the lm4780 board in parallel is a dual polarity board are you saying. I check the consistency of the voltage on the two secondary's and see how closely they are matched to each other (when you say exactly matched what would the level of tolerance be 0.1-0.2 volts?)

Wire the two middle wires, leaving three wires , the middle wire going to 0v and other two going to the Ac inputs. Then on output I would take one plus and one minus to lm4780 board and ground to board. Or if I wanted to run both boards off the same PSU I would connect bother boards, alternatively use another transformer/rectifier PSU board to drive the other channel.

Is that correct?

Thanks heaps for your input



Sent from my 486 IBM WIN 3.1x Desktop...

[Ian.M]

The red board is set up the way I just described, for a center tapped secondary. 
Yes 0.2V difference on a pair of 22V secondaries should be well enough matched - that's under 1% difference.   The phasing needs to be correct, so after connecting the secondaries in series, check you've got about 44V end to end.  If its near zero instead reverse ONE of the secondaries.   

Grounding for audio systems is a whole different can of worms.  Unless all the PSUs are double-insulated , all exposed metalwork needs to be grounded for safety. However ideally you only want the 0V rail/signal ground connected to mains earth ground at ONE point in the whole system.  If its grounded in multiple locations, you can get massive hum problems due to ground loops + any difference between the ground conductor potentials at different mains sockets feeding the system.

See http://sound.westhost.com/earthing

[goldfinger]

Some audio amplifier chips can work with single power supply (a positive voltage) or with a split power supply (a positive voltage and a negative voltage). There are pros and cons to both.

A bridge rectifier converts AC voltage to a DC voltage with lots of ups and down, but basically only positive voltage referenced to the minus pin on the rectifier. Capacitors after the bridge rectifier smooth out this DC wave and try to make it as flat as possible.
So the minus on the bridge rectifier is your 0v, your ground.

If you have a transformer with a center tap (the secondary side has three wires, the center wire beeing connected to the center of the secondary winding), or you have a transformer with two separate secondary windings allowing you to connect the two windings in the middle resulting in a three wire secondary, you can use a bridge rectifier to obtain a positive voltage and a negative voltage at the same time.  The center tap becomes your 0v , your ground.  The minus of the rectifier becomes -v and the positive becomes +v

Have a look at this page, where in the Full Bridge Rectifiers you have on the right side pictures showing you such a circuit, where you get +v and -v  using a center tap transformer: http://www.play-hookey.com/ac_theory/power_supply/ps_rectifiers.html


 
Also, here's some basic formulas.
The peak voltage after it's rectified by a bridge rectifier will be equal to Vac x 1.414 (square root of 2)  - 2 x Vdiode , where Vdiode is the voltage drop of one diode in the bridge rectifier. The Vdiode varies depending on what rectifier you buy and how much current flows through the rectifier, and how hot it is... usually it's anything between 0.8v and 1.1v
It's important to keep in mind this peak dc voltage when choosing capacitors, so that you pick a high enough voltage rating. Ex. for 22v AC you have 22 x 1.414 - 2v = 31.1 -2v = ~ 29v .
Imho this is too close for comfort to pick 35v rated capacitors and I'd use 50v rated capacitors.
The capacitance after the bridge rectifier can be determined depending on how low you're willing to let the voltage go. This formula approximates  C = Current  / [ 2 x Mains Frequency x ( V dc peak - V dc minimum) ]

So for example, going with your 22v AC 400 VA transformer, you have 22v AC and 200VA for each positive and negative voltage, about 200/22 = 9.1A AC so about 0.67 x Iac = 0.67 x 9.1 =  6.A 
Assuming you'll have peak DC of 29v and you want to keep your amplifier with at least 24v all the time and you're in US where you have 60 Hz mains frequency then let's figure capacitance up to a safe 6.5A:  C = 6.5 / ( 2 x 60 x (29-24))  = 6.5  / 120 x 5 = 6.5 / 600 = 0.01083 Farads or 10840 uF  minimum ... so I guess 3 x 4700uF would do, or 4 x 3300 uF, or 5-6 x 2200uF , whatever is convenient.

ps. On the PCBs in your post, the first board it seems instead of one single bridge rectifier (four diodes in a glass passivated package) they're planning to use 4 separate diodes in TO-220 (or similar) packages and have two diodes at a time on a heatsink, probably thinking it would be easier to cool the diodes this way. Still the same thing.

As for the second board, I don't know what the guy that did that was smoking. Seems to me it's some kind of board designed to overcome some limitations regarding height. The board is basically split in the middle, and you have capacitors for the -V on one side, and capacitors for +V on the other side.
It looks like they're making rows of 4 capacitors in series, which results in less capacitance but higher voltage rating. So for example you could put 4  1000uF capacitors rated for 16v and you'd have 250uF and capable of 64v or something like that (  in series, 1/Ctotal = 1/c1 + 1/c2 +... 1/Cn)
But, they screw it all up when they have in the center only 3 capacitors because they had to make room for those holes for V- and V+ on each side sides. Seems silly to me, you could just use a single capacitor rated for higher voltage and mount it horizontally instead of having 4 capacitors of lower height?

Hey Mari

Thanks so much for your detail reply also. Yeh my transformers don't have centre tap and I think that's where my main confusion has come from (well that and I'm very new to electronics!)

Both the transformers I have 2 and 4 wires for 1x and 2x secondary.

The amp runs in range of 26-35v DC so I thought that 22vac would be around the right amount

As for the large board "the guy smoking something " it was advertised as 100,000uf and size of the lead and cap spot seems to be in line with 220uf to 1000uf caps 35-50v.

I didn't think it was in series but maybe it is???


The other board I have which made more sense to me based on having no centre tap is this one attached.



With this board I'm assuming I just put my two secondaries straight onto the board?


A couple other questions also (sorry if I'm a pain)


- what's the pro/con of using small amount of large caps vs larger amount of smaller? Eg, 25x1000uf vs say 3x8000uf - let's say voltage and temp stats are comparable? Esr? Ripple handling?

- ripple- I'm assuming generally ripple rating isn't too much of a worry providing you have enough capacitance to smooth it out. Is ripple rating accumulative eg if ripple rating on cap is 1v and you have 10 it would have 10v rating?

Should I consider some sort of cap for noise filtering? How do you determine best value?


Thanks heaps guys your awesome!

[goldfinger]

The red board is set up the way I just described, for a center tapped secondary. 
Yes 0.2V difference on a pair of 22V secondaries should be well enough matched - that's under 1% difference.   The phasing needs to be correct, so after connecting the secondaries in series, check you've got about 44V end to end.  If its near zero instead reverse ONE of the secondaries.   

Grounding for audio systems is a whole different can of worms.  Unless all the PSUs are double-insulated , all exposed metalwork needs to be grounded for safety. However ideally you only want the 0V rail/signal ground connected to mains earth ground at ONE point in the whole system.  If its grounded in multiple locations, you can get massive hum problems due to ground loops + any difference between the ground conductor potentials at different mains sockets feeding the system.

See http://sound.westhost.com/earthing

Thanks mate appreciated so I'll just use dmm on the two remaining secondary wires (outer wires) and check they are close enough to 44vac.

I'm assuming that because it's a +\- that end result will be roughly 29vdc at the amp? For each side ?




Sent from my 486 IBM WIN 3.1x Desktop...

[goldfinger]

The red board is set up the way I just described, for a center tapped secondary. 
Yes 0.2V difference on a pair of 22V secondaries should be well enough matched - that's under 1% difference.   The phasing needs to be correct, so after connecting the secondaries in series, check you've got about 44V end to end.  If its near zero instead reverse ONE of the secondaries.   

Grounding for audio systems is a whole different can of worms.  Unless all the PSUs are double-insulated , all exposed metalwork needs to be grounded for safety. However ideally you only want the 0V rail/signal ground connected to mains earth ground at ONE point in the whole system.  If its grounded in multiple locations, you can get massive hum problems due to ground loops + any difference between the ground conductor potentials at different mains sockets feeding the system.

See http://sound.westhost.com/earthing

I did see another PSU design that had the 0v going through another rectify bridge then down to the IEC chassis ground is the purpose of that to convert the Ac to DC before grounding eg less dangerous ?


Sent from my 486 IBM WIN 3.1x Desktop...

[Ian.M]

No.  a bridge rectifier wired as in fig 4 of the page I linked, acts as a high current pair of anti-parallel diodes, and prevents high currents flowing unless the voltage driving them exceeeds the diode Vf.   Similarly if you short + to - on the bridge rectifier and connect supply ground to one AC terminal and circuit 0V to the other, it will block up to 2x Vf, easily breaking unwanted ground loops.   

[goldfinger]

The red board is set up the way I just described, for a center tapped secondary. 
Yes 0.2V difference on a pair of 22V secondaries should be well enough matched - that's under 1% difference.   The phasing needs to be correct, so after connecting the secondaries in series, check you've got about 44V end to end.  If its near zero instead reverse ONE of the secondaries.   

Grounding for audio systems is a whole different can of worms.  Unless all the PSUs are double-insulated , all exposed metalwork needs to be grounded for safety. However ideally you only want the 0V rail/signal ground connected to mains earth ground at ONE point in the whole system.  If its grounded in multiple locations, you can get massive hum problems due to ground loops + any difference between the ground conductor potentials at different mains sockets feeding the system.

See http://sound.westhost.com/earthing

Hi Ian,

Thanks for all your feedback, so if combining the 22v dual secondaries, to make 44v vac (approx 60vdc) that would be fine as it would feed approx 30vdc to each rail ? (Amp is rated up to 35vdc)


Sent from my 486 IBM WIN 3.1x Desktop...

[Ian.M]

Thanks for all your feedback, so if combining the 22v dual secondaries, to make 44v vac (approx 60vdc) that would be fine as it would feed approx 30vdc to each rail ? (Amp is rated up to 35vdc)

That should be OK as long as the secondaries actually output 22V unloaded.  If they are significantly higer you wont have much margin between the output voltage and your amp's rated operating voltage. 

However as the LM4780 itself is rated for up to +/- 42V rails, if its over +/-35V out unloaded we'll need to check a schematic for the amp board to see if it can safely withstand a moderate overvoltage at low volume levels. (it will certainly drop on load, probably to within its nominal ratings.

The next issue you need to get right is fusing . . . .