# Problems with a simple power supply

#### JohnP

Oct 22, 2016
31

I have a toroidal transformer with 130v 0v 130v 0V input and 18v 0V 18V 0V output. I have wired the input as 230V. The output is wired to give 18V using both windings. Each single ouput winding is rated at 4.44A and 2 should be 8.88A.

I have used a 10,000 uF cap rated at 63V. The bridge rectifier started at a 10A but I have also tried a 35A and a 50A. The cap has also bee replaced by a 820 uF 200V.

None of this is helping. The no load output from the AC is 19V at all times which is correct because our mains is 245V and not 230V. This should give a DC output of 27V under load. (19V x 1.4141).

A load of 5A ( a car head lamp (60W + 65W) wired to use both filaments is series) drops the DC output to 16V but the AC transformer output remains at 19V. The input to the transformer stays at 245V under load.

I thought that a load of 5A would not have dropped the ouput voltage down to 16V.

I have tried 2 different meters.

I have made 2 other DC supplies with no problems. I am making a simple error but I have no idea what it is.

#### AnalogKid

Jun 10, 2015
2,811
The no load output from the AC is 19V at all times which is correct because our mains is 245V and not 230V. This should give a DC output of 27V under load. (19V x 1.4141).
No, it shouldn't.

27 V is the ***no-load*** peak voltage of the output DC ripple. Under load that peak voltage will sag.

The DC energy out of the supply cannot be greater than the RMS AC power out of the transformer. 19 Vacrms x 5 A = 95 W. 27 Vdc x 5 A = 135 W. That extra 40 W has to come from somewhere.

Some of it can come from the transformer IF it is oversized for the job and you have a much larger filter capacitor. The capacitor is "topped off" by the peak voltage of the transformer secondary, and discharged by the load. The larger the capacitor, the less it sage under load between AC peaks. The extra power comes from the transformer. When it is charging up the cap, its secondary current is much greater than 5 A; when the cap is discharging but still higher voltage than the instantaneous output voltage of the transformer, the transformer secondary current is near zero. Thus, the secondary current is more a series of pulses than a sinewave. Not that this works only if the transformer core can support such high peak current pulses - they can be 10 x the RMS value.

When you push the transformers or capacitors to their limits, linear power supplies are more complex than most people realize. Old timers have nomographs that relate these various aspects of this type of design.

ak

#### Audioguru

Sep 24, 2016
3,650
The peak of the 19VAC is 26.9V but the fullwave rectifier drops the DC output to 24.9V if the capacitance is very high. But since the capacitor is only 820uF then the massive current of 5A is causing high ripple which reduces the average DC voltage. The light bulbs reduce their resistance when the voltage is too low which causes a higher current.

If it also dropped the output voltage when the capacitor was 10,000uF then the capacitor was bad or was a fake.

#### AnalogKid

Jun 10, 2015
2,811
If it also dropped the output voltage when the capacitor was 10,000uF then the capacitor was bad or was a fake.
Without getting into all of that natural logarithm goop, lets use the first order approximation of a constant discharge current.
EC=IT
E x C = i x t
Voltage change x capacitor value = discharge current x time
t = 0.08333 s, 1/120 Hz (full wave rectified AC frequency)
i = 5 A
C = 0.01 F (10,000 uF)

E = (i x t) / C = 4.1667 V

So even if the 10,000 uF capacitor is in perfect health, 4 V of ripple is enough to cause a lower reading on a DVM.

ak

#### JohnP

Oct 22, 2016
31
The peak of the 19VAC is 26.9V but the fullwave rectifier drops the DC output to 24.9V if the capacitance is very high. But since the capacitor is only 820uF then the massive current of 5A is causing high ripple which reduces the average DC voltage. The light bulbs reduce their resistance when the voltage is too low which causes a higher current.

If it also dropped the output voltage when the capacitor was 10,000uF then the capacitor was bad or was a fake.

Thanks for the reply, I can understand it. I have used the same capacitor when using the same transformer on outputting 38V through a bridge at 4A and only lost 2V, that was almost at the limit of the transformer. I didn't expect to loose over 10V when running it at 19V at 5A. I have to say though it's a chinese job but then again most are. I have quite a stack of caps so I'll set up a few in parallel. The voltage was within 0.1V on both capacitors. 16.0V on the 820uF and 16.1V on the 10,000uF. I'll be honest. A few days ago I stuck a screwdriver across it to make sure it was discharged. By gawd it gave out a crack, might have damaged it. I'll let you know with different caps. Thanks again.

#### JohnP

Oct 22, 2016
31
No, it shouldn't.

27 V is the ***no-load*** peak voltage of the output DC ripple. Under load that peak voltage will sag.

The DC energy out of the supply cannot be greater than the RMS AC power out of the transformer. 19 Vacrms x 5 A = 95 W. 27 Vdc x 5 A = 135 W. That extra 40 W has to come from somewhere.

Some of it can come from the transformer IF it is oversized for the job and you have a much larger filter capacitor. The capacitor is "topped off" by the peak voltage of the transformer secondary, and discharged by the load. The larger the capacitor, the less it sage under load between AC peaks. The extra power comes from the transformer. When it is charging up the cap, its secondary current is much greater than 5 A; when the cap is discharging but still higher voltage than the instantaneous output voltage of the transformer, the transformer secondary current is near zero. Thus, the secondary current is more a series of pulses than a sinewave. Not that this works only if the transformer core can support such high peak current pulses - they can be 10 x the RMS value.

When you push the transformers or capacitors to their limits, linear power supplies are more complex than most people realize. Old timers have nomographs that relate these various aspects of this type of design.

ak
Thank you very much for the reply, it's good of you to take the time.
I thought that a load of 5A on a transformer rated at nearly 9A was not pushing the transformer or capacitor to it's limits. I used the same setup before using the '36V' from the transformer at 4A and only lost 2V not 10V. As I replied to Audioguru I may have damaged the cap discharging it with a screwdriver. I'll try using different ones and see what happens. I expect the V to drop but 10V seems an awful lot.

#### JohnP

Oct 22, 2016
31
I forgot to add, I'm in the UK so my 't' is a bit orf as we say.

Last edited by a moderator:

#### JohnP

Oct 22, 2016
31
I think I should explain. I thought I would have a play with some Class D amps from Ebay. Just under 50 years ago I made a sterio pair of Mullard 20W (EL34's) into 15 Ohm amps 0.05% THD and an RIAA. etc. pre amp to go with them. I was and still am a complete novice. I had a great deal of help back in those days but now I have you chaps to take pity on me.

Strangely enough not only am I very rusty but I don't have a clue really. Things have changed. I really need a DC supply of 24V but the TPA3116 has a max of 26V. There is the TDA7498 amp that can run up to 30V. These amps can use over 4A but I doubt I will ever use more than 4W through the 15" Tanny Monitor Golds I have. How I am going to drop the 3V or so for the 24V ones I have no idea, other than a string of diodes or an SMPS (buck).

Sorry I know I am going off thread but it might make a difference to suggestions. I am a bit new to Forums.

Last edited by a moderator:

#### Audioguru

Sep 24, 2016
3,650
Unregulated 24V is too close to the recommended maximum of 26V. The datasheet shows plenty of power with a 21V supply.

#### JohnP

Oct 22, 2016
31
Without getting into all of that natural logarithm goop, lets use the first order approximation of a constant discharge current.
EC=IT
E x C = i x t
Voltage change x capacitor value = discharge current x time
t = 0.08333 s, 1/120 Hz (full wave rectified AC frequency)
i = 5 A
C = 0.01 F (10,000 uF)

E = (i x t) / C = 4.1667 V

So even if the 10,000 uF capacitor is in perfect health, 4 V of ripple is enough to cause a lower reading on a DVM.

ak
At 50 Hz - mine would be 5V - please tell me if I'm wrong. I think you meant 0.008333 s, 1/120 Hz......

#### JohnP

Oct 22, 2016
31
I guess it's a buck convertoron the end of it, they seem very popular with these D Class but I wasn't sure if noise might get through. I don't know anything about them much.

http://www.ebay.co.uk/itm/DC-300W-2...260381&hash=item212b9e684c:g:uI4AAOSwvv9ZgbdA

This should be OK :-

http://www.ebay.co.uk/itm/DC-DC-CC-...910186&hash=item3af0c3a4e0:g:zBoAAOSwt05ZtoNM

Thanks for your help - again. I see your on Electro Tech Online. Spec helped me a lot and then vanished. He had had an accident but I thought he had got over it. I check now and then but nothing since March, any ideas ?

#### AnalogKid

Jun 10, 2015
2,811
At 50 Hz - mine would be 5V - please tell me if I'm wrong. I think you meant 0.008333 s, 1/120 Hz......
Correct on both, dropped 0 typo.

ak

#### JohnP

Oct 22, 2016
31
Don't worry, at my age I drop a lot more than that. I do have a few more questions to add to this thread but my granddaughter has just fallen out of a tree and is in hospital. Looks like we are off to the other side of London. I do have a new laptop but it's Windows10 and I'm XP. Either way I'll be back in a week - or less.

#### JohnP

Oct 22, 2016
31
Unregulated 24V is too close to the recommended maximum of 26V. The datasheet shows plenty of power with a 21V supply.
Thanks for the reply, sorry I'm late but I have been away.
You are perfectly correct. I have a pair of 15" Tanny Monitor Gold that are about late 60's vintage. That's the days when efficiency really counted. I would annoy the neighbors at 5W a channel. During the Queen's jubilee we had a complaint from 50 yards away. The door was open though. Ornaments start walking at 40W, we've grown old together. I'm getting a 300W 20A (?) buck converter, that's what is common on these setups. 20V will do nicely.
I'm also looking at the TDA7498 amps which have a max of 39V so this 27V supply should be OK with them.

#### JohnP

Oct 22, 2016
31
Without getting into all of that natural logarithm goop, lets use the first order approximation of a constant discharge current.
EC=IT
E x C = i x t
Voltage change x capacitor value = discharge current x time
t = 0.08333 s, 1/120 Hz (full wave rectified AC frequency)
i = 5 A
C = 0.01 F (10,000 uF)

E = (i x t) / C = 4.1667 V

So even if the 10,000 uF capacitor is in perfect health, 4 V of ripple is enough to cause a lower reading on a DVM.

ak
I have spent the last week trying to get my head around what capacitors I need for this power supply. It seems a very confusing area. I came to a very rough conclusion that 10,000uF per amp should be OK.. Do you think 4 x 20,000 uF (10,000uF / A) caps would be OK ?

I also have a Monicor 120W line amp (deceased) that I'm using as another power supply project. It's 40V DC at about 7A. (data sheets are not available but the transformer is rather large as are the output windings) It has 4 x 4,700 caps on the power supply. At 120W 40V it's 3A that's just over 6,000uF per.A. They were not exactly high fi as the used another transformer to provide 100V for the line speakers. Would it be worth while adding a few more uF?
I'm converting it into a linear power supply. Almost all the bits are there.4 x TIP35C but I could adapt the heatsink for 2N3055's. The LM-723 for a regulator. I don't need 10A , I think 30W per transistor is quite enough even with a couple of small fans added. (Sorry but I'm a fan addict, i do know a bit about cooling - makes a change)

http://electronics-diy.com/30v-10a-variable-bench-power-supply.php.

No doubt there are better circuits but this seems a simple one to sart with.
Comments on this would be very welcome - should I have started a new thread ?

#### JohnP

Oct 22, 2016
31
Thanks for the reply, sorry I'm late but I have been away.
You are perfectly correct. I have a pair of 15" Tanny Monitor Gold that are about late 60's vintage. That's the days when efficiency really counted. I would annoy the neighbors at 5W a channel. During the Queen's jubilee we had a complaint from 50 yards away. The door was open though. Ornaments start walking at 40W, we've grown old together. I'm getting a 300W 20A (?) buck converter, that's what is common on these setups. 20V will do nicely.
I'm also looking at the TDA7498 amps which have a max of 39V so this 27V supply should be OK with them.

I think you mean 0.0083333 (never mind)
As mine is an easy 0.01 (50Hz) If I use 20,000uF caps I'll have a drop of 2V from the transformer DC supply. I'll also be using SMPS with the TPA 3116 amps. I'll run at 20V thats gives me enough. The TDA 7498 (36V max) I think should be OK at 27V and a 2V drop flat out - I'll never get near that. End terrace 11" solid wall.

One last point here. I have just read this :-
http://www.ti.com/lit/an/sboa020/sboa020.pdf
It states that using mica for TO-220 and T-03, etc is not recommended due to it being brittle.
Further down in a table it states that it has better thermal transfer.
I have used a few in the past with no problems.

Replies
4
Views
1K
Replies
17
Views
2K
Replies
6
Views
218
Replies
8
Views
1K
Replies
14
Views
327