Old and basic linear power supply

[MarkF]

@MarkF
Thanks. Are the TIP3055 and 2N2905 in your drawing for dropping the voltage to zero in case the regulation opens?

No.  The transistors are to bypass the regulator and provide higher current.

I was only pointing out the pot connections. 
If the wiper opens, the pot will pull the ADJ pin low and lower the output voltage since there is no path to the output pin.

[vk6zgo]

Oh! You've got the "Luxury" model!
Mine doesn't have a meter. 

[vk6zgo]

By the way, PP9 is just another number for the same thing as a 276.

It seems "Ever Ready" in the UK stuck to the same numbers as "Eveready" in Aust (& the USA?), although it was no longer the same company.

in Oz, Eveready generally used a totally numeric system, though they "fiddled"with it a bit.

The zinc carbon "D" cells were "950s" if zinc-carbon, & "D50" if alkaline.
"AAs" could be "915s" for penlights, or "1015s" for use in Electronics".

Their competitor, "Diamond" used the same numbers, but prefixed with "3".

Japanese companies used another numbering system & Taiwanese ones, yet another.

[bdunham7]

it's a vintage Signal generator. Generates up to 1Khz sines and square. The system is already a half wave setup and I have 1V ripple on the rails already.

As mentioned before, it seems the PSU is replacing 4 batteries which I am not familiar with - PP9? which seem to be 9V each so 18V both. So maybe an 18V regulator could work. I see rechargeable PP9s are 8.4Vx2 = 16.8. To be honest I feel that 18V could be ok.

yesterday after I posted my sketch I realised that I don't think it's going to work. The way it's configured won't give me a positive and negative output from the bridge, am I right? If I am not mistaken I would need to ground the "central" ends of each 15V winding?

Or, as suggested by someone, I could just keep the half wave configuration - which is what I have now.

The circuit as drawn by MarkF, in the post where he points out that you have put LM317 in the negative rail instead of LM337, is your best option--and should be dirt cheap to implement, adding the current limiting resistors I mentioned if you want.  One big advantage of the LM317/LM337 combo is that if your voltage choice isn't perfect, you can easily change your mind later.  In fact, for sensitive split-rail systems that rely in one way or another on exactly equal positive and negative voltages, they can easily be 'shimmed' up or down by bodging on higher value resistors in parallel to one or the other divider resistor.

I had suggested retaining the half wave configuration under the mistaken assumption that you had two separate transformers.  Since that's not the case, there's no reason to retain that system.   I would point out that the capacitor values might actually be a bit excessive at your current levels, which is OK if you have them on hand.  Otherwise you can likely decrease them each by a factor of 5 and still be OK.

[SilverSolder]

Here are a couple of ideas for fixed regulators:

Fixed positive 18V regulator, 2V dropout voltage. ~$0.50 qty 1
www.digikey.com/product-detail/en/MC78M18CTG/MC78M18CTG-ND/1481494

Fixed negative 18V, 1.3V dropout, ~$0.50 qty 1
https://www.digikey.com/product-detail/en/MC7918CTG/MC7918CTGOS-ND/1481547


These are just examples of what appears to be available.  You'd need to read through the specs to see if there are any "gotchas"!

Using fixed regulators would keep the "spirit of simplicity" of the original design, but using somewhat more modern components.

[MarkF]

One thing that caught me was the tab gauge. 
Carefully read the datasheets when picking the TO-220 style regulators.
The single gauge tabs are very thin and you need to be careful when mounting them.

In your case with a low current draw, the heatsink isn't a real big deal.

[vk6zgo]

Sorry, I missed your original pix of the PSU, so forget my comments about being "home brew".
 
I think it is an earlier design than the 1970s, maybe early to mid 1960s, as single device regulators were m
fairly widely available by the beginning of the 1970s.
I wonder if they used the rather "bodged" system  because,  just using the transformer windings for "standalone" pos & neg supplies did not allow enougn "headroom" for those older regulators.

You may need to keep that in mind if specifying a regulator.

[vk6zgo]

But wait! There's more!
https://worldradiohistory.com/hd2/IDX-Site-Early-Radio/Archive-Wireless-World-IDX/70s/Wireless-World-1971-08-OCR-Page-0003.pdf

This gives you a bit of info about the capabilities of your generator.
It doesn't "top out" at 1kHz as you supposed, but goes out to 1MHz.

Over on the "vintage-radio.net forum", there are several interesting threads on the Levell TG200D series, probably the secnd one being the most useful, as it is about the PSU.

https://www.vintage-radio.net/forum/showthread.php?t=159533
https://www.vintage-radio.net/forum/showthread.php?t=143263
They are all "mad as hatters" on that forum, but are a good (& very knowledgable) crowd, & I strongly
recommend it.

Googling also brought up the "Radio Museum" site which, apart from the one we are looking for has a list
of other Levell devices.
One, in particular, took my eye, which was an RF voltmeter which went up to 410MHz.
That's  HP, Marconi & Rohde & Schwarz territory!

It seems the little Levell instruments, although their appearance seems to shout "hobbyist level", definitely
"punch above their weight" when it comes to performance!

[tony359]

1Mhz! I'll check again but I maxed out the frequency selector and I could read 1Khz - maybe I just missed a step!

Interesting, to be honest I picked that up at my workplace some time ago and never thought it would be usable or useful in any way. I'll check out the other thread, thanks!

Edit: I think I just mis-read the frequency value when I tested last time! Yes, 1Mhz! Wow!

[tony359]

@MarkF

I missed your message when you pointed out the 317 on the negative rail: indeed that was just a typo. I intended to change the label but I must have missed it! Thanks for mentioning.

I guess that 2x18V regulators will be perfectly fine so I don't have to worry about the regulation. I am actually impressed that my schematic was not just a bunch of nonsense  and it only needed minor changes (besides the 317 in the wrong place but that again was not intentional).

Now a question before I struggle more with drawing a circuit: what do you use to make yours? I have tried a couple online but the first would not allow me to save, the second is missing some basic components!

While I find a better software, I guess I can go with what @MarkF posted based on my first sketch (thanks) and
1. Replace the LM317 with 7818 and 7918
2. Add 75R on both positive and negative end of transformers as current limiters as @bdunam7 suggested - May I ask what is the math behind this selection? If I consider 18VDC and 75R I get 240mA and 4.32W.

Regarding the regulators, I was let to believe many years ago by an engineer friend of mine that the heat dissipated by a voltage regulator depends on the supply voltage. That is, if I supply a 7805 with 40V I have a much higher dissipation than if I supply it with 10V. That does not seem to be correct and the dissipation is only linked to the current that passes through it - I grabbed a 7805 from my storage and connected it to my bench PSU: indeed with no load the 7805 would just stay cold up to the max voltage I can supply of 15V.

(Hence the 7815 and 7915 that are screaming hot in my Behringer active speakers are like that because of the current drawn by the circuit and not because the supply voltage is 41V! I have been misled!! )

Thanks again all!

[bdunham7]

I guess that 2x18V regulators will be perfectly fine so I don't have to worry about the regulation.

...but if they aren't, you can't adjust them! 

2. Add 75R on both positive and negative end of transformers as current limiters as @bdunam7 suggested - May I ask what is the math behind this selection? If I consider 18VDC and 75R I get 240mA and 4.32W.

15VAC and 75R = 200mA, 15VAC x 200mA = 3VA and 3W into a purely resistive load.  Thus your worst case scenario, a dead short, puts the transformer and resistor right at their max rating. 

Regarding the regulators, I was let to believe many years ago by an engineer friend of mine that the heat dissipated by a voltage regulator depends on the supply voltage. That is, if I supply a 7805 with 40V I have a much higher dissipation than if I supply it with 10V. That does not seem to be correct and the dissipation is only linked to the current that passes through it - I grabbed a 7805 from my storage and connected it to my bench PSU: indeed with no load the 7805 would just stay cold up to the max voltage I can supply of 15V.

(Hence the 7815 and 7915 that are screaming hot in my Behringer active speakers are like that because of the current drawn by the circuit and not because the supply voltage is 41V! I have been misled!! )

Both the supply voltage and the current matter.  The power dissipation of the regulator will be the product of the voltage drop and the output current, including the voltage divider section whether it is internal or external, plus whatever the regulator dissipates due to internal operation--which is quite small.  So a 7805 delivering 1A from a 10V supply will dissipate approximately 5 watts, while a similar device delivering 500mA from a 15V supply will also dissipate approximately 5 watts. 

[tony359]

I guess that 2x18V regulators will be perfectly fine so I don't have to worry about the regulation.

...but if they aren't, you can't adjust them! 

Honestly I feel it's ok - that unit used to run on 9V batteries.

2. Add 75R on both positive and negative end of transformers as current limiters as @bdunam7 suggested - May I ask what is the math behind this selection? If I consider 18VDC and 75R I get 240mA and 4.32W.

15VAC and 75R = 200mA, 15VAC x 200mA = 3VA and 3W into a purely resistive load.  Thus your worst case scenario, a dead short, puts the transformer and resistor right at their max rating. 

Right, I had forgotten the PSU is 15V!

Both the supply voltage and the current matter.  The power dissipation of the regulator will be the product of the voltage drop and the output current, including the voltage divider section whether it is internal or external, plus whatever the regulator dissipates due to internal operation--which is quite small.  So a 7805 delivering 1A from a 10V supply will dissipate approximately 5 watts, while a similar device delivering 500mA from a 15V supply will also dissipate approximately 5 watts.

Ah, ok. Then I was not misled, thanks for clarifying!

Now, before I attempt to calculate the power dissipation for the regulators, as @silversolder suggested, am I correct in assuming that the output of the bridge is going to be +/-19.8VDC?

[bdunham7]

Now, before I attempt to calculate the power dissipation for the regulators, as @silversolder suggested, am I correct in assuming that the output of the bridge is going to be +/-19.8VDC?

Measure the open circuit AC voltage of one primary, it will probably be higher than 15VAC by a surprising amount.  Take that number multiplied by 1.4 and you'll have your approximate OC voltage.  Yes, it's 1.4142818... and there's a diode drop in the bridge, etc, but that is all nothing to worry about.  Then subtract your output voltage (18) from that number and multiply by your current, which if IIRC was .012A, plus the quiescent current of the regulator due to its internal voltage divider if the datasheet shows that it is significant enough to bother with. 

[tony359]

thanks @bdunham7

I've experimented a bit with a bridge rectifier, a load and a capacitor. Very interesting session!

I have found that with a load and a capacitor the actual voltage out of the bridge is +/- 25.07VDC. Which looks perfect for a 7x18 - I thought I would end up with not enough voltage to be honest.

The voltage out of the transformer with no load is 23.2VAC which after the bridge turns into 31.2VDC (I measured with just a capacitor in) and it's consistent with the math which gives me 32.7VDC.

If my regulated output voltage is 18VDC, that gives me 13.2V for the regulator to drop. Multiplied for the max current of... 200mA I guess (this is the short circuit current) plus 3.5mA as quiescent current, that makes for 2.68W to dissipate. 

I've found this: https://www.ti.com/lit/an/slva118a/slva118a.pdf
And the 7xxx datasheet: http://www.farnell.com/datasheets/1878412.pdf

The datasheet I found does not give a PDmax - as it says the component is internally limited. But it gives a TETA-JAmax of 65 for the TO-221 package. My calculation (page 12 of the document I found) says that my Teta-JAmax is = (Tj-Ta)/PDmax = (150-20)/2.6862 = 48.39 so it looks like the regulator won't reach the 150 degrees temperature even in a short circuit without a heatsink.

THAT SAID, I may attach them to the case just in case, nobody wants a regulator to go up to 100 degrees I guess!

Is the above correct?

If so, can @MarkF please confirm the values of the capacitors he recommended? I have 100mF and 1000mF - is that millifarad?

[bdunham7]

The voltage out of the transformer with no load is 23.2VAC which after the bridge turns into 31.2VDC (I measured with just a capacitor in) and it's consistent with the math which gives me 32.7VDC.

If my regulated output voltage is 18VDC, that gives me 13.2V for the regulator to drop. Multiplied for the max current of... 200mA I guess (this is the short circuit current) plus 3.5mA as quiescent current, that makes for 2.68W to dissipate.

You're overestimating the power because the voltage will drop as the load increases.  You didn't say what load you were using, but if it is close to your real-life 12mA load, look at how much the voltage already dropped.  At 200mA, it will likely drop into the 15VAC range. In addition, this is why I was recommending a 75R resistor on each primary--they'll both limit the short circuit current and dissipate most of the power.

As for capacitors, IIRC you had those in your original drawing and both MarkF and I recommended you swap them, putting the larger in front of the regulators.  I also pointed out that they were significantly larger than needed--in microfarads.  Millifarads would be ludicrously huge for this.  You can use 220uF for the ones right after the bridge and 22uF after the regulators and that's still likely more than needed.

[tony359]

You're overestimating the power because the voltage will drop as the load increases.  You didn't say what load you were using, but if it is close to your real-life 12mA load, look at how much the voltage already dropped.  At 200mA, it will likely drop into the 15VAC range. In addition, this is why I was recommending a 75R resistor on each primary--they'll both limit the short circuit current and dissipate most of the power.

I thought what I was calculating was the short circuit power dissipation - which is the worst case scenario. If I don't need a heatsink in that case, then I am safe. The test with the load was just to have an idea of what voltage I was getting out of the bridge. That was a 170 Ohm resistor if I remember correctly (7x1.2K in parallel). At 25VDC that gives me 147mA which is way more than I would normally use. I did not check what the AC voltage was then but I can do that.

Am I following your trail of thought or am I completely off track?

As for capacitors, IIRC you had those in your original drawing and both MarkF and I recommended you swap them, putting the larger in front of the regulators.  I also pointed out that they were significantly larger than needed--in microfarads.  Millifarads would be ludicrously huge for this.  You can use 220uF for the ones right after the bridge and 22uF after the regulators and that's still likely more than needed.

That's why I was confused! I thought "mF" stood for millifarad! I have a selection of capacitors, I'm sure I can find something suitable, thanks for clarifying.

[MarkF]

Since you are not building a 'generic use' power supply and your current requirements are very low, I would take a minimalist approach.  You don't need a bunch of reverse polarity connection protection, etc.  In addition, I would not worry about inrush current limiting (unless you are going to use very large capacitors.  You needed to worry about current limiting in the original design because of the zener diodes.  Here the regulators do have some limited overcurrent protection builtin.

I redrew the from an earlier post of my with what I would do.
Since you already have some capacitors, I would use 1000μF-2200μF on the regulator inputs and 10μF-47μF on the output.

One thing that has not been mentioned is that your current requirement is close to the minimum for the MC78xx and MC79xx series regulators.  Therefore, I would add power ON indicator LEDs on the output set for a 15mA-20mA current draw to insure you always have the minimum.



You ask about software tools.  I use DipTrace for schematics and PCB layouts.

[tony359]

Thanks @MarkF

I did not notice the 7x18 had a minimum current requirement - good idea for the LEDs, thanks!

And really thank you for the software suggestion, I'll give that a go asap!

[MarkF]

Thanks @MarkF

I did not notice the 7x18 had a minimum current requirement - good idea for the LEDs, thanks!

And really thank you for the software suggestion, I'll give that a go asap!

They don't come right out and specify a minimum current. 
They hit at it when specifying the conditions under which their data is given.

But, I've had problems with ripple in no-load or small load cases.
Some of it depends on manufacturer too.  You may not have a manufacturer choice for 18V regulators.
The 5V, 12V and 15V regulators are more common.

And both anyone asks...  I only buy parts from Mouser and Digikey.

[tony359]

I used to buy from CPC or RS but RS has quite high minimum quantities to purchase for each item. So I recently switched to Farnell where you can buy a single component - for a slight higher price. But I still need to buy at least £20 to have free shipping. So for less important projects I rely on - cough cough - Ebay
I try to stick to good sellers when I can. When the project is really not very important I purchase from China and cross my fingers! I have to say that so far I've been pleased!

I've ordered regulators and resistors. I will fish a suitable capacitor within my stock. I shall keep you posted with the outcome.

For now, thank you very much for being so patient with me, I will try to retain what you taught me! 

[SilverSolder]

[...]
You ask about software tools.  I use DipTrace for schematics and PCB layouts.

Not wanting to divert the thread too much, but  DipTrace looks pretty cool for hobby use.  I use an ancient unsupported product and will sooner or later have to move to something else...   DipTrace seems pretty flexible with what it lets you do?

[SilverSolder]

I used to buy from CPC or RS but RS has quite high minimum quantities to purchase for each item. So I recently switched to Farnell where you can buy a single component - for a slight higher price. But I still need to buy at least £20 to have free shipping. So for less important projects I rely on - cough cough - Ebay
I try to stick to good sellers when I can. When the project is really not very important I purchase from China and cross my fingers! I have to say that so far I've been pleased!

I've ordered regulators and resistors. I will fish a suitable capacitor within my stock. I shall keep you posted with the outcome.

For now, thank you very much for being so patient with me, I will try to retain what you taught me! 

I use eBay a lot for small quantities as well.  -  Digikey is flexible enough to use even for small quantities (they will even mail it as a USPS small package for not much $) and I think there must be a depot near where I live, because it usually arrives in 1-2 days.

[tony359]

If I am over £20 I am going with Farnell hands down. I get everything via DHL for free (*) next working day and I know what I am buying. But if I need 2 capacitors and I am not in a rush, then Ebay is the key

(*) well, everything is more expensive there in small quantities hence the "free" shipping

[tony359]

Hello all

Update: I have received all the components and have rebuilt the PSU! It works like a charm. The regulators get lukewarm with 200mA load so with normal use I am very safe. Under short circuit the 75 Ohm resistors get very hot as expected but regulators stay cool.

Not the most tidy circuit I guess - I have used a board I had already which features "linked" rows and that did not help in making things tidy. But I didn't want to invest any more money on that thing!

The circuit didn't work at the beginning, -15V would start ok and then drop to zero after a few seconds. It turned out I had forgotten to connect the common wire to the central section of the secondary windings! After that everything worked ok. I loaded the rails with 200mA to test it and all went well. I was planning to screw the regulators on the chassis but then realised that the 7918 has INPUT on the tab so rather than insulating the tab I opted for a couple of small heatsink I had around - not really needed anyways! Outputs are stable and clean. To add a bit more protection I swapped the fuse in the mains plug with a 1A fuse.

Once connected, everything worked ok and after fixing a couple of snagging (loose knobs etc) the unit now works like a charm. When the dial is set to the "battery" symbol I didn't realise the needle moves to a symbol that I guess means the battery is ok. Indeed now the needle stays towards the top of the range but within it so happy days. (the picture is just a crop of the larger one, it does not mean the needle is over the maximum when checking the supply - I just didn't take a picture of that)

Thank you all for your help and for being patient with me, I have really appreciated that and I have thoroughly enjoyed this small project which has taught me lots!!