Power supply question

rshuck

Member
2005-11-07 8:16 pm
I have a Charlize and a USB Monica from DIY Paradise that I plan on putting in the same case with a Skynet 8080 SMPS, but I have a question on how to get the SMPS to put out 18v for the Monica.

The SMPS has +12vdc, +5vdc, -5vdc, and -12vdc rails. I fiddled with the trim pot to get the +12vdc rail to put out +13.6vdc, which brings the +5vdc rail to +5.29vdc.

When testing things out, I realized that connecting my multimeter's positive lead to the +12vdc rail and the negative lead to the -5vdc rail gives me +19.6vdc. Is it ok to power the USB Monica from this? I'm not sure if this is usable power or not, and I don't want to destroy anything by just trying it out The SMPS puts out something like 6A on the +12vdc rail and 1A on the -5vdc rail. If that doesn't work, how could I get +18vdc from this SMPS? I thought of a series connection from +5vdc to +12vdc to the positive side of the Monica, with the negative side then leading to 0Vdc on the SMPS.

I appreciate any help anyone can give me.
 
The overall performance of the power supply as you have planned relies on the weekest part, which in your case would be the 5V/1A output. In series with the 12V you just come up with 17V and still only 1A - not more.

If the range of the trimpot on the SMPS is insufficient, there's usually no other chance than to change the secondary transformer winding for the 12V output.

In most cases this would be an impossible task because the transformer cores of commercial products are often glued together so you're not able to get any good access to the windings. Even then the windings themself are impregnated with resin and out of reach anyway. In seldom cases it might be possible to add a (few) windings on the outer rim but this will add some leakage inductance - so not really a good idea though, but could work.

In case you can make it to add a (few) windings you might not need to change the feedback path, because only one output voltage is really regulated and all other output voltages are a result of the secondary winding relationships. For instance in most older power-supply (i.e. for AT-mainboards in the range of up to 250W) the only regulated output voltage is the +5V only! All other outputs rely on the regulation of that +5V output. Nowadays all output voltages may be continuously cobserved for over-/undervoltage levels to prevent any damages (will immediately shut down in case of such a fault condition), but (really) regulated is still just one output.

Now in your case you might have to add just one or maybe two windings on the secondary of the 12V output (the winding count is anyway very low - very often in the range of just 4 to 10 windings, so a single added winding could increase the voltage by 25%, so be very carefully). Just an important reminder: You need to figure out the orientation of the windings to "add" the voltage of the added winding instead of "subtracting" it, because they need to be soldered in series.
Hook up a "small" load to the regulated +5V - I often use a 5.1 Ohm resistor with 7W. But prior to do so, you better check the voltage rating of the +12V bulk capcaitor which in most cases is just a 16V type and to prevent any risk of an exploding cap you better replace them with a voltage rating of (at least) 25V. Make sure it's a low ESR type "specially" designed for SMPS.

All these instructions are for "eletronics technicians" and amateurs with a certain level of skill in that particular area and any newbies and two-left-handed people should never try this - parental guidance might be necessary in the latter case :D

I've modified a couple of those old SMPS to get 14.4V for my Tripath amps (i.e. from www.41hz.com) to squeeze out every possible Watt. For peak current consumption it was sometimes necessary to add some bulk capacitors at the output of the SMPS. Don't overdo this because some SMPS tend to oscillate if the capacitive load exceeds a certain value (even if the +12V output is not the regulated one).

So much for now and good luck with your project.
 
Corax said:
If the range of the trimpot on the SMPS is insufficient, there's usually no other chance than to change the secondary transformer winding for the 12V output. [/B]


I am not that familiar with this power supply -- but you should be able to change the values in the resistive divider for the error circuit to come up with the right output value. NOT FOR THE FAINT OF HEART.
 

rshuck

Member
2005-11-07 8:16 pm
Corax - So, assuming that my current draw is less than 1A at that voltage, I MIGHT be ok, but only if I change those caps to ones with a 25vdc rating.

Scratch this, I want to be safe, so I will just buy another smps for 18vdc. I know my limit!

What you've explained makes sense, but I am nowhere near trying to do the required modifications to make it work. I followed nuuk's guide on this, with the 1Ohm 10 Watt resistor on the 5v rail, and since that's the regulated side, it all starts to come together now.


Jackinnj - I am to poorly skilled and inexperienced to try this. Thanks for mentioning the possibility though.
 
The replacement of the bulk caps with a higher voltage rating is only necessary if you've decided to change (increase) the secondary winding. In all other cases just leave them untouched.

To get an 18V SMPS might be really hard - I've never seen one with that output voltage. The next possible value (according to what I've investigated) is 24V, which in turn might be too much for your Monica.

Well to be honest I've seen some other SMPS with just a single +15V or even +15V and -15V but the current capabilities are not worth mentioning it (at max. a few 100mA) because they're just auxiliary outputs for i.e. low current consumption analog circuits.

How about the thought to use a toroid transformer with rectifier and bulk capacitor(s) and a secondary side switching power supply?
National Semiconductor has a nice switcher called LM2576 or LM2596 and is good for 3A. Low external component count - just six for the adjustable version (one inductor, two capacitors, two resistors and one diode) - makes it real simple for your purpose. I build some myself and can tell you they're "solid" like a rock. Never had any issues with them. Well I abused them as pre-regultors for my self-made lab power supplies to reduce dissipation losses over the series regulator. Works just perfect and keeps everything nice and :cool:.

The secondary side switching solution doesn't consume more space/volume than the primary side switching solution (if I compare the dimensions of the Skynet 8080 with a toroid in the 100...200W class). However the weight will be of course more heavier based on the used toroid.

Well, the final decision, which way to go, is up to you. I'm just giving some hints on how to accomplish your desires. ;)
 
Note that the controller for SMPS picks up the voltage on the secondary, rectified and filtered side of the power supply -- if you change the turns on the secondary the error signal is just going to be larger so the controller will adjust its duty cycle lower to effect the correct output voltage.

it isn't difficult to take one of these transformers apart, btw -- they just have to sit in a moderately hot oven for about 20 minutes and the resin holding the windings in place gets soft enough to take the thing apart.

Skynet 8089 would be a good choice for Class D Amps.
 
Yes it does jackinnj - but from the +5V output and not from the +12V output. I mentioned to increase the corresponding windings on the +12V output not on the +5V output. This way you don't have to take care about any feedback related stuff - it remains unchanged!!!

To unwound the power tranformer wouldn't be a good idea though, even if it works what you have mentioned (to heat it up in an oven). The problem is in most cases that such transformers are usually not winded like tranformers for 50/60 Hz. To minimize leakage inductance the primary and/or secondary windings are not wound on the core one after the other. They're often interleaved, making it almost impossible to unwound just one secondary winding without touching/interfering with other windings.
Therefore I highly recommend not to disassemble/unwound any of the windings unless you know ecactly what you're doing and have advanced skills to reassemble/rewound such a transformer!!!

The easiest way I already mentioned - add just one or more windings (if necessary) on top of the other windings. Figure out the correct polarity/orientation of the +12V secondary winding to solder the added winding(s) to get an increased output voltage. Furthermore you may have to change the bulk capacitor to one with a higher voltage rating. That'll be all - you can use the SMPS without any hesitations concerning the (in)correct regulation. This is definately the easiest way (I know about) to get an +18V output from such a power supply.
 
Assuming that you just need the 18v and not any of the original output voltages, and assuming all relevant capacitors are rated to handle more than 18v, I think I would probably try altering the SMPS's feedback's resistive divider, first, only because that seems easier, to me. However, note that I am not an expert and could be wrong.

Just in case you want to risk trying it: You could test it by using a potentiometer with a max value that's very large compared to the ungrounded resistor in the SMPS's feedback divider. Measure the resistance between the pot's center (or wiper) contact and one of its other contacts and set that resistance to its maximum value by turning the pot shaft. Then connect those two pot contacts across the feedback resistor (Use very short leads. Also: their orientation/polarity doesn't matter.). Put your voltmeter on the SMPS output and see what happens when you power it up (keeping your hand on the switch, maybe ;-). If that still seems OK, try SLOWLY turning the pot. If the output voltage changes in the wrong direction, connect the pot's contacts across the other feedback resistor. If you get to an acceptable voltage, and nothing has melted or exploded yet (or even gotten too warm), then power down, disconnect the pot (don't change its shaft position!), and measure the resistance between the two pot contacts that were used. Then change the feedback resistor, so that its value is equal to what it was with the pot resistance in parallel with it, OR, easier: solder a fixed resistor (or some combination of them) across (in parallel) with the feedback resistor, that has the same value as you found was needed for the pot's resistance. (Ideally, you should be experienced with SMPS circuits, and have an oscilloscope, and check that the SMPS is still functioning well, internally.)

Most importantly: Don't blame me if your SMPS is destroyed. :)

If it were me, I would probably just build a simple little boost-mode SMPS, to get the 18V from 12V (or 5V, or whatever). It would only need a cheap 'off-the-shelf' inductor (probably only a small one, too), instead of a transformer.

But it might be best to 'start from scratch':

How much current do you need, max, for the 18V? And what other voltages do you need, at what max currents? Are any other power supplies already available, besides the SMPS? If so, what are their output voltages and max currents?

You might want to go to http://www.national.com and try putting some numbers into their Webench 'automatic SMPS designer' app. If you don't need a lot of current, you should be able to fairly-easily build your own very-low-parts-count SMPS supplies, for everything you need, or maybe use some combination of SMPS and linear regulator-based supplies.

(By the way, I have a boatload of older National LM78S40 switchmode power supply chips (14-pin DIP, IIRC), if you want some for the cost of shipping. They work pretty-well for getting 18VDC from 12VDC, at up to about 1 Amp, IIRC, and could give almost any higher max current by just adding a power transistor. The datasheet should be at national.com.)

You might also want to go to http://www.linear.com and download SwitcherCad (aka LTspice). Without knowing anything about using that software, you can run it and select "File" then "Switch Selector Guide" and put in your available and desired voltages and the max output current, and the software will design an SMPS using any of the Linear Tech chips that are suitable, and will give you the schematic and run a simulation of it, all automatically. It's somewhat similar to National's Webench, when used that way (but uses LT chips instead of National's, of course), and is much faster to use, but doesn't give you the PCB layout, nor the interesting thermal analysis, that you get with Webench. However, it is also a full-blown spice circuit simulator.
 
jackinnj: Yip, I had those SMPS in mind - with the lovely TL494 - just the standard one for AT-size PCs (in the power range of up to 250W) and of course with multiple outputs PC-motherboards require (+5V, -5V, +12V and -12V, and maybe +5V standby and/or other auxiliary outputs).
I stopped counting those SMPS I fixed or modified till today. :D


gootee: To change the feedback path without changing the secondary windings is of high risk if the change in output voltage exceeds a certain limit. And this limit is, besides other criteria, mainly dictated by the PWM-controller (i.e capabilities and mode the controller can/will run: CCM, DCM, ...). Between no load and maximum load the PWM-controller changes the duty-cycle in a specific range. Depending on the controller and a misaligned winding ratio for the desired output voltage they're not always capable of a duty-cycle range necessary for that specific SMPS design (just to explain it in simple word - to go into details could blow up this thread). Besides this, there're some other things to consider but I won't overdo it right now with excessiv explanations.

Nevertheless I wanna mention the danger you might encounter by doing it your way.
Depending on the load situation of the SMPS and a shifted "operating point" of the PWM-controller you might experience the following circumstances:
1) At no load the output voltage might exceed (by far) the expected voltage and could damage other circuits and the SMPS itself, due to PWM duty-cycle minimum limit - worst case scenario though. :hot:
2) At "full" load - if you still wanna call it that way - the output voltage will drop before the expected maximum output current is reached, due to the fact that the PWM duty-cycle maxmimum limit is reached. :dead:
3) ...

Case 2) will under normal circumstances neither harm the SMPS nor the attached "Monica".

Only with the correct (calculated) winding ratio the PWM controller is capable to work at it's best under all load conditions. The SMPS works much more reliable with the changes I mentioned - even if it's still an "abuse" of a not really designed SMPS for that specific purpose.

As a matter of fact it would almost react like a standard unregulated power supply (made up of a toroid transformer -> rectifier -> bulk capacitor) to load changes on the +12V output (or +18V, after the change) while regulating the +5V output where it "sees" a constant load resistance.
This real reaction is mainly based on the voltage drops over the copper windings and the recifier diodes which also dynamically changes with temperature and current flow.


After all, my recommendation is, to invest some more or less tricky/fiddling work instead of going the easy way (by changing the feedback) with the result of a much more reliable SMPS.
 

megajocke

Member
2003-01-11 8:01 pm
It should be possible to get a 18V capable winding out of the transformer without rewinding it. The +12V rectifier and its connections to the transformer are left alone. All other secondary connections (including 0V) are removed. The former +5V winding gets new big rectifiers and snubbers with anodes going to ground. The filter inductor is modified so that the +5V and +12V are put in series. Filter caps on the former +12V may need to be changed and the feedback divider will need rebuilding. The control circuit power may need to be rectified from the 5V taps now to give proper voltage for it.

Converting to 24V can be done by full wave rectifying the whole 12V winding. Control ciruit power to be taken with a diode from the center tap.
 
Corax said:
jackinnj: Yip, I had those SMPS in mind - with the lovely TL494 - just the standard one for AT-size PCs (in the power range of up to 250W) and of course with multiple outputs PC-motherboards require (+5V, -5V, +12V and -12V, and maybe +5V standby and/or other auxiliary outputs).
I stopped counting those SMPS I fixed or modified till today. :D


gootee: To change the feedback path without changing the secondary windings is of high risk if the change in output voltage exceeds a certain limit. And this limit is, besides other criteria, mainly dictated by the PWM-controller (i.e capabilities and mode the controller can/will run: CCM, DCM, ...). Between no load and maximum load the PWM-controller changes the duty-cycle in a specific range. Depending on the controller and a misaligned winding ratio for the desired output voltage they're not always capable of a duty-cycle range necessary for that specific SMPS design (just to explain it in simple word - to go into details could blow up this thread). Besides this, there're some other things to consider but I won't overdo it right now with excessiv explanations.

Nevertheless I wanna mention the danger you might encounter by doing it your way.
Depending on the load situation of the SMPS and a shifted "operating point" of the PWM-controller you might experience the following circumstances:
1) At no load the output voltage might exceed (by far) the expected voltage and could damage other circuits and the SMPS itself, due to PWM duty-cycle minimum limit - worst case scenario though. :hot:
2) At "full" load - if you still wanna call it that way - the output voltage will drop before the expected maximum output current is reached, due to the fact that the PWM duty-cycle maxmimum limit is reached. :dead:
3) ...

Case 2) will under normal circumstances neither harm the SMPS nor the attached "Monica".

Only with the correct (calculated) winding ratio the PWM controller is capable to work at it's best under all load conditions. The SMPS works much more reliable with the changes I mentioned - even if it's still an "abuse" of a not really designed SMPS for that specific purpose.

As a matter of fact it would almost react like a standard unregulated power supply (made up of a toroid transformer -> rectifier -> bulk capacitor) to load changes on the +12V output (or +18V, after the change) while regulating the +5V output where it "sees" a constant load resistance.
This real reaction is mainly based on the voltage drops over the copper windings and the recifier diodes which also dynamically changes with temperature and current flow.


After all, my recommendation is, to invest some more or less tricky/fiddling work instead of going the easy way (by changing the feedback) with the result of a much more reliable SMPS.

Thanks, Corax. I'm glad you posted all of that, and completely agree with everything you have said and implied.

I had just wanted to try to provide 'another option', and hoped that I had made it clear-enough that it was at least 'risky'. I probably should have also made it more clear that adjusting the feedback ratio could easily cause other possibly-catastrophic problems, and/or some more-subtle less-than-optimal operating point problems, some of which I did not even think about, or know enough about.
 
megajocke: Of course what you've mentioned can be done with the expense of much more work: Removing all unnecessary stuff, changing the feedback divider, using a new rectifier, re-/de-/unsoldering so much components, ...?!? Hey, I'm a lazy guy and adding some winding(s) on top of the power transformer isn't that much work. Of course you have to remove the power transformer once to do that and remove one of the former 12V pins to get the original and added winding in series, but that's all the efford you have to do. However, many ways will lead to Rom ;) and I don't mind if you wanna go your own way :rolleyes:

gootee: Thanks for the flowers. :angel:
Those words have really touched my heart. :)
I guess we're both on the same "wavelength". ;)
Besides the charming stuff, in turn I agree with your statements too.
Nowbody can consider everything at at the same moment. Once in a while you have to take a power nap and think over it again, in case you've missed a very important thing :cannotbe:. Nothing we (or I) have to argue about. Human beings do mistakes and so do I.
Nobody can tell from himself that he knows everything. I got my lack of knowledges too, and I'm more than glad if someone can tell/teach me more to get rid of that lack. I'm always willing to learn - it will always be an improvement of knowledges (if you've inhaled and processed the information) and skills (when you practice it later).
 
Corax said:
megajocke: Of course what you've mentioned can be done with the expense of much more work: Removing all unnecessary stuff, changing the feedback divider, using a new rectifier, re-/de-/unsoldering so much components, ...?!? Hey, I'm a lazy guy and adding some winding(s) on top of the power transformer isn't that much work. Of course you have to remove the power transformer once to do that and remove one of the former 12V pins to get the original and added winding in series, but that's all the efford you have to do. However, many ways will lead to Rom ;) and I don't mind if you wanna go your own way :rolleyes:

gootee: Thanks for the flowers. :angel:
Those words have really touched my heart. :)
I guess we're both on the same "wavelength". ;)
Besides the charming stuff, in turn I agree with your statements too.
Nowbody can consider everything at at the same moment. Once in a while you have to take a power nap and think over it again, in case you've missed a very important thing :cannotbe:. Nothing we (or I) have to argue about. Human beings do mistakes and so do I.
Nobody can tell from himself that he knows everything. I got my lack of knowledges too, and I'm more than glad if someone can tell/teach me more to get rid of that lack. I'm always willing to learn - it will always be an improvement of knowledges (if you've inhaled and processed the information) and skills (when you practice it later).


<GRIN!>

You are quite welcome.

Thank you.

Agreed.

(I would like to say more, but I fear that that might create an unstable positive feedback loop. :)
 

megajocke

Member
2003-01-11 8:01 pm
Sure, if you can fit any more windings in there. :) You still have to make sure the capacitors can take the new voltage and readjust the divider. The choke can probably be left alone if a little more current ripple can be tolerated.

But to me it seems easier adding two diodes than rewinding the transformer. I'd done it that way.
 
Folks,

I guess we can quit at this point any further discussion. It appears to me that rshuck has left us alone with our discussion of his problem. :apathic:
No further questions, statements, ..., from him to our postings. Are you still there? A (further) short notice that we were going too far or not exactly answering your question(s) would have been appreciated.

Maybe it's our mistake, because in his last posting (five days ago) he mentioned the following:

... I am to poorly skilled and inexperienced to try this. ...

If I'm not mistaken we've overrun him with technical details which are far of his skills. Guess we should've read more carfully what he was asking for. :xeye:

"rshuck" - if you're still reading this thread just let us know and tell us if everything is clear by now or if you need any further explanations to make up your mind concerning your SMPS (ab)use for your Monica - I mean, to modify or not to modify your SMPS.