-2
\$\begingroup\$

I want to design a 300 volt DC / 2 Ampere power supply, for a high-power amplifier circuit.

The input can be either 220 volt AC, or 18 volt DC; both of them are available, however, 220 volt AC is preferred, because of 18 volt regulator's current limitations.

I searched the net, but didn't find much.

Any suggestions !?

EDIT:

The output voltage ripple shall be max. 0.5 volt.

EDIT2:

Actually, the goal is to design a circuit which is capable of outputting low ripple +/- 150 volt / 1 A pulses. There are 5 amplification stages. The first 3 stages uses +/- 18 volt OpAmps, and the final two differential stages use power transistors with +/- 300 volt bias supplies. That's why I need +/- 300 volt DC supply!


I searched more, and found the following circuit:

enter image description here

C1, C2, C3 - 0.1 mf 630V

C4, C5 - 0.01 mf 630V

C6, C7 - 100 mf 450V

R1 - 10E 5W Wire Wound

R2, R3 - 220KE 2Watts

D1, D2 - BY127

D3, D4 - BY127

L1, L2 - 12 Turns 18 SWG Wound over 4 Cm long Ferrite Rode.

(source: http://flashwebhost.com/circuit/600_volt_power_supply.php)

Any comments on this circuit !?

\$\endgroup\$
  • 1
    \$\begingroup\$ you need a booster circuit for dc-dc voltage converting. first rectify your ac voltage. then use a booster for increasing your voltage. Have a look at LT3758A. and this article www.ti.com/lit/an/slva372c/slva372c.pdf \$\endgroup\$ – HOPE Oct 8 '15 at 6:31
  • 6
    \$\begingroup\$ Do you have ANY experience with respect to building electronics, power supplies or even high voltage ??? If not then I would very strongly recommend you not to go ahead with this until you have gained more experience. I would not even consider building a 300V supply although I designed and build my own lab supply (30V, 4A) and have plenty experience with electronics. But hey, if you want to electrocute yourself, go ahead just don't say I did not warn you. \$\endgroup\$ – Bimpelrekkie Oct 8 '15 at 7:55
  • 1
    \$\begingroup\$ You'll also need to consider short circuit protection: 600W would make a respectable fire if you drop a screwdriver across the terminals. 300V/2A is an unusual choice for an amp as you're way above the rated voltage of most semiconductors; e.g. electronics-lab.com/project/600w-audio-amplifier-2 uses a 30V power supply. \$\endgroup\$ – pjc50 Oct 8 '15 at 8:59
  • 1
    \$\begingroup\$ Precision 150V pulses? What on earth is this for? \$\endgroup\$ – pjc50 Oct 8 '15 at 9:32
  • 4
    \$\begingroup\$ The schematic attached is a complete bullshit, the diodes short the mains every other half period. Actually, the whole website is full of such "interesting" schematics that could never work, I wonder whether the author at least tried to assemble some of these. He might be very surprised with the fire effects around... \$\endgroup\$ – Ladislav Oct 11 '15 at 16:25
1
\$\begingroup\$

How about this solution?

Connecting several 24 or 48 volt DC power supplies in series, in order to get nearly 300 volt DC !!?

enter image description here

(source: http://www.acopian.com/acopianPowerSupplies/entry.aspx?nsid=17)

Actually, the CL switching power supplies are cheap and easily available.

http://www.czcl-power.com/product/2014-4-30/239.html

I searched their website, and they say:

Q:

If we need a 24V output power supply, but CL does not have this model, can we use two 12V power supplies connecting in series instead of one 24V power supply?

Ans:

YES, basically you can do this to get the right output voltage, but be careful that the rated output current of the series system should be the rating of the minimum one in these series connected power supplies. Furthermore, we like you to parallel a diode at the output of power supply to prevent possible damage of internal capacitors.

(source: http://www.czcl-power.com/news/2014-5-4/333.html)

Any suggestions, comments, or hints on this approach !?

\$\endgroup\$
  • 2
    \$\begingroup\$ This looks simple enough to be feasible. Make sure you pick up isolated PSUs for this, otherwise mounting them on a common base (metal plate, rail etc) may short the outputs together. And remember: just because a 24V PSU is generally safe, several such PSUs delivering 300V are not safe anymore. I remember a DC shock I experienced from 110V battery: it was really hard to let the damn thing go! A 300V DC shock is a sure ticket to a vita chamber. \$\endgroup\$ – Dmitry Grigoryev Oct 8 '15 at 10:20
  • \$\begingroup\$ 300/24 = 12.5... so doesn't work. But a bigger problem would be the output bias. Connecting 12+ power supplies in series could cause voltages in the supplies to go out of voltage spec. I.e. transformer isolation rating or X/Y-capacitor breakdown voltage. So it's not always safe to do this. You should know your supply's circuit. Designing a custom SMPS is a much better way. \$\endgroup\$ – JHBonarius Feb 28 at 9:46
1
\$\begingroup\$

Run poweresim to get a quick and dirty SMPS design, something like this:

enter image description here

The chokes and gate drive transformers should be off-the-shelf, but the main transformer is likely to be a custom part. You can wind one by hand on a suitable core to prototype this; then get someone like Coilcraft to make this custom part for you.

How this works: there is a L-C filter section that prevents the power supply from sending noise into the powerline, then a rectifier and storage capacitor. After that you have a half bridge with a pair of mosfets that drives a custom transformer at some fairly high frequency, say 50-100kHz. The transformer output is rectified and filtered again to produce the output. The output voltage is sensed by a resistive divider and the control IC generates signals to drive the mosfet gates, which are sent to the gates via gate drive transformers (and some extra circuitry). That's it :)

Safety note: most parts of this circuit will kill you if you touch them while it is powered on. The capacitors can kill you even after this is powered off. Put resistors across every capacitor calculated so the capacitor discharges itself over a few seconds through the resistor; to debug, turn off, wait five seconds, then work. I also like having several voltmeters connected across some of the key capacitors eg C10, C15, C7 so I know at a glance what voltage the circuit has in it. Hands off completely while it is on (this includes holding oscilloscope probes or touching knobs on the oscilloscope). Hook up oscilloscope, then step back, then turn on. Plugging this into a GFI socket (or power strip) is also a good idea, but don't count on that alone.

Suggestion: If you have never built a low-voltage SMPS before, a high-voltage SMPS is not a good place to start. Pick an off the shelf supply, even if it is expensive. I recommend TDK-Lambda GENH 300-2.5 (there certainly are other options). Unless you're planning to make many of these (100's) you likely won't save any money by trying to build a custom one.

\$\endgroup\$
  • 1
    \$\begingroup\$ @Omid1989: As far as switch-mode power supplies go, this is a really simple one :) Poweresim will generate the BOM for you. For the most part, if you build it exactly like this, it will work; but you may have to adjust some things to get the right balance of efficiency, ripple and transient response. \$\endgroup\$ – Alex I Oct 8 '15 at 7:24
  • 2
    \$\begingroup\$ This is a good answer... in theory. But I think given the level of EE experience reflected in the question, there's non-negligible chance he'll taze (or even kill) himself building this. Building a HV SMPS is not a project for noobs. \$\endgroup\$ – Fizz Oct 8 '15 at 7:26
  • 1
    \$\begingroup\$ @RespawnedFluff: You're probably right. On the other hand, we all start somewhere. I added some safety notes. Omid1989, I'd recommend you team up with an experienced EE on this. \$\endgroup\$ – Alex I Oct 8 '15 at 7:44
  • 1
    \$\begingroup\$ @Omid1989: Run poweresim, it will answer all your questions :) The ripple can be traded off against step response and efficiency; you can pick any two of the three. It depends on the frequency it runs at, which depends on how the controller IC is configured. Best guess it could be made to be under a volt into a constant load fairly easily by increasing the size of the inductors and capacitors in the output filter. \$\endgroup\$ – Alex I Oct 8 '15 at 7:55
  • 1
    \$\begingroup\$ @Omid1989: This is a relatively classic TL494-based design. You can find out a lot more in various appnotes if you google the chip's name and some ancillary keywords like "design" and so forth. \$\endgroup\$ – Fizz Oct 8 '15 at 8:08
0
\$\begingroup\$

220v to 240v AC with just a rectifier and a big capacitor would be 310v - 340v DC. Isolation issues aside.

How accurate does your output need to be?

\$\endgroup\$
  • 1
    \$\begingroup\$ You will need a really big capacitor for this to work at 2A at all, otherwise you will be well below 300V. Accuracy issues aside. \$\endgroup\$ – Dmitry Grigoryev Oct 8 '15 at 7:04
  • \$\begingroup\$ That's a good suggestion. But how to reduce voltage ripple !? The voltage ripple shall be max. 0.2 volt. \$\endgroup\$ – Omid1989 Oct 8 '15 at 7:04
  • \$\begingroup\$ Ah well now that you put that extra information you will have to look for a different solution. Good luck with getting 300v and 0.2v ripple though \$\endgroup\$ – silverscania Oct 8 '15 at 7:09
  • 1
    \$\begingroup\$ @Omid1989: 0.2V ripple on 300V is <0.1%. That will be very difficult to achieve. It is possible, for example TDK-Lambda GENH 300-2.5 has a 20mV ripple spec on a 300V, 2.5A power supply. However TDK is really good at power supplies, and they're not cheap. Don't expect your first or second try at designing something like this to come close. \$\endgroup\$ – Alex I Oct 8 '15 at 7:11
  • \$\begingroup\$ @DmitryGrigoryev , How big the capacitor should be !? \$\endgroup\$ – Omid1989 Oct 8 '15 at 7:23
0
\$\begingroup\$

I found another solution, which seems reasonable to me.

It consists of 3 main parts:

  1. Transformer
  2. Bridge Rectifier
  3. LCL Filter

For example:

enter image description here

Now, in order to have a very low ripple in the output voltage, we add another LC stage to the filter, making an CLCLC filter. For example:

enter image description here

Of course one can add as many LC stages as desired, but I found out that 3 capacitors and 2 inductors (CLCLC) is enough.

The ripple factor of the output voltage for a simple CLC filter is calculated by:

where

and the f is the frequency of the AC input voltage.

And it can be easily generalized for multiple stages.


Your comments, hints, or opinions on this new solution are fully appreciated.

\$\endgroup\$
  • \$\begingroup\$ any comments on this new design !? \$\endgroup\$ – Omid1989 Oct 11 '15 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.