"Special" power supply


I got a problem. I want to build an amp for subwoofer usage, with an output between 800 and 1200 (best maybe 1000) WRMS into 4 ohm load. Therefore I opened a thread...


...and tried to "jack" (not the right word, I think, cause it´s been a very long and long lasting thread, before I got into) another...


...but noone seems to be really interested to help me. Or whatever the cause for the few answers is -- I don´t know. Now I have to try to arrange with the facts, and gotta try it another way.

1.) I got two pcs. of "SIEMENS 4AM6542-8DD40-0FA0". Rated 2.5kVA(cont.), 13.3kVA(short.) with several primaries, from 550 to 208 Volts, and secondary rated at 2 x 115 Volts. So the given voltages (w.o. diode losses) are (+/-) 70.96, or 74.35, 78.05, 81.32 (8 steps of ~ 3.5 V) 102.72 VDC, but I can´t get 10 "extra" Volts too simple... ^^

But these I would need for the only design I found, which could give so much power by using only the devices, I already have: (looked one more time, so number is different than mentioned in the thread) 30 pcs. of IRFP450 (from Harris, that means, only 150 W Ptot per device), and 40 pcs. of IRFPC50A. So, I got only N-Ch.-devices. The amp schematic is:


...and my transformers got no "extra" 10 Volts... so perhaps I have to add some windings, to get this. What do you think about this ? Eventually 7 Volts additional AC for this secondary, to get the 10 "extra" VDC. Would YOU do that? But it seems to be the only way, and it´s not very dangerous, I think. (Or: Do YOU know another diagram of an amp, using higher rail voltage up to 150 Volts (FETs would stand easily), perhaps with extra output stage? Or any other solution (except bridging, that I do already know) to use the HV-SOA of my FETs adequately? Sorry: Class D, G, H would be too complex for my first big amp.)

2.) And IF theres a solution, for example with voltage of say: 125 Volts - could one use 120 V ELKOs (connected to the rail, e.g. 4 x 15mF per rail, and if every rail has to produce a peak current of 20 Amperes, shared by the supply and the ELKOs, which gives a worst case of 5 Amperes for one ELKO), each one connected in series with, e.g., a 1 Ohm 10W resistor [should give about 5 Volts voltage drop], which is connected to ground/center tap, for extended life of the voltage-under-rated ELKOs? (I know in this case the caps would NOT give as much voltage stabilization as right-voltage-rated ones would give, cause the overall charge in coulombs is lower. But otherwise I have to use the more expensive higher voltage types.)

Please give my some answer, I got more and more hopeless.
Thank you in advance!



2008-03-12 9:59 pm
1) how did you calculate 102.72 VDC ?
on the other hand you have 2 x 115 Volts AC ... can you explain ?

2) did you measure your trafo secondary volts at idle ? what AC did you get ?
if you add windings for 125Vac, you will end up having 175Vdc when idle
Hello, again?

@AndrewT: Which design(s) do you mean?

Repeat 1.) My internet connection most of the time is non-existing. Absolute maximum (really rare, e. g. 1 to 10 peaks of a few minutes each day) is 150 Bits per second. I search nearly the whole time when It´s possible, but can´t find (exactly) what I need. FYI, I got serious problems with english language, too. For this little text, until THIS sentence, I had tu use translation dictionary ca. 10 times.

(My translation software generates an absolute mess, so I don´t want to use. And for translating threads or sites of them any translation of one site would take double the time than to load, which is imo enough without it. [Between 10 minutes (very simple, only html side, at good connection time) and "loading timeout" fault, this fault comes nearly at half of number of sites, cause connection decreases to zero while loading a site.) Big problems to search effectively. What I would need, is depending on...

Repeat 2.) I got nearly no money, and wouldn´t get some next time. I gotta work with what I have: The 2 transformers, 30 x IRFP450 (Harris), 40 x IRFPC50A(IRF) and some little "vegetables" like capacitors, resistors, varnished copper wire, ferrite and iron cores, current-compensated-double-coils, some ELKOs... I WILL buy what I need more, but every month I can only use e.g. 20 Euros (or nothing, when I need more money to live, than I have, then I gotta lend some money from someone, this is often enough the case with effectively 365 Euros/month).

@payloadde: Why do you ask this? :confused:
My transformers are SIEMENS 4AM6542-8DD40-0FA0. Several different rated primaries, secondaries rated at 1 x 230V or 2 x 115V. Voltage at my home growed from 230V to 240V. Feeding this to the different primaries, this secondary direct current voltages [after rectifying and filtering, but for now without diode losses, cause for now I don´t know which rectifier exactly I will use: normal or schottky, cause I GOT both, normal and schottky diodes (schottkys only rated 200V), but the schottkys only would stand 15 or 20 Amperes (some different types), and for now, as long as I don´t know about the rail voltage, I don´t know how much current would it be exactly, so the schottkys eventually arte too small)] would be formed:

P 550 ((550/240)*115*Sqrt.(2) =) S +/- 70.96; P 525 S 74.35; P 500 78.05; P 480 S 81.32; P 460 S 84.85; P 440 S 88.71; P 415 S 94.05; P 400 S 97.58; P 380 S 102.72; and which I probably NOT would use: P 230 S 169.71; P 208 S 187.66 --- all +/-, both transformers rated on 2.5 kVA continuously, and 13.3 kVA at short time/term or whatever that means in english. I don´t know exactly...;) Again: Why did you ask this? Do you know a fitting design for me? Or only doubted my fundamental maths? :D

Greets, Fred
I have used multiple primary Siemens transformers before, they are OK. The maths appears correct. Try a quick fix, use the 380V primary connection which will give you +/- 100V The amplifier should work OK with this voltage. In the mean time look for a 230/24V transformer of at least 200W rating, the 24V secondary winding is connected in series with the 230V supply raising it to 254V this is used to supply the 380V winding. This technique is called buck/boost Your English is good.


2012-12-19 7:24 pm
Hi Fred,
looking at the schematic of the Actrk600 it seems to me you need a good 10V xtra on top of the V+.The current needed is only 20...30mA.
If you can add a winding to your transfo to have say 20V rectified followed by a 7812 stabiliser you get a 12V floating power.Just connect it on top of V+ and you have it.

@metalsculptor: Thank you :) , but: Sorry. My english is really less good, as one may think (although I believe, it´s better than what is generated sometimes from "google translate"). I try hard to go ahead (what was much more easy at an age of 15 years, I realized...). But for now it only *seems* to look better, than it is in reality. Therefore: I don´t understand what "In the mean time" means. I looked into online dictionary, but there was no exact "goal", and so I can´t...

Why did you recommend me to clamp such a 200VA - transformer in such a way between mains and the big one? I know that this would work good (often I´m on "esp" sites, and Rod Elliott teached me many things, I didn´t know before), but why should I do this? (At first I gotta repeat: For security, I have to calculate with a mains input of 240Veff, cause voltage at my home often rises to 240V) This would give me a secondary voltage (after the 200VA one) of max. >265V, and feeded in the SIEMENS´ 380V taps, a final secondary of max. +/- 113,43VDC (again w.o. diode losses).

I´d REALLY LIKE to use this high voltage (which in another way I couldn´t get - next higher voltage w.o. this config would be 169,71VDC, and this would require 180V or 200V caps - uhhh...) for rails of my planned poweramp, cause I GOT a 300VA transformer with one secondary of 24V (and this would work fine), and YET, I got no ELKOs. And I would need SUCH HIGH CAPACITANCE to optimize a +/- 90V rails amp (maximum of Actrk600) for such an amount of deliver-current-ability (needed power is ca. 1000 WRMS!), that the voltage WOULD NOT decrease so much (maybe more than 10 Volts) at the peaks, that 1.) max. power is only <800 WRMS into 4 Ohms, and 2.) eventually the waveform of the sound would be "cutted" on the tops and bottoms, so (although I don´t know, but expect so) some distortion would be generated.

(Or not? I only suppose this, cause I read some time, that if one would feed an amp with too high input signal, then output waves are "cutted", what leads to high distortion level. In the text this effect was called "clipping". And wouldn´t falling rail voltage, holding the input height, lead to a similar effect?)

Oh god, this capacitors are extremely expensive, and I got nearly no money (only 375 Euros at one month - called "Hartz IV" in Germany). I only have the two big transformers, some smaller ones, the mentioned FETs (30xIRFP450 with 150W Ptot; 40xIRFPC50A with 180W) and some commonly useable components. Therefore a configuration similar like some PA-amps, smaller capacitance but more than enough rail voltage, wouldn´t be as false. Usage of 2 x 15 mF 120V caps would be cheaper than usage of 2 x 50 mF 100V caps. But I don´t know much about amps, so I´m not sure, if this would work without other problems. And, btw, my question from above (usage of resistors for voltage drop between caps and ground) is not answered till now. Hmmm... I got 7 x 4700µF 350V, and 6 x 1500µF 420V caps (all components are the rest of former projects, when I had still money), but I don´t know if it is allowed to connect center tap only between 1 or 2 pairs of the smaller ones, the other caps connected straight from + to -, or so --- I´ve never seen that, although it could work eventually.

@Ketje: Oh, yes! Thanks! Now I think I could do this. :smash:
Or: Do YOU know another diagram of an amp (fitting as good as this, or better (?) with parts) using higher rail voltages? Cause I can´t find substitutes with higher voltage rating for the 200V - transistors in the driver stage, and don´t know effect other ones with more different parameters would may generate.

(At the search of "alltransistors" not enough parameters are listed at the overview, and cause of my extremely bad www connection I can´t load many datasheet pdfs.) So if, then I would need types with very similar parameters, and maximium voltage increased to 250V. Rail voltage over 90V would be too much for the original driver stage, cause Quasi said (if I remember right) in one thread, it would be the case, that nearly the double of one rail voltage is present on one transistor. I don´t know. And I don´t understand the circuit enough to simply increase values of resistors, or add some.

Meanwhile I´m not more aversed to try other circuitry than AB, but I can´t find a schematic/configuration which uses only similar N-channel-FETs like mine. I´d like to build e.g. a topology that uses similar devices like mine at high voltage like 2 x 200V (or higher), maybe with IR gate driver (for another project I already worked with IR2153). But only if one could "transform" this output, e.g. simply through a 50Hz transformer with primary center-tap to low voltage at high current. This could give acceptable results at the frequencies I want to use the amplifier, which would be widely under 100Hz (ca. 15-80Hz). But such a topology is nowwhere to find, I think... :D So I would be happy, if Quasi himself would have been answering my questions.

What do YOU think about my description, how to use voltage underrated (which would be cheaper to buy) electrolytic caps (I mentioned low resistors between GND and caps)? Or more than double overrated ones, that I already got, (what otherwise would be simply unadequate - they´re real big buckets!) in a connection only part of them (or right rated ones) with, paralleled part of them without GND contact? (I know, probably this all (HV topology as good as cap (ab)usage) is unrealistic, but It´s interesting, cause I simply GOT them...;) )


Sorry, guys, I only tried "dict.cc", cause until I tried "in the meantime" the generated translations always were easily understandable. And loading for translation was not very long, although my connection is horrible.

I got fed up with gt while repeating one mess after another, when tried to translate little complexer text parts. So I skipped to translate only the single words or text parts, which I got problems with, and used dict.cc therefore.

Maybe not the best choice. :no: Sorry.


Hello, KVP!

Thanks very much, glad to read this! :) My sorrow about that decreases gradually. And it seeems to be easier bit by bit, with every post.

One more mistake: In the thread, I mentioned (where Quasi warned of higher rail voltage) there was another schematic than Actrk 600, it was named "Atrk-nmos". There P-channel mosfets rated at 200V were used.
This two diagrams I have mixed up - sorry! I also don´t know which of the two would be easier to convert. On the Actrk diagram there are MJE350 used as drivers, they´re rated higher (-300V), but I don´t know if higher voltage is possible therefore, cause the small signal BJTs are rated rather low, and at the schematic it´s marked again, that maximum is 90 Volts.
Please, couldn´t anyone give me tips for usage of other transistors, or values and position for adding resistors, to make say maximum of 115 V possible?