Voltage overhead.

[davelectronic]

Thanks for taking the time with the expansion, I know roughly what you mean, but couldn't see it unless I put it on paper. Maths is not my strong point, visualisation to product is. My best option I think is to use a simulation program. I'm a Linux user, so will have to look in to what's going to work on that system. I should have known better than to get cheap cable off ebay. I'm going to have one more go with the 25 Amp cable, same as the 40 Amp stuff I already have. Just can't get the turns count on this 700 watt transformer core with the 40 Amp cable. I'm starting to question cost now, it's only that I want to see the finished viability of using a MOT in a low voltage power supply. Another £16.00 on cable coming up, I must be mad...

[Ian.M]

LTspice generally runs pretty well under WINE on Linux

[davelectronic]

I will have a look, give it a go, cheers Ian.

[davelectronic]

I've not managed to get LTSpice working on linux, what I've got now is the transformer in the chassis and the 35 Amp bridge rectifier in circuit. This drop 1.58 Volts from AC 14.7 Volts down to 13.12 Volts. I'm using a low drop out voltage regulator with minimum input of 13.60 Volts. The circuit has two MJ11015 darlington high power transistors, what I'm hoping is to use some snap in 25 Volt 6800uf capacitors x 9 capacitors, so 61200uf of capacitance to keep the voltage up for regulation.
These capacitors are reasonable price for that value. I was wandering if that is a high enough capacitance to regulate 12 Volts at a maximum of 12 Amps current, i can fit 10000uf capacitors x 9 so 90000uf total. But these capacitors are nearly twice the price of the 6800uf capacitors. No other change in components. So I'm looking to find out if 61200uf of capacitance for 12 Volts 12 Amps is high enough. Any help appreciated.

[mariush]

 Capacitance (Farads)  =  Current /  [ 2  x Mains Frequency x  ( Vdc peak - Vdc min) ]

Use LT4320 and four mosfets like Hero99 (and i think me) recommended to get smaller losses in the bridge rectifier - you'd have around 15w dissipated in the bridge rectifier right now.
It's not THAT expensive and you may be able to get by with just 3-4 x 10k uF 25v instead of 9 6800 or whatever. 

If you don't have the bridge rectifier making heat there, you may even be able to use cheaper 85c rated capacitors

For example, you can get Samwha 22000uF 35v from TME.eu for 2 uk pounds each : http://www.tme.eu/gb/details/hc1v229m35045ha/85c-snap-in-electrolytic-capacitors/samwha/

Or 33000 uF 35v for 2.8 uk pounds  (3000h@ 85c) : http://www.tme.eu/gb/details/hc1v339m40050ha/85c-snap-in-electrolytic-capacitors/samwha/

In 105c you have 10000uF 50v rated for 1.6 uk pounds : http://www.tme.eu/gb/details/he1h109m30050ha/105c-snap-in-electrolytic-capacitors/samwha/

[davelectronic]

Thanks for the reply, I've committed hardware to the case now for the bridge rectifier. There's precious little room inside the case as it is, I'm not sure what else is needed in the fets rectifier circuit. My two boards are stacked, one for the capacitors, the other the control board. I could always just go with a separate regulator input from the transformers AC for the regulator. That shouldn't sag with a less than 1 Amp regulator load.

[davelectronic]

So I've mounted the transformer and bridge rectifier, added 10000uf just as a test, getting 21 Volts no load. Might be ok yet with a moderate level of capacitance.

[Ian.M]

Put a load on it, (ideally at least 2/3 the final max load) and scope it.  Measure the slope of the ripple between halfwaves and the minimum and maximum voltage.  That should give you enough data to determine if its practical and how much more capacitance it will need.

[davelectronic]

It's the ideal thing to do for sure, but I've not got the use of a scope at the moment. I'm going to take a chance on 60000uf, I could go to 90000uf but that would cost double what I can get 60000uf in 6800uf snap in high ripple current capacitors. Should be ok up to 12 Amps...

[Ian.M]

It's the ideal thing to do for sure, but I've not got the use of a scope at the moment.

That's a pity.  You *really* need that data to avoid throwing more money away. 
What I am concerned about is whether the diode drop becomes excessive at your required load current - I expect the average output voltage to be under 13V and that adding more capacitance to maintain a higher voltage will be ineffective due to the increased peak current in the diodes increasing their voltage drop.

The discharge rate of the bulk capacitor can be calculated so you don't have to actually measure the slope, but it would be worth setting up to measure the peak and trough of the ripple under load using a diode and a capacitor (Peak reading voltmeter is easy. For the trough reverse the diode and use a 1Meg pullup to a higher voltage.  Calibrate out the diode drop using known DC input voltages)

Also measure the mean DC voltage, and ripple directly using the capacitor for DC blocking as a cross-check.  You'll need to know whether the DMM is true RMS or half or full wave average (try both polarities!), calibrated as RMS equivalent for a sinewave, as you need to convert the ripple to a pk-pk voltage.

Another option would be to use an Arduino (or other MCU with ADC and UART) as a 'pooor man's' oscilloscope.

[davelectronic]

Wish I had a scope..
Probably just use 9 x 10000uf capacitors. When think how many times I've rewound the MOT etc, it's all bolted in now with 14 AWG on it. I've used the same case as my previous project psu (now in use) This one's just a bit of fun really. I can't run it 24/7 it would rack up my electricity bill. Last MOT primary I tested was 700 watt unit, it drew 3 Amps no load. I know there wasteful in the power department, it's an occasional use DX HF radio psu for high power linear amplifier.

I've thought about building it so the MOT only powers up on HF linear use, and a more modest transformer runs long term RX radio state. But it was only a thought. I've put so much in to it I might as well do it properly and see it to the end.

[davelectronic]

That's 90000uf 9 x 10000uf capacitors, it's giving 21 Volts. Let you know the end voltage result when I've built the control board. It's got the low drop out voltage regulator under the small heatsink and 40mm fan. Should be interesting to see the loaded voltage.

[davelectronic]

Capacitor bank, only one picture posted in first try.

[davelectronic]

Finally finished the MOT project, as I only had 14.70 Volts secondary voltage to work with, I lowered my expectations to 10 - 12 Amps at 12 Volts. In practice I've got that easily, and fairly sure I could push 15 Amps at a 50% duty cycle out of it. It's to power a HF linear amplifier, so duty cycle anyway. Just wanted to say thanks for all the help, much appreciated. 
PS. Front posts, rear sockets unloaded voltage reading.
And 150 Watt halogen lamps load. Current and voltage reading. After 90 minutes continuous use, the transformer reach 76°C the transistors 52°C and rectifier 48°C. A total capacitance of 94700uf was used.