Amps
[Top] [All Lists]

Re: [Amps] Direct rectification of AC mains to drive the amp, VDD Supply

To: amps@contesting.com
Subject: Re: [Amps] Direct rectification of AC mains to drive the amp, VDD Supply
From: Manfred Mornhinweg <manfred@ludens.cl>
Date: Mon, 23 Sep 2013 16:07:13 +0000
List-post: <amps@contesting.com">mailto:amps@contesting.com>
Dear all,

it's funny to see how so many people discuss this matter, with some of them bringing up good technical reasons in favor or against, and others just giving their opinion without any good support.

Well, there is an old and wise adage:

------------------
The men who say it can't be done, should abstain from interfering with the men who are doing it.
------------------

Based on this adage, I will continue using direct mains-connected, non-isolated power supply schemes, in those situations where it makes technical and economic sense.

I have build a fair number of such devices, and have never regretted it. The direct line connected supply I have been designing duirng the last days for my legal limit amp, is closely based on a smaller (500 watt) unit I designed and built about two years ago, and which I used among other things to power several versions of test amplifiers using low cost MOSFETs. By "low cost" I mean "less than 5 dollars".

This power supply uses just one IC, one power MOSFET, a few diodes, an inductor, a moderately sized electrolytic cap, and about a dozen small parts. It delivers tightly regulated output voltage, with adjustable voltage, current limit and short circuit protection, wide input voltage range, and features an efficiency of roughly 93% when using cheap parts, which could be increased to about 96% just by using better, more expensive parts, and a slightly more complex circuit. The input power factor at full load is about 0.7 to 0.8, depending on the impedance of the mains supply. That's not great, but quite usable. An active power factor correction circuit could of course be added, at the cost of nearly doubling the complexity and cost of the whole supply, and reducing the efficiency to slightly under 90%.

The design I'm doing now is for 105VDC output, at 20A continuous duty, current limit at 23A, taking 180-260V AC input. It has an opto-isolated enable/disable input too, so the potentially noisy switching regulation works only during TX. Total cost for the electronic parts, bought in single quantity at Digikey, is around 70 dollars. In mass production the whole thing would probably cost less than 30 dollars, but of course we hams don't do mass production...

I think this is pretty good, compared to any other kind of regulated 2kW power supply. I'm willing to accept the slight complication of having no insulation from the mains.

My 500W direct line connected supply, which has basically the same design but with smaller parts and no enable input, has been working flawlessly since I built it. And until now I haven't electrocuted myself nor anyone else, nor have I damaged anything. Having line-connected electronics is more a matter of inconvenience: One cannot simply connect any oscilloscope to it! Either one has to use an isolation transformer during testing, or one has to float the oscilloscope too, which is quite inconvenient. Some scopes have insulated inputs, and these are great, but mine isn't like that.


Now a few comments about a linear amplifier powered at about 320VDC from the rectified and filtered, but unregulated 220-240VAC line, using two ARF1505 devices:

I looked into this matter when those MOSFETs first appeared. I do see some problems with it. The main one is some reports about high voltage MOSFETs having trouble with hotspotting when used in linear modes. The issue is this: In saturated condition (switching service), a MOSFET's conductivity has a negative temperature coefficient, which leads to the current distributing itself evenly over the die, and all is fine. But the transconductance has a positive temperature coefficient, so that in linear mode the current will tend to crowd in the hotter spots, further heating them, and making the devices fail. A certain amount of resistance distributed inside the devices is effective for avoiding this effect, up to a specific combination of voltage and current. Below that voltage, the device operates safely, while above it, it tends to hotspotting and failing. Now the question is if these specific MOSFETs can actually work at 300-350VDC in linear class AB, close to their full ratings, without having this trouble. I have read some papers telling that they cannot, but I have never tried it muyself - they are too expensive to fry them!

If anybody here has actually used them in linear service at 300V or more, and high power, for a prolongued time without failures, I would absolutely love to hear about it! If nobody has done it, the first who does will be wading uncharted territory.

In switching service (class D, class E, class F, even saturated class C) there should be no problem, but linear service in class AB or class A might be expected to cause reliability problems when running at 300V.

And the other, very minor problem: Power factor. To get an acceptable power factor, at least an inductor should be used in series with the input to the rectifier. Otherwise the power factor will be really low, when using a filter capacitance large enough to get a low ripple by brute force. A pretty good approach would be to use an active power factor correction circuit instead of the simplistic supply. That would give 350 or 400VDC, loosely regulated but much better than having no regulation, and the input power factor would be excellent. But this power factor correction circuit is just as complex as my direct off-line, lower voltage supply, which has tight regulation, overcurrent protection, and reasonable power factor!

OK, back to my design work, or I will never finish it...

Manfred

========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>