2,000W into 1 ohm
I mentioned in a thread a while ago about doing an exercise of engineering an amp capable of delivering 2k watts into a 1 ohm load, with distortion aiming at the magical ppm figure. This is still happening, and making progress ... key problem as I saw it was managing crossover distortion intelligently - I'm looking at a conventional class AB output stage at the moment - and it didn't make sense to try and control it using a classic global negative feedback approach.
By nature I'm an excellent scavenger, I look to see what ideas are already out there - so I'm trying out some concepts in using local negative feedback. This is evolving, step by step - and showing promise: in a simulation of the output stage only, getting effectively 2kW in 1R, at 200kHz with reasonable stability - and the waveform at this stressful frequency looks pretty good, there is still some crossover glitching, but I'm reducing the visible level of it steadily.
.
By nature I'm an excellent scavenger, I look to see what ideas are already out there - so I'm trying out some concepts in using local negative feedback. This is evolving, step by step - and showing promise: in a simulation of the output stage only, getting effectively 2kW in 1R, at 200kHz with reasonable stability - and the waveform at this stressful frequency looks pretty good, there is still some crossover glitching, but I'm reducing the visible level of it steadily.
.
Total Comments 19
Comments
-
Posted 12th February 2014 at 08:44 AM by abraxalito -
Of course that will be a major issue, but I'm taking it a step at a time. In this first stage I'm using perfect power supplies, and developing the amplifying topology to work correctly within those ideal conditions. Once I'm happy with the core behaviour then I'll introduce more and more realistic PS elements, and do what is needed to keep the core circuitry sheltered, or robust, so that the performance is still maintained.
It's the principle of divide and conquer ... ;)Posted 12th February 2014 at 11:03 PM by fas42 -
Posted 13th February 2014 at 10:32 AM by abraxalito -
Purely as a rule of thumb, I would be unhappy if anything more than roughly the same level of distortion was injected through the non-perfect behaviour of the PS - via non-zero impedance and noisy mains, say well within the same order of magnitude ... as that contributed by the amplifying topology. I'm aiming at 120dB for the latter, so you could say that number is the target ...
Posted 14th February 2014 at 12:24 AM by fas42 -
OK then just putting some numbers to this - from one or two DSs that I've looked at, a typical EF output stage (single transistor pair) will have a PSRR between 60-70dB. Presumably to handle 2kW you'll be needing of the order of 20 pairs of devices - unless you have something a bit special up your sleeve in terms of advanced tech. With at best 50dB PSRR (degraded 20dB due to the paralleled devices) in your composite OPS then to achieve <1ppm it looks as though your supply impedance needs to be of the order of 0.3mohm. Does this seem reasonable?
Posted 14th February 2014 at 02:53 PM by abraxalito
Updated 14th February 2014 at 02:57 PM by abraxalito -
Those figures look reasonable in themselves, but remember I'm trying ideas in local feedback, which helps a great deal. In fact, I just tried seeing where the design is at, with regard to supply rejection on the output units - this at a still very rough stage, and will obviously depend on the accuracy of the models ... promising on first peek, well above 70dB at 200kHz, and 110dB in the audio band.
Posted 15th February 2014 at 05:46 AM by fas42 -
Steadily moving along ... the OPS is doing 2KW into 1R, at 200kHz, with a very clean waveform, no visible crossover artifacts, bias about 125mA per device. Harmonics are down at the 55 and 60dB level or so, and improve stongly once over a MHz, but still a little resonance in the drivers in the MHz range at maximum output to knock over ...
Posted 18th February 2014 at 09:45 PM by fas42
Updated 18th February 2014 at 09:47 PM by fas42 -
Changing my approach a bit - worrying about behaviour at 200k is not helping distortion figures at 20k, so will focus on the latter for a bit. Next "issue" is that optimum performance, at the moment, comes with the penalty that the conventional means of maintaining thermal stability is compromised - 'tradeoff' time, or a new way?
... some reading and thinkin' to be done ...Posted 22nd February 2014 at 03:44 AM by fas42
Updated 22nd February 2014 at 03:46 AM by fas42 -
Back to business ... as is usual when investigating achieving certain goals, the really interesting stuff emerges when one is considering how to deal with, best optimise some behaviour that may be a problem. And one of the considerations is the world of temperatures, thermal behaviour. At the moment I'm looking at a bipolar output stage, and these things love losing the plot if you're not careful, thermal runaway.
There are standard ways of dealing with this - but how about thinking outside the square? Why not give the devices excellent thermal inertia, so that no matter how electrically active they are their internal temperature varies as little as possible - this can only lead to improved overall performance.
Partially inspired by the Symphonic Kraft 400 priorities, I'm looking at some materials engineering at the moment ...Posted 26th February 2014 at 12:20 AM by fas42 -
:D ...
Posted 3rd March 2014 at 07:56 AM by fas42 -
Okay, this blog has inspired me to try and build a stand alone linear power supply that can deliver 1000W into 1 ohm for 4 channels. I have a slightly used 5kVA house transformer (frame type +/-24V) that I will rubber mount in an aluminum box with the diodes and 4x35mF caps. Which rectification scheme and diode selection is likely to get me the best supply?
Posted 28th August 2014 at 06:54 AM by yldouright -
Good on you! As is obvious from the lack of further input from me here I managed to get sidetracked from pursuing this - this is an endemic problem for me these days - but I encourage you to go for it!
As regards the best way to produce the rails, in part I would want to know the type of amplifier circuitry it was driving - if a typical circuit I would separate the rails feeding the output stage from all the prior circuitry all the way back to having separate rectifiers for each part. If I wanted to go all the way, I would go an extra step and use a separate, optimal and obviously quite small transformer just to generate the rails for the pre-output circuitry.
Also, I would like to determine the parasitics of the transformer fairly well, there are methods outlined for determining these; these parameters would be very useful to know.Posted 28th August 2014 at 11:15 AM by fas42 -
Noted. I had the krill and Aleph-X in mind for this supply but since it will remain in a separate box, I'd like to use it as my universal reference supply for 30-35VDC rails. Having it in a separate box allows me to use a high capacitance mains wire which should remove HF parasitics from the delivered power line and get me a respectable S/N ratio. I don't want to use additional transformers in it. As far as the rectification scheme goes, dual half or full wave and what kind of diodes for either? How do you feel about regulation after the diodes? What are the best tests to describe my transformer and in what order should I perform them. My equipment may not be up to the task, no distortion analyzer but I do have a 35MHz/2mV scope and 6.5 digit meters.
Posted 28th August 2014 at 11:53 AM by yldouright -
A separate box is good, and bad. Good, because it keeps some of the nasties away from the amplifying circuitry; but bad because it means that further filtering, regulation is required inside the amplifier boxes to prevent significant voltage fluctuations, caused by current transients through the impedance of the umbilical cord - that cord should be the lowest resistance, lowest inductance, highest capacitance type you can devise, for that reason ...
The 'goto' thread, IMO, is here: [URL]https://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size.html[/URL] - a huge amount of very useful input, and wrangling about what's important, and why. In there is good discussion of ways to "measure" transformers.
Note that there are a lot of people in diyAudio with very extensive, hands on experience with what devices are best to use in various areas - I don't have anywhere near that sort of experience; my approach is to start with a target performance, and use what's needed, that's easily available, to give me that - I don't do casual experimenting, if it works it works. In other words, I don't like to give advice on the 'right' part to use; I work from the angle of a troubleshooter, rather than a designer.
Regulation is good, but the power you're dealing with is great - a smart design will be needed be to it make happen efficiently.
Hope this helps a bit :)!!Posted 29th August 2014 at 11:10 PM by fas42 -
This sorta died, didn't it?! :o Trouble is, I run out of mental puff easily ... so, now I'm trying a new tack: taking an advanced design that's already out there, by someone else; diving into it and completely getting to grips with how it behaves; and then optimising it, and also extending its performance. I'm also feeling inspired at the moment to possibly design and produce PCBs for such - which will have a lot of my thinking and finessing incorporated.
The aim, again, would be to have something that operates at a top level technical performance, and also delivers excellent sound - would there be any interest if I produced a few boards, etc?Posted 12th December 2014 at 09:42 PM by fas42 -
Learning quite a bit from this exercise - at the moment tackling the dark underbellies of output stages ... as has been stated many times these are not straightforward, thermal and balancing issues are very much part of the "battle" to get the best from the circuit - again, I'm looking at making the circuit be well behaved at 200kHz, which then highlights where the weaknesses are; ensures the focus of optimising is directed into the right areas ...
Posted 15th December 2014 at 10:00 PM by fas42
Updated 15th December 2014 at 10:03 PM by fas42 -
Looking nice ... how about 1.6kW into 2 ohms, at 200kHz, eh, :D ? Not interested in the distortion numbers, but rather whether the circuitry is behaving itself, operating correctly at these levels - and it does, the waveforms that matter all are very clean doing this sort of power.
There is no point in having a real unit do this, the thermal issues are not worth the fuss needed - but it shows that the circuit has solid integrity in how it behaves - it has a very wide safety margin ....Posted 18th December 2014 at 11:12 PM by fas42 -
Been thrashing a bit with looking at the behaviour of multiple output devices in parallel - this uses MOSFETs so the usual means of bias current equalising won't work ... and there are a number of ways of dealing with this. Why I worry about this is that the competent behaviour will be compromised if the output devices are uneven in their current handling, so I'm looking at the "smartest" way of handling this in the real world.
Anyway, at the moment I feel the best, but possibly interim solution - if this ends up being put out there - is providing matched output stage combinations of parts along with the PCB to ensure that a built unit will be up to scratch.Posted 31st December 2014 at 03:08 AM by fas42 -
Out of curiosity, and for those who like numbers, I just did a THD at 20kHz, 1.6kW into 2R with current config, using perfect voltage rails - pretty nice, no harmonic worse than 130dB or so down, means about 0.00003% THD - trick will be to convert that into something working in the real world, that loses as little of such competence as possible.
Posted 1st January 2015 at 06:50 AM by fas42
Updated 1st January 2015 at 06:53 AM by fas42