Found an entry into the "rail2rail OPA crowd" literature-
Apparently US Patent 4,570,128 has been fairly influential, search all the subsequent patents that cite it.
Best wishes
David
That's not the problem, the problem is the stray capacitance of the power supply.
Put some stray capacitors from the supply to earth, these are in parallel with the load, they load the OPS.
This does not happen with a non-floated supply.
Best wishes
David
No. There are effectively two supplies each with (we assume here) identical capacitances to ground. This is the same for both cases. The only difference with floating supplies is that the output signal appears across them. But this does not imply any current flowing through them is apparent in the load. (Trace the current paths on the diagram).
Hi!
this remains me this: http://www.diyaudio.com/forums/solid-state/189766-square-law-class-amps-2.html#post2956849
and my working amp: http://www.diyaudio.com/forums/atta...69-tcj-evolve-simply-good-amps-idea-first.pdf
The architecture has been around for many decades now but for some reason has been overlooked by the mainstream. It is particularly useful for high power MOSFET amplifiers. Another thread discussing current drive at the moment ("Current drive amplifier? ") makes further rightful mention of its performance advantages in such a case - most of which apply to more conventional amplifiers too.
Somewhere buried in this thread (or another cited within it) is also a reference to the use of a non-switching bias arrangment. My personal experience is that simple implementations relying on complementary MOSFET square law transfer functions fall well short of expectations. Voltage driving the gates in parallel often leads to issues of stability and/or inefficient simultaneous conduction, let alone returning compromised distortion measures.
My own preference uses a square law derived from matched diodes driving separate n/p feedback loops. It uses the resultant currents to drive the MOSFETs independently and in their more-linear biased region which aids performance measures further. Importantly in current drive applications it is possible to guarantee non-switching since output voltage and current remain in phase as the amplifier drives a resistive load. (For ultimate performance, however, some means of error correction need also be implemented to counter the non-linear output impedance of the matched diode cell).
For voltage drive applications there are no doubt more complex and probably better non-switching examples but these require separate current monitoring for each MOSFET to ensure guarantee non-switching with reactive loads.
Dave, I have made a common source amp that I presented here on diyaudio some months ago. The thread is a bit uninspired perhaps and no one seemed to care about it. It has a floating PS and is single ended class A and uses a special power saving techniqe.
http://www.diyaudio.com/forums/solid-state/296251-description-my-single-ended-class-amplifier.html
I have a prototype standing beneath my desk and it sounds really like a single ended class A. The sound is light and velvety and very relaxed and detailed.
Regards.
http://www.diyaudio.com/forums/solid-state/296251-description-my-single-ended-class-amplifier.html
I have a prototype standing beneath my desk and it sounds really like a single ended class A. The sound is light and velvety and very relaxed and detailed.
Regards.
It is indeed a shame since the topology offers so much for so little complexity - such as the feedback around the output devices as your circuit also shows.
The circuit is a rather neat way of reducing dissipation - although adds a fair bit in the auxillary circuit too.
On the down side, the class A operation is limited to just 36Wrms into 8R for the case of a purely resistive load and therefore somewhat less for real loudspeakers. I remain unconvinced of the merits of single ended class A in any application.
The arrangement I described above, for 150mA bias with single n/p "dual die" 16A lateral MOSFET devices will allow several 100's of watts - I say this deliberately vaguely as practical experience has shown me that single devices employed much over 100W are not so reliable. Nevertheless such an arrangement delivers non-switching performance for much greater power outputs when designed appropriately. The ratio of non-switching current sharing is relatively easy to accomplish over four decades.
Hopefully we might encourage some more contributions...
Padamiecki. I löooked at that square law thing. In theory that works well, but when you start measuring the circuit in a simulator the thing turns out to be not so good that it might look. The transfer funstion of a mosfet is indeed square lawed, but in real life there are obviously other mechanisms in action.
That "dual die" circuit mentioned, where can I read more about it? I haven't really read all the replys carefully.
That "dual die" circuit mentioned, where can I read more about it? I haven't really read all the replys carefully.
Hi
Class A makes more sense in Sweden if it helps heat your house, but I definitely need efficiency, like the CE rail-to-rail op-amps.
Nice to see that the post here seems to have sparked some interest in your own thread.
Best wishes
David
My take from the article is that you can have an amp where only the output devices need to be high voltage/power.
The rest of the circuitry is sorta 'conventional'.
As you know my goal is always supa performance with supa simplicity. Once you get near 200W@8R, you find you need to complicate matters just to deal with high voltages in the 'low level' stages ... cascodes, Unobtainium devices etc.
Vanderkooy's approach means you can do all this fancy stuff at +/- 15V.
It's an approach that hasn't had even 1/10th the work done on it compared to even 'simple' amps like astx's superb piece of work.
If I was designing & building another 'conventional' amp this Millenium, I'd be looking at this ... if I didn't just use Hypex modules.
The cost ... a power transformer for each channel.
I'm a disciple of Bob Carver .. one of the few who have taken on the Golden Pinnae brigade in DBLT ... AND WON too. He advocates single supply for stereo.
Of course 2 lumps of iron in a stereo amp will appeal to the 'hand carved by Virgins from solid Unobtainium' brigade too ... though I would probably use SMPSs
Yes indeed to all the above.
I initially looked at this for theoretical feedback optimization reasons.
But the practical benefits look nice too, best of both worlds.
Only one quibble, Cherry pushed this idea way before Vanderkooy.
This is not essential, they are separate choices.
Partly it seems that people who like unusual solutions use both CE OPS and also floated supplies.
They can actually combine neatly but there is no reason to prevent a CE OPS with conventional supply, as Cherry initially did and the rail-to-rail op-amps do.
That is what I plan too, have started to study how to bias it.
Just don't want to mess up the basic simplicity with a complex bias scheme.
Will start to simulate later today.
Best wishes
David
Yes it does. Although Vanderkooy's amplifier was published some 30 years after the ATC amplifier circuit was published (in their own manuals) and is devoid of the output stage gain balancing that I have described previously.
The ATC amplifier is credited to Tim Isaac and I believe it was in turn derived from his own high power PA amplifiers that were manufactured maybe a decade before that (?) - maybe someone can supply a better account...
No it isn't. Separate windings on a single 'lump of iron' is more than sufficient. Dual shielded windings for each output stage (and also for the low voltage rails) are more than adequate - as discussed already in this thread, charging currents are like a common mode signal that does not appear in the load (the speaker).
Not only is it sufficient in regards of inter-channel effects if engineered properly, we are also in danger of entering a whole new debate regarding the lack of understanding stereo within the industry and beyond.
Indeed! An approach that in my experience provides exceptional results - and at high powers where as you say, conventional approaches are not optimal. But also an approach largely ignored I suspect because there are basic engineering principles that are not widely understood.
.. and Great Guru Baxandall sent it to this grasshopper in simplified form as a thought experiment to school me into the equivalence of CE & EF circa 1979
Please ask Prof Cherry for permission to post, at least, the gist of his circuit.
Of course an LTspice ASC or GIF. of yours would count as 'new' work and fully avoid copyright from the AES & IEEE mafia etc
Last edited:
I should clarify Guru Bob Carver's recommendations.
He advocated 'single' supplies with relatively poor regulation to get higher 'music power'. Also connecting one speaker out of phase (with suitable correction at the input of the amp) so the 2 channels did not draw power simultaneously from each rail for a 'mono' source.
'Music Power' is a very misunderstood concept abused to hell by Marketing. But for Bob's relatively large amplifiers, it makes a lot of sense.
And I'll remind people of Bob and DBLTs ... as at the other end of the spectrum ... I'll remind people ad nauseum of JC and sighted tests
________________
You could do a Vanderkooy type amp with separate windings on a single transformer .. but you would lose the 'music power' advantages ... and the kudos from the Golden Pinnae brigade from having 2 lumps of iron.
Last edited:
Cherry's IREE work was published 1978, submitted 1977.
But his book has similar, 1968.
Will do, I plan to email him shortly.
Best wishes
David
Hello Dave,
Can you tell us about what amplifier and speakers you currently use to play music.
Good to see krglee is also interested in seeing schematics and the second person to ask. I find discussions on amplifiers without a schematic not a very satisfactory way to have a discussion on the subtleties of amplifiers.
Can you tell us about what amplifier and speakers you currently use to play music.
Good to see krglee is also interested in seeing schematics and the second person to ask. I find discussions on amplifiers without a schematic not a very satisfactory way to have a discussion on the subtleties of amplifiers.
This is two questions.
Lateral Mosfets make the bias very simple but have poorer availability and consequently inflated price, and it will only become worse.
Vfets may be a reasonable option and I plan to check this.
Bias may still need some work.
A floated supply makes the most sense if the cases are connected to source and no emitter/source resistors are used.
In that case no thermal washers are needed and the heatsink can be earthed.
But that is only feasible for Lateral FETs AFAIK.
Problematic with present status and prospects of lateral Fets.
For other transistors there is no benefit from the elimination of heatsink washers but there remain all of the electrical problems I have already mentioned.
I know you dispute these but I am more inclined to rely on my own analysis, backed by "Alley and Attwood", than a pseudonymous internet post.
At least until I see a circuit + references to an explanation that supports your claim.
Best wishes
David
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.