A little Class D remedial help?

[Andy H.]

This one really isn't about class A, okay? (My _yping isn't _hat bad...)

Did a quick forum search here for info on class D operation, finding just the one thread jumping out at me. From 'amps and cabs' as it happened:
http://www.thegearpage.net/board/showthread.php?t=98440
Didn't find the answer to my question though, which after briefly resurfacing is clearly sinking fast now. With a fast diminishing chance of some lucky net cast trawling it up again just for me...

I also noticed the recent poll on keeping these two forum sections (Amp and Cabs / Amp Technical) separate:
http://www.thegearpage.net/board/showthread.php?t=197120
Leave as is...and even more, don't answer any tech questions posted in the amp & cab section...they'll either learn to post properly, or don't deserve the answer...
Actually though perhaps a little extreme, I'd basically not deny a grain of sense there. Sooo anyway...

Clearly I'm taking leave of my senses posting there then. Oh dear! :crazyguy

Well I'm going back there first to redirect any serious responses here.

And if I may beg your indulgence, I'll rephrase my own query from the bowels of page 4 of that thread, making this my guaranteed once only, first and final near double forum post...


MY QUESTION:

Looking at the specs for the (class D operation) Crate PowerBlock amp, I see that it is rated (bridged, into an 8 ohm load) at 150wRMS output at 10% THD.
(Hit specs here):
http://www.crateamps.com/html/product.cfm?pid=119 (http://www.crateamps.com/html/product.cfm?pid=119)#

As solid state amps are normally rated at very much lower distortion values, is there any good reason for Crate to rate at this high a THD figure? Is that anything to do with the class D operation, and what would it likely mean in real world terms? Is this really a 150w amp or is that an optimistic rating?


Also Randall Aitken was saying that class D amps are rather impedance critical. Would running this guitar amp into a 16 ohm load likely be a genuinely bad move, causing major tonal shifts or worse?

Also if not a terminal problem, how might mismatching it affect output power? Would running it into a 16 ohm cab be comparable to running a class A/B amp into double the impedance - ie an output power reduction of about 1/3? Or are we talking a very different story?

So the real world example is Crates PowerBlock. Whats the real story though?

Oh BTW while I'm not an amp tech or frequent poster here myself, I do come here a lot for info. I do post a fair bit on the Orange Amps forum, trying hard to keep the worst technical mis / disinformation from blowing up peoples nice new MM OTs. And though I know very little of this 'class D' stuff of which you speak, I do know that it doesn't stand for "digital".
Just as 'class A' unfortunately seems to mean "Well people, accounts and marketing just got together, and you boffins just became our class A design... appendix!" :NUTS


From what a couple of you were saying in the other thread, some of you guys really appreciate the analogue solid state side of things too. All enlightenment on real world class D use would be much appreciated.

Why in fact the :) then Mr Phillips?
Class D is inferior to Class C which is inferior to Class B which is inferior to Class AB which is of course inferior to Class A.

:)

Cheers and a belated Happy New Year to all!

Andy Heatlie.

[Old Tele man]

..maybe this (http://users.ece.gatech.edu/~mleach/ece4435/f01/ClassD2.pdf) will help?

[John Phillips]

Why in fact the :) then Mr Phillips?

Andy Heatlie.
Just a little dig at the way marketing copy writers manage to twist innocuous technical definitions into judgemntal implications about quality :).


Adspeak (Translation)...

Class A = first class, Class B = second class, Class AB = compromise. (Even when Class A is actually Class AB.)

Running tubes at 100% of maximum dissipation = getting their best and fullest performance. (Even when since they're actually in Class AB, running them at 100% at idle means they will exceed it the minute you play anything.)

Negative feedback - oh dear, that's really bad, that's feedback (squeal, nasty, fingers in ears) and it's even worse, its negative. (Or is that just on eBay?)

Cathode, AKA self biased. This is really good, you can put in any tubes and it will bias itself. No nasty technical knowledge required, and far better than being stuck with a fixed bias. (Oh wait - you are stuck, with self-bias, unless you change the cathode resistor. Even changing the tube rating won't make a lot of difference, because it will just self-adjust right back to where it was! Even with a truly fixed, fixed-bias amp, you can still choose different tube ratings for different bias points... and if you have a trimmer, well... why do they call that 'fixed' again?)

Class D? Of course Class D is inferior, it's three classes below even that awful compromise, Class AB!

:)


I actually don't know the technical answers to your questions, but Mr. Aiken surely will...

Have you seen that new Fender 'Jazzmaster' - no, it's not a guitar with a complicated switching system this time ;) - Ultralite amp? Class D too, I'm pretty sure. Seems to get good reviews.

Happy New Year, Andy!

:D

[reaiken]

]lso Randall Aitken was saying that class D amps are rather impedance critical. Would running this guitar amp into a 16 ohm load likely be a genuinely bad move, causing major tonal shifts or worse?[/color]

Also if not a terminal problem, how might mismatching it affect output power? Would running it into a 16 ohm cab be comparable to running a class A/B amp into double the impedance - ie an output power reduction of about 1/3? Or are we talking a very different story?


Well, Andy, it's like this...class D has come a long way in recent years, especially in the several years since I've messed around with it. There are some really exciting (okay, I need a life) new integrated class D amps and controllers out there, and higher voltage, faster MOSFETs coming out practically every day.

Here's the deal: In a class D switching amplifier, the signal is PWM modulated (pulse-width modulation, where the width of fixed-frequency pulses is varied depending on the level of the input signal), or modulated with some slick variation of PWM. This modulated signal is at a high frequency, typically 200 kHz or more. An output filter is then used to remove the high frequencies, leaving on the modulating waveform to drive the load.

Since the active devices, the output MOSFETs, are either on or off, and they have extremely low on-resistance, they don't dissipate any appreciable power except when they are transitioning between the off and on states.

There are a couple problems with this;

(1) The faster you switch, the more transition losses you have, so the efficiency goes down, and the heat goes up. Unfortunately, the slower you switch, the less bandwidth you get (lower frequency response, less high end), because you need to have a switching frequency at least ten times the highest frequency you want to pass, better yet, 20 or more times. The lower the distance between the highest frequency and the switching frequency, the more complex the output filter has to be to remove the switching frequencies.

(2) The output can only swing as far as the power supply rails. This can be implemented as a dual-rail supply or a bridge-tied load with a single supply. For example, if you have a supply of 50V, you can get an output of 100V p-p at most (actually less, because you can't get 100% modulation). This means that at most, with a 50V supply, you can get a 100V p-p output signal swing, which results into 312.5W output power into 4 ohms (50V peak = 35.2Vrms, which gives 312.5W into 4 ohms). Since the supply rails are fixed, you can get at most 100V p-p into 8 ohms, which gives an output power of 156.25W, and into 16 ohms, you only get 78.125W. If you wanted the same 312.5W into 16 ohms, you'd need a power supply of 200V p-p. MOSFETs that are rated for 200V are a lot slower than those rated for 50V, so they aren't as efficient, and you can't switch them as fast (the bandwidth issue I spoke of earlier). If you did design for 16 ohms/312.5W, and you connected a 4 ohm load, the amp would put out a whopping 1250W, and draw enough current to fry it, if you didn't put in some current limiting. Because of this, they don't lend themselves very well to wide impedance ranges, much like normal solid-state amps. You could put an impedance-matching transformer at the output, but it would weigh more than the entire amp.

(3) The output filter is usually a two-pole LC type, which must be designed to fit a certain response, such as Butterworth, Chebyshev, etc. These filters have to be terminated in their characteristic impedance, or the response won't be flat - it will peak up or roll off at the upper frequency limits, or have ripples in the passband. It is very difficult to design a filter that looks good into a 4:1 load resistance change.


Now, the good news is this: Guitar amps don't need 20-20kHz response of typical stereo gear, since the speakers in most guitar cabs don't have response above 5kHz or so. This allows you to switch at lower frequencies, or it allows you to not care about filter characteristics near cutoff if the switching rate is high enough to put the filter transition band up at 20kHz or higher, so you can keep a high switching frequency and the correspondingly smaller filter, and not worry too much about the effects of impedance mismatch.

Hope this helps clear it up for you.

Randall Aiken

[Andy H.]

First of all an apology to Randall Aiken. I've just noticed that I've been mis-spelling your name. Probably quite a lot as I've been linking to your excellent articles over at the Orange forum for ages. Ironically (considering my first post above - honestly a coincidence!), I've been adding a 't'!
Very sorry about that Mr. Aiken!!! http://www.orangeamps.com/forum/icon_smile_blush.gif (http://javascript<b></b>:insertsmilie('[:I]');)

Thanks Old Tele man. Spotted that article on the other thread.
Also Googled a few other useful ones:
http://en.wikipedia.org/wiki/Switching_amplifier
http://www.irf.com/technical-info/appnotes/an-1071.pdf
http://sound.westhost.com/articles/pwm.htm

Still digesting some of that stuff, but one thing which keeps jumping out is the low distortion levels achievable...
Oh, yeah - one more thing about class D: it doesn't distort worth a damn (guitar-amp-wise, that is). You must run the output stage below clipping to avoid nasty artifacts. Fortunately, you can get such high powers at such high efficiencies, this isn't usually a problem. They'll excel in re-amping situations, and the switching frequencies are getting high enough to give beyond-20kHz bandwidth, so they can reproduce guitar-range square waves faithfully.

Randall Aiken
...which takes me back to my original question. Why is Crate quoting the 150w at 10%THD? (And is it really a 150wRMS amp?)

Seems a strange one. When I first came across the amp and looked up the specs I dismissed to myself it as cheap rubbish - well I thought they were doing something like the old 'PMOP' peak power rating. You know, "make the numbers look impressive!" nonsense.

But the amp seems to work well and be popular. (Just as here, quite a few of the 'Orange people' - amps, not land - seem to be using it too.) So it clearly isn't just cheap rubbish, but quite an effective amp at a very good price which pro / semi pro musicians are more than happy to use.

Still puzzled on that one I have to admit...

I've played one of those Acoustic Image amps as well. Don't recall which combo it was, but when I got my Marleaux Nyelectric (solid bodied nylon strung) guitar from the nice people at Overwater Basses in England, that was what they let me loose on. I was most impressed with the sounds from that little box, not to mention the power. (The guitar wasn't too shoddy either, ie sold!)

John, I thought for a moment you meant the new Jaguar bass. Its "not a guitar with a complicated switching system" either... http://www.orangeamps.com/forum/icon_smile_wink.gif (http://javascript<b></b>:insertsmilie('[;)]');)

No I hadn't noticed that new Fender amp. 3.4kg 250w? You could very well be right about it being another class D amp. I guess we'll be seeing a lot more of them popping up in the near future - one reason I'm keen to get a handle on this stuff!

Think I'd still prefer the 'ultra lightweight' (ie well sub 20 kg) 250w all valve Koch bass head we were discussing though... :drool

A few more :) 'ironies' (technical and otherwise) which have come up on the Orange forum lately:
Quality MFD built speakers. (As in high quality...!)
'Hi-fi'. (Since Hi-fi speakers are usually tweaked to sound nice rather than accurate, why isn't it 'Ni-fi'?)
Old amps which haven't been played for a long time, so the filter caps "will be fine"...http://www.orangeamps.com/forum/icon_smile_wink.gif (http://javascript<b></b>:insertsmilie('[;)]');)
"No gain, no pain" was a great quote there I thought!
Shure SM57s being much better amp mics than '58s. (I've always found that 58s work pretty much like '57s when you take the shield off - as near as matters anyway. Justified according to the FAQs on Shures site...)
http://www.orangeamps.com/forumtest/topic.asp?TOPIC_ID=7802&whichpage=2
Nothing rhyming with Orange. (True AFAIK, though Kurt Vonnegut managed 'door hinge' in one of his books!)
Oranges being available in black...http://www.orangeamps.com/forum/icon_smile_cool.gif (http://javascript<b></b>:insertsmilie('[8D]');)

Thanks for the feedback - much appreciated! Starting to get to grips with it a little bit now. Any more thoughts on the Crate?

Andy.