Well, we're on subjective territory now, so there is a good chance we may not all agree.
I agree with you that most P.A. systems sound utterly awful. In my experience, the single worst item in the typical small P.A. chain is the utterly horrid Shure SM58 microphone. It is simultaneously the world's most popular vocal microphone, and IMO, the world's most awful-sounding vocal microphone. I've never been able to figure out how it became so popular.
Also in my experience, the second worst item in the live P.A. chain is the compression driver, horn-loaded mid/tweeter used in the vast majority of P.A. speaker cabinets. Most horns sound utterly horrid, so I suppose they are a good match to the equally horrid Shure SM58. I have heard good P.A. speakers using horns, but they usually have prices with four digits in them (in American dollars.)
So my recipe for a good-sounding small P.A. starts with these two steps: No Shure SM58s allowed. No compression horn tweeters allowed.
What's left? That depends on how loud you want to be. I use a 30 watt Acoustic AG30 as my P.A. system. It uses an 8" woofer with a coaxial post-mounted dome tweeter, like many better in-wall home sound systems, or a larger version of the speakers in your car. It is certainly not as good as the studio monitor speakers I use for music listening at home, but it sounds far better than the majority of P.A. systems I've heard.
The catch is that if you want ear-bleeding SPL levels, you are not going to get them from this P.A. But I've used it for acoustic guitar and vocals to a crowd of about 75 people, indoors in a fairly large room, and it worked fine in that situation.
And what about the "No SM58s" rule? I use a quite affordable handheld condenser mic for vocals: Amazon.com: Nady SPC-25 Condenser Microphone: Musical Instruments
Brand-name-snobs will snicker at the idea of using a Nady mic, but if they use their ears instead of their eyes, they will find out the SPC-25 sounds so much better than the usual SM58 that it is almost shocking.
Certainly, but there is more to it than that. Singing loudly compromises many aspects of the human voice. That's why classically trained tenors sound like angry bulls, and classically trained sopranos sound like a ruptured high-pressure gas main, and you cannot make out the words that either of them are singing, even if you know that language.
In recent years, the decline in music education, and the rise of TV shows such as "The X-factor" have popularized belters - singers who shriek, bellow, and scream their way through songs.
But there is more to singing than belting. Take Sarah McLachlan's "Building A Mystery" as an example ( YouTube ) - that song would sound utterly ridiculous if she had belted it. Not only that, she would not have been able to access those beautiful airy high notes in head-voice if she had been belting.
What about male singers? Yes, there is more than belting for male singers, too. Chris Martin of Coldplay is one example - he is one of the few male singers to frequently use high notes in head-voice, and that is one of the reasons he can access higher notes than most male belters.
Incidentally, singing loudly over a period of years very often leads to a damaged voice - typically hoarse, harsh, and limited in range. If you've heard Don Henley's harsh croaking in recent years, you're hearing the result of belting out high notes for years. Quite tragic.
I agree that a lot of P.A. systems make singers and instruments sound worse.
However - personally, I'm glad singers now have the choice to sing operatically, to belt, or, if they choose, to sing with more subtlety and expressiveness, using a microphone to assist. Those who prefer to sing without electronic assistance still have the option to do so.
-Gnobuddy
Strange, I managed to make a SM58 sound good through a PA, mind you it was a real PA with compression horns, horn loaded mids, folded horn bass cabinets. I have been to many a concert where SM58's reproduced the voice just fine. Lastly at a house concert for about 30 people through a Yorkville system. I do not know which model it was but I am pretty sure it was a 10" low frequency driver. Vocals with a guitar and drums, the guitar player also had some bass going at times using his feet and a synth. Not sure how they did the bass.
Yorkville
Yorkville
That's a very perfectionist point of view.
Do you really hear 0.01% rise to 0.02%? I say no.
In fact, from decades of 'professional' observation of real listeners, several % distortion often goes unremarked.
And in many situations, a system is "not loud" unless it is run past 5% THD. They were sometimes as happy with 10W as 100W, as long as they got the obvious "loud" from distortion.
10% is of course generous to the Marketing Dept. 5% primarily 2nd (naked SE triode amp) is really inaudible. 3% 3rd (naked P-P triode) doesn't really sour the music. The more complex distortions of high NFB amps do deserve lower THD numbers for the most critical listeners. BUT if a listener is THAT critical s/he should buy an amp so big that it never gets near clipping or high THD. (Ah, but battery size/cost/runtime set very strict limits on "so big".)
Also: your "abrupt kick-up" gives a heavy advantage to un-clean amplifiers. That "clipping rise" is very nearly the same for any modern amp: just clipping. However if a sloppy amp does 1% at best (rarely audible) before it rises, it will earn a higher "honest Gnobuddy" rating than an amp which gets down to 0.01% before the clipping-rise.
Attachments
Last edited:
Possibly. Maybe you just discovered something about my personality.
I disagree. Unclean amplifiers are unclean because of low (or no) negative feedback; they never show the abrupt "hockey-stick" bend from a nearly horizontal THD vs power curve, to a nearly vertical one.
As an example, Printer2 has posted THD vs power curves for a 51 year old (1967) SS class B power amp design called the Lil Tiger a few times. I'm attaching the relevant graph to this post as well. Negative feedback is low by modern standards, because of the limited open-loop voltage gain of the design - and the result is a THD curve with no abrupt knee in it at all. It just curves smoothly upwards.
So, power at the onset of clipping, or power at 10%? In a way it comes down to which form of unfairness you prefer: be unfair to the excellent low-THD abilities of the amplifier, and so advertise the 10% THD power output? Or be unfair to the maximum possible amount of extremely dirty power that can be squeezed out, and instead be fair to the excellent low THD performance the amplifier is capable of if you turn it down just a couple of dB?
The question is largely rhetorical, in the sense that the marketing world has spoken, and they have plumped for the "how much very dirty power can we possibly squeeze out?" approach.
(Anyone else finding multiple blank lines are now being automatically inserted between paragraphs when you hit "preview" or "submit" on this forum? That's what caused the enormous whitespace between paragraphs in this post. I'm using Firefox 60.0.1 / Xubuntu Linux.)
-Gnobuddy
Attachment needs new paperclip.
What is your rating of the 1952 Hafler-Keros? 1W? 13W? 19W? 29W?
(FWIW--- I do agree it looks like it was plotted by tracing a counter-edge, not actual points...)
Attachments
Last edited:
I built guitar amps with tubes from the age of around 10. I had tinkered with solid state designs but never found anything that I liked well enough top make a cabinet for. That changed when I built a Lil Tiger power amp driven by a DIY germanium preamp. It sounded better than some of my tube stuff to my 15 year old ears.
Note that "better" probably meant "different" since most of my tube amps were somewhat similar. I made a few more guitar amps based on the Lil Tiger, the Plastic Tiger and the Universal Tiger, but none had the character of that first one. There were two unique elements in the first one that were not in the subsequent amps. The original used all PNP germanium transistors. It also used a 15 inch field coil speaker lifted from an old console organ (probably a Hammond since I remember that two of us couldn't lift it). The field coil was energized in a manner that can not be discussed here. You can figure it out.
The germanium imparted a soft clip distortion characteristic all its own, and the speaker had it's own sonic quirks as well.
Note that there has been much discussion about the audibility and characteristics of THD, but the main nastiness distortion component in many guitar amps is IMD. This is why you can play lead (one note at a time) through a Marshall cranked to 30% THD, but chords (3 to 6 simultaneous notes) sound really nasty. Power chords (2 or 3 carefully chosen simultaneous notes) are designed so that their primary IMD components are close to, or part of the scale being played.
It is almost impossible to have IMD without THD, but some levels of THD can be seen with relatively low IMD. The character of the active components and their biasing has a lot to do with this. I haven't seen many IMD plots for class D amps, but I bet that they turn into blenders rather abruptly, possibly before their THD gets out of hand.
I spent a lot of time measuring and optimizing IMD in linear cell phone and cell tower RF power amps. IMD is a major design criteria, while THD is not.
This is one of those debatable things. If you provide zero bias to a complimentary symmetry transistor output stage, it actually operates in class C, since it remains cut off till the input swings to roughly +/- one Vbe of the output devices.
So if you now bias it to just eliminate the dead spot in the output, technically you have just achieved class B operation.
Ideally you adjust bias for nearly uniform summed transconductance throughout the crossover region, and that usually involves a few mA of quiescent current - in devices that will be flowing several amps of peak current at full output. If the bias current is 2% of the peak current, (and the output devices are therefore conducting over, say, 185 degrees instead of exactly 180 degrees), is it class B, or AB?
Things are very different for the usual valve class AB output, where the quiescent bias current is as much as 70% of the peak current. Now there is no doubt at all that operation is class A at low power, shifting to class AB near full output power.
Before the audiophools got in the act and started worshiping class A, every audio electronics textbook or design book I read called the typical push-pull solid-state output stage - biased to just minimise crossover distortion - class B. Now people call them class AB, the implication presumably being "not as good as class A".
I'm not emotionally attached to either term, but to me, "class B" is a lot closer to the engineering reality when the device only conducts for the barest whisker more than 180 degrees of a cycle.
-Gnobuddy
Sorry, apparently I need a new brain. (It's been busy around here!)
Three thousand Jumbo Watts. With fries.
The '52 Hafler-Keros curve looks a lot like the '67 Lil Tiger one as far as overall shape goes. The signature of relatively low negative feedback designs.
-Gnobuddy
Attachments
So I think that the main objective here is to AVOID any clipping in our little Class-D battery-powered amplifiers because it sounds so bad. How is this done simply and effectively? Is a pair of Zener diodes in the input enough to eliminate the awful Class-D clipping? Or is there another circuit that can be used?
Class-D clipping looks awful on the scope due to hi-frequency artefacts - that are above the audible frequency band.
But what you see is not what you hear. So I generally would not call it sounding awful per se.
To avoid class-D clipping there are certainly a bunch of options.
I prefer a long tailed pair differential amp driving the symmetrical inputs of TPAxxxx.
But what you see is not what you hear. So I generally would not call it sounding awful per se.
To avoid class-D clipping there are certainly a bunch of options.
I prefer a long tailed pair differential amp driving the symmetrical inputs of TPAxxxx.
See post 92
Jim explains it there. The circuit is only slightly old.
There's a summary of clipping techniques in my post to this thread (#7). Plus some interesting variations on the soft-clipping-op-amp on Page 2.
Read those and then come back with questions. Too lazy, too sick and too tired to retype it all, sorry
Last edited:
Speaking only for myself, I have heard extremely awful-sounding clipping from two entirely different class-D boards.
I did not put either on the scope to see what the waveforms looked like during clipping.
-Gnobuddy
You can choose to bias a SS amp anywhere from full Class A to full Class B. The thing is that Class AB SS amps sound ok when you get rid of crossover distortion. If you bias a tube amp at that point, maybe not so much. But just think how much power you can get out of a pair of piddly little tubes if you did.
I have one speaker like that. It is a very cheap 8" dual-cone, "full-range" P.A. speaker. In reality, of course, it is anything but full range - I measured the fundamental bass resonance at around 130 Hz, and of course it doesn't do proper high treble either. Really it is just a midrange speaker, just as you'd expect from its low price and intended usage (background music and public address voice announcements in shops and businesses, at the lowest possible cost.)
The thing is, with the right EQ curve, a clean preamp, and my little 2-watt 6AK6 push-pull power amp section, that particular speaker makes my Les Paul clone "speak" with a very vocal, almost human-like tone, a bit like what I think is a very expensive archtop jazz guitar in the opening music to the old TV show "Frasier".
In this clip, listen to the guitar lick at around 0:50, just as Kelsey Grammar sings the words "Frasier has left the building" : YouTube
That's close to the sound I was getting with the super-cheap 8" speaker. Not what you expect from a Les Paul type guitar, is it? But it worked remarkably well with some of the music played at my regular jam meetings, and was certainly the most unique guitar tone at any of those jams.
I couldn't even come close to that tone with any other speaker I tried.
Back on topic: for the portable guitar amp I'm building, I made sacrifices to the gods, read my tea-leaves, turned over my tarot cards, and decided to take my chances using a lithium-ion cordless tool battery I already have. Because these batteries can be very dangerous, and these days there are bizarre legal risks involved if you provide Internet information that, when used by someone else, causes harm to person or property, I won't go into any more detail than that on what exactly I did.
It's probably okay to add one thing: I settled on Anderson Powerpole connectors to use with the battery pack (and, of course, for the mating connectors I will attach to my class D amp board.)
A bit more info on Powerpoles here:
1) Anderson Powerpole Installation Secrets
2)Anderson Powerpole and SB Connectors | Powerwerx
-Gnobuddy
I think one of the reasons people bias output tubes "hot" is because grid current flow through the usual control-grid coupling cap biases them colder and colder as you start to overdrive them, and crossover distortion rises enormously as a result. The crude and simple fix is to bias them far into class AB when quiescent.
Perfect for 1935...but what about today? I think this is one of the reasons why George (Tubelab) can get so much power out of piddly little tubes...bias them just enough to get rid of crossover distortion (like transistors), then drive them with MOSFETs so there is no grid coupling cap, no bias shift, and crossover distortion doesn't soar when you start to overdrive them.
-Gnobuddy
What are you doing to prevent the damage caused by over-discharge (reported by TubeLab George): "Lithium Ion (LiIon or Lion) was the original lithium chemistry used in phones and laptops. They all have a minimum discharge voltage. The cells will be irreversibly damaged if discharged below this voltage, and any apparatus using these cells should self disconnect before the battery reaches this voltage."
But you are assuming overdriving a tube amp. I am talking tube amps in general. Back in the day when tube amps for audio reproduction was the norm the amount of bias is roughly the same as today. The datasheets almost never recommend a bias of under 50%.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.