"Single Ended Push Pull Amplifier"

Started by March 2, 2014
Just for the hell of it I decided to post this today. It's a little brainwr=
eck of mine. I have looked around enough that I THINK it is unpatentable bu=
t I THINK it is a different implementation than usual, or at least I haven'=
t seen it.=20

I have flippantly or lovingly (take your pick) dubbed it the "Class T" ampl=
ifier for reasons beyond the scope of this text. (details available on requ=
est)

https://dl.dropboxusercontent.com/u/29948706/classt02.jpg

One thing not shown there is the line from the power oscillator to the driv=
e circuit which is needed to sync it. The idea is not complicated, the tria=
c is triggerred just before zero crossing. The oscillator runs at about 60 =
Khz or so to facilitate use for the audio spectrum.=20

To drive positive, the trigger pulse during that phase is advanced with res=
pect to the output of T1. To trigger negative of course the trigger is adva=
nced before the input power switches back to positive.=20

In the waaay back burner of my mind is the drive circuitry. I would like to=
implement it totally in the analog domain. Basically a ramp generator and =
comparators or something along those lines. Sample and hold of course, I gu=
ess it would be a good idea to do that at the power oscillator frequency eh=
?=20

I think I can make this thing even cheaper, but for certain applications. F=
or example a powered subwoofer. If you adhere to proper design for a "hot" =
ground, T1 could be elimintated and the squarewaves from the oscillator cou=
ld be fed directly to Q1. For triggering, optocouplers could be used with w=
hatever circuitry (surely doable)and we're down to three power devices, not=
a one of them spending a microsecond dissipating, a coil and a few capacit=
ors. The cap up there coupling into T1 would feed Q1 directly but would hav=
e to be large enough to bypass the lowest frequency to be amplified. The wh=
ole thing would be easier to design with T1, but if push came to shove it c=
ould go, and then the whole thing needs to be isolated. If it's an amplifie=
d speaker, that would entail satisfying the UL or whoever regarding the enc=
losure. Still doable.

I could see this as useful for sound reinforcement, like bands on the run s=
etting up at different venues. The amplified speakers do eliminate another =
thing for the roadies to carry. The cabinet would have to be ruggedized jus=
t to make it safe, as internally the components would have path and potenti=
al to earth. But it could be done that way.=20

In this case, like with most (audio) power amps, I think it would be a wast=
e of time to try to regulate the power oscillator. The only efficient way t=
o do that is by running it above resonance and controlling the frequency. T=
his would cause complication in the drive circuit I think, and believe that=
a constant frequency is very desirable. Also, that matters in the choosing=
of L1 and C2.=20

Down there, C1 and R1,2 are to assure the triac turns off on time. The resi=
stance is there strictly to keep the peak currents down, and there are two =
because I think that R1,2 opening would not be a really desirable failure m=
ode. In fact detecting that for a shutdown would be good, but let's make it=
work first before figuring out how to make it stop... (one of my sayings, =
number 23 I think, I could look it up if you insist)

I'm pretty sure this is unpatentable because there is enough of this going =
on in directly fed VFDs for motors. Instead of plunking down the hundred bu=
cks for a provisional, I looked and found there are VFDs that do not even b=
other to rectify the AC. They run on three phase and convert frequency with=
triacs. Actually I was kinda impressed, and the thing is super efficient. =
I state now though, that those VFDs were not the inspiration for this, I ha=
d no idea such a cicuit existed until I actually looked, as I was wondering=
if this "Class T" doodad was worth the hundred. Once I saw that I figured =
that if someone hadn't done it yet, they coud easily enough, so it would be=
somewhat like trying to patent the wheel.=20

Of course when corporate gets through, they WILL patent the wheel and anyth=
ing that rolls is going to generate a royalty for them. Look man, I live in=
the US and every time I say "Now I've seen everything" they come up with m=
ore shit, especially in law. That is also beyond the scope of this text bec=
ause it would be a BODY of text akin to the works of Hamurabi. You folks in=
the good countries have no idea of the crazy shit that is illegal here.=20

Anyway, I guess it goes without saying that I submit this figuring someone =
will at least comment. You might call me a madman, but remember Madman Munt=
z ? People could not believe his TV worked with so few tubes. Now I have an=
amp with only one output device per channel, but not that old inefficient =
class A, and it can't really be called class B, or AB. It is certainly not =
a conventional class D, and it certainly isn't class G or H.=20

Thus - class T.
Always nice to see new ideas!

Not just VFDs, I have seen telephone ring generators syntheise 20Hz 90V ring buss from incoming 60/50Hz AC using a triac. So your idea can work at low frequencies.

Your LC filter may not work as you intend, you may find that C/R1/R2 needs to a bidirectional switch, think of how a buck convertor works.

Finally many triacs have problems commutating at 400Hz let alone 60kHz. Back to back mosfets would be more workable.

I guess others will have much more to say too.

On Sun, 2 Mar 2014 07:54:48 -0800 (PST), jurb6006@gmail.com wrote:

It has a problem with sustaining the inductor current when the triac is off. It would work better if the triac were replaced with mosfets, working synchronous-rectifier style. Lower losses, too. Similar things have been done in multiple-output switching supplies, to improve the regulation of outputs that aren't in the main feedback loop, but those are unipolar outputs so they can use a catch diode to sustain the inductor current. -- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation
jurb6006@gmail.com says...
> > Just for the hell of it I decided to post this today. It's a little brainwreck of mine. I have looked around enough that I THINK it is unpatentable but I THINK it is a different implementation than usual, or at least I haven't seen it. > > I have flippantly or lovingly (take your pick) dubbed it the "Class T" amplifier for reasons beyond the scope of this text. (details available on request) > > https://dl.dropboxusercontent.com/u/29948706/classt02.jpg > > One thing not shown there is the line from the power oscillator to the drive circuit which is needed to sync it. The idea is not complicated, the triac is triggerred just before zero crossing. The oscillator runs at about 60 Khz or so to facilitate use for the audio spectrum. > > To drive positive, the trigger pulse during that phase is advanced with respect to the output of T1. To trigger negative of course the trigger is advanced before the input power switches back to positive. > > In the waaay back burner of my mind is the drive circuitry. I would like to implement it totally in the analog domain. Basically a ramp generator and comparators or something along those lines. Sample and hold of course, I guess it would be a good idea to do that at the power oscillator frequency eh ? > > I think I can make this thing even cheaper, but for certain applications. For example a powered subwoofer. If you adhere to proper design for a "hot" ground, T1 could be elimintated and the squarewaves from the oscillator could be fed directly to Q1. For triggering, optocouplers could be used with whatever circuitry (surely doable)and we're down to three power devices, not a one of them spending a microsecond dissipating, a coil and a few capacitors. The cap up there
coupling into T1 would feed Q1 directly but would have to be large enough to bypass the lowest frequency to be amplified. The whole thing would be easier to design with T1, but if push came to shove it could go, and then the whole thing needs to be isolated. If it's an amplified speaker, that would entail satisfying the UL or whoever regarding the enclosure. Still doable.
> > I could see this as useful for sound reinforcement, like bands on the run setting up at different venues. The amplified speakers do eliminate another thing for the roadies to carry. The cabinet would have to be ruggedized just to make it safe, as internally the components would have path and potential to earth. But it could be done that way. > > In this case, like with most (audio) power amps, I think it would be a waste of time to try to regulate the power oscillator. The only efficient way to do that is by running it above resonance and controlling the frequency. This would cause complication in the drive circuit I think, and believe that a constant frequency is very desirable. Also, that matters in the choosing of L1 and C2. > > Down there, C1 and R1,2 are to assure the triac turns off on time. The resistance is there strictly to keep the peak currents down, and there are two because I think that R1,2 opening would not be a really desirable failure mode. In fact detecting that for a shutdown would be good, but let's make it work first before figuring out how to make it stop... (one of my sayings, number 23 I think, I could look it up if you insist) > > I'm pretty sure this is unpatentable because there is enough of this going on in directly fed VFDs for motors. Instead of plunking down the hundred bucks for a provisional, I looked and found there are VFDs that do not even bother to rectify the AC. They run on three phase and convert frequency with triacs. Actually I was kinda impressed, and the thing is super efficient. I state now though, that those VFDs were not the inspiration for this, I had no idea such a cicuit
existed until I actually looked, as I was wondering if this "Class T" doodad was worth the hundred. Once I saw that I figured that if someone hadn't done it yet, they coud easily enough, so it would be somewhat like trying to patent the wheel.
> > Of course when corporate gets through, they WILL patent the wheel and anything that rolls is going to generate a royalty for them. Look man, I live in the US and every time I say "Now I've seen everything" they come up with more shit, especially in law. That is also beyond the scope of this text because it would be a BODY of text akin to the works of Hamurabi. You folks in the good countries have no idea of the crazy shit that is illegal here. > > Anyway, I guess it goes without saying that I submit this figuring someone will at least comment. You might call me a madman, but remember Madman Muntz ? People could not believe his TV worked with so few tubes. Now I have an amp with only one output device per channel, but not that old inefficient class A, and it can't really be called class B, or AB. It is certainly not a conventional class D, and it certainly isn't class G or H. > > Thus - class T.
Triacs have a tendency to fire on the other phase angle when you don't want them too. Jamie
<jurb6006@gmail.com> wrote:
I think I fiddled around with something like this about 25 years ago, but got stuck. My problem was that if the audio output was not being fed equally by the positive and negative portions of the high frequency square wave, then the transformer is essentially passing audio and it's size grows to that of a regular output transformer, negating the performance benefit of going to the 60 kHz frequency. I'm not sure if that same problem applies here, but it's something to consider. And I'm pretty sure Class T has already been used by someone else (maybe a bridged Class D amp where the PWM is realized by shifting the phase of one square wave drive against the other side of the H bridge).
Den s=F8ndag den 2. marts 2014 21.14.57 UTC+1 skrev Ralph Barone:
> <jurb6006@gmail.com> wrote: >=20 > > Just for the hell of it I decided to post this today. It's a little >=20 > > brainwreck of mine. I have looked around enough that I THINK it is >=20 > > unpatentable but I THINK it is a different implementation than usual, o=
r
>=20 > > at least I haven't seen it.=20 >=20 > >=20 >=20 > > I have flippantly or lovingly (take your pick) dubbed it the "Class T" >=20 > > amplifier for reasons beyond the scope of this text. (details available=
on request)
>=20 > >=20 >=20 > > https://dl.dropboxusercontent.com/u/29948706/classt02.jpg >=20 > >=20 >=20 > > One thing not shown there is the line from the power oscillator to the >=20 > > drive circuit which is needed to sync it. The idea is not complicated, >=20 > > the triac is triggerred just before zero crossing. The oscillator runs =
at
>=20 > > about 60 Khz or so to facilitate use for the audio spectrum.=20 >=20 > >=20 >=20 > > To drive positive, the trigger pulse during that phase is advanced with >=20 > > respect to the output of T1. To trigger negative of course the trigger =
is
>=20 > > advanced before the input power switches back to positive.=20 >=20 > >=20 >=20 > > In the waaay back burner of my mind is the drive circuitry. I would lik=
e
>=20 > > to implement it totally in the analog domain. Basically a ramp generato=
r
>=20 > > and comparators or something along those lines. Sample and hold of >=20 > > course, I guess it would be a good idea to do that at the power oscilla=
tor frequency eh ?=20
>=20 > >=20 >=20 > > I think I can make this thing even cheaper, but for certain application=
s.
>=20 > > For example a powered subwoofer. If you adhere to proper design for a >=20 > > "hot" ground, T1 could be elimintated and the squarewaves from the >=20 > > oscillator could be fed directly to Q1. For triggering, optocouplers >=20 > > could be used with whatever circuitry (surely doable)and we're down to >=20 > > three power devices, not a one of them spending a microsecond >=20 > > dissipating, a coil and a few capacitors. The cap up there coupling int=
o
>=20 > > T1 would feed Q1 directly but would have to be large enough to bypass t=
he
>=20 > > lowest frequency to be amplified. The whole thing would be easier to >=20 > > design with T1, but if push came to shove it could go, and then the who=
le
>=20 > > thing needs to be isolated. If it's an amplified speaker, that would >=20 > > entail satisfying the UL or whoever regarding the enclosure. Still doab=
le.
>=20 > >=20 >=20 > > I could see this as useful for sound reinforcement, like bands on the r=
un
>=20 > > setting up at different venues. The amplified speakers do eliminate >=20 > > another thing for the roadies to carry. The cabinet would have to be >=20 > > ruggedized just to make it safe, as internally the components would hav=
e
>=20 > > path and potential to earth. But it could be done that way.=20 >=20 > >=20 >=20 > > In this case, like with most (audio) power amps, I think it would be a >=20 > > waste of time to try to regulate the power oscillator. The only efficie=
nt
>=20 > > way to do that is by running it above resonance and controlling the >=20 > > frequency. This would cause complication in the drive circuit I think, >=20 > > and believe that a constant frequency is very desirable. Also, that >=20 > > matters in the choosing of L1 and C2.=20 >=20 > >=20 >=20 > > Down there, C1 and R1,2 are to assure the triac turns off on time. The >=20 > > resistance is there strictly to keep the peak currents down, and there >=20 > > are two because I think that R1,2 opening would not be a really desirab=
le
>=20 > > failure mode. In fact detecting that for a shutdown would be good, but >=20 > > let's make it work first before figuring out how to make it stop... (on=
e
>=20 > > of my sayings, number 23 I think, I could look it up if you insist) >=20 > >=20 >=20 > > I'm pretty sure this is unpatentable because there is enough of this >=20 > > going on in directly fed VFDs for motors. Instead of plunking down the >=20 > > hundred bucks for a provisional, I looked and found there are VFDs that >=20 > > do not even bother to rectify the AC. They run on three phase and conve=
rt
>=20 > > frequency with triacs. Actually I was kinda impressed, and the thing is >=20 > > super efficient. I state now though, that those VFDs were not the >=20 > > inspiration for this, I had no idea such a cicuit existed until I >=20 > > actually looked, as I was wondering if this "Class T" doodad was worth >=20 > > the hundred. Once I saw that I figured that if someone hadn't done it >=20 > > yet, they coud easily enough, so it would be somewhat like trying to pa=
tent the wheel.=20
>=20 > >=20 >=20 > > Of course when corporate gets through, they WILL patent the wheel and >=20 > > anything that rolls is going to generate a royalty for them. Look man, =
I
>=20 > > live in the US and every time I say "Now I've seen everything" they com=
e
>=20 > > up with more shit, especially in law. That is also beyond the scope of >=20 > > this text because it would be a BODY of text akin to the works of >=20 > > Hamurabi. You folks in the good countries have no idea of the crazy shi=
t
>=20 > > that is illegal here.=20 >=20 > >=20 >=20 > > Anyway, I guess it goes without saying that I submit this figuring >=20 > > someone will at least comment. You might call me a madman, but remember >=20 > > Madman Muntz ? People could not believe his TV worked with so few tubes=
.
>=20 > > Now I have an amp with only one output device per channel, but not that >=20 > > old inefficient class A, and it can't really be called class B, or AB. =
It
>=20 > > is certainly not a conventional class D, and it certainly isn't class G=
or H.=20
>=20 > >=20 >=20 > > Thus - class T. >=20 >=20 >=20 > I think I fiddled around with something like this about 25 years ago, but >=20 > got stuck. My problem was that if the audio output was not being fed >=20 > equally by the positive and negative portions of the high frequency squar=
e
>=20 > wave, then the transformer is essentially passing audio and it's size gro=
ws
>=20 > to that of a regular output transformer, negating the performance benefit >=20 > of going to the 60 kHz frequency. I'm not sure if that same problem appli=
es
>=20 > here, but it's something to consider. And I'm pretty sure Class T has >=20 > already been used by someone else (maybe a bridged Class D amp where the >=20 > PWM is realized by shifting the phase of one square wave drive against th=
e
>=20 > other side of the H bridge).
http://www.crownaudio.com/media/pdf/amps/bcapaper.pdf kinda like a buck sink and buck source connected -Lasse
OK, so what I was thinking of appears to be Class I. I guess everybody wants their amp to be "in a class all its own".
>"Always nice to see new ideas! "=20
Ummm, if I need to go to the patent office and get a provisional, say so no= w !=20
>"Your LC filter may not work as you intend, you may find that C/R1/R2 need=
s to a bidirectional switch, think of how a buck convertor works. " I'm not getting what you're saying here. That C1 and the resistors are only= there to make sure that Q1 turns off at zero crossing. The resistors are o= nly there to limit peak current when Q1 turns on, the cap makes it turn off= on time and not lag letting the inductor change the turnoff time. If you s= ee the circuit acting differently, please elaborate.=20
>"Finally many triacs have problems commutating at 400Hz let alone 60kHz. B=
ack to back mosfets would be more workable. " I have actually drawn that out, even SCRs might work. I could just use a br= idge recifier and use one device. I have also thought about guaranteed turn= off of any thrysistor type device, dependent upon the absence or reveral of= current to turn off. I could easily do that to this configuration if neces= sary. The problem with going MOSFET is that then the only advantage to the = cicuit is that the drive need no be optimized. It is just a little bit more= advanced class D circuit. I was hoping for more notariety.=20
>"It has a problem with sustaining the inductor current when the triac is o=
ff. It=20 would work better if the triac were replaced with mosfets, working=20 synchronous-rectifier style. Lower losses, too. " All I'm really concerned with is the current required to turn off the triac= .=20 Using the KISS theory, figure how a light dimmer works, or used to on incad= escents. It shifts the phase enough, and the pot advances it. think of that= simple ass circuit there, but instead of controlling it with a pot, keep t= he pot down to "bias" level and impress a voltage on it. Of course it is on= ly operating at 60 Hz, but you could put out DC, positive or negative, your= choice.=20 You do bring up one of the hangups about this circuit though, the turnoff h= as to be reliable. the drive circuit cannot help, all it can do is turn on.= If we are talking audio, we are talking about loads that vary more than th= e time of day and shit. I mean you can't count on anything - IF this is to = be a commercial amp.=20 Of course if the output is FET or anything other than a thrysistor type dev= ice, turnoff is again in our ballpark. This is one of the things I hoped to= avoid. Any switching device has that problem, when it turns off. The diffe= rence though,m is that when supplied with AC, or pulsed DC, it turns off wh= en it damwell pleases kind and it does not burn it up.=20
>"Triacs have a tendency to fire on the other phase angle when you don't=20
want them too. " Really. Not that I ever tried to design stuff like this, no matter what I n= ever noticed that. In a light dimmer who woud care ? Even in a Carver 1.5T = who would care. It is either heating a skinny strip of tungsten (Wolfram) i= n an argon (Argon) environment. They have either thermal or capacitive mass= . thing here is, do you know how to suppress this errant behavior ? Or mayb= e are there units that are "guaranteed" to not do this ?=20 Really, if I go with regular devices it takes half the fun out of it. Hell,= I could just use regualr bipolars...
>"My problem was that if the audio output was not being fed=20
equally by the positive and negative portions of the high frequency square= =20 wave, then the transformer is essentially passing audio and it's size grows= =20 to that of a regular output transformer, negating the performance benefit= =20 of going to the 60 kHz frequency." As I pondered the usage without T1, it did occur to me that the coupling ca= p would need to pass the lowest frequency desired. It would be a trick to u= se this config for a servo or something that requies DC and then find the 1= nF or whatever just charges up and... What a bitch, though I am suire many = have stories of things like that happening. With T1 in circuit, I think you= might be able to disregard some of that, but not all. At really low freque= ncies, T1's core could saturate.=20 Either way, this all makes for problems reproducing low frequencies at larg= e current levels. Perhaps BTL is the way to go. It only doubles the ONE out= put device, they can both run off the same power oscillatior, and while I w= ould not even try to just flip T2 over, it won't work that way, it would NO= T dounble the complexity of the drive circuit.=20 More later. There are ways to get this done. It could be at 400 Hz and just= to a sub, or DC or motors in some machine, it could do a bunch of shit. th= e money is not in this configuration though, it is in designing a c hip to = run it. The trigger circuit.=20 But it does qualify as a single ended push pull circuit. Quite efficiently = I might add.
"Ralph Barone"
> > And I'm pretty sure Class T has already been used by someone else
**See: http://en.wikipedia.org/wiki/Class_T_amplifier T = Tripath. .... Phil
"Phil Allison" <phil_a@tpg.com.au> wrote:
> "Ralph Barone" >> >> And I'm pretty sure Class T has already been used by someone else > > > **See: > > http://en.wikipedia.org/wiki/Class_T_amplifier > > T = Tripath. > > > > .... Phil
So a high power output delta-sigma ADC/DAC combination?