| amplifierguru |
Conventional power amplifiers commonly feature a low voltage gain input differential stage followed by a very high voltage gain stage then usually a unity gain emitter or source follower. This design suffers from a distinct topological deficiency that degrades the sound and is responsible for inferior sound quality.
The typical gain distribution is shown in the thumbnail where the weakness of PSRR of the input stage ( not amplifier) due to the divider with stage output Z vs load Z determines intrusion of Class AB PS artefacts which are magnificently amplified by the very high gain Vas to output, blurring the stereo image detail and spatial info and raising the dependance for performance on the PS quality!
Better sound quality from such a design centres around fiddling at the edges of this flawed topology rather than adopting a new one. |
|
|
| amplifierguru |
Ignoring the 1K input impedance glitch in the hastily prepared drawing, feedback theory will dictate that this level of intrusion or sensitivity is -
Vs out proportional to A2 x A3/A1 and also to stage load Z/collector Z.
So to reduce this harsh power supply induced output commutation distortion from degrading the sound we need to -
1. Increase A1 without increasing load Z proportionally so CMs are out!
2. decrease A2A3 so the intrusion is amplified less.
So ideally most of the voltage gain should be in the input stage and not in an after injection stage and especially not in a following high gain miller comp stage which degrades load Z as well and results in the typical thumbnail response.
So far so good |
|
|
| AKSA |
But Greg.....
Your input stage here is still a diff amp!!
The conventional LTP is a voltage in/current out block; a transconductance configuration. It is the current output which normally drives the voltage amplifier. I find myself agreeing with some of your comments, but your block diagram still does not obviate use of a long tail pair.
Cheers,
Hugh |
|
|
| amplifierguru |
Hi Hugh,
We can't throw out the baby with the bathwater here - diff'l amps have many advantages , not the least being good linearity and low temp sernsitivity in comlementary topologies.
Just about stumps for me till tomorrow.
Cheers,
Greg |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by amplifierguru
We can't throw out the baby with the bathwater here
|
True, but "diff amps" are the bathwater. They are about 50% of the problem of the generic Amp topology (the rest is the VAS and teh Output stage).
Why not go for complementary folded cascode current feedback topology.
You can get huge amounts of gain in effectively one stage, no common mode induced distortion, you can use good quality complementary J-Fets (2SJ74/2SK170) in the input and if you wanted follow the (single) Input/VAS stage with a BJT Buffered (Emitter follower) Mosfet Output stage
Sayonara |
|
|
| Kuei Yang Wang |
Konnichiwa,
A picture speaks 1000 words, here a piccie of what I have in mind, principle diagram.
Sayonara |
|
|
| darkfenriz |
I see that positive feedback gets very fashionable these days.
:D :D
sorry for joking. |
|
|
| MikeB |
That's a good one ! (but darkfenriz was faster :D)
You might have overlooked that fc-topology is inverting, means you applied
a positive feedback, your circuit is not functional at all...
Mike |
|
|
| pinkmouse |
:cop:
Thread for Jam's new topology is here. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by darkfenriz
I see that positive feedback gets very fashionable these days.
|
| quote: | Originally posted by MikeB
That's a good one ! (but darkfenriz was faster :D)
You might have overlooked that fc-topology is inverting, means you applied a positive feedback, your circuit is not functional at all... |
Well, not so much really overlooked, it is a result of using "cut & paste" in a graphics programme to get a general indication.
Simply remove the NFB and keep the amplifier open loop, change the feedback application point to the common gates OR change the output stage from common drain to common source for one more inversion, probably a good idea anyway to swap common drain for common source.
Anyway, attached with changed feedback arrangement, you can still use easily 50K...100K input resistors (according to Borberly's work) and then scale the feedback resistor to suit. |
|
|
| mikeks |
| quote: | Originally posted by Kuei Yang Wang
........but "diff amps" are the bathwater. They are about 50% of the problem of the generic Amp topology........ |
But this is completely untrue.... |
|
|
| peranders |
| .. hm, but what is true then? |
|
|
| amplifierguru |
Hi Kuei Yang Wang,
G'day mate.
I disagree with your comment re diff'l amps and find your FET folded cascode impractical to DIY, wasteful of power, requiring multiple supply complexity, ...
Just for some more positive feedback!
Moving on. From my initial thumbnail pic in post 1, it is fairly obvious that the degradation in PSRR (bode plot) shown in post 2 thumbnail is due to the 3200 x miller C paralleling with the A1 load Z causing the atrocious -30dB at 10KHz. While 2 pole comp instead of the miller C can help, real improvement can be obtained by reducing the gain of the Vas e.g. with R to ground. This intuitively raises the signal level at the injection point relative to the PSR injection. This is probably why the low TIM followers perceive a benefit in loading the Vas.
It becomes fairly obvious that we are just fiddling at the edges of a flawed topology and that the Vas is 50% of the problem.
Cheers,
Greg |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by amplifierguru
It becomes fairly obvious that we are just fiddling at the edges of a flawed topology and that the Vas is 50% of the problem. |
You are familiar with Douglas Selfs work?
If acoording to me the differential amplifier is 50% of the problem and according to you the VAS is the other (which is probably about right, if we keep the output stage sufficiently identical) then if we remove the differential and the VAS we have solved 100% of the problem, works for me.
Sayonara
PS, if we are looking for really high performance (as opposed to tinkering at the edges) seperated, regulated supplies for the frontend several volt in excess of the output stage maximum rails are neccesary, as are suitable means to limit the amount of open loop gain available in case of output overload, so that recovery from clipping is clean, rapid and without "dead zones", so no matter what the rest of my discrete amplifier circuit would look like, the frontend powersupply would allways be seperate as I consider this at least highly desirable, if not essential. |
|
|
| amplifierguru |
Au contraire Kuei Yang Wang,
I find the dill'l input stage perfectly acceptable but I flexible on your complementary SE inputs (mostly needing a servo and have higher THD) and it's the EF or SF output stage that's the other 50% of the problem!
They, in combo with the Vas, are wasteful of supply voltage , and in a power amp that's power. More heating - larger heatsinks, larger x'former for a given power, higher voltage supply C's . Typicall 5V of output power wasted and more with 4 ohms like sub amps where power is a big issue.
But also it's the drive - they need to be driven full swing as they have no voltage gain. No Vas , how to drive them? Simply and from a low impedance that can pull them off for speed.
Cheers,
Greg |
|
|
| mikeks |
....the so-called generic topology.....,properly executed,......is in fact the most versatile and least flawed voltage gain-block out there...
This is ice-cold, hard, unyielding fact.......not idle speculation, or anecdotal gossip.... :) |
|
|
| AKSA |
Mikeks,
I find myself agreeing with you. I have built and tested them all, and a standard diff pair, with single ended VAS using a CCS as collector load, is the best overall gain block I've ever tested.
It has the huge advantage of impeccable offset control, essential with direct coupling; low phase shift; easily compensated high nfb with a ready made fb node, and general uncriticality of configuration. And while the bipolar is not a perfect device, the peculiar combination of current injection and Vbe change at the base of the VAS is very well catered to....
This is but a clear example of technological convergence - which also brought us the front engine-rear drive automobile, the gasring stove, the poppet valve internal combustion engine, the cathode ray tube. That is not to say these approaches cannot be dismissed - JLH, Jean Hiraga, Citroen, Bristol Aero engine company and a host of others have done these things other ways, but you'd better have a very good reason to deviate from the norm. In this connection there is much to be said for nibbling at the edges of existing technology and improving on present topologies with better dimensioning and component selection. It's not sexy, but it can still push the performance envelope.
Cheers,
Hugh |
|
|
| amplifierguru |
Hi Mikeks,
Every topology has it's shortcomings but the pedestrian conventional diff'l-Vas-EF/SF has more than it's share - and in particular, in an energy efficient world of the future (should've been yesterday!), the waste of 7V in an output stage or a 300VA transformer when only a 150W is needed or 30,000uF when only 15,000uF is needed because of PSRR weakness cannot be entertained. Hopefully the days of the audio snake oil salesmen are nearing an end and good design will out.
I would rather listen to the amp below (as I am now) with PSRR shown as I know I'm listening to the music, not the power supply.
That's 40dB or 100 times better than the prev. example and less expensive to build.
:D :D :D :D :D
Cheers,
greg |
|
|
| amplifierguru |
Not only that the sound is effortless, transparent and at 0.005% 10K THD is 1/10th that of the simple Vas cct.
I'm in heaven.
Cheers,
greg |
|
|
| CBS240 |
In playing around with a triple Darlington design, for lower freqs. and lower impedance for a sub and woofer combo, less than 1 or 2 KHz. I find that having a second PS seems to achieve better results. The output stage uses +/-55V. For the front end, I have used a separate much smaller 400mA transformer in a voltage doubler(half wave rectifier) to produce +/- 72V. This then goes through a small regulator circuit to make a clean +/-60V. The 12V difference allows plenty of room to absorb the added ripple from the halve wave. Since this is low current here, there is very little heat to dissipate. Given this circumstance, I don't think the half wave rec. should affect the PSRR that much. The first pre-driver utilizes this PS source. The second driver and outputs use the +/-55V source, unregulated of course. This also can compensate a little for the voltage lost in the 3 pn junctions, re of each device, and the emitter degeneration resistors, and at least keep the first driver out of heavy saturation. well, it you don't push it into hard clipping.
BTW, I'm using a simple single diff pair, well with mirrors, and a CCS loaded VAS. I feel that as ASKA stated, this gives reliable DC offset, which is important here. Since I am not looking for speed in this design, added nfb and miller comp. is not such the drag. The idle sound is quite clean, even through a 5 way speaker! Not much FFIIIZZZZZZZZZ at all. A seperate smaller amp will be used for higher freqs. so this circuit will not be exposed to tweeters, which is were you would here more deplorable fffiiiiizzzzzzz.:D |
|
|
| amplifierguru |
Hi Hugh,
I missed you there.
Hi CBS240,
Sure you can go to all this complexity - but I'm emphasising the simple solution, topologically(?).
It's not about Fizzzz in the background. It's the distortions from the PS that are generated and enter during dynamic operation, that add the edge.
DC is not an issue. The design can be AC coupled for similar tolerances, but my preference is for DC coupling , so I'm not listening to capacitor smearing and that involves setting at 0mV cold, running it up to see where it goes when hot. If it's +30mV then you reset it at cold for -15mV so it averages around 0mV. That's the setting up on DC coupled. Not much pain, for no FB cap.
But just as you're not listening to the PS with all you're PS mods, I'm not listening to it with more simplicity than a conventional simple Vas but I canget some 50W more power or headroom for less cost and 1/10th the THD at 10KHz.
Cheers,
greg |
|
|
| The Saint |
Hi greg
What is the spice sofware that you use.
Are you going to show us the circuit or just keep teasing us with it? :D
Looks like CFP output stage with a bootstapped VAS..
Cheers Mate |
|
|
| amplifierguru |
Hi Anthony,
No worries. The conceptual topology is posted http://members.dodo.com.au/~gregball/guru_003.htm
There are some 7 claims in the patent application covering the bias adjustment, DC offset adjust and thermal compensation. The devils in the detail. Obviously there are zeners on the gates to supply. All the details on the functional schematics are in the kit.
Note the 30mW 200MHz+ small signal drivers that pull the gates
off.
Luckily I have an audiophile lawyer mate who's really litigious and into international IP. A pit-bull.
Cheers,
greg |
|
|
| Steve Eddy |
| quote: | Originally posted by amplifierguru
Sure you can go to all this complexity - but I'm emphasising the simple solution, topologically(?). |
If you want to emphasize the simple solution topologically, as I said over in jam's split portion of the thread, it don't get much simpler than this:
So I might well say, "Sure, you can go to all this complexity - but..." :)
se |
|
|
| Workhorse |
Congrats Greg..on your mos##amp , I like it...
On More thing you too use Circuitmaker for simulation ..is'nt its great....
By the way if i want to parallell more mosfets at output,shouln't they require an extra buffer to compensate high drive loss at high frequencies...
K a n w a r |
|
|
| Ultima Thule |
| quote: | Originally posted by amplifierguru
Hi Anthony,
No worries. The conceptual topology is posted http://members.dodo.com.au/~gregball/guru_003.htm
There are some 7 claims in the patent application covering the bias adjustment, DC offset adjust and thermal compensation. The devils in the detail. Obviously there are zeners on the gates to supply. All the details on the functional schematics are in the kit.
Note the 30mW 200MHz+ small signal drivers that pull the gates
off.
Luckily I have an audiophile lawyer mate who's really litigious and into international IP. A pit-bull.
Cheers,
greg |
Hi Gregory,
so, when are you going to file an associated application to gain any intellectual property rights for the invention? :scratch2:
btw, It's a neat design, I like it!
Cheers Michael :) |
|
|
| amplifierguru |
Hi Steve,
You're a #U*&^% genius! Why didn't I think of that.
Hi Kanwar,
Thanks, sometimes it's the simple things. You should see the bias/offset/thermal solutions!!! Yeah, I think I like circuitmaker - except when it breaks tracks when you make changes. Very frustrating trying to fault find a straight line that reads different at both ends.
Gate current draw is about 1.5mA/pair at 10KHz full power from an 8-10mA biassed driver giving a fair slew margin. But at 30mW on a small signal pkg, what the hell - I could run them harder!
So it's R2R and GREAT PSRR - I can use a 170W x'former and 15,000uF a side for a 150W monoblock - and it'll hit 210W before clip (rms) on music crest factor. And it's 100 times less sensitive than a Vas type to those supplies so they can bounce around.
Cheers,
greg |
|
|
| amplifierguru |
Thanks Michael,
The legals/IP are being handled by my lawyer mate, he's setting up a home theatre and wants 7 monoblocks. So he's earning them.
It's the simple things in life - like a handful of gain spread BC's, some hexfets and too much time and passion.
Cheers,
greg
:D |
|
|
| planet10 |
| quote: | Originally posted by amplifierguru
in an energy efficient world of the future (should've been yesterday!)... |
... we'd all be using digital amplifiers....
dave |
|
|
| Ultima Thule |
Greg,
HEXFET's, do you use (IRF) vertical FET's in the output? In such cae I wonder wethere you have Q5 and Q6 thermaly tied to the output FET's Q7, 8 9 & 10 because of their positive tempco, if thermaly tied together it will compensate a bit for thermal runaway in those output FET's.
When thinking of thermal compensation, it hits me that also if Q2 and Q4 would be thermaly connected to the output transistors mentioned above will give thermal compensation as well.
And there's more posibilities too actually when one get's passionated thinking over it... :D
The above applies to the attached schematic here below fetched from your link in your earlier post in this thread, hope that's ok for you Greg.
What is the overal gain for the whole circuit before FB is applied?
Tell me about C2 and C3, are they small values in order of pF's or so, or is it such a high value it can be considered as a bootstrap capacitor?
Cheers Michael |
|
|
| amplifierguru |
Hi Michael,
I designed it around IRFs! The 3-4V gate voltage. In fact the tests I did covered 100 prs of 2 batches and determined tempco which are compensated for in the final. Yes - looking for the tempco comps was fun and, if you look at the blurb on my website I indicated a variety of tempco comps. At least three.
Q5,6 only provide a fraction of the measured tempco so I have devised simple, largely existing, alternatives.
The C's 2 andf 3 are serious bootstrapps. They are larger than needed for response boost , for PSRR boost!
Done the hard yards.
Cheers,
Greg |
|
|
| lumanauw |
Hi, Amplifierguru,
Isn't that bootstrapped is bad from PSRR point of view? Your design stresses on good PSRR, but you use bootstrapped? |
|
|
| AKSA |
Hi Greg,
I really like your topology. I believe it solves many of the usual problems...... And the use of emitter followers to drive the gates is very appealing to me, and the simple use of a bootstrap enables constant current operation, high linearity, while accommodating very good offset control.
Try using different values for the complementary bootstrap caps. It will sound quite different and you might even like it.......:devilr:
Yes, it looks very good, congratulations!
Just how vigilant is your lawyer, and how does he perform in out of state jurisdictions??? :clown:
Cheers,
Hugh |
|
|
| Allexx |
Guys,
Schematics in post #6 is absolutely OK. It is amplifier with NEGATIVE feedback. Darkfenriz and MikeB are mistaken. Common base stage in cascode DOES NOT INVERT the phase of the signal. If it is so hard to see it just draw schematics of usual cascode and try to prove 360 degrees phase inversion. |
|
|
| Allexx |
| Sorry, I was wrong because for feedback signal injected into source it will be positive... |
|
|
| Kuei Yang Wang |
Konnichiwa,
I rather like it in many ways.
But I retain that the differential Amp's are troublesome.
If you can do me the honours and just forget for a moment that they are "differential amplifiers", which in your case is not the case.
Let me restate for you your Input/VAS topology:
"A pair of rail symmetrical single ended amplifiers are used as voltage amplifiers with the feedback signal applied to the emitters using emitter followers."
One may ask why not return the feedback signal to the emitter without buffering? And why use BJT Inputs?
If do that we arrive at any one of a number of variation of Mosquite/Buzzquito/Zenquito et al which are popular with the French, with rail symmetric differential or rail symmetric single ended frontends....
http://perso.wanadoo.fr/jm.plantefeve/
Sayonara |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by Allexx
Sorry, I was wrong because for feedback signal injected into source it will be positive... |
It will work if we do it that way....
There is BTW a very neat possible qiescent current compensation scheme possible in this circuit (not drawn), by making the current sources suitably active with a DC loop back from the source resistors, bootstrapping could also be employed, but is probably not needed.
Sayonara |
|
|
| CBS240 |
Hi Guru,
This topology was obviously designed for Fet outputs, I am using BJT's, mostly because I have plenty of them. I have grown fond of the EF stage instead of feedback pair(although yours is not exactly FP) for lower impeadence and thermal stability because of the very positive tempco of BJT's. But your topology idea could be used there as well i suppose. Drivers could probably be smaller.:xeye: and freq responce will be higher. With FET's you loose the Vgs with SF, which is much greater than Vbe in EF, so it makes lots of sense to your design. :) |
|
|
| MikeB |
| quote: | Originally posted by Kuei Yang Wang
Konnichiwa,
It will work if we do it that way....
|
I don't think so. This time the feedback is okay, but your biasing is in
no way defined ! You can't have work 2 currents against each other
and expect a defined voltage... It would work if you replace the 2 ccs
with simple resistors or use your mentioned "dc-servo".
Sorry to be annoying,
Mike |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by MikeB
I don't think so. This time the feedback is okay, but your biasing is in no way defined ! You can't have work 2 currents against each other and expect a defined voltage... |
First, the circuit as drawn is meant as "simplified principle" diagram (as are all the others), secondly, if I select the CCS J-Fet's for sufficiently identical and appropriate Idss (the frontend FET's BTW also would need selection) I CAN expect defined biasing just fine. The exact biasing would then be determined by the frontend source resistors.
| quote: | Originally posted by MikeB
Sorry to be annoying, |
You are not annoying, you are merely completely wrong, which is no problem to me.
Sayonara |
|
|
| jam |
Hi Mike,
I am working on the exact same circuit (without the current sources) as you describe, except with mosfets and a cascoded input fets.
MikeW even made PCB's for me and I have a working prototype which still needs some work. I will try to post the schematic this evening.
I will have to disagree with most of you about the validity of this circuit, I think that it has a lot of potential and have to agree with Kuei Yang Chan on this one..........Sorry! :angel:
Regards,
Jam |
|
|
| mikeks |
| quote: | Originally posted by AKSA
Mikeks,
I find myself agreeing with you. I have built and tested them all, and a standard diff pair, with single ended VAS using a CCS as collector load, is the best overall gain block I've ever tested.
|
True in all respects.....
I fear the main reason the generic (Thompson) design is derided by audiophiles is it doesn't look as 'sexy' as the so-called complimentary designs, or push-pull TIS configurations....
Goes to show...looks aint everything...!!! |
|
|
| jam |
Here is what I have so far.................................:xfingers:
I am sure Carlos has waited for this for a while..........;)
Comments anyone?.....................This is still a work in progress. :o
Regards.
Jam |
|
|
| jam |
| This is the board that MikeW laid out.....................;) |
|
|
| amplifierguru |
Thanks all for the positive comments.
Hi CBS240,
I really designed this around the IR MOSFETs which, if driven in the Vas/SF fashion would waste about 7V of output c.f. < 2V here.
All the other transistors are small signal 500mW high Hfe such that the EF driver stage can be 300MHz and Hfe >500 needing only 20uA base drive current for 10mA Class A standing current and dissipates avg 35mW!
Hi Kuei Yang Wang,
I have covered complementary single ended FET (and BJT) input stage in the patent application as a possible configuration. But it is more problematical for DC offset even in an AC coupled feedback config and really requires a servo which adds complexity. I might add that the complementary FET se input stage you favour will result in much higher THD (due to poorer transconductance /gain) and degraded PSRR due to higher collector impedance of BJT.
The higher power version GB300D uses a 'modulated bias' standard cascode with BJTs and can be FETs.
However, due to the ready availability of monolithically matched FET cascode chips, I prefer to develop the nested/compound feedback design using one such chip for the front end to a bipolar discrete stage as above.
Hi Hugh,
I have explored all manner of variation in the resistors being bootstrapped and leave that subjective stuff to constructors.
Hi Lumanauw,
If you do a sim of the topology you'll find that the size of the bootstrap C's directly affects the LF PSRR but the topology does it for the mid and high.
My aim here was to derive the topology that could provide a viable alternative to the conventional - without the glaring weaknesses. I don't expect it to be acknowledged - even Douglas Self could not acknowledge that PSRR was a problem, even when using a regulated supply dropped THD from 0.04% to 0.006% from memory. If that's not the flag... But maybe, just maybe, Wireless World were right when they sub-titled my article on PSRR "Audio Distortion - Holy Grail Located!"
High Mikeks, If your conventional topology is to work as well you will need more cost/complexity and even in it's basic form is more wasteful of power - it's a dinosaur!
Cheers,
greg
PS thanks for all the emails guys. |
|
|
| Workhorse |
Ok Greg,
I am convinced about high frequency operation of your circuit, it is really good, But having complementary pairs of mosfets at output poses on problem....If someone feels to use this circuit above 1KW++ level .....There would be a Hell with P-channel Mosfet, as they are not available in high power versions, thats why i am die-hard "Cieling Fan" of nvmos.......;)
Could you please elaborate in a better view , how the power supply hash doesn't contaminates this circuit as in traditional circuits with seperate VAS?
K a n w a r |
|
|
| thanh |
Hi amplifierguru ! I saw your schematic . VAS is a common collector stage . Iq through VAS will depend on +-Vcc , .Will this decrease PSRR ?
Are you using Multisim professional version ? I only has got student version . This version is too bad but orcas pspice can't measure PSRR |
|
|
| amplifierguru |
Hi Kanwar,
Yes it's really been designed to get the most from IRFP p and N channel but it might be possible to configure a CFP for the P channel.
The earlier pics I showed the sim results for the 2 topologies and the gain breakdown for the diff'l - Vas- EF/SF conventional approach, showing the inherent PSR at the diff'l /Vas node due to the Rl/Zc divider.
Below pic shows the very different gain breakdown within the loop of the new topology. It can be seen, if we look back from the output at, say, 20V rms output, the bootstrapped load sources 0.5V signal so PSR is relative to this whereas with the very high gain of the Vas, for a similar output level, the signal level at the Diff'l out/Vas in node is 160 times lower giving a poorer ratio of PSR intrusion to signal level. One way of looking at it. In effect increasing A1 while reducing A2A3 is the key - putting less of the gain after the intrusion point and more of the gain before.
hope that helps.
Notice that the drivers turn on to pull the MOSFETs off ?
Cheers,
Greg |
|
|
| X.G. |
hi,guru
good idea!
before about 10 years,I saw that a janpanese DIY expert had a series articles of DIY amp which less sensitive for the power supply in MJ magazine.his method is same as yours by increasinng the gain of the input stage and decreaing the gain of VAS....But he more likes complemetary topology:D ;)
IIRC,his last one of this type amp is complemetary J-FET (about +/- 70 Vcc) cap couple+ MOSFET source follower(about +/-30V Vcc),only one stage for amp volt:eek: ....his topology is so simple!:D
sorry for my direct words.....in my eye,
your amp=japanese DIY amp vari + EF + boostap+ Nelson Pass A5 without active current source;)
regards
X.G. |
|
|
| amplifierguru |
Hi Thanh,
I said I liked things simple - yes it's easy to see the standing current in the Vas is sensitive to the supply, but, being such a late entry point (so much gain x2500 before) any effect is small. It simply maintains the right Vgs to ensure the MOSFETs are doing what the FB loop tells them. The current in this EF Class A stage at 8-10mA is typically 10 times whats needed by the MOSFET gate C.
If the supply goes low, optput needs is lower.
I've been using CircuitMaker which has a downloadable student version. You can do a bode plot with the sig gen on the input then move it to modulate one or other or both supplies, then the bode plot magnitude is the transmission of a signal on the supplies, to the output i.e. PSRR (but must add amp gain , say 30dB to the down figures as that's the gain for a signal at the input).
hope that helps,
Greg |
|
|
| thanh |
| Thanks amplifierguru ! You are ready to answer every question .Good ! |
|
|
| amplifierguru |
Hi XG,
Thanks.
MJ magazine - I don't know that.
I tried KIS (keep it simple) AND achieve great efficiency, low distortion and DC coupled option all at less cost than conventional 'garden variety' design.
"your amp=japanese DIY amp vari + EF + boostap+ Nelson Pass A5 without active current source"
Haha - don't tell me I'll have NP on my back?
Cheers,
greg |
|
|
| mikeks |
| quote: | Originally posted by amplifierguru
If your conventional topology is to work as well you will need more cost/complexity..... |
I really fail to see the truth in this....
| quote: | Originally posted by amplifierguru
........and even in it's basic form is more wasteful of power.... |
How so... :scratch2: |
|
|
| mikeks |
| quote: | Originally posted by amplifierguru
..........degradation in PSRR (bode plot).........
due to the 3200 x miller C paralleling with the A1 load Z ...........
real improvement can be obtained by reducing the gain of the Vas e.g. with R to ground.
This intuitively raises the signal level at the injection point relative to the PSR injection.
This is probably why the low TIM followers perceive a benefit in loading the Vas. |
I am sorry Greg, but low PSRR is dependent on having high feedback, and therefore foward-path gain particularly at ripple frequency.
This foward path gain is (to a first approximation) the product of transadmittance gain (input stage), and transimpedance gain (second stage, alias the lamentably misnamed 'VAS').
The later, (TI gain), is roughly the product of current gain, and the net load at the second stage-output buffer interface.
Therefore, loading down the output of the second stage (TIS), can NOT cause, or even appear to cause an improvement in PSRR...
| quote: | Originally posted by amplifierguru
It becomes fairly obvious that we are just fiddling at the edges of a flawed topology and that the Vas is 50% of the problem.
|
D. Self has demonstrated clearly that this need not be the case.... |
|
|
| amplifierguru |
Hi Mikeks,
You seem to have misinterpreted -
"..........degradation in PSRR (bode plot).........
due to the 3200 x miller C paralleling with the A1 load Z ..........."
This refers to the effective multiplication of the miller C by the gain of the Vas stage to appear across the diff'l load which makes for a lower impedance path at HF from power supply to the Vas input node, of the voltage divider with the output Rc (collector impedance) of the diff'l stage.
"real improvement can be obtained by reducing the gain of the Vas e.g. with R to ground. "
this refers to the practice of resistive loading to ground of the Vas collector to lower gain of the stage which reflects ,say, 320C (10 timers lower) back to the diff'l load. The signal level will then be 10 times higher at the Vas input, for the same output so PSRR is improved 20 dB! Could be perceptible!
Douglas Self doesn't seem to understand PSRR.
As far as cost/complexity goes - i took the same parts and needed a beefier Vas driver stage (more expensive), higher voltage supplies (more expensive power x'former and PS C's) for the same power and more intricate compensation (more C's,R's) to come close to 0.005% at 20KHz to match.
Power efficiency was poorer for the Vas type requiring larger heatsink due to extra MOSFET waste voltage, driver heatsinking and x'former size for more current/voltage. Then there's the overbuilding (wasteful) necessary to compensate for the POOR PSRR - double x'former VA and double PS C - all for 6dB audible improvement.
Check your intrusion equations using A1 A2 A3 with injection at A1A2 node and you will find , for a given OL /CL gain the PSRR is proportional to A1/A2A3.
Cheers,
Greg |
|
|
| amplifierguru |
Hi Lumanauw,
I don't think so. I've measured and sim'd output Z and it's very low, but misplaced it. My previous Eidetic with very high PSRR also used conventional Vas and SF output to achieve it's 150W and rave review/acceptance.
Cheers,
Greg |
|
|
| zinsula |
| quote: | Originally posted by amplifierguru
I don't think so. I've measured and sim'd output Z and it's very low....... |
I believe lumanauw was alluding that the output stage (and thus the whole amp) is a transconductance stage (U in; I out), contrary to a source follower output stage, before closing the feedback loop.
As the transconductance is high, after closing the feedback loop, of course you get low output Z.
But eg. Mauro Penasa stated that this topology sounds better because of better interaction with real loudspeaker loads. There was also a thread in which this was discussed.
Tino |
|
|
| amplifierguru |
Ok zinsula,
I'm listening to the prototypes of the 150W and they're sounding pretty good to me - but I built and designed them.
Yes, I think I've seen that discussion. What more can I say - it's subjective.
Cheers,
Greg |
|
|
| darkfenriz |
| quote: | Originally posted by jam
Comments anyone?.....................This is still a work in progress. :o
|
One question: what is the circuit close to Vgs complemetany multilier to geter with feedback? I mean R49/50/51/52 and C13. A combination of lacal feedback from VAS and from output stage?
regards |
|
|
| jam |
Hi darkfreniz,
I wanted the option for global feedback - in which case you used R52 and R49 or if you wanted to keep the output stage outside the feedback loop - use R52, R50 and R51.
C13 was added for compensation but was not needed in practice.
Regards,
Jam
PS. I suppose you could use all four resistors if you wanted to use nested feedback loops. |
|
|
| Bensen |
| quote: | Originally posted by amplifierguru
Conventional power amplifiers commonly feature a low voltage gain input differential stage followed by a very high voltage gain stage then usually a unity gain emitter or source follower. This design suffers from a distinct topological deficiency that degrades the sound and is responsible for inferior sound quality.
The typical gain distribution is shown in the thumbnail where the weakness of PSRR of the input stage ( not amplifier) due to the divider with stage output Z vs load Z determines intrusion of Class AB PS artefacts which are magnificently amplified by the very high gain Vas to output, blurring the stereo image detail and spatial info and raising the dependance for performance on the PS quality!
Better sound quality from such a design centres around fiddling at the edges of this flawed topology rather than adopting a new one. |
amplifierguru,
So, we want much gain in the stage with the error correction, mostly differential. But with Jfet's it's hard to have many gain, and with BJT's you will bias the VAS stage to havily, with many gain in the first stage.
One question, can I assume that a dual differential (output off the first connected to the gates off the second diff pair) is one stage?? This way I can have relative high gain (X200) and have no problem biasing the next stage to high.
To me it looks that way. Also, can anybody tell me what a normal gain (not gain actually but amplifying factor) is in an amp? I've seen design's with X1000 and X20000.
See attachment.
Cheers
Ben |
|
|
| amplifierguru |
Hi Benson,
Yes you can certainly achieve similar gains over two non-bootstrapped stages but look at the added complexity. Also you need to tailor the responses of these two stages into ideally a single pole response - if to meld with something like my complementary EF driver and SF output - which has a response pole at 100KHz.
Regarding gain you could easily set your first two stages at 5000 or 25000 depending on your need for low or high NFB. The output stage is only average 40 but declines from 100KHz.
Cheers,
Greg |
|
|
| mikeks |
| quote: | Originally posted by amplifierguru
.........real improvement can be obtained by reducing the gain of the Vas e.g. with R to ground. |
No...
| quote: | Originally posted by amplifierguru
Then there's the overbuilding (wasteful) necessary to compensate for the POOR PSRR - double x'former VA and double PS C - all for 6dB audible improvement. |
Complete red herring i fear....one merely has to use a simple RC filter to drop ripple below noise... :smash: |
|
|
| amplifierguru |
Hi Mikeks,
You're one of Omar's spawn, right? Sniping again?
RC filtering just doesn't work at power demand supply bounce frequencies - but you know that!
Have you come to terms with the A1A2A3 scenerio yet, no mention of that?
Cheers,
greg |
|
|
| peranders |
Contest: What do we have to say or claim in order to get a YES from Mike? :scratch:
Mike, there are many NO's from you, without any explanation. |
|
|
| Ultima Thule |
Mikeks meaning RC filtering of supply voltage feeding A1A2... yes? :scratch2:
Greg,
what IRFP are you using, I guess IRFP240/9240, or actually the N versions as the other are obsolete I guess.
Cheers Michael |
|
|
| jam |
Greg,
Are your kits ready for sale and if so a photo of what comes in the kit would be great.
Thanks.
Regards,
Jam |
|
|
| mikeks |
| quote: | Originally posted by Ultima Thule
Mikeks meaning RC filtering of supply voltage feeding A1A2... yes? :scratch2:
|
Yes...Thanks Ultima...(at last i get to say a 'yes!' :) )
...covered to perfection by D. Self of course...no mystery there.... |
|
|
| amplifierguru |
Yes Ultima Thule IRFP240/9240 (or 9140<50V supplies) . What's obselete?
Hi Jam just waiting for some parts - maybe a week.
Well hi Mikeks,
lets see a schematic then of a SIMPLE diff'l (A1) and Vas (A2) with 40dB of 20Hz filtering of the supplies that hasn't lost output swing and doesn't thump at turn on requiring a relay? Mine doesn't.
Are you up to it?? :whazzat:
Cheers,
Greg |
|
|
| Ultima Thule |
| quote: | Originally posted by amplifierguru
Yes Ultima Thule IRFP240/9240 (or 9140<50V supplies) . What's obselete?
Cheers,
Greg |
Hi Greg,
ok, 9140 should be more complementary to 240's...
Obsolete - sorry for the inconveniences, I have mixed up some things here... :blush:
In general:
Don't know, the "problem" to see the PS commutation arteffects in an amplifier circuit of general type is maybe because people forget that the signal refference amplified is actually swaping from earth to rail voltages after A1, and from here to the "other side" of A2 the signal refference is again swapped back to earth without thinking that A2 will be balancing between two refferences that are not at all constant, and with H2 and H3 of 20 kHz means... yeah that goes strait through the miller capacitor.
With a VAS stage with huge gain and output impedance the tiny miller capacitance will still have a noticeable effect on A2 stage performance with respect to PS commutation, but the problem can be solved of course, eg. the general VAS type can be improved, right?!
Normaly clever use of clamping technics with simple solutions can very well take care of thumping I believe, perhaps zener diodes in parallel with R in the RC filter, well it's non-issiual things.. just annoying the "KISS guy" if the omni-amp-design-smartness is not inherent in the circuit without adding more components, ultimately it should be just one amplifying component, eg. a wire with gain? ;)
But if we would design with a general VAS stage, what would you do to cope with PS commutation arteffects?
I mean I would add a ridiculously small ~10VA (2x12V or so) transformer ontop of the heavy PSU feeding A3, and actually use high performance regulation feeding A1A2, yes it adds up to complexity, but I'm not afraid of it, not any beefy things really... :)
Your thoughts?
Cheers Michael :) |
|
|
| amplifierguru |
Hi Michael,
You had me worried there for a moment as I've just characterised the differential tempco's of IRFP240/9240 at 50mA!
You have to appreciate where I'm leading with this - an amp that is not only simple, low parts count and cost that delivers performance, but doesn't need relays for de-thump and is highly insensitive to LF as well as HF artefacts - PS borne. This is because, once this is improved by say 10-100 times, the supply can be relaxedso you have a 150W amp with a 150W supply and loads of headroom. A bouncy supply. If the amps 100 times better for PSRR the supply could be 10times more variation and you're still 10 times better for supply borne hash but with transient power of a much larger amp.
I did this in 1990 with the Eidetic by running a similar comp diff'l , comp Vas and SF MOSFET (Toshiba) outputs - oh and an OPA627 chip front end all in a nested FB loop. Distortion was essentially unmeasurable and PSRR >150dB due to the high LF gain of the OPA as A1. But, using a 300VA transformer it still needed +/- 68V for 150W/ch both driven - but dynamic range allowed 240Wrms short term. All from a loaf of bread size.
It would easily blow away ML's best - the agent even wired the speakers out of phase and with the RCA input rings un-attached (until we corrected it) - and it still blew them away.
The approach here is to save some 7V using the common source config and a different chip. Due to Home Theatre market I've decided to do it as a monoblock.
Cheers,
greg:D |
|
|
| Ultima Thule |
Hi Greg,
no worries! :)
Actually I have had some thought's too about driving FET's in CS mode directly from diff stage with some kind of a buffer of course in between, I do believe too there's some merites in this kind of topology leaving lot of complexity at the same time, your bootstrapped E follower was a smart thing I have to say hitting many flies in one hit the way it's used in your circuit!
Cheers Michael |
|
|
| amplifierguru |
Thanks Michael,
Yes the comp EF interposed really did the trick, I was able to use it for bootstrapping, resulting in the really high first stage gain and linearity. If I take out the bootstrap C'c it still functions but THD goes up over 50 times to about 0.25%, the diff'l stage being near it's 70mV input limit (52mV + Vre).
The Vgs saving over the conventional topology has allowed me to select the next transformer down in volts in the same VA range resulting in better current for the regulation so win, win.
Cheers,
Greg
Cheers,
Greg |
|
|
| Workhorse |
Hi Greg ball..................guru
What is the Damping Factor of your Killer!
Does the bootstrap enhance the output swing to touch the rails much closer than with a difference of 2V.......in Killer, though just asking because our nvmos just touched the rail with a margin of 300mV only ......high efficiency is'nt it.....
Cheers,
K a n w a r |
|
|
| amplifierguru |
Hi Kanwar,
You must remember that the IRFP240/9240 have higher Rds on than your HD devices - particularly the Pch! So about 1.7V is typical for +rail and lower Rdson for Nchannel goes to about 0.7V of -rail.
IRFP9140 is better for +rail but only 100V rated.
Damping Factor is > 1000 to 1KHz. Have to do!!! :D
Cheers,
Greg |
|
|
| Workhorse |
Hi Greg,
Oh goodness! I forgot that we are using the Damn Heavy Duty N-channel Devices or you can say Monsters in our designs...by the way your damping factor is E x c e l l e n t !...
Congrats once again on your design..
Have you Tried IRFP350/IRFP9350 from IXYS Corporation.....400V 38/32A devices thats too in complementary fashion.....but large gate capacitance but you can insert a buffer as well.....
Cheers,
K a n w a r |
|
|
| peranders |
| quote: | Originally posted by amplifierguru
Damping Factor is > 1000 to 1KHz. Have to do!!! :D |
Nelson Pass claims that damping factor isn't particular important. :nod: Something to think about. I don't say it's true but I'll guess there is a limit surely but where is it? |
|
|
| Workhorse |
Oh, Mr Anders first time I have seen you disbelieving NP in terms of DF great or say amazed from your mindset.
NP builds Audiophile gear only, Guru builds excellent Audio gear which serves both communites and you build the Headphone Amps which serves the Head and Eardrums !:D :D :D :D :D ;)
And we build Pro-audio gear to serve the people who use our equipment to earn their living...
Headwise,
Peranders cheers,
K a n w a r |
|
|
| peranders |
| My personal feeling is that the loadspeaker should be driven from a stiff voltage source but I have never felt I would like to test driving the speaker via a small resistance. As we know many Gainclones have 0.2-0.3 ohms at the output and the builders are really pleased. In that case there isn't any debate really is a L//R filer is better and a pure R. Some things are fashion as you may have noticed. |
|
|
| djk |
My rule-of-thumb is that the DF is of no concern until lower than about 20.
When it is below 20 it begins to change the Q of the driver enough that box tuning needs to be examined. With wide impedance swings it will also affect frequency response. While the later makes a more measureable difference, the former may be more audible (since in general the box size is proportional to the square of the Qts). |
|
|
| ilimzn |
| quote: |
Originally posted by peranders
Nelson Pass claims that damping factor isn't particular important. :nod: Something to think about. I don't say it's true but I'll guess there is a limit surely but where is it?
|
It is only unimportant if your speakers are designed for that eventuality ;)
In particular, when designing a speaker, you can do it for three general cases of signal source:
1) Idealized pure voltage (Zout -> 0)
2) Idealized pure current (Zout -> infinity)
3) A known Zout
There is only one way to design a speaker to work virtually identically on all of these sources: it's impedance has to be completely compensated to appear as a constant resitance.
This is certainly possible, but with anything but a single driver, it becomes progressively complex. Even for a single driver, this driver has to have an exceptionally linear motor if the reactance is to be compensated properly.
On the other hand, designing a speaker for a voltage source (or reasonably close to a voltage source) puts the 'zero' impedance of the source in aprallel with what would be compensation networks on a fully reactance compensated design, which means they can be omitted, to the tune of a rather hefty amount of $ saved. It also alowes a more relaxed design of the loudspeaker motors (in fact, this is where sometimes inexcuseable attempts at saving money go).
In practise, I would have to agree with the assesment that DF becomes progressively inconsequential as it increases over about 20. I cannot immagine how one would design a speaker which would require DF of say more than 50, and perform drastically different if it was not provided by the amp.
That being said, there are sometimes advantages of having a (relatively' high driving impedance, but if the amp and the speaker are designed as two separate components, without knowing in advance that they are going to work together, given that series crossover components in the speaker already have resistance, it is a better policy to leave any additional correction of amp output impedance inside the speaker - it's mostly as simple as just using thinner wire on the inductors.
The psychological effect of the DF definition deserves a special mention, namely, that DF~~Zload/Zout gives such pleasing large numbers ;). In fact, a much better indicator would be something like Zload/(Zload + Zout), preferably expressed as a percentage. Unless the source has negative impedance, you can never electrically damp the speaker cone more than it's own impedance alowes. |
|
|
| vAD |
| Jam, Please, *.cir file, your circuit is very interesting. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by peranders
My personal feeling is that the loadspeaker should be driven from a stiff voltage source |
Oh NO. NOT THAT OLD CHESTNUT AGAIN.
DAMPING FACTOR.
Boy, does nobody ever actually bother to think things through.
I have, strictly for fun, simulated a highly reactive alignment of subwoofer choosing to vary the damping factor (actually source impedance) from infinite damping factor to a daping fatcor of 2, I then proceeded to simulate strictly boundary effect (that is without room mode effects) of moving the same enclosure in a 1m square within the room.
Here the difference between a damping factor infinite, 6.4, 3.2 and 2:
Worst case variation is a 3db lift at 60Hz which is quite broadband, audible but hardly "horrible", considering the variation between an infinite DF and one of only 2.
And here the difference of golden mean positioning and movement in 20" intervals from 20"-60" wall distance:
We find as much as 7db variation in the 60-100Hz region.
Now which one is worse?
Case closed on damping factor per se.
A VERY DIFFERENT story is if the damping factor is variable with signal (typhical example is the reactive load sending so much current back into the Amp that the SOAR protection briefly activates, completely removing ANY damping).....
Sayonara |
|
|
| peranders |
| quote: | Originally posted by Kuei Yang Wang
Boy, does nobody ever actually bother to think things through.
I have, strictly for fun, simulated a highly reactive alignment of subwoofer... |
Sometimes you have a gut feeling but I'm open minded for this. Simulations is one thing, real life is an another. What do these curves say to you in pratical terms? Which sounds the best? Which speaker, which amp etc etc. |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by peranders
Simulations is one thing, real life is an another. What do these curves say to you in pratical terms? Which sounds the best? Which speaker, which amp etc etc. |
My point is simple. In terms of FR and "damping of resonance" a quite large amount of output impedance has an impact that is no worse than moving the speaker not all that far (~ 10") in most rooms.
What I think leads to the claims for certain SS Amplifiers having "great bass on difficult speakers because of high damping factor" has in fact more to do with the ability of these amplifiers to retain their damping factor even if the amplifier has to cope with a lot of dynamic & transient current being shoved back up it's tailpipe by the speaker.
Of course, this is not a problem for any amplifier that attains it's output impedance by means of using a device with naturally low internal impedance (triode) instead of attaining it by large amounts of looped inverse feedback, which blows up big time if the amplifier inside of the loop cannot keep up with the combined demands of "damping" the resonant system and the signal.
Sayonara |
|
|
| peranders |
| At the moment we agree :nod: Things are more complex than only fousing on one thing. There are more dimensions to the problem. |
|
|
| AndrewT |
Hi Kuei yang wang,
re | quote: | | the reactive load sending so much current back into the Amp that the SOAR protection briefly activates |
Can you explain how this mechanism operates on both a badly designed and a well designed VI limiting circuit? |
|
|
| amplifierguru |
Hi All,
I only stated DF as > 1000 because Kanwar asked. I'll go with consensus on this that anything over 20 is fine and >100, good.
Of course 10cm of internal output cabling and external speaker cables will dominate an output Z of 0.008 ohm!
Cheers,
greg |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by AndrewT
Can you explain how this mechanism operates on both a badly designed and a well designed VI limiting circuit? |
Even with a "well designed" VI limiting (I prefer the term SOAR protection) you may still get problems, as the SOAR Protection is dependent on the output transistor, not on the speaker.
My point is that a conventionally dimensioned and supposedly "well dimensioned" output stage may still be unable to cope with the back EMF from a highly reactive speaker and the signal, when high levels are present.
The best solution is of course to build the amplifier so that even a dead short to one rail cannot exceeed the output transistors SOAR....
Sayonara |
|
|
| jacco vermeulen |
| quote: | Originally posted by Kuei Yang Wang
The best solution is of course to build the amplifier so that even a dead short to one rail cannot exceeed the output transistors SOAR.... |
Thorsten,
limited by the transformer untill the primary fuse blows ? |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by jacco vermeulen
limited by the transformer untill the primary fuse blows ? |
Something like that. You could add a few small value decoupling resistors between multiple smaller value main filter capacitors as well.
I build many years ago several massive "Borg Cube" PA Amplifiers (basically 19" wide > 20" tall and deep).
I used completely "silly buggers" industrial darlington switching transistor modules (something like 300A @ 380V when switching inductive loads) in a circlotorn arrangement.
No shortcircuit protection was needed when running on around 120V Rails (we DID TRY HARD to blow up the Amp's). The PSU wiring and transformers had enough impedance to stop the currents from rising higher.
Sayonara |
|
|
| Ultima Thule |
| quote: | Originally posted by jacco vermeulen
Thorsten,
limited by the transformer untill the primary fuse blows ? |
Jacco,
suppose some designer uses a Zillion Farad capacitor bank after the mains transformer... ;)
Cheers Michael |
|
|
| jacco vermeulen |
That is why i am asking, Michael.
Would need a zillion output devices as well ?
That is supposing there are no resistors needed to protect the capacitors from failing ?
Capacitors can not be loaded or unloaded with unlimited current values, if i got the story right. |
|
|
| AndrewT |
Hi Kuei yang wang,
re quote:
the reactive load sending so much current back into the Amp that the SOAR protection briefly activates
Can you explain how this mechanism operates on both a badly designed and a well designed VI limiting circuit?
Can you answer the question and if need be refer to back emf from the speaker or exclude it?
What mechanism causes the limiting to adversely affect the sound quality? |
|
|
| peranders |
| Have you come to any conclusion? Is it possible to make a killer amp which is simple? There are maybe contradictions? |
|
|
| amplifierguru |
Hi Peranders,
An interesting issue -
..."Is it possible to make a killer amp which is simple?"
I recall jamming a piece of toast in the toaster at 6am one morning with my 1yo daughter sitting in a high chair and running out to pick up the morning paper from the driveway. When I returned the kitchen was full of smoke and the toaster had charred and flamed a jammed piece of toast, searing overhead cupboards. I bought a more expensive safety toaster that day!
While a simple amplifier can be designed to limit excessive currents throughout the stages, many don't, and, in a low cost design without ancilliary measures in place, catastrophe can be serious! However simple designs with limits on each stage are considerably more forgiving.
Take, for example the topic topology. The input stage is differential, desirably using current sources - so currents here are limited to 2 x quiescent. Next stage is running 8-10mA typically and from the emitter resistors = MAX. The drivers actually pull current out! Given gate zeners of Vz just above operating range the output current cannot sensibly be catastrophic, and with fused collectors, inherently safe!
I doubt this amp could go up in flames, no matter what you did at the outputs. Most people expect some problem to result from errors as with most appliances. As long as they're not dramatic or little more than changing a fuse, it comes as a warning/deterrant.
I don't see that a simple amplifier should be expected to incorporate every conceivable protection so it continues to sing during/after prolonged abuse - in a domestic scenerio. A human wouldn't! Different for PA.
My take.
Cheers,
Greg |
|
|
| darkfenriz |
| quote: | | "Is it possible to make a killer amp which is simple?" |
By simplicity one can understand many things: low number parts, counting IC as 1 part or not, simple electronic circuits only (CE, EF), some devices in signal path and some not and for some low number of poles and zeros in forward path.
So we all love simplicity, some must only complicate here and there to achieve the expected level of simplicity.
cheers |
|
|
| Kuei Yang Wang |
Konnichiwa,
| quote: | Originally posted by AndrewT
Can you explain how this mechanism operates on both a badly designed and a well designed VI limiting circuit?
Can you answer the question and if need be refer to back emf from the speaker or exclude it? |
I did. It so blindingly obvious that I cannot explain any better. It is REALLY, REALLY basic. Read a decent book on how a moving coil speaker works if you don't get it.
| quote: | Originally posted by AndrewT
What mechanism causes the limiting to adversely affect the sound quality? |
Oh come off it. If the protection circuitry engages the amplifier clips. THAT is what adversely affects the sound.
Sayonara |
|
|
| Greg Erskine |
| quote: | Originally posted by amplifierguru
I recall jamming a piece of toast in the toaster at 6am one morning with my 1yo daughter sitting in a high chair and running out to pick up the morning paper from the driveway. When I returned the kitchen was full of smoke and the toaster had charred and flamed a jammed piece of toast, searing overhead cupboards. I bought a more expensive safety toaster that day!
|
Hi toasterguru,
Any advice on toasters. I'm having trouble getting my raisin toast to brown just right. I either burn it or its underdone. I also find the toast doesn't seem hot enough to melt the butter properly. ;)
Just to keep things on topic, I often eat raisin toast while listening to my AKSA (a "simple killer amp" IMHO).
Thanks |
|
|
|
