Dead serious. Amps with very stiff supplies degrade MUCH more gracefully when overdriven. So much so that you might not even notice the onset of clipping.
25 watts of single ended class A "sounds" like much more because the rails don't budge with signal level. So you get less garbage conducted through saturated outputs when it clips compared to a smallish class AB with the rails bouncing off the ceiling.
25 watts of single ended class A "sounds" like much more because the rails don't budge with signal level. So you get less garbage conducted through saturated outputs when it clips compared to a smallish class AB with the rails bouncing off the ceiling.
What happens when an unregulated power supply runs out of juice? Does clipping occur?
Wouldn't it be better to have an oversized output stage that could never be driven into clipping, coupled to a modest supply, instead of the opposite?
Wouldn't it be better to have an oversized output stage that could never be driven into clipping, coupled to a modest supply, instead of the opposite?
Of course it's best to always avoid clipping. But consider 2 amps, each testing 200W/8R unclipped sine wave. One has +/-65V rails and the other has +/-85V rails. Both drop to +/-60V with a full signal sine wave at 1kHz. Drive each 6dB into clip with music. The better regulation amp will sound "less dirty". The poorer supply will have higher peak power (2.3dB more), but for insignificant duration. If you never clip either, you'd never be able to tell the difference. The poorer regulation unit will be cheaper to build. The trafo will cost less than half and you might want an extra output pair or two per channel. But if I'm cranking up tunes at a party, the better regulation amp will play cleaner louder, and heat up less. Differences between them will be magnified many times driving low impedance (or multiple paralleled) speakers.
I'm referring to a domestic application where a 50W amplifier would suffice and probably never clip. The 200W output stage is specified merely to provide extra headroom (electronics are cheap) and flexibility to drive low impedance or difficult loads. The 500va transformer would more than suffice for a class ab 2x50w amp. Same application, same transformer VA rating, just a bigger cap bank and output stage for extra headroom. See the logic?
an amplifier that does this is a failure.
It cannot drive current into difficult loads.
agreed, exactly my experience....
sizing power traffos depends on a couple of factors and priorities...
the best for music, get the biggest you can afford and have room in the chassis for...
best for output trannie protection, about 1 times power output...
so depending on your priorities, traffo sizing can take many forms, there is one that suits your needs...
big power traffos will require big heatsinks as well...
Ajt,
your question is valid.
Consider the amplifier that has rails of +-85Vdc.
What maximum unclipped output voltage would one expect when this amplifier is tested into a 1M dummy resistive load?
Plot that value on a graph of Output Voltage (dbV) vs load resistance (log of ohms).
Progressively reduce the load resistance to approximately 1/4 of the minimum rated nominal speaker impedance (i.e. 2r0 for an 8ohms amplifier, or 1r0 for a 4ohms amplifier).
Now repeat the exercise for the other amplifier that starts out with supply rails of +-65Vdca.
Now compare the two plots.
The +-85Vdc amplifier cannot deliver output current into difficult loads. This shows as a collapse in output voltage, or a collapse in output power (ref the doubling expected when load is halved), when the load gets lower.
your question is valid.
Consider the amplifier that has rails of +-85Vdc.
What maximum unclipped output voltage would one expect when this amplifier is tested into a 1M dummy resistive load?
Plot that value on a graph of Output Voltage (dbV) vs load resistance (log of ohms).
Progressively reduce the load resistance to approximately 1/4 of the minimum rated nominal speaker impedance (i.e. 2r0 for an 8ohms amplifier, or 1r0 for a 4ohms amplifier).
Now repeat the exercise for the other amplifier that starts out with supply rails of +-65Vdca.
Now compare the two plots.
The +-85Vdc amplifier cannot deliver output current into difficult loads. This shows as a collapse in output voltage, or a collapse in output power (ref the doubling expected when load is halved), when the load gets lower.
Andrew,
mental exercise is fine, but this is a diy'ers board...
members here are eager to share actual experiences
rather than theoretical or imaginary amps....
let us see an actual warm amp for a change...
mental exercise is fine, but this is a diy'ers board...
members here are eager to share actual experiences
rather than theoretical or imaginary amps....
let us see an actual warm amp for a change...
Folks,
Most has been said, but the interpretation might evade, as Ajt said. So let us go to practical audibility (but Ajt, please let us not make too much of a point of difference between theory and practice; there is none. It is simply a matter of correct understanding. This business of such alluded differences comes from uninformed and misleading subjective interpretations.)
Thus: The figures mentioned were correct - but of what audible implication? There are several. Firstly a sag of say 85V to 70V is considered serious. All of 18%!! But converted to output and thus loudness, that is 3dB. Most will agree that we regard that as a difference not worth writing home about - audible but just. In practice: Play your system at a reasonable volume and then turn down the volume by as little as you can definitely notice. Except for those of the super hearing species, chances are that that will represent a halving of the output energy. (And as Andrew suggested, electronically that is already on the border of questionable power supply design. Drawing a capacitor-input power filter from peak value to just about depleted, results in a 0,707 reduction; the well-known r.m.s. factor. Most half-decent power supplies will show <20% reduction.)
Yes, there is the clipping monster. Scaly dealers will compare a 100W and 150W amplifier by adjusting the input such that the 150-watter will just not clip; the 100-watter then naturally will. "Listen to the distortion!" sir will exclaim. Sure. But turn the input down to where the 100-watter will also just not clip, and listen if the difference is worth the extra (of course!) cost. It will not be.
Similarly concerning headroom, I fear 3dB is equally not worth vacillating about. Depending on one's music preference, some 6 - 8dB if not more is desirable. 6dB up from 100W is 400W, and 8dB up is 600W! It was said elsewhere that one forgets hearing is logarithmic. You do pay for that little bit of extra audible volume/headroom, friend! Only don't pay for what is not worth-while.
Finally from that perspective, sober power supply design involves a decent filter capacitor bank and a reasonable power transformer; no more (but it is your money). I think I did mention that, respecting both, my own 2 x 80W amplifier has a 200W (C-core) power transformer, the rail voltage drops from 39 - 35V maximum under load, and I cannot hear the difference between a moderately clipping and a pure output (there is a peak clipping indicator on board - something any power amplifier should have). And even under heavy symphonic music the transformer gets only luke-warm.
Hopefully some useful perspective here.
Most has been said, but the interpretation might evade, as Ajt said. So let us go to practical audibility (but Ajt, please let us not make too much of a point of difference between theory and practice; there is none. It is simply a matter of correct understanding. This business of such alluded differences comes from uninformed and misleading subjective interpretations.)
Thus: The figures mentioned were correct - but of what audible implication? There are several. Firstly a sag of say 85V to 70V is considered serious. All of 18%!! But converted to output and thus loudness, that is 3dB. Most will agree that we regard that as a difference not worth writing home about - audible but just. In practice: Play your system at a reasonable volume and then turn down the volume by as little as you can definitely notice. Except for those of the super hearing species, chances are that that will represent a halving of the output energy. (And as Andrew suggested, electronically that is already on the border of questionable power supply design. Drawing a capacitor-input power filter from peak value to just about depleted, results in a 0,707 reduction; the well-known r.m.s. factor. Most half-decent power supplies will show <20% reduction.)
Yes, there is the clipping monster. Scaly dealers will compare a 100W and 150W amplifier by adjusting the input such that the 150-watter will just not clip; the 100-watter then naturally will. "Listen to the distortion!" sir will exclaim. Sure. But turn the input down to where the 100-watter will also just not clip, and listen if the difference is worth the extra (of course!) cost. It will not be.
Similarly concerning headroom, I fear 3dB is equally not worth vacillating about. Depending on one's music preference, some 6 - 8dB if not more is desirable. 6dB up from 100W is 400W, and 8dB up is 600W! It was said elsewhere that one forgets hearing is logarithmic. You do pay for that little bit of extra audible volume/headroom, friend! Only don't pay for what is not worth-while.
Finally from that perspective, sober power supply design involves a decent filter capacitor bank and a reasonable power transformer; no more (but it is your money). I think I did mention that, respecting both, my own 2 x 80W amplifier has a 200W (C-core) power transformer, the rail voltage drops from 39 - 35V maximum under load, and I cannot hear the difference between a moderately clipping and a pure output (there is a peak clipping indicator on board - something any power amplifier should have). And even under heavy symphonic music the transformer gets only luke-warm.
Hopefully some useful perspective here.
Last edited:
Pioneer Spec 2
can be seen here....Pioneer Spec-2 | Owners Manual, Service Manual, Schematics, Free Download | HiFi Engine
on page 30, looking at the schematic, it shows 91 volts(75v)...
now what could this mean?
i interpret it as 91 volts standby and 75 volts fully loaded..
now that's about 18% sag....
and till today the Pioneer Spec 2 still sells for over U$2k....
that tells a lot about amps....
the notation on the schematic implied FTC sine wave power testing...
in actual everyday music listening, sag will never go anywhere near that level...
can be seen here....Pioneer Spec-2 | Owners Manual, Service Manual, Schematics, Free Download | HiFi Engine
on page 30, looking at the schematic, it shows 91 volts(75v)...
now what could this mean?
i interpret it as 91 volts standby and 75 volts fully loaded..
now that's about 18% sag....
and till today the Pioneer Spec 2 still sells for over U$2k....
that tells a lot about amps....
the notation on the schematic implied FTC sine wave power testing...
in actual everyday music listening, sag will never go anywhere near that level...
Build an amplifier with a proper PSU.
You will find the same as me.
If the amplifier is current crippled, the output voltage will drop excessively as load (test) resistance is reduced.
You have posted in the past that very high Vdrop in the PSU is acceptable, even in the tens of volts. I have tried to point out to you that much better performance is possible and beneficial to sound quality, if the Supply rails drop by much smaller amounts. You have consistently refused to accept my findings, you even argued without any evidence that my quoted Vdrops in my typical amplifiers are impossible. Do the tests, measure the amplifiers with good and bad current delivery. Then compare them in listening tests.
Last edited:
you should practice what you preach, show us what you have done...
please see my post above, that is an actual amplifier....
please see my post above, that is an actual amplifier....
The penny pinching example goes to show that it's more common than you think, even among respected brands. Of course it would sound better with a stiffer supply, but that doesn't prevent manufacturers (or DIY'ers) from using the cheaper and more common solution. You're free to build any way you want, and so is everybody else. I've done it both ways depending on what I expect out of the result and will continue to do so.
Hi Guys
Been working on a few board layouts for this circuit. Here is an updated schematic and a PDF of the component layout. The board is 7.5x3.25", 191x83mm.
The gerber and excellon files are in the .rar
Oop! The forum won't allow me to upload the .rar, so I'll have to post that on my own site over the weekend.
Note that the usual file acronyms have been changed to plain English - my own PCB house's preference. They should work just the same with any other manufacturer.
As it is, the PA can be used as is without the no-clip circuit, which I'll post a board for shortly. It just plugs onto the PA board.
The bias generator BJT is on a little break-away board and bolts onto the front surface of one of the output devices. Just uses the same mounting bolt. The front of the TO-264 package changes temp more quickly than the heat sink, so bias tracking should be better this way.
The driver heat sink footprints are for the largest type that Aavid makes. The smaller sizes use the same pin spacing, so any you can find will work. Note that the heat sinks are grounded for lowest noise, and the drivers must be insulated from them with mica and shoulder bushings.
I changed the small signal BJTs to 2N5400-2N5550 types, as these are more linear than the MPSA42/92. 2N5401-2N5551 are fine, too, with just a little more voltage capability. The specified TO-92s have the standard EBC pinout. If you sub BCxxx you will have to face the parts opposite to the silk-screening.
The electrolytics are Nichicon HE-series, 105C long-life types. The 18mm dia units have 10khr life; smaller ones are 8khr and less. The input cap is a bipolar Panasonic or equivalent. All Rs are metal film, 600mW 1%, although 400mW will work in most places.
There is a pair of pads to poke with meter probes for setting bias. These are isolated from the circuit being tested by 10k resistors. This introduces negligible error.
Note also that although this amp was designed for 250W at 8R and will easily double that into 4R given a proper PSU and heat sinking, this same board can be used for lower outputs. The large output stage promotes low THD. At the lower supply voltages, one could easily reduce the 330mR emitter resistors to 100mR and accept the higher idle currents without a serious increase in idle dissipation.
For example, at +/-90V, the idle dissipation is about 7W per device, so 56W for the whole output stage. With 0.3C/W heat sinks, there is just a 17C rise at idle. This is with the 330mR emitter resistors. If you lowered these to 100mR at the same supply voltages, diss would rise to 187W and temp by56C.
On the other hand, if the supplies were reduced to provide 100W at 8R, then the 33mR/100mR dissipations become 31W/9.5C and 104W/31C, respectively.
Have fun
Kevin O'Connor
Been working on a few board layouts for this circuit. Here is an updated schematic and a PDF of the component layout. The board is 7.5x3.25", 191x83mm.
The gerber and excellon files are in the .rar
Oop! The forum won't allow me to upload the .rar, so I'll have to post that on my own site over the weekend.
Note that the usual file acronyms have been changed to plain English - my own PCB house's preference. They should work just the same with any other manufacturer.
As it is, the PA can be used as is without the no-clip circuit, which I'll post a board for shortly. It just plugs onto the PA board.
The bias generator BJT is on a little break-away board and bolts onto the front surface of one of the output devices. Just uses the same mounting bolt. The front of the TO-264 package changes temp more quickly than the heat sink, so bias tracking should be better this way.
The driver heat sink footprints are for the largest type that Aavid makes. The smaller sizes use the same pin spacing, so any you can find will work. Note that the heat sinks are grounded for lowest noise, and the drivers must be insulated from them with mica and shoulder bushings.
I changed the small signal BJTs to 2N5400-2N5550 types, as these are more linear than the MPSA42/92. 2N5401-2N5551 are fine, too, with just a little more voltage capability. The specified TO-92s have the standard EBC pinout. If you sub BCxxx you will have to face the parts opposite to the silk-screening.
The electrolytics are Nichicon HE-series, 105C long-life types. The 18mm dia units have 10khr life; smaller ones are 8khr and less. The input cap is a bipolar Panasonic or equivalent. All Rs are metal film, 600mW 1%, although 400mW will work in most places.
There is a pair of pads to poke with meter probes for setting bias. These are isolated from the circuit being tested by 10k resistors. This introduces negligible error.
Note also that although this amp was designed for 250W at 8R and will easily double that into 4R given a proper PSU and heat sinking, this same board can be used for lower outputs. The large output stage promotes low THD. At the lower supply voltages, one could easily reduce the 330mR emitter resistors to 100mR and accept the higher idle currents without a serious increase in idle dissipation.
For example, at +/-90V, the idle dissipation is about 7W per device, so 56W for the whole output stage. With 0.3C/W heat sinks, there is just a 17C rise at idle. This is with the 330mR emitter resistors. If you lowered these to 100mR at the same supply voltages, diss would rise to 187W and temp by56C.
On the other hand, if the supplies were reduced to provide 100W at 8R, then the 33mR/100mR dissipations become 31W/9.5C and 104W/31C, respectively.
Have fun
Kevin O'Connor
Last edited:
Hi Guys
Hasn't power supply sizing and sag been discussed to death?
The economic fact is that most people do not push their systems anywhere near full output, so a saggy supply is quite acceptable. This is especially true given the poor quality of synthetic surround sound that most people have willingly or unwillingly accepted as the "norm" along with the whole "theater " brain washing that THIS is what one should have.
Most people will say that they are "serious" about their sound, but have for the most part given up trying to buck what the system has to offer. Those who really are serious buy better equipment that does not sag under load - or sags so little that they are deaf long before they could hear it.
An amp built to suit the original poster's needs - a party amp for someone who makes money providing loud clean sound - has to have a stiff supply. The no-clip circuit will save tweeters and keep things from getting too out of hand distortion-wise, but a burly design keeps it pumping night after night.
There is no economic reason why you should not have a "big" output stage, as there are compelling reasons to do so in the name of low-THD. Given the actual amount of power most people need, it is easy to have a very stiff supply without spending a tonne of money and without resorting to anything as ridiculous as regulated supplies.
Have fun
Kevin O'Connor
Hasn't power supply sizing and sag been discussed to death?
The economic fact is that most people do not push their systems anywhere near full output, so a saggy supply is quite acceptable. This is especially true given the poor quality of synthetic surround sound that most people have willingly or unwillingly accepted as the "norm" along with the whole "theater " brain washing that THIS is what one should have.
Most people will say that they are "serious" about their sound, but have for the most part given up trying to buck what the system has to offer. Those who really are serious buy better equipment that does not sag under load - or sags so little that they are deaf long before they could hear it.
An amp built to suit the original poster's needs - a party amp for someone who makes money providing loud clean sound - has to have a stiff supply. The no-clip circuit will save tweeters and keep things from getting too out of hand distortion-wise, but a burly design keeps it pumping night after night.
There is no economic reason why you should not have a "big" output stage, as there are compelling reasons to do so in the name of low-THD. Given the actual amount of power most people need, it is easy to have a very stiff supply without spending a tonne of money and without resorting to anything as ridiculous as regulated supplies.
Have fun
Kevin O'Connor
Last edited:
yes, even if your car can do 200km per hour....you just cruise way way below that.....😀
to me, as long as any amp can pass the FTC regulations without sustaining damage, then that amp is okey in my book....
to me, as long as any amp can pass the FTC regulations without sustaining damage, then that amp is okey in my book....
there are other considerations, a smaller power traffo protects the output devices from damage better than a bigger one...precisely because of the sag, it is a balancing act really...
Hi Struth
Let me be the first to say "thank you" for your valuable contribution by way of this circuit schematic and layout. I'm sitting here with Bob Cordell's book and flipping between the Dx MkIII circuit and yours, trying to make sense of the differences!
Regards
Let me be the first to say "thank you" for your valuable contribution by way of this circuit schematic and layout. I'm sitting here with Bob Cordell's book and flipping between the Dx MkIII circuit and yours, trying to make sense of the differences!
Regards
- Status
- Not open for further replies.