Simpel amp - symetric or not?
Hi all,
A friend of mine needs amps for his 5.1 system and has asked me to find out the best solution looking at spec vs price.
The amp schould be arround 100 - 120 W pr. channel to have some headroom and to make the amps powerfull enough to drive his system at high levels.
For the sub I'm thinking a bridged amp?
I'd like to make a descrete amp, the problem is the VAS stage.
If I go for a symetric configuration I need to double the number of transistors in the input stage but are symetrics designs better sounding? - is the money well spend?
The result will be avalible for free download at my website
\Jens
Hi all,
A friend of mine needs amps for his 5.1 system and has asked me to find out the best solution looking at spec vs price.
The amp schould be arround 100 - 120 W pr. channel to have some headroom and to make the amps powerfull enough to drive his system at high levels.
For the sub I'm thinking a bridged amp?
I'd like to make a descrete amp, the problem is the VAS stage.
If I go for a symetric configuration I need to double the number of transistors in the input stage but are symetrics designs better sounding? - is the money well spend?
The result will be avalible for free download at my website
\Jens
I like the symmetric especially if the outputs are complementary. It can provide twice the drive current. I know the dilemma about needing to work with more parts. The cost is often not the main thing, for me it is the extra complexity and effort. For the sub, bridging is good because it allows more output power and some distortion cancels, so the circuits may be able to be simpler.
Symmetric vs nonsymmetric
Hi,
If I remember correctly D. Self only has non symmetrical designs in his book about power amps, and Slone has only symmetrical in his book – why is this?
Thinking about the operation of the symmetrical design it must be superior to the other type because both the positive and negative signal has the same signal path (Assuming of cause that NPN and PNP transistors behave the same)
But since there are several nice non symmetrical amps out there (P3A I have heard and ASKA witch I regrettably haven’t heard – among others) are these as good as a well build symmetrical amp all other things being equal?
In the end, is the choice made to satisfy the eye or the ear?
\Jens
Hi,
If I remember correctly D. Self only has non symmetrical designs in his book about power amps, and Slone has only symmetrical in his book – why is this?
Thinking about the operation of the symmetrical design it must be superior to the other type because both the positive and negative signal has the same signal path (Assuming of cause that NPN and PNP transistors behave the same)
But since there are several nice non symmetrical amps out there (P3A I have heard and ASKA witch I regrettably haven’t heard – among others) are these as good as a well build symmetrical amp all other things being equal?
In the end, is the choice made to satisfy the eye or the ear?
\Jens
Hi,
could it be to tickle the desire for both of those aspects or just to have something unique to set it apart from the rest. The main things I would strive for are the best performance for with the least cost and complexity regardless what the actual topology is.
Honestly, being more of a pure tinkerer, I have never tried any pre-designed amps other than the Tiger ages ago. In constructing quite a few amps for one project like a 5.1 system, I would even consider audio chip use in some of them. But I can't say I know much of anything about home theater systems.
could it be to tickle the desire for both of those aspects or just to have something unique to set it apart from the rest. The main things I would strive for are the best performance for with the least cost and complexity regardless what the actual topology is.
Honestly, being more of a pure tinkerer, I have never tried any pre-designed amps other than the Tiger ages ago. In constructing quite a few amps for one project like a 5.1 system, I would even consider audio chip use in some of them. But I can't say I know much of anything about home theater systems.
Hi,
those are good criteria to me. One thought that comes to mind for a powerful class AB sub amp is not even symmetrical, but uses only N-channel mosfets. It is the schematic X-pro posted not very long ago. It is an output stage with constant current setting the bias. X-pro's output stage
You drive it with a class A or AB driver stage of your choice and can include it in the gobal feedback loop.
those are good criteria to me. One thought that comes to mind for a powerful class AB sub amp is not even symmetrical, but uses only N-channel mosfets. It is the schematic X-pro posted not very long ago. It is an output stage with constant current setting the bias. X-pro's output stage
You drive it with a class A or AB driver stage of your choice and can include it in the gobal feedback loop.
Over thirty years ago I built my first SWTPC Tiger .01, symmetrical from input to output.
For many years I felt that this was the only topology that sounded good. The big Luxman, Accuphase, Bryston, SAE, GAS, all had the same basic topology.
The amplifiers with single diff pairs didn't sound as good.
The amplifiers with quasi output stages didn't sound as good.
Though similar, the Leach seemed to sound the best.
Why?
Several reasons, the first being capacitors. The Leach had power supply bypass caps, signal path bypass caps, and no ceramics. Leach didn't like an active current source for the input pairs, he uses resistor to a regulated voltage.
SWTPC came out with a lower cost version of the .01, gone were the meter, input level conrol, active current sources, and pre-drivers. While it didn't measure the same, it sounded the same.
McIntosh went from a quasi output stage to a symmetrical type output stage driven with a single diff pair and a single VAS. No CCS, just resistors to a regulated supply. Stock, the amplifier sounded very soft, lacking in detail, and muddy in the bass. Everyone always blames the output transformer and the barely out of the single digit damping factor for this sound.
Dead wrong!
The McIntosh has no power supply bypass caps, no signal path bypass caps. Adding these changed the sound so much that the frequency response was measured! Dead flat!
What does this tell me WRT home theater?
I would build a small supply for the front end for all five channels, run it a few volts higher than the outputs.
Front end?
The McIntosh is sure attractive in its simplicity. A diff pair plus the VAS, it makes the AKSA look complicated. But keep in mind that the original Leach symmetrical front end is only three more transistors that add less than $5 to the cost of all five channels. You may have to flip a coin on this one.
Output stage?
Emitter follower with MJ21193/21194 outputs, one pair for each 125W of output required. The QSC USA1310 does 1KW with eight pairs. You want to use 30Mhz plastic, fine, just use twice as many. 100W at 4 ohms is 7A collector current. At 2A the plastic devices can do 50Mhz, what do they look like at 7A or higher? The MJ21193/21194 are only 4Mhz, but they can do it at 7A+, they are faster and have higher gain at high currents. Plastic doesn't handle the power as well, but if you use twice as many you can get the current down and the speed up too. I use metal.
http://homelf.kimo.com.tw/skychutw/ampzilla/schematics/sonAmpzilla_sch.jpg
The SON is 150W/4R, 250W/2R and bridges to 300W/8R or 500W/4R. The MJ21193/21194 parts are much heavier than the original outputs, you could probably lose the current limiting.
The MJ21193/21194 have such good SOA at ±100V that it makes no sense to bridge for high power, just raise the voltages and parallel enough outputs.
For many years I felt that this was the only topology that sounded good. The big Luxman, Accuphase, Bryston, SAE, GAS, all had the same basic topology.
The amplifiers with single diff pairs didn't sound as good.
The amplifiers with quasi output stages didn't sound as good.
Though similar, the Leach seemed to sound the best.
Why?
Several reasons, the first being capacitors. The Leach had power supply bypass caps, signal path bypass caps, and no ceramics. Leach didn't like an active current source for the input pairs, he uses resistor to a regulated voltage.
SWTPC came out with a lower cost version of the .01, gone were the meter, input level conrol, active current sources, and pre-drivers. While it didn't measure the same, it sounded the same.
McIntosh went from a quasi output stage to a symmetrical type output stage driven with a single diff pair and a single VAS. No CCS, just resistors to a regulated supply. Stock, the amplifier sounded very soft, lacking in detail, and muddy in the bass. Everyone always blames the output transformer and the barely out of the single digit damping factor for this sound.
Dead wrong!
The McIntosh has no power supply bypass caps, no signal path bypass caps. Adding these changed the sound so much that the frequency response was measured! Dead flat!
What does this tell me WRT home theater?
I would build a small supply for the front end for all five channels, run it a few volts higher than the outputs.
Front end?
The McIntosh is sure attractive in its simplicity. A diff pair plus the VAS, it makes the AKSA look complicated. But keep in mind that the original Leach symmetrical front end is only three more transistors that add less than $5 to the cost of all five channels. You may have to flip a coin on this one.
Output stage?
Emitter follower with MJ21193/21194 outputs, one pair for each 125W of output required. The QSC USA1310 does 1KW with eight pairs. You want to use 30Mhz plastic, fine, just use twice as many. 100W at 4 ohms is 7A collector current. At 2A the plastic devices can do 50Mhz, what do they look like at 7A or higher? The MJ21193/21194 are only 4Mhz, but they can do it at 7A+, they are faster and have higher gain at high currents. Plastic doesn't handle the power as well, but if you use twice as many you can get the current down and the speed up too. I use metal.
http://homelf.kimo.com.tw/skychutw/ampzilla/schematics/sonAmpzilla_sch.jpg
The SON is 150W/4R, 250W/2R and bridges to 300W/8R or 500W/4R. The MJ21193/21194 parts are much heavier than the original outputs, you could probably lose the current limiting.
The MJ21193/21194 have such good SOA at ±100V that it makes no sense to bridge for high power, just raise the voltages and parallel enough outputs.
Caps
Hello,
Thanks for a great reply
>
The McIntosh has no power supply bypass caps, no signal path bypass caps. Adding these changed the sound so much that the frequency response was measured! Dead flat!
>
Are the supply caps you refer to the small caps across the bigger caps?
Are the signal bypass caps the one across the VBE thermal transistor in the output stage?
I have alvays used 1µF film caps across the big eletrolytic caps, and 1 µF across the VBE transistor in the ouputstage for better speed.
BTW can you please check the link? I can't get it to work
\Jens
Hello,
Thanks for a great reply
>
The McIntosh has no power supply bypass caps, no signal path bypass caps. Adding these changed the sound so much that the frequency response was measured! Dead flat!
>
Are the supply caps you refer to the small caps across the bigger caps?
Are the signal bypass caps the one across the VBE thermal transistor in the output stage?
I have alvays used 1µF film caps across the big eletrolytic caps, and 1 µF across the VBE transistor in the ouputstage for better speed.
BTW can you please check the link? I can't get it to work
\Jens
"Are the supply caps you refer to the small caps across the bigger caps?"
Yes, Leach uses a pair of 22µf for each channel. I used a single pair of 47µF in the McIntosh for its mono supply.
"Are the signal bypass caps the one across the VBE thermal transistor in the output stage?"
In addition to this one, don't forget the cap in the feedbck loop to ground. Most people remember to do the input cap
"I have alvays used 1µF film caps across the big eletrolytic caps, and 1 µF across the VBE transistor in the ouputstage for better speed."
Not big enough, see above.
"BTW can you please check the link? I can't get it to work"
Try the home page:
http://home.kimo.com.tw/skychutw/
go to:
Ampzilla page:
Then to SON schematic
From the home page again, go to:
Circuits, and then:
Super Tiger and the MK31B
Note that the SON has a small ±52V shared supply for the front ends and does 150W/4R from only ±45V
Yes, Leach uses a pair of 22µf for each channel. I used a single pair of 47µF in the McIntosh for its mono supply.
"Are the signal bypass caps the one across the VBE thermal transistor in the output stage?"
In addition to this one, don't forget the cap in the feedbck loop to ground. Most people remember to do the input cap
"I have alvays used 1µF film caps across the big eletrolytic caps, and 1 µF across the VBE transistor in the ouputstage for better speed."
Not big enough, see above.
"BTW can you please check the link? I can't get it to work"
Try the home page:
http://home.kimo.com.tw/skychutw/
go to:
Ampzilla page:
Then to SON schematic
From the home page again, go to:
Circuits, and then:
Super Tiger and the MK31B
Note that the SON has a small ±52V shared supply for the front ends and does 150W/4R from only ±45V
Open loop gain
Hi,
Ok, imagine a 3 stage NFB amp. Should I go for high or low open loop gain?
If I make the amp have less open loop gain I can move the dominant pole longer out, giving the same amount of feedback “power” to a larger frequency range – right? This will give more distortion at low frequencies – because of less NFB – but give the same mid and high band distortion?
The thing that bugs me is the low freq distortion. Will this be audible because its harmonics are low freq as well?
Thanks for any input
\Jens
Hi,
Ok, imagine a 3 stage NFB amp. Should I go for high or low open loop gain?
If I make the amp have less open loop gain I can move the dominant pole longer out, giving the same amount of feedback “power” to a larger frequency range – right? This will give more distortion at low frequencies – because of less NFB – but give the same mid and high band distortion?
The thing that bugs me is the low freq distortion. Will this be audible because its harmonics are low freq as well?
Thanks for any input
\Jens
I recommend D.Selfs "Audio Power Amplifier Design Handbook"
if you want to build a straightforward cost effective design.
Though I don't agree with all he says and the design compromises
he makes, there is a massive amount of useful info in this book.
It shows (if you can read between the lines) how to build
an extremely high performance amplifier for barely more
than a mediocre one.
sreten.
if you want to build a straightforward cost effective design.
Though I don't agree with all he says and the design compromises
he makes, there is a massive amount of useful info in this book.
It shows (if you can read between the lines) how to build
an extremely high performance amplifier for barely more
than a mediocre one.
sreten.
sreten said:I recommend D.Selfs "Audio Power Amplifier Design Handbook"
if you want to build a straightforward cost effective design.
Though I don't agree with all he says and the design compromises
he makes, there is a massive amount of useful info in this book.
It shows (if you can read between the lines) how to build
an extremely high performance amplifier for barely more
than a mediocre one.
Well said. I strongly agree, the book is about the best tutorial out, and yet I disagree with some of Self's conclusions. Still, a must read for amp builders today.
The original poster asked for advice on a cost effective, well performing amp. Ask 10 different designers and you will probably get 10 different answers. I'll give you several suggestions, with sources for bare boards available today.
Self's book makes a solid case for single ended designs (except for the complementary outputs of course). Slone's book (High Power Audio Amp Design) states that the symmetric designs tend to cancel 3rd order distortions, but he compares a primitive single ended design with a fairly refined symmetric one, not really a fair comparison.
I tend to agree with Self, that a well implemented single ended design will be excellent indeed, and that the extra parts for a symmetric design only add cost and decrease reliability. In fact, I have seen better implemented single ended designs that outperformed Self's best. I would go with a well implemented single ended design with EF outputs for a reliable, well proven design. That said, hardly any designers are doing them, because the symmetric designs have more marketing appeal today.
Here is a quick list of available boards for amps:
Slone sells boards ($32) and kits for most of the designs in his book, on the web at Seal Electronics. By all accounts his designs sound excellent and are reliable, solid performers.
http://www.sealelectronics.com/
Professor Marshall Leach offers boards ($25) for the famous Leach amp which everyone seems to love, at:
http://users.ece.gatech.edu/mleach/lowtim/
The excellent Borbely 60 watt and 100 watt symmetric designs are available from AudioXpress, fairly inexpensively ($11-16). Stay away from the DC-100 unless you are sure you can get the matched JFETS, which are long obsolete.
http://www.audioxpress.com/bksprods/pcbs/solidstate.htm
Of the amp boards now available, there is one notable exception to the symmetric design camp: the Pass A40 board from AudioXpress. At $6 per board, this is a Best Buy. It is a basic but elegant design which does not sound quite as good as the previous ones, but you would not expect it to for this price. (I warn you it has a PCB design error which the good people at AudioXpress will not fix. They just document it and let you make a cut and jump, not a big deal but rather.... unprofessional.)
http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
The original 26-year-old design article showed the correct artwork and is on the web at:
http://www.passdiy.com/pdf/a40.pdf
Mr. Pass is far too modest about this little design, stating that it is only suitable for Class A use and no part substitutions permitted. On the contrary, you can easily change the bias from class A to class B; it becomes a versatile single ended driver board for a broad range of uses. The power output depends on your supply voltages, but you can easily get 100 watts class B on a +-50v supply (but only with upgrades to all voltage sensitive components, for instance PNP -> 2n5401, NPN -> 2N5551, except for the VAS -> 2N5416 and CS -> 2N3440, FET -> 22K resistor ).
The A40 will drive class A bipolar outputs, class B bipolar outputs and especially class B lateral or vertical mosfets quite nicely, with proper bias changes. The bipolars must be darlingtons or have separate drivers, but you can use mosfets barefoot.
The trick is to set the bias properly, as needed for the outputs. Use as designed for class A darlingtons (but good luck finding PMD semiconductors today, perhaps you can sub 2N5684 and -86 outputs with TIP41-42C drivers, which unfortunately may be marginal for Class A operation at only 300 watt dissipation per output device
. Leave out the autobias diode and change resistors for class B bipolar outputs. For lateral mosfets, bias is 1-1.5v, for vertical mosfets, bias ranges around 6 volts, so you will need to put in a pot for one of the resistors in the Vbe multiplier (the 750 ohm one from base to emitter) and of course pick resistors to get you close to the right range.
OK, that's a starting point. Definitely check out the Self book and the Slone book. Probably any one of the suggested boards is a good starting point for your amps, but you should read those books and decide for yourself.
BTW you can get boards for some of Self's designs although I kind of get the idea they are inspired more his desire to explore an ideas than just to provide a good sounding amp. That's another way of saying the part count is a bit high. I've never tried one but I suspect they actually are good sounding.
He uses CFB output sections which can be difficult to get right but if anyone can get a layout that works with these Self is likely to be the one. I get some dark amusement from how he tosses off the advice the CFB can be subject to oscillations that are hard to pin down. I wonder how many folks have read right past that and then found themselves caught up in a frustrating exercise to find the cause of these "benign" oscillations.
If you layout your own board it's no big trick to substitute an EF output which may leave you nerves less frazzeled. I must be fair, however, a CFB an be done. I've done three but it was no picnic.
He uses CFB output sections which can be difficult to get right but if anyone can get a layout that works with these Self is likely to be the one. I get some dark amusement from how he tosses off the advice the CFB can be subject to oscillations that are hard to pin down. I wonder how many folks have read right past that and then found themselves caught up in a frustrating exercise to find the cause of these "benign" oscillations.
If you layout your own board it's no big trick to substitute an EF output which may leave you nerves less frazzeled. I must be fair, however, a CFB an be done. I've done three but it was no picnic.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=17524&highlight=
We had this symmetric question and discussed it rather much in the thread above. The conclusion was that a symmetric amp has more distortion, at least more odd harrmonics but we could agrre on which type really is the best. I personally like symmetrical best, by the looks at least
We had this symmetric question and discussed it rather much in the thread above. The conclusion was that a symmetric amp has more distortion, at least more odd harrmonics but we could agrre on which type really is the best. I personally like symmetrical best, by the looks at least
Re: Simpel amp - symetric or not?
Simple answer here is I think not.
Certainly when discussing cost/ performance/complexity issues.
Cascoding the input stage output, using this as the reference
for compensation and utilising two pole compensation look
pretty cost effective to me.
sreten.
JensRasmussen said:
Simple answer here is I think not.
Certainly when discussing cost/ performance/complexity issues.
Cascoding the input stage output, using this as the reference
for compensation and utilising two pole compensation look
pretty cost effective to me.
sreten.
"The AKSA is, in fact, very like the McIntosh - simplicity itself!
I do not favor symmetrical designs. I believe a single diff pair, solitary VAS and PP output stage can be made to sound best."
If I didn't already own about 40 large power amps I would probably build an AKSA, just to see how it stacks up.
Another single diff pair amp that turned my head was the Threshold S150. It, on the other hand , is not simple.
I do not favor symmetrical designs. I believe a single diff pair, solitary VAS and PP output stage can be made to sound best."
If I didn't already own about 40 large power amps I would probably build an AKSA, just to see how it stacks up.
Another single diff pair amp that turned my head was the Threshold S150. It, on the other hand , is not simple.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.