THIS is what I've been looking for, a clear personal opinion - thank you, Andrew.
Nige, I don't believe in special modes of operation, I believe in getting it right in the first go. Sit down, think what would be a nightmare for the amp and make sure that even it comes his way, he survives and does not become unstable, does not overheat, does not need to push out unholy quantities of current and/or voltage (meaning choose wisely according to your needs), and so forth.
Where the industry gets away with as little as one pair of power trannies, or in better sample two pairs, or in very good samples three pairs, I use 4 pairs. Sure, it's more complicated, more matching to do, and it surely costs money, but it pays with dividends in the end. If I can get away with two toroids rated at say 300 VA, I will likely use two 500 or 600 VA toroids.
All the while KNOWING full well that because of the easy load my speakers present to the amp, and their relative effciency, I will never get even close to its limits. It's not needing the resources, it's the pride of owning them. Of KNOWING that it can do a hell of a lot more if ever required.
So I try to keep it as simple as I possibly can and know how to do. As few switches, detectors and sensors as possible, but if required, don't think about it, just do it.
In other words, if you know EXACTLY the conditions it will work under, AND you trust the user (you do trust yourself, don't you?
), you will let it ride for all of the money.But if you're unsure of either, or both, you will add some bandwidth limiting?
Good logic - I agree.
Right. If I trust the user I mean that he will not blame on me when uses the gear improperly. Like his precious vinyl player without rumble filter damages his precious subwoofer because my amp honestly amplified the signal. Or his boutique DAC fried his super-tweeter.
As Albert Einstein once remarked - there are only two things I know which have no limits - space and human stupidity. I'm not sure about space.
As everyone who has ever developed an electronic product for the general market knows, half of the development time goes to idiot proofing the product.
Got you Wave, thank you.
back to the advantage of single pole passive filtering vs two pole filtering.
Adding a single pole passive filter to the input of every receiver is a real pass band adjuster.
But it is not the sole filter in the feed to the input.
Look at any Source to Power Amp connection.
The Source is likely to have an output impedance.
The cable will have some capacitance and some inductance.
The power amp has some input impedance.
All of these add up to at least two pole filtering and more likely a staggered three pole filter before the power amp.
Our job is the try to get the stagger about right that the wanted audio signal gets through and as much of the unwanted signal gets attenuated. That's where the advantage of multipole filtering comes in:- more attenuation.
I suspect we all already have at least two pole passive filtering at or before the input to our Power Amplifiers.
Adding a single pole passive filter to the input of every receiver is a real pass band adjuster.
But it is not the sole filter in the feed to the input.
Look at any Source to Power Amp connection.
The Source is likely to have an output impedance.
The cable will have some capacitance and some inductance.
The power amp has some input impedance.
All of these add up to at least two pole filtering and more likely a staggered three pole filter before the power amp.
Our job is the try to get the stagger about right that the wanted audio signal gets through and as much of the unwanted signal gets attenuated. That's where the advantage of multipole filtering comes in:- more attenuation.
I suspect we all already have at least two pole passive filtering at or before the input to our Power Amplifiers.
I believe this was not so much a problem in the past, when the transistors we had tended to roll off rather early anyway, but these days, with very high transistor Ft, I agree your above scenario is more than likely.
On the other hand, we actualy do do some subsequent filtering, like it or not. Eventually, we need to curtail the response of one or more stages, and in the process, we also do some filtering. The typical comensation capacitors ultimately reduce the transistor gain and their in situ response.
Stability issues make us do that.
Completely agreed with you that we need to investigate the entire signal path from the preamp output to the amp input. A simple example - the output impedance of preamps varies wildly from say 47 to say 600 Ohms, a 13:1 ratio. When deciding what your input filter should look like, you have to be resigned to the fact that if you work it out for say 120 Ohms (as per the IEC recommendations) and place your -3 dB cutoff point at say 200 kHz, with 47 Ohms it will be higher, and with 600 Ohms it will be lower than that.
I remember many years ago, the Norwegian company Tandberg had an interesting approach to this. In their 2075 top of the line receiver (cca 1977), they had a classic input filter at some relatively wildly high point, say like 300 kHz. Then, between the input differential pair and the VAS differential setup, they installed an RC filter with a lower cutoff point, say 200 kHz, both in inverting and noninverting lines. The ultimate effect would be a second order filter at say something like 150 kHz, but the input pair was given more leeway for better slewing purposes, I'm guessing.
However, in their prime time 3003 power amp (cca 1980), they avoided this setup and did some VERY serious input differential stage compensation, with a rather complex RLC setup instead. I'm still trying to understand the logic and purpose of it all. Unfortunately, they used some long gone BC transistor (447), which I do not have in my simulator, and frankly which I have never even heard of, so it's not easy.
This amp isn't as wrong as it looks .The simplicity is so close to disaster as to invite it . Many things here look badly wrong or you have read many times are wrong . Forgive any errors as it is to show an idea and might be weak on detail or things the wrong way round etc .
What is right with it ?
The input pair is reasonably low noise .
DC offset is reasonable , better than 25 mV if a little care is taken .
Input degeneration is not bad and is half resistor and half re of the transistor ( about 50R per side ) .
The input pair is reasonably balanced . Any imbalance will cause second harmonic distortion which might be nice to have in small quantities .
Slewing should be OK with real music
The VAS transistor has about the right input impedance and is running about 60% it maximum ( 400mW max ) . The input impedance is better than 1k3 ( { 25/5.3} x 280 )
VAS degeneration is minimal ( no emitter resistor ). Linearity is reasonable as the Cob is low using the external capacitance to dominate . Price is 4 devices / $
The bootstrap seems wrong on two counts . 10 uF seems too low , try it you might get a surprise as MOS FET's are odd animals . It can be a polyester 100 V 10 uF if so . Also the gain of the MOSFET's seems wrong to drive a bootstrap , not so . Be careful the other way , above gain 0.95 you might need a series resistor . Doubtless a constant current source ( CCS ) will be preferred . Some still prefer these bootstrap versions . Forget simulations , give it a try sometime . Like savory ice cream it might be OK ? Remove the capacitor and see how life changes .
No tail CCS ? Some poor chap put a tail resistor on a simulator the other day and was shock how good it was . It is only by going to JC's perfected CCS that he got a dramatic difference ( Thanks John , I never suspected it was for that reason the extra resistor added ) . Instead try a clean supply . It will cost peanuts to do . Use CCS and clean supply , why not ?
The filtering of input looks all wrong . As long as your preamp is conventional it should be OK . If using a transformer volume control as I did recently it might be ideal . I was very pleasantly surprised . It was a loan , glad I had a listen .
The 1000 uF is so as to be close to DC . That means only first order input filtering at the experimental stage . You might be surprised . You might keep it this way . As far as my ears can tell this is where being too cleaver and buying expensive capacitors might be a mistake . If you get your spectrum analyses out and measure this area you might find it is not the world you imagine . Mostly 100% DC feedback is the requirement .
The MOS FET's are like a little world of components ( 2 MF = 7 bipolar if including vbe bias , also many resistors ) . They love feedback like it is Christmas every day . Thus what seems so wrong with them is not so bad . They will not measure good open loop . However they love closed loop . Bias is not critical and can be set by ear . What a shame we don't listen to amps . If we did that might be an option ( ? ) . Sorry that was naughty . 100 mA is very correct as it is a design parameter known by manufactures .
Gain of 40 is about right . Could be used with a passive pre-amp and CD .
47 K input is reasonable , can be reduced . The 2K2 might be reduced .
The other by ear parameter is the 47 pF input cap . As John says there might with Vinyl or poorly designed equipment be some slewing issues . If not going to 50V /uS slewing we should adjust this capacitor to taste ( As Andrew said ) . Helps keep radio out . If marketing a product how good your ears are will be judged . Ultimately this can not be taught . I think it can be learnt . Sometimes you have to say I don't care what the simulator says I don't like that . 20 years later you find out why . Good news is 90% of the time if it measures well it will be fine . Excellent is a bit more work than fine .
The power supply will interact with the filtering choices . 2 x 22 000 uF I think is a minimum and 250 VA per channel .
It is assumed MOSFET's need more driving ( 60 V rails to VAS ) . Be careful and listen . The driver clipping first seems a virtue to me . Even my oscilloscope agrees .
From here one can get back to the usual school of design . Some might find usual is not to their liking . The Hitachi amp which this is a reduction of should be tried also .
I think some amps I see are the equivalent of playing 3D chess , fine if you like . I more than admire the ones who do it . Yes we can do much better than this design . I suspect 90% do worse when they play 3D ? I must thank Roy Gandy of Rega for teaching me this . He said make something as simple as you can then build from that . Never change more than one thing at a time . If not you never have truely learnt as you have no real idea what happened in the transition . Roy tested 2 Nylon screws verses 2 steel screws on the Rega turntables . His staff heard in seconds it was OK . It took him weeks they said . These screws were to stop people tightening them which caused problems . He was careful to be sure he liked it . Say what you like about Rega he is no idiot . Roy told me people like to look at the bearing . He then said no idea why . For that reason alone he made it 1 micron finish when 6 is OK . He said it was easy enough to do. I agree with John that when it is easy to do and has no down side we should go further than theory suggests . Sorry John if I misinterpret , It has long been thought less than 0.1 % distortion etc is the limit of human perception ( I have no idea if correct , I think it might be true ) . However if one amp sounds great and is 0.1 and another 0.0000001 % why not have the latter . The customer most likely will not hear a difference . If they do it is more likely bandwidth or stability / power delivery .
If MOSFETs are your cup of tea ...
I have yet to hear a MOSFET design whose bass lines are as convincing as those of the bipolars.
As for Roy Gandi's advice, it seems quite simple and reasonable. If you simultaneously change two items in anything, and a difference occurs, you have no way of knowing whether the difference was made by both, or just one, and if so which one, or by some interaction between the two.
Regarding our limit of hearing THD, I read somewhere it could be as high as 2% if the decay of the harmonics is as it should be, but could be as low as 0.1% if not. As for IM dostortion, the same text argued that 0.3% was already audible by quite a number of people. But I must say I think this was Otala's original text in IEEE, and that was in 1976, quite possibly we have learnt otherwise since then.
My view is that proper decay of harmonics is far more important and audible than simple THD as such. I base this on my own experience - I've heard amps rated at 0.1% sound way better than many rated at 0.001%. Otala's own amp, as proposed in the said IEEE paper, sounds much better than most offerings with both more power AND lower THD specs. And it's rated at 0.3% THD across the audible band, poor by today's standards.
I have yet to hear a MOSFET design whose bass lines are as convincing as those of the bipolars.
As for Roy Gandi's advice, it seems quite simple and reasonable. If you simultaneously change two items in anything, and a difference occurs, you have no way of knowing whether the difference was made by both, or just one, and if so which one, or by some interaction between the two.
Regarding our limit of hearing THD, I read somewhere it could be as high as 2% if the decay of the harmonics is as it should be, but could be as low as 0.1% if not. As for IM dostortion, the same text argued that 0.3% was already audible by quite a number of people. But I must say I think this was Otala's original text in IEEE, and that was in 1976, quite possibly we have learnt otherwise since then.
My view is that proper decay of harmonics is far more important and audible than simple THD as such. I base this on my own experience - I've heard amps rated at 0.1% sound way better than many rated at 0.001%. Otala's own amp, as proposed in the said IEEE paper, sounds much better than most offerings with both more power AND lower THD specs. And it's rated at 0.3% THD across the audible band, poor by today's standards.
BTW, Nige, I suggest you look over the old (1980 and onwards) issues of ELEKTOR. They had a fully complementary power MOSFET design which they revisited several times and by now, I reckon what was there has to have been squeezed out of it. A mature design, I'd say. And very good sounding. It was very popular locally, so I've seen and heard quite a number of them, certainly enough to say it sounds good.
You already know I believe thet, as John would say "all other things being equal", a fully complementary design will usually sound better than a classic design you show above. Not guaranteed, but certainly possible.
You already know I believe thet, as John would say "all other things being equal", a fully complementary design will usually sound better than a classic design you show above. Not guaranteed, but certainly possible.
Hi DVV , I've got the ELEKTOR somewhere .
If the output stage is changed to conventional vbe bias and Complimentary Feedback Pairs it generally improves . Equally adding FET's gives better bass power . I know what you mean about FET bass . However all the amps I heard like that were bad amps . Did you try them in your design ? No source resistors required if buying from Exicon who match at no cost . Having looked closely at your amp I suspect it would be great with them in place . 3 sets 300 mA bias .
I suspect complimentary will be a marginal improvement as the curves say it will . Without ever trying it it seems high order distortion is reduced .
I think 0.1 % IM is the target .
My design is saying what if we had FET's earlier . We might have had this . It's like we went from Monkey to Human in one step . In amps we did . This is the Neanderthal if you like . We now know they were not so stupid from the tools they used . Us lot probably killed them off .
This is my Boeing 707 if you like . 747 is generic . I hear there is a move to go just below supersonic with conventional designs . About time . One hour or better saved to NY .
If the output stage is changed to conventional vbe bias and Complimentary Feedback Pairs it generally improves . Equally adding FET's gives better bass power . I know what you mean about FET bass . However all the amps I heard like that were bad amps . Did you try them in your design ? No source resistors required if buying from Exicon who match at no cost . Having looked closely at your amp I suspect it would be great with them in place . 3 sets 300 mA bias .
I suspect complimentary will be a marginal improvement as the curves say it will . Without ever trying it it seems high order distortion is reduced .
I think 0.1 % IM is the target .
My design is saying what if we had FET's earlier . We might have had this . It's like we went from Monkey to Human in one step . In amps we did . This is the Neanderthal if you like . We now know they were not so stupid from the tools they used . Us lot probably killed them off .
This is my Boeing 707 if you like . 747 is generic . I hear there is a move to go just below supersonic with conventional designs . About time . One hour or better saved to NY .
That's only until they construct a jet engine which uses cow dung for fuel.
Norweigian style, I understand they are already using cow poo to manufacture bio diesel within your local friendly bio diesel pump station.After that, Mach 3.15 should be easy peasy.
Saves like 4 hours London to NY. Hey man, if you've got it, flaunt it. You may laugh . My friend Pippa ex BBC World Service and secretary to Alan Bond . He has just perfected a rocket / jet hybrid . UK to Australia in 45 minutes possibly in out lifetimes . To go to Spain would take longer as it would circle the Earth first . Perhaps the cost will be competitive ( $20 000 / kg I think at present ) . The main idea is to transport into space cheaply or was when I last took an interest .
http://www.reactionengines.co.uk/pr...JET ENGINE - PRESS RELEASE - 10 JULY 2012.pdf
http://www.reactionengines.co.uk/pr...JET ENGINE - PRESS RELEASE - 10 JULY 2012.pdf
Talking of bandlimiting, here is the old and famous (much questioned and criticized) study from Bell Labs.
http://www.alcatel-lucent.com/bstj/vol10-1931/articles/bstj10-4-616.pdf
I find data on pages 621-625 as important.
George
>Edit:This opens the door to a most valuable archive:
Bell System Technical Journal, 1922-1983
http://www.alcatel-lucent.com/bstj/
http://www.alcatel-lucent.com/bstj/vol10-1931/articles/bstj10-4-616.pdf
I find data on pages 621-625 as important.
George
>Edit:This opens the door to a most valuable archive:
Bell System Technical Journal, 1922-1983
http://www.alcatel-lucent.com/bstj/
Last edited:
Thanks for that BSTJ link! There are a number of interesting things I'd wanted to follow up on.
Nige, please don't misunderstand me, it's just that I find the idea of manufacturing fuel, on a local level, from cow poo so fascinating and yes, a bit amusing as well.
But this is by no means a "mad scientist" idea, the Norwegians are actually developing a network of fuel pump stations across the country which do just that, and of course, they have developed one model and are developing more small city cars which use that fuel. It's no joke, no far off vision, it's today's reality. And if it works, who am I to complain?
My only real fear is that the price of beef will rise, and then some. Baby beef will become a true luxury item. Who wants to slaughter a power manufacturing organism?
As for the development of new modes of traveling, I only took a short look at the link you supplied (I'll read it in detaila later), but it seems to me this is the slingshot theory at work. The idea being to shoot a projectile into space, which takes some fuel to do, and then almost sail back down to your desired destination. Overall, one saves quite a bit of fuel in comparison with our rather wasteful jet planes of today, AND you gain in traveling time.
I don't know if it will happen in my lifetime, but I sure am hopeful. I will always support an idea which does its job and saves in natural resources. This may sound whacky, but I am angry at GM's tardiness in introducing small, eco friendly turbocharged 1.4 litre engines in the car model I own, Chevrolet Cruze. I was forced to buy a classic version of the model, with a naturally aspirated 1.8 litre engine simply because they could not make enough of the smaller turbo engines to offer them across the board. So that energy and exhaust friendly engine if offer only in the rich West (USA, Australia) and not in the poor East. Sure, it's an Euro 5 compliant engine, very clean, but the smaller one is cleaner still, AND uses less fuel.
Dvv . It wasn't that . Pippa and I were walking one day . I have known her for years . She suddenly says " you would have liked my old boss ". I had no idea she did that , she is BBC Pippa not space scientist Pippa . Alas all of the things she typed , not one fact had stayed with her ! Very frustrating . It was always a mystery to me why a certain power in 1943 didn't strap pulse jets onto their aircraft ? What a stupid use they did put them to . My dad was delivering news papers and hid behind the wall when hearing one stop . A minute or two later two old ladies were puling him out of the dust and rubble . The street was gone ( Sutton , London ) . When I was a baby I used to play on one in the back garden in Rye Sussex of my grandfather . apparently when taken to a museum years later it was still live ! I even twiddled its propeller distance measuring thing !
I don't know if anyone noticed the amp I drew had almost no bandwidth limiting and fair bit of feedback ? I have built these amps and they are totally OK and can be tweaked with two pole VAS compensation to further increase high frequency negative feedback .
The big deal is the lack of low frequency restriction seems highly desirable . Even with warped records no real problems found . Each choice shown was about using the best ideas for each section . I suspect very low bass enhances a feeling of space in the music ? This is not bass power , just the corner frequency . OK it might not be in the recording . That is not an excuse to say don't bother . In fact if already a problem don't make it worse .
Going back to pulse engines . My interest in electronics came from my dad teaching me the basics aged 4 . As I was highly interested in my grandfathers flying bomb he taught me that also . Locomotives and motorcycles came next . I was allowed to clean the carburetor of his BSA . I took my dad to the steam railway the other day . They insisted he drive the train and not me . That was correct . To be honest I didn't want to . He insisted I was the train lover . To which I said not me . He is 81 .
The Bell labs paper . I will print that as printed seems to work better for me .
About engines . Why not a double 4 cylinder V8 synchronize ? One engine switches off when not required via a clutch . In addition hybrid electric drive ( 50 kW ) ? The nice thing about hybrid is it can provide 4 wheel drive when needed . The engines can be direct drive via a clutch to the wheels . The beauty is the V8 becomes available when true performance required . For example a long drive in Germany . Now the master card . The car uses the engines in rotation . Every time one starts the car the engine least used is put into use . In addition in failure mode one engine could be switched off . The two engines would share a water supply so as to have the unused one ready for use . These could be no heavier that a usual V8 if discounting the extra crank . An economy version with one engine could be made . I would suggest 2000 cc per engine to make it a true super car . One could even have a V8 and a four together if wanting to be exceptional . That could be a 6 litre T12 as in Napier circa 1930 .
I don't know if anyone noticed the amp I drew had almost no bandwidth limiting and fair bit of feedback ? I have built these amps and they are totally OK and can be tweaked with two pole VAS compensation to further increase high frequency negative feedback .
The big deal is the lack of low frequency restriction seems highly desirable . Even with warped records no real problems found . Each choice shown was about using the best ideas for each section . I suspect very low bass enhances a feeling of space in the music ? This is not bass power , just the corner frequency . OK it might not be in the recording . That is not an excuse to say don't bother . In fact if already a problem don't make it worse .
Going back to pulse engines . My interest in electronics came from my dad teaching me the basics aged 4 . As I was highly interested in my grandfathers flying bomb he taught me that also . Locomotives and motorcycles came next . I was allowed to clean the carburetor of his BSA . I took my dad to the steam railway the other day . They insisted he drive the train and not me . That was correct . To be honest I didn't want to . He insisted I was the train lover . To which I said not me . He is 81 .
The Bell labs paper . I will print that as printed seems to work better for me .
About engines . Why not a double 4 cylinder V8 synchronize ? One engine switches off when not required via a clutch . In addition hybrid electric drive ( 50 kW ) ? The nice thing about hybrid is it can provide 4 wheel drive when needed . The engines can be direct drive via a clutch to the wheels . The beauty is the V8 becomes available when true performance required . For example a long drive in Germany . Now the master card . The car uses the engines in rotation . Every time one starts the car the engine least used is put into use . In addition in failure mode one engine could be switched off . The two engines would share a water supply so as to have the unused one ready for use . These could be no heavier that a usual V8 if discounting the extra crank . An economy version with one engine could be made . I would suggest 2000 cc per engine to make it a true super car . One could even have a V8 and a four together if wanting to be exceptional . That could be a 6 litre T12 as in Napier circa 1930 .
- Status
- Not open for further replies.