Nigel
There is no need for interstage transf. to drive 2 x 813 UL / OPS in AB1 class, just one little modified Mullard 5-20 circuit will be OK ., and no need to drive those 813 in AB2 class of operation since than output power will be way over 300W , but 813 tube reliability well be compromised , and THD also .
Sorry that you don`t have three phase power system at your home , it is very good way for supply of even relative small power SS Amplifiers .
There is no need for interstage transf. to drive 2 x 813 UL / OPS in AB1 class, just one little modified Mullard 5-20 circuit will be OK ., and no need to drive those 813 in AB2 class of operation since than output power will be way over 300W , but 813 tube reliability well be compromised , and THD also .
Sorry that you don`t have three phase power system at your home , it is very good way for supply of even relative small power SS Amplifiers .
Last edited:
Nigel
Personally I don`t like performance of standard tube SE /A1 class circuit , basically they can sound Excellent & Ultra fluid in midrange and treble region , but usually they are without of any serious deep bass performance .
Actually my personal opinion that to DIY some SET circuit is OK , but only up to 10W of output power , normally using those extremely linear direct heated Triodes ( 2a3 , 3ooB .... ), open loop , without any GNFB , everything above that ` magic `10W is difficult & expensive to DIY .
Maybe some biamping can help in that situation , with some powerfull Singleton based 2 x 50 W SS AB class Amp`s as helpers for lowest bass spectrum driving some quality woofers .
Last edited:
There seems to be a way of making SE amps that dictates the usual sound. I am not sure what I did but even in the measured sense less problems than ones I compared. I spent 6 months plotting everything I could 8 hours a day and suddenly found the problems were reduced. I seem to find 8 watts is an absolute maximum and then one must change horses. I strongly prefer EL34 over 300B. I had both the state of the art Sowter transformers and low cost Danbury. I have to say if doing it blind Danbury would win.They exceeded their modest spec. I comfortably had 20 Hz to 47kHz and usable 15 Hz to 66 kHz. Speaker matching is vital as loop feedback not used. Fluid, exactly. Some think it is distortion. I doubt that. If so real music has it also. Transistor amps have subtraction distortion and valves addition. On the whole addition is the slight winner. Why does recorded Beethoven sounds so weird. False brightness which simply isn't true to real life. Same orchestra , same building and Mahler does not show it. Mostly Beethoven sounds thin, boring, overrated and musically poor on recordings. HMV Ricardo Muti No7 although still too expansive sounds uplifting. 78's are much closer to the truth. I love all music, but judge classical. Quad ESL 63 mainly as reference.
The little SS amp I showed today in link seems rather well thought out. Tiny bit more current perhaps to get it to slew faster. The HF distortion is not bad. The mechanisms that make SS AB bad are not the same as SE valve so we must have very low HF distortion .
The little SS amp I showed today in link seems rather well thought out. Tiny bit more current perhaps to get it to slew faster. The HF distortion is not bad. The mechanisms that make SS AB bad are not the same as SE valve so we must have very low HF distortion .
The non polar electrolytic I know of shine both sides of the foil. Very subtle if looking for it. Keep them below 0.4 V and they work well. Often that is not too difficult, for example feedback loops. To be very simplistic at 0.4V they become mildly awful and deserve their bad reputation. That sound is dull, enclosed and slightly harsh. I built an amp with a gain of 100 and 100 watts. It sounded very wonderful. Maybe that's why? Gain of 60 is very good.
Nigel
I suspect that you prefer EL34 over 300B is mainly because of different G1 driving swing requirments ,
any way I think that very obvious fluid sound presentation from SE Amps is result of many simple coincident inside of those simple SE tube circuits , from specific electro-magnetics behaviour inside of SE - OPT , A class of operation of all stages , No GNFB , .....
BTW , before one month or so I was driving my car trough Belgrade streets , on one intersection was the red light and I stop waiting for green light , about 20 or 30 meter from me one guy standing alone by the wall was playing some unknown Jazz notes with his saxophone , than I spoke to my self :
If I could 100% faithfully reproduce only just that single saxophone live sound at my home via some Audio system ..... , that will be something ! .
I suspect that you prefer EL34 over 300B is mainly because of different G1 driving swing requirments ,
any way I think that very obvious fluid sound presentation from SE Amps is result of many simple coincident inside of those simple SE tube circuits , from specific electro-magnetics behaviour inside of SE - OPT , A class of operation of all stages , No GNFB , .....
BTW , before one month or so I was driving my car trough Belgrade streets , on one intersection was the red light and I stop waiting for green light , about 20 or 30 meter from me one guy standing alone by the wall was playing some unknown Jazz notes with his saxophone , than I spoke to my self :
If I could 100% faithfully reproduce only just that single saxophone live sound at my home via some Audio system ..... , that will be something ! .
I had a 12 inch 45 RPM disc recorded in the Corn Dolly Oxford of the Sax solo Alfie. Lots of hiss and lots of reality. It is lost now. Have you noticed Sax rarely has the hoot and bass colour on recordings? Sorry to say 78's often do. This Charlie Haden link only fails on Sax so they and I didn't show it, the rest not given is a melodic masterpiece. This bit is for the real Jazz interlectual. And to think A&M and the Tijuana Brass gave us it. ECM more often recorded C H.
https://www.youtube.com/watch?v=mEPw9FHo_ZI
https://www.youtube.com/watch?v=aUvWeeTmjNk
https://www.youtube.com/watch?v=eXFaJpb-reg
https://www.youtube.com/watch?v=mEPw9FHo_ZI
https://www.youtube.com/watch?v=aUvWeeTmjNk
https://www.youtube.com/watch?v=eXFaJpb-reg
Last edited:
Nigel
That`s very good observation ! , recorded & later reproduced sax sound usually don`t have that natural hoot and bass sound color , but it is far away from live sax sound , with very limited tonal spectrum , as somebody during recording session turn of the bass pot on minimum .
Thanks for your tracks suggestion , special for Fathead Newman music , and here is one of my favorite jazz sax recording : https://www.youtube.com/watch?v=OI6usmV625Y, 1959 recorded , far from ideal , but sax sound is pretty good .
That`s very good observation ! , recorded & later reproduced sax sound usually don`t have that natural hoot and bass sound color , but it is far away from live sax sound , with very limited tonal spectrum , as somebody during recording session turn of the bass pot on minimum .
Thanks for your tracks suggestion , special for Fathead Newman music , and here is one of my favorite jazz sax recording : https://www.youtube.com/watch?v=OI6usmV625Y, 1959 recorded , far from ideal , but sax sound is pretty good .
813 SE triode amps
Champ 1000 Watt Tube Amp
Here are two examples of 813 amp . One is 12 watts on a good day and the other is 1000. SE people seem happy with expensive output transformers that do not seem as good as the cheap Danbury ones I used. That is if these specs are right.
The RH series amps seem excellent and on paper, no worse than this 813 amp allow for small power difference. RH34 is my favourite.
The odd thing to say is my EL 34 amp was not 1000 000 miles different in concept yet seems to out perform this 813 amp in SE. I have no big argument with this except to say there seems to be a way of doing things which repeats in each design I see. I had 1% THD 8 watts and 0.2% 1 watt with no loop feedback. I also used pentode input and 82% UL near triode. I guess I should be happy. In fact I was identical and possible had marginally nicer distortion curve(very similar) . What I didn't have was the slight bit of extra output. Lets be clear, 8 watts to 80 watts is a nice difference. 20 watts is not enough to bother. As said , 8 watts SE is when to change horse. 500 watts PP seems a better horse. As you say 200 watts should be easy to get. My SE UL driver would be excellent if my little test was representative. Alas I can not show my design as I let the rights go to another who has not bothered to make it. Apparently his business has been overly successful since I did it. I supect it knocks what he makes sideways. His 300B amp was not as good. It was loose and distorted and slow. Romantic but........
Champ 1000 Watt Tube Amp
Here are two examples of 813 amp . One is 12 watts on a good day and the other is 1000. SE people seem happy with expensive output transformers that do not seem as good as the cheap Danbury ones I used. That is if these specs are right.
The RH series amps seem excellent and on paper, no worse than this 813 amp allow for small power difference. RH34 is my favourite.
The odd thing to say is my EL 34 amp was not 1000 000 miles different in concept yet seems to out perform this 813 amp in SE. I have no big argument with this except to say there seems to be a way of doing things which repeats in each design I see. I had 1% THD 8 watts and 0.2% 1 watt with no loop feedback. I also used pentode input and 82% UL near triode. I guess I should be happy. In fact I was identical and possible had marginally nicer distortion curve(very similar) . What I didn't have was the slight bit of extra output. Lets be clear, 8 watts to 80 watts is a nice difference. 20 watts is not enough to bother. As said , 8 watts SE is when to change horse. 500 watts PP seems a better horse. As you say 200 watts should be easy to get. My SE UL driver would be excellent if my little test was representative. Alas I can not show my design as I let the rights go to another who has not bothered to make it. Apparently his business has been overly successful since I did it. I supect it knocks what he makes sideways. His 300B amp was not as good. It was loose and distorted and slow. Romantic but........
Last edited:
No, never. I feel it complicates things a lot, and I know it makes things a lot more expensive.
But I have considered the possibility of extending the existing regulator to the driver as well. This is much more effcient and easily achieved. In fact, I tried it a few times. AS things stand now, it seems to add little to the sound quality for a not insignificant price hike.
The driver gobbles up 3-7 times more current than everythig before it, thus requiring more powerful transistors in the regulator, which would need their own drivers.
Another approach uses a simple resistor followed by a larger capacitor especially for the drivers, but still fed off the unregulated line. This can be heard as sligtly improved under only the most stressful situations for the driver, like feeding the output stage for 2 Ohms and say -45 deg load phase shift. Short of that, not worth the time and trouble.
I used 22 Ohms 2W and a 2,200 uF/63V capacitor.
Actually, I own no such speakers. Worst case phase shift with my regular speakers is -25 degrees at 250 Hz, while the minimum impedance occurs just below that and is 6.5 Ohms.
My worst case load would be my AR94 speakers, which dip down to 3.3 Ohms with a phase shift of -45 degrees. That works out to an equivalent impedance drop to 2.34 Ohm equivalent. By the time this become fully relevant, which would be at around 26 Vrms output, the SPL in the room is unbearably high on an average level, which is to say that even those would be only peaks, certainly not continuous.
This will depend on the amp's ultimate purpose. If I was making it just for myself, than 300VA trafos would be just fine and with room to spare. However, if it is to be general purpose, then I am in the same fix everybody else doing the same would be in, not knowing what the amp might face somewhere.
Agreed. The only rationale for a high power amp is if one has a known difficult load of low efficiency and insist on concert hall levels. For te rest of us, the only rationale might be that we will have peaks here and there and that we insist on unchenged amp tonality even when under relative stress. In other words, we simply want a good headroom.
In my particular case, the trafo load regulation is 5% or less, i.e. 5% for 300VA trafos and just 3% for 500VA and above. Manufacturer's specs quoted.
Also, I do not intend to commerialize it, so I am not under any pressure to have to quote wild specs to boost sales.
The question is - if I realistically need say 300VA trafos, will it hurt if I use say 500VA trafos? This because locally, the price difference between 300 and 500 VA trafos is rather small, if memory serves something like €20 or so?
Frank, after my experimenting with fully regulated power supplies, I have come to the conclusion, right or wrong, that the regulator speed is a major player in the whole kaboodle. If it's not as fast as, or preferably faster than whatever it feeds, all bets are off.
Thus, my choice of MJE 15030/15031 is no freak accudent. If you look at their data sheets, you will notice that in the region I am using them in they have an effective Ft of about 60 MHz, i.e. they are fast.
To regulate the driver transistors, I'd have to use something even faster than they are, and once again, they are MJE 15030/15031. So, I'll either use two of then im parallel for power dissipation reasons, or perhaps say IRF 240 MOSFETs which have the required muscle and are faster than the MJEs.
Whatever regulates them, it will need a LOT of development work to make it work properly.
Lastly, don't forget that everything I do is fed off filtered power lines, meaning that my reference baseline is initially way higher than usually/normally. This does free me from a fair numer of considerations others have make, but of course at a price premium of the filter's price. And, after my hands-on experiences with even very much High End devices, which that same filter improved upon even when it shouldn't have, I honestly see no particular benefit in further complicating the audio electronics. Too little benfit - if any.
In addtion to the fact that Jan is right in observing that the normal average power levels we use at home are really fairly low. Since my filter is limited by EU safety rules and regulation to just 10A 240V, I'd say I have an adequate safety margin (on its own, it shows first signs of saturation ate around 14A 240VAC standard model, 28A 240VAC super model - take a pot shot guess at which version i'm using to feed the amp).
SLEWING
All this slewing distortion nonsense was put into focus by Ben Duncan. Ben suggests that all slewing is about is the feedback loop lets go and shows crossover distortion. That would probably be when driving a reactive load. Now would that ever happen??? Whether Ben meant me to see it that way I don't know. TID might simply be exactly that and no complicated reason. Thus a very low slew rate amp with sensible input limiting ( 80 kHz) might outperform monster slew rates for the more minor interest of how the amp sounds if this loop is strong.
Fitting a VAS resistor might help as the loop is not going to be so very much damaged at HF doing that . Better to have that loop strong whilst loosing a bit of bass control. Could even be that bass damping control is the cause of instability higher up ? I dare say if VAS re is increased Cdom can be better tuned? I speculate that 16 R 8 mA about right if VAS u=100. Roy Gandy said an experiment needs at worse two variables. Better make it one. That's why I like simple designs. As Roy said you never know in a complex design exactly what you did. The staff at Rega said he drove them mad as one new screw was a 6 weeks test. That was metal to nylon to prevent people doing up the motor too tight for goodness knows what reason. He told me he polished the bearing to 1 micron as it costs no more if done in quantity. The reason he said was people look at them and inferred something ( pretty?). If not 6 micron would have been fine.
All this slewing distortion nonsense was put into focus by Ben Duncan. Ben suggests that all slewing is about is the feedback loop lets go and shows crossover distortion. That would probably be when driving a reactive load. Now would that ever happen??? Whether Ben meant me to see it that way I don't know. TID might simply be exactly that and no complicated reason. Thus a very low slew rate amp with sensible input limiting ( 80 kHz) might outperform monster slew rates for the more minor interest of how the amp sounds if this loop is strong.
Fitting a VAS resistor might help as the loop is not going to be so very much damaged at HF doing that . Better to have that loop strong whilst loosing a bit of bass control. Could even be that bass damping control is the cause of instability higher up ? I dare say if VAS re is increased Cdom can be better tuned? I speculate that 16 R 8 mA about right if VAS u=100. Roy Gandy said an experiment needs at worse two variables. Better make it one. That's why I like simple designs. As Roy said you never know in a complex design exactly what you did. The staff at Rega said he drove them mad as one new screw was a 6 weeks test. That was metal to nylon to prevent people doing up the motor too tight for goodness knows what reason. He told me he polished the bearing to 1 micron as it costs no more if done in quantity. The reason he said was people look at them and inferred something ( pretty?). If not 6 micron would have been fine.
Dejan . You don't need fast transistors to regulate stages in class A. That's the answer. As I showed a day or two back if dissipation is OK all except dumpers can be class A and that will make the dumpers much faster into the bargain. Also separate PSU is possible using it's own rectifier and no regulator (fast soft recovery diodes). Mike Creek did that. Only dumpers need high speed and they can use any old crap PSU ripple if HT is high enough (I am being a bit naughty to say that). The dumpers become regulators in themselves if lets say 5 V is lost. We are not talking money here. That 5 V can be brought back if book spec is required. Call it party mode.
Nigel
Both 813 amps look OK , but that 1KW Champ is way to powerfull , it will be good to use them as modulator for some powerfull AM transmitter .
Your 8W EL34 amp is OK with distributed THD specs , just run for some high efficient ( bass ) loudspeakers , in the past UK made some very efficient ones .
And I did not say that is easy to get 200W , I actually say that 2 x 813 tubes in PP/UL /AB1 class will get around 250W with easy and will last for decades , but associated hardware around tubes will be Very Expensive ,
actually the cheapest things in that 250W amps is the tubes with associated sockets , but OPT need to be very good and expensive one , maybe some double C core OPT with S= 30cm /square or more , pretty big main transf .to , PSU chockes , PSU oil capacitors , chassis , ..... good to spend around 800E or more per mono block .
But the main question still remain : Who need 250 W / Ch for home use ? , I don`t .
Both 813 amps look OK , but that 1KW Champ is way to powerfull , it will be good to use them as modulator for some powerfull AM transmitter .
Your 8W EL34 amp is OK with distributed THD specs , just run for some high efficient ( bass ) loudspeakers , in the past UK made some very efficient ones .
And I did not say that is easy to get 200W , I actually say that 2 x 813 tubes in PP/UL /AB1 class will get around 250W with easy and will last for decades , but associated hardware around tubes will be Very Expensive ,
actually the cheapest things in that 250W amps is the tubes with associated sockets , but OPT need to be very good and expensive one , maybe some double C core OPT with S= 30cm /square or more , pretty big main transf .to , PSU chockes , PSU oil capacitors , chassis , ..... good to spend around 800E or more per mono block .
But the main question still remain : Who need 250 W / Ch for home use ? , I don`t .
You have just voiced the talk of British audio industry. Limit the input signal and you don't need any big deal of an amp slew rate.
What is completely missed here is the phase shift effect of that filter on nominal 20 kHz. Search the web and you'll find that if this phase shift is mentioned at ll it will be at 20 kHz, as the nomnally upper limit of our hearing.
Most don't even mention it, presumably because they have nothing to say. Some do, mostly High End companies, and of the more popular makes, Harman/Kardon is the only one I know of. It's usually declared as square wave tilt at 20 kHz, either in degrees or in percentage, the accepted conventional limit being 4 degrees or 5%. No idea where that came from.
This may or may not be true, but I see it like this: it certainly can't hurt to be precise. And the only way to get it is to have a widebandwidth amplifier.
If you now introduce an input filter, which is usually first order or 6 dB/oct., it does have a small phase shift, but put it at 80 kHz, and it will filter downwards to 20 kHz as a larger square wave tilt. Or, if you want to have it, take a good look at how low you can place it before it starts to have more serious effect at 20 kHz.
I do use it, and I agree with what Sony and reVox/Studer usually did or do, which is to put it at around 200 kHz. That's still low enough to kill some unwanted HF, but high enough not to do too much damage at 20 kHz.
80 kHz is a very much British limit, look and you'll see that many, even most British amps do it at just around that point. In my view, that's too low, which is probably why I find many British made amps to lack some sparkle up there.
Elsewhere, John Curl said it: giving an amp a slew rate of 100 V/uS is easy enough if you go for it straight away in the design stage. That translates to about 280 kHz bandwidth for a nominal 28.3 Vrms of output, which is not too high, I think.
The point, Nige, I think is not in the slew rate as such, but at what it is a product of, bandwidth. The wider it is, the less phase shift you have.
Indeed, slew rate problems are a thing of the past, provided the designer is half way competent. Use fast devices, degenerate the input pair (which ultimately is the source of the max current to charge the comp cap) and you can have a stable amp with relatively small comp cap.
No more slew rate problems.
You can use an input filter for bench test purposes, but since music is inherently band-width limited, you don't need it for the intended use.
Jan
- Status
- Not open for further replies.