As far as I can see you have come rightly to see that the amp needs a VAS and not a TIS. The difference is the TIS is fed from a resitances ( impedance ) higher than it's input base impedance ( or emitter if you prefer ). VAS has the feeding impedance lower. This is having your cake and eating it. I discovered this when making old British amps sound good. The insistance on saying it is a current driven stage ignored the rather better idea of pretending it is a valve. The only trouble then is to make certain Cdom can be driven. If a single VAS we often need far more drive just so that sourcing and sinking is enough ( it is not symetrical ). If we were that one legged cyclist we need a Tour de France leg.
MOS FET's not having minority carriers do not mind a simple driving circuit. Ironically the audio FET has the ideal bipolar driver. That is a resistor between gates ( or bases ). Nothing better to clear the bases of charge.
Now there is a thought. Sinclair used a resistor. It's voltage controlled by modulating the CCS ( heat ) . If the CCS has a surplus why not do that ? Then the Bipolar amp has what it needs. That Vbe multiplier most use might not be ideal.
MOS FET's not having minority carriers do not mind a simple driving circuit. Ironically the audio FET has the ideal bipolar driver. That is a resistor between gates ( or bases ). Nothing better to clear the bases of charge.
Now there is a thought. Sinclair used a resistor. It's voltage controlled by modulating the CCS ( heat ) . If the CCS has a surplus why not do that ? Then the Bipolar amp has what it needs. That Vbe multiplier most use might not be ideal.
Lol, now you have a reason...BTW, JFets have the lowest 1/f noise of all transistor types......JC ain't stupid !.
Another interesting point in the article that I referenced is that PNP bipolars are half as 1/f noisey as NPN bipolars, and one tenth as noisey as Mosfets.
Early in my audio career, I recall hearing some early transistor amps with PNP quasi output stages and noted something 'right' about them compared to complimentary outputs.
This is an area for future revisit/investigation.
It might take you a few amp swap in/outs to get a handle on what I am saying....you could try brown noise, or pink noise.
OTOH I have some FNAB (Quantum Purifer Formerly Known As Bybee *) like filters that I can instantly swap in and out of the system on the fly.
This makes such comparisons trivially easy and unequivocal.....of course they stay back in the system once such listening comparisons are completed.
Dan.
* 20 years back the performer Prince (Punce) declared that he should no longer be called Prince, and instead should be referred to as Person Formerly Known As Prince.
One of the local radio stations made great fun of this and conducted a phone in competition to decide a new name for him.
The comp ran all week with all manner of suggestions, some good, some stupid.
On the eleventh hour, a listener phoned in and said "I reckon you should just call him Davo".
That suggestion was declared the instant winner, and to this day the moniker Davo is still mockingly used by this network radio station.
** In Aus just about any name gets tagged with the 'o' suffix....Johnno, Stevo, Wino, Garbo etc...
Last edited:
Do you think you know how Bybees work? I'm very curious. I thought that, at the price, they must contain some very expensive - even magical - component.
By the way, I think it was "The Artist Formerly Known as Prince" - a pretty effective tactic, since we are still amused by it 20 years later. In spite of his foibles, I think he is a pretty talented mix engineer, judging by a mix he did 20 years ago of Mavis Staples. Sounded great.
Ooooooh, I feel guilty now, I have obviously not brought you up t date where I have come to stand. Mea culpa. Give me a day Nige and I'll sen you the latest schematics. Simplier than before, but just as effective.
Please, just send me an empty note ro dvv333@yahoo,com, I've had to change e-mail adresses. And write this one down.
There have bee hints dropped over the years.....magic is not mentioned.
The construction is a white ceramic tube with a metallization band at each end, and coated with a conductive/resistive dark layer joining the connection bands.
Afaik, this layer is applied as a paste/slurry and then fired in a kiln for extended period...several days iirc.
A 'special' very low ohm resistor is fitted down the centre of the tube, and copper end caps are fitted/soldered to connect the two conductors in parallel.
I think the expense is in the 'rare earth' oxides used, the resistor, and the processing/labour costs.
Long before I heard of FNAB's, I experimented with various oxides and other compounds and found all to cause colourations/signatures of differing sorts, and various mixtures caused further signatures....some pleasant, some running from the room awful.
I have three FNABs, two complete, and one dismantled.
The resistor and the tube have individual signatures, and the assembly has another.
Subjectively, the FNABs do cause a subjective 'quieting'/calming in the system, perhaps partly due to a 'dithering' like quality, but they also cause diminishing of transient attack/immediacy/liveness and and an overall signature that whilst I recognise and respect, do not like long term.
I am currently running an interesting mixture that cleans/reduces VLF noise, with the resultant being outstanding system solidness, liveness and dynamics, clarity and really nice extension in lows and interestingly the highs.
This also works with recording systems and stadium live sound reinforcement systems.
So in answer to your original question, no, I do not fully understand how FNABs work.
That said, all materials have a 'signature'...the trick is in noting, understanding and combining these individual signatures.
Thanks, I stand corrected...BS Artist more like.
Agreed, technically good mixes, I just can't stand his brand of music....drives me out of the room.
Dan.
Here you go.
Note that this is all about when a bipolar is driven into saturation, where (for an NPN say) the collector-emitter voltage is significantly lower than the base-emitter voltage. This is NOT the typical situation for a complementary bipolar emitter follower output stage, unless the drivers can pull above and below the main output power rails. Note the prescriptions to avoid storage time altogether.
Attachments
The transient at the start of a bass strike is what it's all about; steady, harmonically spare, 'pure' bass is incredibly boring - it's all the "accompaniments" to the fundamental bass frequency that actually make this part of the spectrum interesting. And a lot of the time the accompaniments are so wrong it's ridiculous - for me, downright irritating, a complete turn-off. As bad as hearing some stupid kid driving down the street with the ground throbbing away - "duff, duff, duff, duff, duff ..."
I listen for the harmonic correctness of the bass notes being produced, and hence hear a bass richness in even, yes, microscopic laptop speakers - I don't need the throbbing undertone to be happening all the time to get the sense of bass being part of the mix ...
I listen for the harmonic correctness of the bass notes being produced, and hence hear a bass richness in even, yes, microscopic laptop speakers - I don't need the throbbing undertone to be happening all the time to get the sense of bass being part of the mix ...
I guess you might be running a bias current of 350 milliamps or more, meaning three or four pairs of output devices. (1watt =2*bias current amperes^2*4 ohms)
You can do this with multiple output devices, however I feel concerned the mention of 1W/4 Ohms might inspire others to attempt this in a lower powered amplifier using single output pairs. Even 1W/8 Ohms would need 250 milliamps
Taken too far, increasing bias will displace the crossover point from an ideal position near the cusp of conduction to a steep point up the transfer slope where the discontinuities between the output halves will mess up at the levels where much listening is done.
You are right, Christophe, the word I should have used is probably "sudden". Like a ot of bass lines coming on at once in an instant, first nothing then at full blast.
AS a matter of fact, I often use The Blue Man Group first CD. Two songs stand out, No.5 which alternates between slow and easy and fast an thunderous, but the real deal is song No.8. It has a deathly silence and the deepest and most forceful bass I ever heard fom any CD. That's one of them hitting a drum with a diameter of some 8 feet (app. 2.5 meters). Anything less than a capable amp will instantly dry up and you will hear it as a poorly reproduced bass. It makes the bass cone visibly move, so I suppose its actual frequency is below 40 Hz. It really socks it to you. A wonderful CD to have to beat off those who believe 5W of tube audio is enough, just play it on their system and hear the amp choke and gag.
Or, if you prefer Vangelis, just play the title theme from Bladerunner, called "Metallic rain". About 1/3 in, there is a thunderous joining of a whole band at almost full force. Hear them choke and gag.
For the mids, Hevia (a Spaniard who plays electronic bagpipes, for which he holds the patents). If your speaker can deliver that, then it has a very good midrange at the very least.
Then there's the classics, like Billy Cobham's "Quadrant 4". Never fails.
I guess you might be running a bias current of 350 milliamps or more, meaning three or four pairs of output devices. (1watt =2*bias current amperes^2*4 ohms)
You can do this with multiple output devices, however I feel concerned the mention of 1W/4 Ohms might inspire others to attempt this in a lower powered amplifier using single output pairs. Even 1W/8 Ohms would need 250 milliamps
Taken too far, increasing bias will displace the crossover point from an ideal position near the cusp of conduction to a steep point up the transfer slope where the discontinuities between the output halves will mess up at the levels where much listening is done.
Your comment is well taken. How much bias can be applied is EXTREMELY varied from amp to amp. A specific example here - I played around with my wife's HK 680. It's factory setting is 80 mA per device, and it uses two pairs. I took out their junk quality trimmer and installed a multiturn Amphenol trimmer, then raised the bias current in 10 mA steps. Some gain was obtained at first, a slightly more fluid sound, but I could hear no difference between 130 and 140 mA, or rather, there was a difference but in the unwanted direction. The sound became still more fluid, but some detail started getting lost or subdued. So I turned it back down to 120 mA, that's still 50% more than specified, but a bit better than specified.
This also happened with other amps. There are limits one should not cross. For really higher up, say 130 mA and upwards, an amp has to be designed for it, or you pay the price. Not surprised, add more bias and you actually upset the whole current gain line, predriver, driver and output stage. What the output transistors might take in their stride may not sit well with the predriver and driver. It's a field you have to tread lightly to improve without paying the price.
Oh yes, I forgot - at 135 mA per device, and with four pairs, I am actually running the amp at (4*135) 540 mA. The price I pay is that my predrivers will require local heat sinking, the drivers are OK since they are mounted on the big system heat sink anyway, to keep their thermal conditions as similar to the output devices as I can.
So where can-be the problem, Dvv ? Bass attacks like kick drums are high energy, but the slope will be in the midle/trebbles harmonics. So, where can be the differences (on *your* speakers) ? Damping factor, I believe, and power.
With my impedance compensated speakers, I'm unable to make a difference in the bass region, between a Class D, a class AB with voltage feedback, and a class AB curent feedback, equiped with BJTs or FETs. As long as they have enough power and same power supplies.
About quiescient currents, i have the same feelings than you. There is a point where, increasing the class A range don't give any benefit at my ears. I usually set my amps to 150mA per device. Happily, it is the point where Laterals have a flat temperature factor, above which it becomes negative: no need for thermal compensation.
Last edited:
Christophe, the problem is in the sudden burst of energy much above what is normally used, which leaves many amps current starved. Those with good PSUs just sail through it, no questions asked. And I do not mean MY speakers, which are both impedance and phase corrected and are reasonably efficient at 92 dB/2.83V/1m. Change them for say AR94, and it's another story, lower efficency by 4 dB, lower nominal impedance, equals more power needed. JBL Ti600 are an easier load to drive, so not too bad with them.
I'd like you to name some amplifier costing below €1,000 which has a really good PSU. PSUs in them are usually the first victims of "saving" money in mass production. Next on the savings list are output stages. Exactly where it counts the most.
I'd like you to name some amplifier costing below €1,000 which has a really good PSU. PSUs in them are usually the first victims of "saving" money in mass production. Next on the savings list are output stages. Exactly where it counts the most.
Dejan. This is a speculation I did on the old Quad 33 runing up to 80 V DC per rail. The 0R1 and 0R22 allow the Quad protection circuit to work. A LED doing the work of a silicon diode to raise the current limit. I like the second idea best. It is a circuit ideal for reduced loop feedback. I suspect a shunt cap between CB on the T03 device to make stability like that of a triple Darlington. The -ve is just mirror image. This is concept and values from a Quad 303. I would have changed them if it became real no doubt ( 4R7? ) . The TF 2000 VA. Call the idea Darlington-Sziklai triple. I would not think this circuit to be less robust.
Some UK manufacturers use a similar topology,e.g. Audiolab. I must admit I never saw the advantage of using Sziklai type of predriver/driver, despite the fact that theoretically you have soe local feedback with it. I find that speed is of ultimate importance in the output stage, and thus prefer emitter-follower topology, as the fastest around, even if not with least distortion. Also, it's easily programmable and cn be suited to almost anything we need.
Last edited:
I tried the idea with no loop feedback. It was excellent. Strange thing is with a little noise from the input pair it works better. Sort of dither. I can agree with Dan. 1/F is the noise to reject. Pure white noise is OK. In the 1960's someone suggested a supersonic wave and zero bias. Call is Class BC with carrier. It could be OK. The thing to grasp is for very little complexity a class D advantage taken into class BC. The output choke made to suit. The sounds will be a bit brighter is my guess. This will mask the choke a bit. I doubt it needs much power to work ( 2 V rms or less ? ).
Something very weird has happened with my OB speakers and I have no idea how it happened. FM was sounding very old fashioned. Simply because it fails in the same way as the speakers. Last night whilst listening to a wonderful BBC Radio 2 program on Nick Drake I went from digital to FM ( FM is digital since 1972, special system I suspect still in use, A type of Nicam ). Yes it was mildy coloured, it was also simply better. More depth. Listening to Ravel on Radio 3 FM right now. At last a sound that makes me want to listen. It sounds much like 1980's FM through Quad ESL 57's in texture. That is the speakers are not obviously there. No antiseptic quality which Radio 3 sometimes has. Like the Quad's a nice, not very obvious treble quality. The cup distortion I am sure is still there. I have listened so long I have tuned it out.
On Colleen's system we played The Three Tenors on Decca on CD and LP via JVC L3-E. The presentation almost identical. The JVC is like a broadcast device in this ( derived from TT 71 ). The Audio Technica PU if that's what it is can be highly accurate ( AT 95 had the lowest distortion figures I have seen regardless of price, Linn K18 is a posh version ). The CD was supurb and yet the LP was better. The LP was as new. It was hard to think it wasn't the mastertape as the JVC is that good. This is bonkers as the CD should be closer. One engineer said the stupid conversion to 44.1 kHz might be to blame. It was said the owners of copywrite don't want direct access to the mastertape. The specific differnece was depth around singers to other things in that space. CD was cardboard cut out sound. The LP was if you like 16 bit and the CD somehow MP3. Is it possible CD is this bad? MP3 I don't mind as it is an honest way to do a good job in limited file space. MP3 is good tap water if you like. In tonality slightly preferable to FM. MP3 will record the way 78's sound. That's not bad. I have a few FM tuners so speak for all when saying the medium of sound is a bit coloured. Better than cassette, less good than a Revox.
Something very weird has happened with my OB speakers and I have no idea how it happened. FM was sounding very old fashioned. Simply because it fails in the same way as the speakers. Last night whilst listening to a wonderful BBC Radio 2 program on Nick Drake I went from digital to FM ( FM is digital since 1972, special system I suspect still in use, A type of Nicam ). Yes it was mildy coloured, it was also simply better. More depth. Listening to Ravel on Radio 3 FM right now. At last a sound that makes me want to listen. It sounds much like 1980's FM through Quad ESL 57's in texture. That is the speakers are not obviously there. No antiseptic quality which Radio 3 sometimes has. Like the Quad's a nice, not very obvious treble quality. The cup distortion I am sure is still there. I have listened so long I have tuned it out.
On Colleen's system we played The Three Tenors on Decca on CD and LP via JVC L3-E. The presentation almost identical. The JVC is like a broadcast device in this ( derived from TT 71 ). The Audio Technica PU if that's what it is can be highly accurate ( AT 95 had the lowest distortion figures I have seen regardless of price, Linn K18 is a posh version ). The CD was supurb and yet the LP was better. The LP was as new. It was hard to think it wasn't the mastertape as the JVC is that good. This is bonkers as the CD should be closer. One engineer said the stupid conversion to 44.1 kHz might be to blame. It was said the owners of copywrite don't want direct access to the mastertape. The specific differnece was depth around singers to other things in that space. CD was cardboard cut out sound. The LP was if you like 16 bit and the CD somehow MP3. Is it possible CD is this bad? MP3 I don't mind as it is an honest way to do a good job in limited file space. MP3 is good tap water if you like. In tonality slightly preferable to FM. MP3 will record the way 78's sound. That's not bad. I have a few FM tuners so speak for all when saying the medium of sound is a bit coloured. Better than cassette, less good than a Revox.
- Status
- Not open for further replies.