Any ideas as to why a zener ref CCS might sound more dynamic?
The tail of a differential pair is supposed to present a high dynamic impedance.
What might be the diff if any between a 0.5 watt zener and a 1 watt one?
I am wondering whether the perceived quality is related to the amount of silicon or the characteristics of a zener diode under dynamic conditions.
I remember building a phono pre-amp published in Wireless World in 1977 - I can't remember the author Eric .. someone? The CCS was referenced by a zener (5.6 volts? by coincidence).From memory this allowed a input option where the MM cartridge was in series with the feedback connection.
On a slightly different tack a power amplifier I built from ETI magazine (Australian edition) had a CCS in the tail of the diff input stage.
The reference was a transistor but there were two feed transistors with unequal resistors in the emitter connection to the supply rail (100R and 82R). It drew a lot of current. This strategy was said to optmise the currents in the halves of the diff pair.
Michael J.
My GUESS is that transistors prefer a slightly higher voltage than available from say 2 diodes, or 1 LED, etc.
In fully complementary input stages, the lower the CCS transistor's voltage, the more pronounced the differences between NPN and PNP, but vice versa also holds true. By about 5V6, this difference becomes very small, and by 8V2, it's practically gone, being relegated to the second decimal pace.
Honestly - I don't know. I tend to overdo things, so I rarely use less than the ZY series, which are 1.3W. They cost just a wee bit more than the ZPD and similar 0.5W diodes, so why the hell not?
I have seen this a few times. H/K 680 has the same in case of the current mirror for the VAS, 82 and 100 Ohms.
I referenced that to Bob Cordell's statement that current mirros did make the VOLTAGES more even, but could display some instability at bias current level, and to solve that, we needed what he called a "helper transistor".
Since that means more parts, even if only two, 1 transistor and 1 resistor, the industry will always go for the cheapest solution, and using 82 Ohms does in fact improve the offset somewhat.
Has anyone tried a very long tail on a long tail pair ? If could be zener protected for start up . Put it up at 300 V . Only problem I can see is hiss . 150 K if 2 mA . Without doing the maths I guess the usual CCS to be superior ? The zener would be miles below operating in usual running .
Trouble is, I was referring to the sound of the music...
There's ordinary competence, and then there's the quality that live conveys, a sparkle, a sense of real people creating what you're hearing. One astute company had 3 separate setups in 3 rooms, and every one of them had that extra quality, that made me perk up when I heard them.
It's only months later that I discover that the amps used, none of which I had ever heard of before, are recent ones that are getting top level praise around the traps ...
I think boring or bland is wrong . Real music isn't . I have heard hi fi with as near zero distortion as its worth discussing that is far from boring or bland . Best example would be everything Quad made after 1967 . 606 was not bland , the others were . 303 being better than 405 to my ears . As far as Quad were concerned all were the same if played at a real for classical music level .
Point taken about the zener DF 96 . It would need something ( gas tube or neon regulator ) . How bad is the capacitance when not activated as I intended . I have just measured a 47 V 1.3 W zener on a 9 V tester at 139 pF ( 111 pF the wrong way ! ) .
Point taken about the zener DF 96 . It would need something ( gas tube or neon regulator ) . How bad is the capacitance when not activated as I intended . I have just measured a 47 V 1.3 W zener on a 9 V tester at 139 pF ( 111 pF the wrong way ! ) .
Here's the three types of CCSs I have been using for the last 20+ years.
Frankly, I never could detect any sonic difference between B.) and C.), but Perhaps I never made an amp good enough to convey any possible differences. It's not like I built hundreds of them, less than 10 over the last 20 years.
I'm not saying these are the best there are, but I am saying that I reckon they are very good for the money.
Frankly, I never could detect any sonic difference between B.) and C.), but Perhaps I never made an amp good enough to convey any possible differences. It's not like I built hundreds of them, less than 10 over the last 20 years.
I'm not saying these are the best there are, but I am saying that I reckon they are very good for the money.
Will do it in the morning . Basically a long tail pair circa 1935 had the cathodes of two valves joined together and the connection between them would be the tail ( Alan Blunlein and others ) . The concept was to take the tail to infinity . Thus the tail would see an infinite resistance . The higher the resistance the longer the tail ? Often in valve circuits the tail was very short . With transistors we can have artificially long tails using CCS . My idea is make it long in reality .
With due respect to precise technical definitions as per Amos' Electronics Dictionary, if the general discussion has been in relation to the input stages of a transistor amplifier then a Differential Amplifier will mean a long-tail pair arrangement with a single output.
In valve(vacuum-tube) days where the term long tail pair originated, this would mean a phase splitter with two outputs working in push-pull.
You would find this for example V2 in the second stage of the Leak TL12 amplifier - Schematic - Leak TL-12 Plus Tube Amplifier @ AmpsLab.com
Not withstanding that you have used the term long-tail pair in it's correct context I take it you are theorising over an input stage involving a Differential Amplifier.
Michael J
I know of no amplifier better than the Hitachi to show some variations . I didn't spend much time on this and even less thought . I choose 430 V as being cheaply available using a valve HT transformer . For fun did the maths for supplying the VAS . Now that's as long a resistor tail as we might get .
The other example is showing that there is no great voltage swing so equally possible .
Harley Davidson I suggest would never have been made if designed by simulator . How say you about Long Tail Pairs ? Not interested in right or wrong , just why .
Last edited:
With all due respect, Nige, what I see on your illustration is what I have seen countless times, and what I, all by myself, own in two versions, once from HK 680, and once from Toshiba SB-45, both integrated amps. The HK was relatively expensive, the Toshiba was cheap.
In a highly modified mode, I have a third sample, HK 6550, but to even recognize it as a later evolution of the basic circuit, you have to know them all and have some imagination.
But the first two are immediate hits everyone can easily recognize.
All that said, I would now need to be told what is it that makes their tails so long, because as far as I can see, in all cases a differential amp is driving another differential amp, in an attempt to make a single ended input differential amp appear more symmetrical to the output stage.
In all cases, i am very znhappy with the input stage CCS, batteries and all. I have for years maintained - and still do - that instead of fancy input circuits and so forth, the single most beneficial thing you can do for your amp is to separate the power supplies for the VAS from those of the output stage, run them at somewhat higher voltages to compensate for drops across transistors, and allow your current stage to work at slightly lower voltages, so as to move more to the left of the SOAR curve of the output devices.
Not even to mention the fact that full electronic voltage regulation (assuming it's properly done) gets rid of much line noise before it ever reaches the circuits.
In a highly modified mode, I have a third sample, HK 6550, but to even recognize it as a later evolution of the basic circuit, you have to know them all and have some imagination.
But the first two are immediate hits everyone can easily recognize.
All that said, I would now need to be told what is it that makes their tails so long, because as far as I can see, in all cases a differential amp is driving another differential amp, in an attempt to make a single ended input differential amp appear more symmetrical to the output stage.
In all cases, i am very znhappy with the input stage CCS, batteries and all. I have for years maintained - and still do - that instead of fancy input circuits and so forth, the single most beneficial thing you can do for your amp is to separate the power supplies for the VAS from those of the output stage, run them at somewhat higher voltages to compensate for drops across transistors, and allow your current stage to work at slightly lower voltages, so as to move more to the left of the SOAR curve of the output devices.
Not even to mention the fact that full electronic voltage regulation (assuming it's properly done) gets rid of much line noise before it ever reaches the circuits.
Spot on Brad, that IS a serious concern.
Now that you mentioned it, how would you go about it, in principle?
I haven't seen a single "best" approach. Schottky clamping is impractical because those diodes are generally too-low reverse breakdown voltage. Having diif amps sensing each collector-emitter voltage, comparing that to a reference, and rapidly limiting base drive is cumbersome. Sensing rails and clamping input signals is difficult to do without introducing other distortions below clipping levels.
This storage time problem with bipolars is one of the reasons DMOS appeals, despite other suboptimal attributes depending on configurations.
The problem of keeping a bipolar transistor out of saturation in digital logic was solved with a Shottky diode as used in Shottky TTL:
Schottky transistor - Wikipedia, the free encyclopedia
But power transistors tend to have higher saturation voltage, so this won't work. Surely there's something not much more complicated that will. Following a link from the above article, maybe the Baker Clamp could be used - if this isn't enough to keep it out of saturation, make D1 a Shottky and/or add diodes in series with D2 until it is:
Baker clamp - Wikipedia, the free encyclopedia
Schottky transistor - Wikipedia, the free encyclopedia
But power transistors tend to have higher saturation voltage, so this won't work. Surely there's something not much more complicated that will. Following a link from the above article, maybe the Baker Clamp could be used - if this isn't enough to keep it out of saturation, make D1 a Shottky and/or add diodes in series with D2 until it is:
Baker clamp - Wikipedia, the free encyclopedia
- Status
- Not open for further replies.