Sansui or somebody had a patent on that back in the '70s, I think. No, that's not why I didn't use it; the patent's long since expired. The idea's been around for a while and there are folks who like the idea, claiming that it reduces distortion. Others feel that it increases distortion, though I forget the line of reasoning. In the end, I didn't use it because it constitutes a form of feedback.
That's a modification you could make pretty easily and it won't cause any grief that I can see as far as the rest of the circuit goes. If it makes you happy, then give it a shot.
Grey
That's a modification you could make pretty easily and it won't cause any grief that I can see as far as the rest of the circuit goes. If it makes you happy, then give it a shot.
Grey
maney said:
"THEORY"
Posts: 3,8,13,15,19,26,31,35,38,41,45,48,50,54,56,58,71,82,88,89,99.
the·o·ry Audio Help /ˈθiəri, ˈθɪəri/ Pronunciation Key - Show Spelled Pronunciation[thee-uh-ree, theer-ee] Pronunciation Key - Show IPA Pronunciation
–noun, plural -ries. 1. a coherent group of general propositions used as principles of explanation for a class of phenomena: Einstein's theory of relativity.
2. a proposed explanation whose status is still conjectural, in contrast to well-established propositions that are regarded as reporting matters of actual fact.
3. Mathematics. a body of principles, theorems, or the like, belonging to one subject: number theory.
4. the branch of a science or art that deals with its principles or methods, as distinguished from its practice: music theory.
5. a particular conception or view of something to be done or of the method of doing it; a system of rules or principles.
6. contemplation or speculation.
7. guess or conjecture.
Bruce
Is there a question or an observation in there.BC said:
"THEORY"
Posts: 3,8,13,15,19,26,31,35,38,41,45,48,50,54,56,58,71,82,88,89,99.
the·o·ry Audio Help /ˈθiəri, ˈθɪəri/ Pronunciation Key - Show Spelled Pronunciation[thee-uh-ree, theer-ee] Pronunciation Key - Show IPA Pronunciation
–noun, plural -ries. 1. a coherent group of general propositions used as principles of explanation for a class of phenomena: Einstein's theory of relativity.
2. a proposed explanation whose status is still conjectural, in contrast to well-established propositions that are regarded as reporting matters of actual fact.
3. Mathematics. a body of principles, theorems, or the like, belonging to one subject: number theory.
4. the branch of a science or art that deals with its principles or methods, as distinguished from its practice: music theory.
5. a particular conception or view of something to be done or of the method of doing it; a system of rules or principles.
6. contemplation or speculation.
7. guess or conjecture.
Bruce
Or simply a comment that I can't fathom.
I must get more sleep.
AndrewT said:
Hi AndrewT,
Observations. It appeared to me that someone was wondering were there was any Theory behind Gray's design. I apologize if I miss understood. I thought Grey has been very thorough with his explanations of design.
Another possible misunderstanding on my part:
Grey-
There are 4 outputs. Which one(s) for the + input of the transducer, and which one(s) for the - input of the transducer.
Bruce
I'm here...just the usual kids and such...add to that a quick trip up to the mountains to check on my land. Hadn't been up there in ages and, amongst other things, there was a severe wind storm--needed to check on damage to trees, etc. But I'm back, and...
Am I to understand that by transducer you mean the output stage?
Conceptually, the front end plugs directly into the output stage. Note that the front end is drawn fully balanced, whereas I only show one output bank, that being sufficient for a single-ended 25W amp. If you build two output banks and hook one to each side, you can run the circuit as a fully balanced 100W class A circuit.
The top drive of the front end (either side...your choice as to whether you want an inverting or non-inverting amp) goes to the top of the bias circuit and the bottom drive goes to the bottom of the bias circuit--just exactly as it would look if you overlapped the two schematics and matched up the lines. (I may or may not have drawn the schematics so that the lines actually match, but you can tag them together in your mind.)
I built my front end fully fleshed out for two reasons:
--I intend to do a 100W amp
--The current draw for the fully balanced front end sums to DC, which is a good and marvelous thing in that the power supply drops out of the equation if all it has to supply is DC. It's when you ask a power supply to deal with AC that things get ugly.
Grey
P.S.: There are lots of people who feel that Science=Theory=Mathematical Formulas. If you don't supply the formulas, they don't feel that it's proper science.
I've even heard Bible-thumpers around these parts argue (apparently seriously) that since there's no math in Darwin's The Origin of Species, then it can't be science, hence evolution didn't happen, hence the world is only 6000 years old. Meanwhile the rest of the world has entered the 21st century...
I'm being called away--this post may be a bit ragged...
Am I to understand that by transducer you mean the output stage?
Conceptually, the front end plugs directly into the output stage. Note that the front end is drawn fully balanced, whereas I only show one output bank, that being sufficient for a single-ended 25W amp. If you build two output banks and hook one to each side, you can run the circuit as a fully balanced 100W class A circuit.
The top drive of the front end (either side...your choice as to whether you want an inverting or non-inverting amp) goes to the top of the bias circuit and the bottom drive goes to the bottom of the bias circuit--just exactly as it would look if you overlapped the two schematics and matched up the lines. (I may or may not have drawn the schematics so that the lines actually match, but you can tag them together in your mind.)
I built my front end fully fleshed out for two reasons:
--I intend to do a 100W amp
--The current draw for the fully balanced front end sums to DC, which is a good and marvelous thing in that the power supply drops out of the equation if all it has to supply is DC. It's when you ask a power supply to deal with AC that things get ugly.
Grey
P.S.: There are lots of people who feel that Science=Theory=Mathematical Formulas. If you don't supply the formulas, they don't feel that it's proper science.
I've even heard Bible-thumpers around these parts argue (apparently seriously) that since there's no math in Darwin's The Origin of Species, then it can't be science, hence evolution didn't happen, hence the world is only 6000 years old. Meanwhile the rest of the world has entered the 21st century...
I'm being called away--this post may be a bit ragged...
maney said:
I've asked John to name it--surely his privilege, since as far as I know he was the first to come up with it--he has yet to do so.
In principle, you could run the "Curlode" (okay, that looks a little ungainly. I'll keep working on it.) harder if you wanted to. I don't know that I mentioned that earlier as an avenue of exploration for those who want lots of negative feedback. If you increase the value of the load resistor that drives the Source, you'll get a bigger signal there. That will in turn give you more output. Of course, you'll also have to increase the value of the resistor and/or pot that provide the signal for the Gate to get it to bias properly, which will also produce more gain...
You'll reach a stopping point when the front end load is sitting half-way from the rail to the Drain of the front end cascode at zero signal, but since that's what all the textbooks recommend for maximum voltage swing anyway, it'll probably suit folks seeking NFB just fine.
Grey
BC said:Another possible misunderstanding on my part:
Grey-
There are 4 outputs. Which one(s) for the + input of the transducer, and which one(s) for the - input of the transducer.
Well, no reply..... hmmmmmm.... sorry
Thanks Grey for your kind reply.
This was a misunderstanding.
1. RIF (Reading Is Fundamental)
2. The use of the term transducer was completely incorrect.
BruceGRollins said:
One of the principles on which Cray's original number cruncher was based, in fact. If you think it helps at audio, imagine the effect when you have many thousands of fussy, sensitive gates switching at tens of MHz (I don't recall exactly, but I doubt that early ECL went much past 50MHz).
Of course they're confused - without malleable formulas it's not engineering, but of course science is the leading process for getting from not understanding to engineering.
There's no good way to do equations in this medium, is there? Ignoring nonlinearity, the net output for one half (the top or bottom), viz., i1-i2 is the product of two terms (I'm going to try this without the markup, it's too time consuming):
(gm1 + gm2)(R1 + R2)i
where i is the drive to the VAS, and so the drive to the source is -i. Aside from the sum of the transconductances, this is just what you'd expect if R1 and R1 were merely in series, rather than having the source connected to their junction. The other term is
1 / (1 + (gm1 + gm2)R2)
Which is, again, just what you'd expect from a common source stage with R2 the source resistor and without any interesting connection - except that the sensitivity is doubled by the presence, again, of the sum of the transconductances. For R2 considerably larger than 1/gm, as is the case here, I believe (100mA bias in the VAS?), increasing R2 will decrease the gain - the degeneration term will shrink almost inverse lineraly, but since R1 >> R2 to begin with, the gain term will rise only slowly. I don't think you do in fact need to increase R1 when increasing R2 (to keep the same current) as that causes no decrease in the gate-source bias; so likewise you can't increase R1 to make up for the loss from a larger R2 and still have the same VAS bias.
Martin "I sure hope I used the same R1 R2 naming in the earlier one" Maney
Having a few minutes this morning, I started to make a cleaned up copy of my derivation (also to sort out which page of chicken scratches was the correct one - there was at least one false start, and they got thrown into one heap)... and I found I'd omitted one term at the start of that derivation. Luckily it doesn't make a very large difference - the above should have been:
(gm1 + gm2)(R1 + 2 * R2)i
The missing R2 * i is exactly the semi-cascode's contribution, so it's embarassing to have left it out, but at least for the values in the GR25 it has little effect.
[added in edit] I just went back to the actual GR25 schematic, and I see I'd forgotten about the non-R2 source degeneration resistors which haven't been in the (over?) simplified circuit I was working with. With those the relative contribution of the semi-cascode would be increased because they would decrease the effective gm; OTOH they also allow a larger R1 for a given DC bias, so I'm not sure what the net effect in this specific circuit would be.
There's not really a way to make a "simplified" version of the bias circuit. If you take anything out it just quits working. Given that there are only two transistors and three resistances, it's not all that complicated.
Using the parts nomenclature from the output stage schematic, the bias circuit is comprised of R1, R2, V1, R3, Q1, and Q2.
The trick to a Vbe multiplier is that it's based on the voltage between the base and emitter of the transistor. You'll see different approximations of Vbe depending on what book you're looking in; anywhere from .5 to .7V, most frequently in the range of .6 to .65V. The idea that there is always a semi-predictable voltage between the base and emitter is inherent in a bipolar transistor's nature. No, there isn't anything quite like it for tubes or FETs. They have voltages that cause things to happen once they're applied to the grid or Gate, but they aren't locked in the way Vbe is for a bipolar. Incidentally, Vbe isn't really as constant as you might wish. It does vary somewhat with the current through the device and the temperature, and both of those effects come into play in the GR-25 because the driver stage is run hard and the amp as a whole runs hot.
The bias circuit in the GR-25, like most of the topology, came from the JC-3. The basis of the circuit is two "normal" Vbe multipliers, back to back. The Vbe from Q1 and Q2 add together, creating (in the case of the MJE15030 and MJE15031) an aggregate voltage of something like 1.15V. That voltage, when applied across R2 and V1 results in a current which, once defined, also flows through R1 and R3, setting the voltage from the collector of Q1 to the Q2, hence the output stage bias.
The 500 Ohm resistors in the driver stage keep that voltage from collapsing, which is why there are two outputs from the front end. See? It all ties together. (Sooner or later.)
As you turn V1, you're adjusting the resistance between the bases of Q1 and Q2. The Vbe of the two transistors remains nearly constant so what you're really adjusting--at least as far as the bias circuit is concerned--is the current through R1 and R3. R2 is there simply to tighten up the range of the bias pot, V1. It's the sum of R2 and V1 that does the trick.
There's nothing particularly odd about the output stage. It's a conventional follower output. Once you come to terms with the fact that it uses MOSFETs rather than bipolars, and arrange the bias circuit accordingly, it's on to the races.
Grey
Using the parts nomenclature from the output stage schematic, the bias circuit is comprised of R1, R2, V1, R3, Q1, and Q2.
The trick to a Vbe multiplier is that it's based on the voltage between the base and emitter of the transistor. You'll see different approximations of Vbe depending on what book you're looking in; anywhere from .5 to .7V, most frequently in the range of .6 to .65V. The idea that there is always a semi-predictable voltage between the base and emitter is inherent in a bipolar transistor's nature. No, there isn't anything quite like it for tubes or FETs. They have voltages that cause things to happen once they're applied to the grid or Gate, but they aren't locked in the way Vbe is for a bipolar. Incidentally, Vbe isn't really as constant as you might wish. It does vary somewhat with the current through the device and the temperature, and both of those effects come into play in the GR-25 because the driver stage is run hard and the amp as a whole runs hot.
The bias circuit in the GR-25, like most of the topology, came from the JC-3. The basis of the circuit is two "normal" Vbe multipliers, back to back. The Vbe from Q1 and Q2 add together, creating (in the case of the MJE15030 and MJE15031) an aggregate voltage of something like 1.15V. That voltage, when applied across R2 and V1 results in a current which, once defined, also flows through R1 and R3, setting the voltage from the collector of Q1 to the Q2, hence the output stage bias.
The 500 Ohm resistors in the driver stage keep that voltage from collapsing, which is why there are two outputs from the front end. See? It all ties together. (Sooner or later.)
As you turn V1, you're adjusting the resistance between the bases of Q1 and Q2. The Vbe of the two transistors remains nearly constant so what you're really adjusting--at least as far as the bias circuit is concerned--is the current through R1 and R3. R2 is there simply to tighten up the range of the bias pot, V1. It's the sum of R2 and V1 that does the trick.
There's nothing particularly odd about the output stage. It's a conventional follower output. Once you come to terms with the fact that it uses MOSFETs rather than bipolars, and arrange the bias circuit accordingly, it's on to the races.
Grey
lineup said:hmmm
GR-25 by GRollins
I'd better do some sims of this circuit
.. to see what's the major drawbacks/shortcomings.
To get to know the good things I need only to read all posts here.
thanks
Lineup
Well, gee, that's a real toughie...
Amplifiers without feedback tend to have somewhat higher distortion and lower damping factor. Both of those have already been mentioned. The (potential) DC offset won't show up in a simulation (I will refrain from commenting), but that's been covered, too.
If you think there's something hidden, then it's hiding very well. But then, you never know about that Grey guy, he's pretty sneaky.
Grey
I believe I had mentioned earlier the idea of building a 100W version. Given that the water-cooled idea bubbled to the surface again (ahem) recently and I remembered (finally) to place some pictures in my old thread about same, it started me thinking about the whole water-cooled thing. If I've got enough doo-dads in stock, I make start building one in the near future. After all, winter is coming and I might as well make the heat serve two purposes at once.
Or I may delay long enough to try out a bias idea I've been wanting to take a whack at.
Grey
Or I may delay long enough to try out a bias idea I've been wanting to take a whack at.
Grey
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.