john:
i've used the "jfet follower behind a nice volume control" also and been pretty happy with it for the situations where i did not need voltage gain. i even tried a balanced version (needed 4 followers - painfully matched) with a pot as shunt volume control, kind of like figure 3 here from borbely. Sounded pretty to me.
thanks for the background and please don't stop now ...
hmmm ...
only if it's got jfets
mlloyd1
i've used the "jfet follower behind a nice volume control" also and been pretty happy with it for the situations where i did not need voltage gain. i even tried a balanced version (needed 4 followers - painfully matched) with a pot as shunt volume control, kind of like figure 3 here from borbely. Sounded pretty to me.
thanks for the background and please don't stop now ...
hmmm ...
only if it's got jfets
mlloyd1
The folded cascode not only creates gain--which isn't really necessary in the Blowtorch, as the differentials can easily manage the modest levels required by themselves--but it level-shifts the signal from the opposing ends of the complementary differential back towards ground. A follower won't do that by more than the Vgs (assuming MOSFETs) of the follower devices.
There are ways around this, however. One is to simply use a resistive divider from one end of the differential to the other and tap it in the middle. The downside is that you've raised the Zout of the circuit and you won't end up with much in the way of drive current. As a variation, you could use resistors to bias the followers and take the signal from a tap at the center, but you're increasing the Zout again. Another possibility is to capacitively couple the differentials to the followers, but then you're inserting capacitors into the signal path. This isn't a fatal flaw, after all scads of tube designs manage to sound pretty damned good in spite of multiple coupling caps in their innards, but it does violence to John's original concept. Yet another possibility is to use low enough rails that the DC offsets from the front end just match the Vgs of the output MOSFETs, thus managing to have everything at once...well, as long as you weren't expecting much in the way of voltage swing (an objection that holds true for all of these ideas, for that matter). Might work for the initial stage of a moving coil phono circuit, although the distortion and noise would be higher than necessary. Of course, there's always adding a follower output stage, but that's cheating.
Or...
You can go ahead and use John's folded cascode as-is and use feedback to lower the Zout. See the JC-3 schematic for an example of this strategy. But the simplest way to lower the Zout is to drive the ever-living pee-widdle-dookey out of the folded cascode, then load the output with the resistor value of your choice. This has the added benefit of ground-referencing the output, which somewhat reduces DC drift.
I had another idea that I was going to toss out, but somebody asked me something and it flew out the window. Oh, well.
Grey
P.S.: I've got an idea involving a floating follower that I've never actually tried to build, since it involves adding a stage and I am reluctant to do so unless absolutely necessary. Might get around to trying it as a front end for the New-Tron someday.
There are ways around this, however. One is to simply use a resistive divider from one end of the differential to the other and tap it in the middle. The downside is that you've raised the Zout of the circuit and you won't end up with much in the way of drive current. As a variation, you could use resistors to bias the followers and take the signal from a tap at the center, but you're increasing the Zout again. Another possibility is to capacitively couple the differentials to the followers, but then you're inserting capacitors into the signal path. This isn't a fatal flaw, after all scads of tube designs manage to sound pretty damned good in spite of multiple coupling caps in their innards, but it does violence to John's original concept. Yet another possibility is to use low enough rails that the DC offsets from the front end just match the Vgs of the output MOSFETs, thus managing to have everything at once...well, as long as you weren't expecting much in the way of voltage swing (an objection that holds true for all of these ideas, for that matter). Might work for the initial stage of a moving coil phono circuit, although the distortion and noise would be higher than necessary. Of course, there's always adding a follower output stage, but that's cheating.
Or...
You can go ahead and use John's folded cascode as-is and use feedback to lower the Zout. See the JC-3 schematic for an example of this strategy. But the simplest way to lower the Zout is to drive the ever-living pee-widdle-dookey out of the folded cascode, then load the output with the resistor value of your choice. This has the added benefit of ground-referencing the output, which somewhat reduces DC drift.
I had another idea that I was going to toss out, but somebody asked me something and it flew out the window. Oh, well.
Grey
P.S.: I've got an idea involving a floating follower that I've never actually tried to build, since it involves adding a stage and I am reluctant to do so unless absolutely necessary. Might get around to trying it as a front end for the New-Tron someday.
mlloyd1 said:
One and only Mr Lloyd,
i suppose tempting you to the Dark side of a HifiZen Blowtorch line-amp by posting a source for 75$ cent Toshiba A1349/C3381 single substrate matched dual bjt's is out as well ?
Here Kitty
Piccy, are we ?
Attachments
bogdan_borko said:
It's still in my head--not drawn up. I have thus far resisted the temptation to post schematics of things I haven't tried yet. Don't intend to succumb today. It may or may not work, and enough people here think I'm nuts (Egad, Margaret, the boy listens to his circuits! I just don't know what this world is coming to...) without further annoying them by posting potentially unworkable schematics.
The DC offset thing is really only half the problem. The other part is the limited voltage swing. This one, fortunately, has a relatively simple solution--just raise the rails to an arbitrarily high level in order to make the swing across the front end load resistors match your intended output swing. The problem here is that any of the JFETs you might want to use have too low a voltage rating (although you could cascode). MOSFETs and bipolars can do the job easily...as can tubes, but you're only exacerbating the DC offset problem.
Caps work, of course. I threw together a circuit similar to the one in the "Pumpkin" thread a couple of years ago, except that mine used a follower output. Sounded pretty good. One of the "tricks" that I used was that I drove the follower hard (the cap was on the output). People drive caps with dinky currents, then wonder why the don't sound good. Harrumph. Push 'em harder and a lot of the complaints dry up and blow away in the wind. However, I'll be the first to admit that it's better to have no cap at all; I was just using the cap in order to keep from having to level shift the output signal. Worked just fine--well enough that I've kept that circuit together to this day instead of cannibalizing it for parts.
The question is: Can it sound better?
Yep.
Grey
I might point out something that might not be obvious in the 'Blowtorch'.
The point to point wiring that looks so 'direct' rather than neatly bundled, is a real advantage that we have with a totally enclosed shielded environment, without any power AC in the box. We don't have to shield, which lowers the capacitance of the wire, as well as hang it in the air, so that we don't have much dielectric absorption to worry about, we don't have much internal air circulation, so we don't get much outside contamination from pollutants in the air.
The only thing that we have to watch out for is crosstalk from various inputs and keep the left and right channels separated from each other.
Most preamps just use circuit boards to route the switching. I do it both ways, and the Blowtorch is the best way, even if it looks a lot like a bunch of 'clothes lines' in the box.
The point to point wiring that looks so 'direct' rather than neatly bundled, is a real advantage that we have with a totally enclosed shielded environment, without any power AC in the box. We don't have to shield, which lowers the capacitance of the wire, as well as hang it in the air, so that we don't have much dielectric absorption to worry about, we don't have much internal air circulation, so we don't get much outside contamination from pollutants in the air.
The only thing that we have to watch out for is crosstalk from various inputs and keep the left and right channels separated from each other.
Most preamps just use circuit boards to route the switching. I do it both ways, and the Blowtorch is the best way, even if it looks a lot like a bunch of 'clothes lines' in the box.
You shield only against magnetic component of the field. Electric HF field gets along wires into the AL box, because the connectors are insulated on the panel. Only in case you had connectors body electrically connected to panel (like BNC, SMA ...) and used shilelded cables, then you would prevent from HF electric field component. In case of insulated connectors you get all the electric field inside.
These are very basics of shielding and do not want for me to tell you for how many years I have designed instruments for High Voltage laboratories and power testing plants.
These are very basics of shielding and do not want for me to tell you for how many years I have designed instruments for High Voltage laboratories and power testing plants.
PMA said:
And aluminum doesn't do a good job of magnetic shielding. From experience, most attempts at shielding low frequency, low impedance magnetic fields are pointless. It's like the little boy plugging the hole in the dike with the water all around him.
And, as described above, the electric field common knowledge doesn't quite paint a clear picture of the real world. RF doesn't really pay attention to anything less than a seamless encloser, mumetal is great until you bend or join it.
The best approach to interference is careful attention to the high impedance nodes in your circuit and the return paths.
Regards, Mike.
- Status
- Not open for further replies.