The inputs are a balanced switched attenuator followed by balanced cathode followers. There is 22k series resistance in each leg of the attenuator with the adjustable element 4k7 at maximum attenuation. Each leg is referenced to ground via the usual 1 Meg grid leak
Usually the value of a differential system is that noise appearing identically on both legs is subtracted out by the receiver. Using the Concertina as a driver defeats this in two ways.
First, the Concertina’s power supply noise is present virtually unattenuated at the plate, but is attenuated by more than the triode’s mu at the cathode. This means that the differential receiver cannot cancel it out.
Second, stray magnetic fields (power lines, etc.) can induce voltages along signal lines. If the lines are routed similarly, and the line impedances are identical, the induced voltages will be similar and will be mostly canceled out by the differential receiver. But you can prove to yourself that this won’t happen with the Concertina by inserting identical high impedance signals into both legs. The different signals occur because the plate and cathode impedances are quite different.
I think that there are simple and better alternatives, but we discuss those, let’s see if you agree with what I wrote above.
First, the Concertina’s power supply noise is present virtually unattenuated at the plate, but is attenuated by more than the triode’s mu at the cathode. This means that the differential receiver cannot cancel it out.
Second, stray magnetic fields (power lines, etc.) can induce voltages along signal lines. If the lines are routed similarly, and the line impedances are identical, the induced voltages will be similar and will be mostly canceled out by the differential receiver. But you can prove to yourself that this won’t happen with the Concertina by inserting identical high impedance signals into both legs. The different signals occur because the plate and cathode impedances are quite different.
I think that there are simple and better alternatives, but we discuss those, let’s see if you agree with what I wrote above.
Naturally, I agree with all this. I had considered it as an after the fact modification for CD input only-98% of my listening is LP/78's which my main system is totally dedicated to, and as I said, fully differential from cartridge to output. Space is tight, so I was attempting to keep the component count low and additionally, I do not need any significant gain so I was trying to avoid another 2 differentials pairs with their usual CCS's. It may be a better solution to build a separate amplifier for the CD as was my original intention.
I guess it depends on how far you want to go. I'd be a little concerned about the moderate amount of distortion that any triode I know would present when its P and K were loaded with a resistance as low as 1.5K and it were driven to line levels. But if you're OK with that and you're not concerned with power supply noise or stray magnetic fields, go for it!
Then we are in total agreement; 1k5 would be a stupidly low load feedback or nay; an input cap would be the better option using cathode bias and the available HT shared more or less equally. I am confident that I can get the HT sufficiently quiet, hum in the microvolt range. As for stray fields.........
Many thanks for your input!
Many thanks for your input!
Hi, hope this question is on topic,
what to do when constraints forced You to use sub optimal tubes for the function? I'm thinking specifically in a direct coupled concertina/input stage, made solely of *AX7 tubes. (HT limited to < 320V but very easy output tubes to drive)
what to do when constraints forced You to use sub optimal tubes for the function? I'm thinking specifically in a direct coupled concertina/input stage, made solely of *AX7 tubes. (HT limited to < 320V but very easy output tubes to drive)
What constraints are you dealing with? Can you be more specific about what you mean by "what to do"?
Hi, I mean you got these tubes, and can't change for some better suited ones. So You need to found the less worse solution, in the given context. But nevermind , I jump in a highly theoretical discussion with a tiny practical problem issue. "What to do" was meant to "what design objetives you can relax, to got the circuit still working, even if it will be far from perfect , given the wrong tubes"
The 12AX7 as cathodyne will be marginal, but for reasons Chris hinted at.
Let's put some approximate numbers to it. You'll want at least 120V on the plate of the first stage for linearity, assuming a +/-10V (or so) required swing, a common number for EL84 output stages. That means the cathodyne cathode will have to be at 122V or so. For a 0.75mA DC current, that works out to about 160k for the load resistor. With a 320V B+, a 160k plate resistor will drop the same 122V. That leaves the tube with 80V cathode-to-plate, which ain't great.
An alternative would be to put a voltage divider between the plate of the first stage and ground, with the cathodyne grid attached to the tap. The upper leg would be bypassed with a capacitor. You can see an example of this in the ImPasse article on my website (though with different tubes). Let's choose a 200k and 470k in series, with the 200k bypassed. That puts (470/{200+470})(120) = 84V on the cathodyne grid, so about 86V on the cathode. For 1mA current, plate and cathode resistors are 86k (82k is standard and close enough), dropping 172V. This leaves 148V cathode-to-plate, which is pretty good.
Let's put some approximate numbers to it. You'll want at least 120V on the plate of the first stage for linearity, assuming a +/-10V (or so) required swing, a common number for EL84 output stages. That means the cathodyne cathode will have to be at 122V or so. For a 0.75mA DC current, that works out to about 160k for the load resistor. With a 320V B+, a 160k plate resistor will drop the same 122V. That leaves the tube with 80V cathode-to-plate, which ain't great.
An alternative would be to put a voltage divider between the plate of the first stage and ground, with the cathodyne grid attached to the tap. The upper leg would be bypassed with a capacitor. You can see an example of this in the ImPasse article on my website (though with different tubes). Let's choose a 200k and 470k in series, with the 200k bypassed. That puts (470/{200+470})(120) = 84V on the cathodyne grid, so about 86V on the cathode. For 1mA current, plate and cathode resistors are 86k (82k is standard and close enough), dropping 172V. This leaves 148V cathode-to-plate, which is pretty good.
Hi SY, one of my many faults which you specifically may have noticed is that I never figured out how to post a "thumbnail sketch." I've been getting by with attachments, which are unwieldy. How do you post a "sketch" instead of an attachment, and maybe more importantly, where can I find a document which describes this?
Well, you can upload it. Or you can put it somewhere else and link it via the "image" tag (the little square dealie above the message box that has a little triangle and a blob, sort of like a mountain and sun).
Thanks SY. I'll work out around your suggestions. The voltage divider trick never crossed my mind, so many thanks.
Under the message box, scroll down to the "Manage Attachments" button, press it and you'll get a popup window. Browse to the file, upload, and then close the popup. Easy as pi.