Page 1 of 1

New Phono Stage

PostPosted: Mon Sep 13, 2010 9:28 am
by Shannon Parks
Six years after building my first phono preamp, I finally listened to it this morning. Loved it! It started out as a 5687 version of the 6DJ8 passive designs (like Bottlehead and others). Obviously, I had miscalculated the losses in my filter and cut-n-jumpered my board to accommodate the 6DJ8, getting about 37dB of gain at 1kHz. The tricky bit was the power supply, which is a 12V power supply (not a wall wart, but a computer type switcher) and a National Semi LM3478 running at 220V and 400kHz. The noise on the supply was just too much, and I fiddled with it back then and then gave up. Just returned to it yesterday and actually listened to it. Not bad at all, really. I think some tweaking can get the supply fairly respectable. One other note: the four triode halves use the IXYS 10M45S as plate loaded CCS. Kinda cool.

Now to add a 2SK170BL to the first 6DJ8 in a cascode configuration ala the old Tek scopes and Allen Wright's FVP5A preamp. Probably will add two more 6DJ8s (paralleled) as the third stage with a switch to let this act as a linestage.

PostPosted: Mon Sep 13, 2010 3:56 pm
by Geek
Sounds great! [:)

How is the noise interference from the supply?


PostPosted: Tue Sep 14, 2010 6:22 am
by Shannon Parks
The RFI issues will probably be one of the last things I get to iron out with this prototype. I'm thinking this will be in a small metal box to protect the rest of system. I like the Pomona boxes (in particular: I have a few pulled from the NSA's trash and they are as solid as anything I've seen. The tubes would poke out, so hopefully shielding won't be necessary.

Right now tweaking the operating points on the 6DJ8 in the current RJM-configuration and tweaking the LM3478 settings. I may just leave it in this low gain RJM config and through it in the Pomona and get several hours listening time on it before proceeding much further.

Hey - hasn't anyone tested the 6DJ8 at different currents to see how much that affects its gm? Even a tube datasheet somewhere? My searches are turning up nothing. The IXYS parts let me run these at 10-12mA easy, but if 5mA does the job, why bother? Not worried about linearity at these levels.


PostPosted: Tue Sep 14, 2010 2:34 pm
by Geek
separks wrote:Hey - hasn't anyone tested the 6DJ8 at different currents to see how much that affects its gm?

I don't think officially, but 5mA drops the gm and increases noise in my tests.

The IXYS parts let me run these at 10-12mA easy, but if 5mA does the job, why bother? Not worried about linearity at these levels.

When I was (well, still am) LED biasing them, there's a 12-15mA low noise, high gm "sweet spot" with between 90-105V on the anode when I was making my passive phono stage.


PostPosted: Wed Sep 15, 2010 6:29 am
by Shannon Parks
Thanks, Gregg! Those were the numbers I was thinking. I may experiment with LED biasing way down the road, but my tests last time seemed to just lower the gain a touch and have less predictable frequency response (than bypassed resistors). But I wasn't using the HP LEDs et al either (standard reds).


PostPosted: Wed Sep 15, 2010 2:54 pm
by Geek
Hi Shannon,

Yeah, those cheapie little red LED's as Morgan Jones specifies, really do work best ;)

There's a trick to keeping LED's stable, consistent, low impedance and linear, even at low current levels.... run a resistor from the LED anode/tube cathode to some B+ that'll run the LED at 10mA+

Look at the LED as a constant-voltage device... the current through it in this method will not affect the current through the tube, because the cathode of the tube will just see say, 1.85V, regardless of what wieredness we're using to increase the LED bias :))


PostPosted: Fri Sep 17, 2010 6:18 am
by Shannon Parks

PostPosted: Fri Sep 17, 2010 2:46 pm
by EWBrown
I've used "cheap red LEDs" as well as yellow ones as cathode resistors, the color isn't important, as is the actual voltage drop.

the small (1.9 mm dome) red HLMP-600s (available from Jameco and others) is the perennial Doc Bottlehead favorite for cathode biasing 12AU7s, 6SN7s, 12AT7s and 6DJ8s, As hese particular LEDs have a very consistent forward voltage drop, generally 1.56 to 1.57 volts, and those in the same manufacturing lot are generally "matched" to better than 0.2%. They will newer win any awards for high luminosity, but for cathode biasung, brightness is of no real concern.

I've used yellow or amber LEDs when I need 1.8 to 2VDC on the cathode. High-brightness yellow is OK for this use, and I've also observed very consistent voltage drops across these. In a 12AT7 driver stage, with 6 mA plate CCS (10M45s) and some bright yellow LEDs I got off the internet , from All of those which I tried in this circuit have a very consistent 1.873VDC Vf drop.

The older "lime green" LEDs are also good, with slightly higher 2.0 to 2.2V forward drop.

Using a resistor from B+ to the LED anode is a good suggestion, if the tube's cathode current is on the low side, say under 3 mA, or so. I'd use this additional resistor to add another mA or 2 through the LED, and 10 mA would be mostly just wasting B+ current,

The older types of low-brightness, low-efficiency type of LEDS are the best for this application , and the newer "superbright" "true green", blue, white, UV, etc have a different chemistry and are no nearly t as good, as their operating impedance tends to be higher, and they can generate more noise. Save them for using as pilot lights or visual "bling" ;) :))

I haven't yet tried them yet, but IR LEDs have a much lower forward voltage drop than the color ones, typically , between 0.9 and 1.2VDC. These should work well for those applications requiring lower cathode voltages.

/ed B