Gear porn - cute little IPC tube preamps

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
> The chassis ground is almost always the same as the signal ground.

In this case it isn't.

Chassis=Signal ground works for small hi-fi systems, but in big studio systems will give you big ground loops.

See where all the cathode resistors go to. That is normally B-.
 
I would turn that chassis into your signal ground . Seems safer to me.
And install some 220k 2 watt bleeders on the caps, sepecially the one with the 22k resistors soldered to it.
 
Connecting the chassis and signal grounds together inside the preamp is a bad idea. The chassis and signal grounds are separate at the preamp for a good reason, as PRR pointed out. It's not a self-contained entity like a guitar amp.

Chassis and signal grounds should connect together, but only at one point in the entire system, usually back at the power supply.
 
B3Groover, pin 7 on the 5879 is the screen grid, and there should be a positive DC voltage there. But it wouldn't hurt to replace R3 and C2 just for laughs. If R3 has drifted high, your screen voltage may be low, which would affect the gain and distortion of the 5879 stage. As a matter of fact +40V sounds kind of low at the screen; the voltage there should be not much less than what you're seeing at the plate.
 
> +40V sounds kind of low at the screen; the voltage there should be not much less than what you're seeing at the plate.

Low screen voltage increases voltage gain, so is commonly done in low-level stages. 40V does not surprise me.
 
Good point. More electrons flying past the screen grid and ending up on the plate equals more gain.

I was thinking of power pentode practice, where it is usual to run the screen at a potential near the plate voltage. I don't mess around with small-signal pentodes much.
 
I had to play a gig last night, so I'm getting back to this now. Thanks again for everyone's continued help. I really appreciate it.

These are the new voltages I'm getting:

12AT7

pin 1 (plate) = 113
pin 2 (grid) = 0.46
pin 3 (cathode) = 1.18

pin 6 (plate) = 341
pin 7 (grid) = -.33 (fluctuating pretty wildly... and yes... it's negative... even when I reverse my leads)
pin 8 (cathode) = .008

5879

pin 1 (grid) = 0
pin 3 (cathode) = 1.4
pin 7 (grid) = 47
pin 8 (plate) = 65.8
pin 9 (grid) = 1.45


The negative voltage on pin 7 of the 12AT7 is strange, isn't it?
 
Ok, so I answered my own question. I wired the heaters back in series and found a 120->20vac transformer. The 12AT7 was getting about 12.9v and the 5879 was getting 6.5v and all of a sudden I had undistorted gain.

It's also noisy as hell which, I assume, is because the heater voltage is AC and it should be DC. I'm also picking up a radio station which I'm assuming is because the heater transformer is completely unshielded.

So... I think I'm getting somewhere! :)
 
> More electrons flying past the screen grid and ending up on the plate equals more gain.

That may be one way to look at it. Another way: in resistance-coupled, voltage gain goes up when you increase the the load resistance and reduce the plate current. Also the maximum transconductance happens when the signal grid is slightly negative. But at "normal screen voltage, these are 5mA-10mA tubes, and with 100K-500K plate resistor we can only use 0.25mA-1.0mA plate current. So we have to put G1 -WAY- negative to get the current down to something we can use with our high-value load resistor.

Put G1 at cathode voltage. Reduce screen voltage until the plate saturates around +50V. Then make G1 negative (or add a cathode self-bias resistor) until the plate is about 1/2 V+. That's best gain.

> I was thinking of power pentode practice, where it is usual to run the screen at a potential near the plate voltage.

Right. Here we want the cost and heater-power of a 100mA tube, and the power output of a 200mA tube. So we generally set the screen voltage up near maximum rating, and accept a lesser voltage gain. 6L6 needs about 40V of grid drive in typical power-amp use; we could raise load resistance to 100K and drop the screen voltage to like 25V, get about 3 times the voltage gain or only 13V drive. However we'd also get only 3 or 4 Watts of power, which is awful lame for a big pair of tubes and all that heater power.

pin 6 (plate) = 341
pin 7 (grid) = -.33 (fluctuating pretty wildly... and yes... it's negative... even when I reverse my leads)
pin 8 (cathode) = .008


All wrong. Look for shorted cathode-cap. Wobbly grid could be oscillations, leaky coupling cap, gassy tube. However it may clear-up when you figure out why the cathode isn't up at a couple volts.

> the 12AT7 wants 12v for the heater, it's all right to give it 6.3, right?

10V is about as low as I'd want to see on these tubes. 6V is likely to be WAY too cold, electrons won't boil-off, nearly no function at all. (As you've proven.)

> noisy as hell which, I assume, is because the heater voltage is AC and it should be DC.

Noise(hissss) or hummmmmm? Hiss isn't about AC heat.

This amp really should be low-hum even with AC heat, just a little better with DC.

> picking up a radio station which I'm assuming is because the heater transformer is completely unshielded.

You have grounding problems. Heater supply is unlikely to be where the radio is getting in. Are you sure you have the input and output windings wired right?
 
I'm not sure about the input/output windings being wired right. I have not changed them.

This is how the HA-100x (input) is currently wired:

Ground pin: going to signal ground, also pin 2 on the Cannon connector,
also chassis ground.

Pin 1 -- unused

Pin 2 -- jumpered to pin 3 on HA-100x
Pin 3 -- connected to signal input (pin 11 on the Cannon)

Pin 4 -- jumpered to pin 5 on HA-100x
Pin 5 -- connected to other signal input (pin 12 on the Cannon)

Pin 6 -- unused

Pin 7 -- going to pin 1 of the 5897 tube

Pin 8 -- jumpered to pin 9 on HA-100x
Pin 9 -- jumpered to pin 8 on HA-100x

Pin 10 -- negative B supply, also connected to pin 5 of the Cannon (at
least, that's how I have it hooked up)

The A25 is hooked up as follows:

Ground Pin - chassis ground

Pin P -- to pin 6 of the 12AT7 (plate) and through .25mf cap to gain pot

Pin B -- B+ supply (pin 1 of the Cannon connector)

Pin 1 -- signal output to pin 8 of Cannon connector

Pin 2 -- unused

Pin 3 -- jumpered to pin 4 of A25
Pin 4 -- jumpered to above

Pin 5 -- unused

Pin 6 -- signal output to pin 10 of Cannon


So far the only caps I have not changed are the two 20mf caps in one of the cans (that I labeled "Can A-1 and A-2 on that schematic) and the mica caps.
 
I'm actually getting a pretty good clean signal now (although with a severe lack of high-end... perhaps this quote from Mark explains why?

[quote author="Mark Burnley"]I wonder if these were designed for use as head amps for some kind of tape machine? IPC= "International Projector Corp" =magnetic film soundtrack?? And if so, is the CR network from the anode of the 5879 an equalisation network of some form?[/quote]

Other than that and the radio station, I'm getting quite a bit of clean gain. I wonder where the radio signal is coming from?

BTW, here are the latest numbers:

12AT7

pin 1 (plate) = 74.31
pin 2 (grid) = 0.208
pin 3 (cathode) = 1.285
pin 6 (plate) = 368
pin 7 (grid) = 0.07
pin 8 (cathode) = 2.838
 
Eureka! I started thinking about the radio problem as soon as I typed this above:

Ground pin: going to signal ground, also pin 2 on the Cannon connector,
also chassis ground.

Well, it's already been said in this thread that signal ground and chassis ground should be separate. So I cut the signal lead and attached it to the signal ground (pin 10 on the HA-100x) instead of the ground pin on the tranny and voila! No more radio!

Damn, I'm learning something!

:green:
 
> I cut the signal lead and attached it to the signal ground (pin 10 on the HA-100x) instead of the ground pin on the tranny and voila! No more radio!

I don't have HA-100x pinout memorized. But it is surely a floating input. Ground may not be essential if you just use it right. Maybe CJ or one of the other old-iron geeks here can help you get it really-right.

I suspect, for general use, you want to UN-ground the center-tap on the input winding. If I read that right (check with a HA-100x expert) you are wired 150Ω and grounded center-tap, you want 150Ω or maybe 600Ω floating.

> a severe lack of high-end...

Pentode connects to first triode with the usual cap and grid-resistor, plus a 2C+1R network that probably is (as Mark says) a tape/film-head playback EQ network. Cut that where it connects to V1 Plate.

This may give you WAY too much gain. Turn pot R10 to minimum resistance. Clip the wire to V2a's cathode cap Can B-1 (but be ready to put it back if it breaks into a squeal). If still too much gain, just reduce V1's plate resistor. Tack a 100K across it as a test. V2a/V2b have so much gain, you can throw-away a lot of signal on V1's plate and still not be in danger of overloading V1. (It helps that V1 is a pentode, so distortion won't rise as plate resistor is reduced.)

OR: clip one end of C1, Screen cap, and short 1K from V1 screen to V1 plate. That makes V1 into a triode. That will reduce V1 gain from about 125 to about 15, a major reduction. But 1:10 in the transformer, 15 in V1, 200 in V2, and 6:1 in the output, is gain of 5,000 or 74dB, still a LOT of gain for most microphones.

You really have too many tubes there. (It looks like they needed a LOT of gain to make-up EQ loss.) You could just skip V1, feed the input transformer to V2a's grid, and be about right for most mike-work. Or put a 100K audio-taper pot after V1 in V2a's grid circuit and call it a gain control. Or wire both sections of V2 in parallel, which would give a better match to the output transformer, and work it as pentode input parallel-triode output with feedback from triode plate to pentode cathode. Classic 2-tube mike-amp. Or re-wire V2 socket to take another 5879 as triode to drive the output. Plate dissipation would be high, or current would have to be low.... you could resistance-load the output and capacitor-couple that to the output iron, which might improve bass response. So many ways to go.
 
Thanks, PRR. I am getting a ton of gain, but also a lot of buzzing. It sounds like a 60 cycle low hum and then an upper "buzz" as well. Weird. No more radio, though. Maybe my power supply is noisy.
 
Got rid of the hum and clipped the 2C+1R network and like you said, now I have SUPER gain. Way too much. But it did fix the lack of high end (sounds really good now!)

I'm going to try and bypass the 5879 and input the signal into V2a's grid, just for shits and giggles. If that works, I might try your 100k audio taper idea.
 
That worked very well. The gain is much more manageable now. Looks like I'll go hunting for a 100k audio pot tomorrow. The local guitar repair shop should have them, no?
 
The other mod to consider is to rewire the input stage in triode mode. You'll get less gain and less distortion. Pentodes running without feedback are very nonlinear; but strapped as triodes, they become a different animal entirely. The 5879 is specifically intended as a low-noise input stage in preamplifiers; but in pentode mode, they're usually used with feedback to correct the nonlinearity of their plate curves.

Here's a 5879 datasheet:
http://www.mif.pg.gda.pl/homepages/frank/sheets/079/5/5879.pdf

Here's a little Ampex 4 x 1 mixer, as an application example of 5879 input stages run in triode:
http://ftp.ampex.com/ampex/manuals/audio/3761man/3671schm.gif
 

Latest posts

Back
Top