6L6 mono amp questions

[rickl]

I recently picked up a Harmon Kardon CA-35 PA amp. This is a 30watt mono 6L6GC PP amp with 2 5814-A (12AU7) driver/tone control tubes. It also has 2 12AX7 mic drivers. A missing 5AR4 rectifier.

I added a 5U4 and brought it up slowly on a varaic. B+ 440v.

It doesn't work. It has been worked on and some noticable things wrong include a 47uF on the output of the 5AR4 rectifier tube (which might explain why the tube is missing), various "upgraded" coupling caps, and other tweaks.

My plan is to strip this down to a simple mono amp with a volume control.

I'd like to use the power (Stancor #138410E) and output (Woodward-Schumacher #606305) transformers, 6L6s, and one (two?) 5814A for a driver. I have a 5U4GB and a 5R4GYB I could use for the rectifier. PS will be either CLC or CRCRC where first cap is around 10-20uF.

Questions:
Anyone have a schematic of the HK CA-35?

How about a simple 6L6 PP with 12AU7 drivers?

I've search the archives and googled but nothing jumps out to me. What about a Heathkit w-5m? I was thinking of only one 12au7 is needed to drive the 6L6s yet the Heathkit uses two. A simple Williamson design?

Any specs on the Schumacher output transformer?

Go back to my SE84 amp?

Thanks for any help. I've been sitting on this post for weeks searching archives and the Net. How hard can this be?

rick

 

[Dr. Photon]

Those old PA amps are pretty cool. What you have is *PERFECT* for conversion into a kickin' guitar amplifier!

For a schematic, I wouldn't exhaust much effort looking for one since it isn't needed. Schematics aren't necessary for repair of simple tube circuits, and are definitely not needed if you just got the chassis. One thing I ususally do is track out the schematic by hand onto notebook paper. I actually find this quite enjoyable - like I'm unlocking the secrets of the amplifier or a geeklike exploration into the unknown of sorts - I can usually get it done in a few hours for something of that complexity. Good lighting is a must. I also make heavey use a 2-AA miniMagLite. I typically redraw the original to look better (and also sanity check it to see if what I drew could possibly work....if not, look again) since the first is pretty random since I don't know what the circuit is doing at first.

The problem with making it into a HiFi amp is that for stereo use, you'd need to find a second one, which might be very difficult. Just be sure to use brand new capacitors. I reuse tubes, iron, switches and pots (if still good), and power resistors. I chunk the old caps and small carbon resistors. I ususally leave the cap cans on the chassis for aesthetic apeal (even though mine are butt ugly!)

 

[PRR]

> How about a simple 6L6 PP with 12AU7 drivers?

6L6 with 440V B+ will need about 40V grid drive. Say standard hi-fi input sensitivity is 2V peak. Then you need a gain of 40V/2V= 20 just to boost the inputs to the grids. Using one half 12AU7 as a cathodyne (works great), the other half has to give a gain of 20; 12AU7 won't do much more than 15.

AND this is a pentode. For speakers you really (or traditionally) want enough negative feedback to kill the high output impedance. 4:1 NFB will do for a start. Now the first stage needs a gain of 20*4= 80. 12AU7 won't come close; 12AX7 with cathodyne load will.

If you self-bias the 6L6 so you can use large cathode resistors, one 12AX7 will do. This was a fairly popular plan.

For fixed-bias, you need half of AX and half of AU. There used to be a tube like this. Very popular in the day but getting hard to find.

> What about a Heathkit w-5m? I was thinking of only one 12au7 is needed to drive the 6L6s yet the Heathkit uses two. A simple Williamson design?

As shown above: for standard hi-fi sensitivity, with ~400V on 6L6, one 12AU7 is barely enough, and inadequate if you like damping.

Williamson is not simple. With two stages of 12AU7, the gain is over 200. We only need 20. The rest goes to negative feedback, but even 10:1 NFB is hard to manage with a transformer and two R-C coupling networks. And Williamsons are unstable, if not done exactly like the original (including a
mighty complicated transformer).

The H-K was a better than average PA amp. You will be fine just restoring the parts you want to keep. You know where the 6L6 are. Verify reasonable DC voltages: ~400V plates, ~250V-~350V G2, and either +~35V on kathodes or -~35V on G1. Follow the coupling caps to the nearest 12AU7, how is it connected? Cathodyne is obvious; push-pull volt-amp (not unlike a Williamson) is an alternate. How is that stage driven? When you get to tone or volume control, you can probably stop, add a volume pot, and call it an amp.

 

[Tim Wyatt]

>
For fixed-bias, you need half of AX and half of AU. There used to be a tube like this. Very popular in the day but getting hard to find.

JJ ECC832

 

[Miles Prower]

How about a simple 6L6 PP with 12AU7 drivers?

12AU7A's for audio work: YYYYEEEECCCCHHHH!. I considered that possibility and dismissed it once I ran some calculations and kept coming up with some horrendous THD estimates, some 15% (as bad as an uncompensated BJT) or more. Looking up examples of where the 12AU7A is commonly used turned up lots of examples of push-pull RF class C drivers and frequency multipliers, digital apps such as memory cells and R-S latches, monostables, Schmidt triggers, and non-sinusoidal oscillators, or sinusoidal oscillator + buffer. Precious few audio apps.

How about this instead: 6SF5 (or 6F5 if you like those funky grid connections to metal buttons) as a voltage amp. I considered this as a possible preamp and came up with these figures:

• V(PP)= 250V
• R(P)= 330K
• V(gk)= -1.0V
• I(P)= 0.44mA
• V(o)= 136.46V(p-p)
• THD= 2.6% (est)
• A(v)= 68.23

Given the magnitude of that possible output, 2.6% THD is pretty good. (And better than the THD claims for some "miracle" circuit that was intended to "undistort" BJT voltage amps that someone on another forum was crowing about.) Since 6L6's don't need anywhere near that much, cutting down the V(o) would improve on that.

For a cathodyne stage, you could use a 6J5. The characteristic looks a good deal more linear than that of a 12AU7A or a 6C4 (half of a 'AU7A) and would make a better cathodyne or follower.

 

[nuvistor]

HK250 is stereo 25W/ch with 1 6CG7 and 2 6L6GB per channel, owner's manual with schematic is available here, unfortunately no manual for your CA-35 amp.

 

[rickl]

> The H-K was a better than average PA amp. You will be fine just
> restoring the parts you want to keep. You know where the 6L6
> are. Verify reasonable DC voltages: ~400V plates, ~250V-~350V G2,
> and either +~35V on kathodes or -~35V on G1. Follow the coupling caps
> to the nearest 12AU7, how is it connected? Cathodyne is obvious;
> push-pull volt-amp (not unlike a Williamson) is an alternate. How
> is that stage driven? When you get to tone or volume control,
> you can probably stop, add a volume pot, and call it an amp.

I need to pull the 6L6s again to verify voltages but notes show for V5/V6:
G1 at +6.6v/+0.42v
G2 442/444
cathodes (pin 3) 70.4/70.8 (seems 2X too high)

I didn't measure any negative voltages so I assume self biased.

Cathodes are tied together (pins 8), G2 (pins 4) are tied together
and a 3.3k 5w to the 'red' lead of the OP, plates are tied to the OP
(red and brown leads).

Tracing back to the 12au7 are difficult. Black soot everywhere.

> HK250 is stereo 25W/ch with 1 6CG7 and 2 6L6GB per channel,
> owner's manual with schematic is available here, unfortunately no
> manual for your CA-35 amp.

Thanks for looking. The schematic is similar to the CA-35. The bias circuit is buired under large electrolytics and covered with black soot. However, looking at the 'left' schematic and comparing to my notes:

There is a "C2" 0.047uF/600v between V5 G1 (pin 5) and the plate of the 12au7 (pin 1) and a "C5" 0.047/600v between V6 G1 (pin 5) and the cathode of the 12au7 (pin 3). I need to continue to follow back the circuit.

Thanks again for everyone's help. I'll dig into this more tonight.

rick


~
~

 

[PRR]

> cathodes (pin 3) 70.4/70.8 (seems 2X too high)

Sure does. Also if I understand correctly, the cathodes are strapped together yet you cite a 0.4V difference.

With power off, check resistance from 6L6 cathodes to ground. (If there is a cathode capacitor, you may have to disconnect it.) 250-350 ohms would be typical for self-biased 6L6 at not-insane plate and G2 voltage. Wonder if you have a burnt resistor or shot-pot.

The grid voltages should be mighty similar. Nominally zero, but if they used the hi-fi HK250 design there could be a couple volts. I'd note the position of the bias balance pot and then set it to the center, see if both grid voltages come out the same. If similar to the hi-fi plan, +6V on a 6L6 grid suggests a very gassy 6L6. A volt or two in 1Meg might getter-out as the tube cooks. 6V in 330K sounds like a frightening amount of gas.

> Black soot everywhere.

So something burned. Not a small resistor, probably not a cap. What burned? I have a sickening feeling that "black soot everywhere" is tar burned out of a transformer, which is a deal-breaker.

You are going to have to clean that soot sooner or later. Soot "everywhere" is causing leakage in high-impedance paths, totally screwing-up bias. Try alcohol and those little brushes that plumbers use to apply flux.

> 0.047uF/600v between V5 G1 (pin 5) and the plate of the 12au7 (pin 1) and a.. 0.047/600v between V6 G1 (pin 5) and the cathode of the 12au7 (pin 3).

Cathodyne. Perfectly good plan for 6L6 in Pentode.

Miles> 12AU7A's for audio work: YYYYEEEECCCCHHHH!

You are welcome to your opinion. But many fine amps use 12AU7. Most of the US-made Williamsons including the mighty HeathKits. Fairly common as the output driver in classic-era pro audio broadcast and recording equipment.

> kept coming up with some horrendous THD estimates, some 15%

Must be a very odd loadline. Steal the circuit values and signal levels from the RCA Resistance Coupled Amplifer charts, or GE 1963 12AU7, you should not see more than 5% THD. If you are SPICEing, doubt your model. I think I am using Koren's models and the numbers seem similar to real-life (I've abused my share of 12AU7 in my days). Note that the book-value for 12AU7 Characteristics are MUCH too hot for good audio performance: think 2mA not 10mA.

Yes, old-type 6SN7 are more linear than the 12AU7 that came from 6SN7. IIRC, by the 1970s you could not tell a 6SN7 from a 12AU7; they probably put the same guts in different bottles.

You don't see 12AU7 everywhere because it is an unexciting tube. For sheer Gm/Ik, 12AX7 beats it with ample current for hi-fi needs. For gain, 12AT7 and 12AT7 are the first-choice tubes. When 12AU7 is almost-enough, the several super-12AU7s like 6CG7 and 12BH7 get the call.

> 6SF5 ... as a voltage amp. V(PP)=250V R(P)=330K V(o)=136.46V(p-p) THD=2.6%(est)

At what load (what is the next-stage grid resistor)? Even 1Meg is a significant load on a tube with 330K plate resistor and ~120K plate resistance. The hi-fi HK250 uses only 330K resistors on the 6L6 grids, and you sure won't get 136V p-p with 330K plate and 330K next-grid.

Also what is the bandwidth? Take 30pFd grid and stray capacitance unless you have a specific load in mind. (A Cathodyne on the same socket as a volt-amp can be built with less than 30pFd grid-and-stray.)

 

[rickl]

Thanks PRR for your help.

I'll check the resistance tonight or tomorrow. Regarding all the black soot, my WAG is wires burning. There are new bypass electrolytics and some non-stock orange drops so a few components have been replaced. The tubes "look" good but I haven't tested them (I have a friend with a tester).

I took a picture of the underside but I can't remember how to get the size down to the 100k limit :-(.

The reason for the Phillips 5814-A driver tubes is because they are there. I do have 12at7 tubes and a lot of different 12ax7 tubes. I thought it would be good to try various 12ax7 tubes to see why these seem to be hated but instead I went with the 5814A/12au7 driver tubes.

FWIW, my pre is a DIY 6sn7 and my main amps are Dyna Mark IIIs with 6cg7 drivers (Curcio boards). Thought this amp might make a nice center channel if I ever go to 5.1.

Thanks again. I dug more into the schematic but now need to convert it from pin numbers to plate/cathode/grid.

rick

 

[PRR]

> all the black soot, my WAG is wires burning.

It is unusual for a tube amp to have enough available current to burn even small hook-up wire. Hardly impossible, because I've done it. But it isn't usually the wire that burns first. More likely is something else got red-hot (and ruined) and set the wires on fire. So be dubious of everything. OTOH, it is unlikely that many parts were ruined: one or two take the brunt of the overload/short and the others rest easy. (Though the soot is still bad.)

> how to get the size down to the 100k limit

Depends on your tools. Crop and re-size to ~600 pixels of the important stuff. Save as JPEG and adjust the compression ratio, not too jaggy but not so perfect that the file-size is huge. Takes some practice, and tools that let you decide what is best instead of dumb defaults.

Alternatively, if you have webspace (many ISP accounts have some, though finding them is not supposed to be easy) just put the big file on the web and post a link in the message. There are "free" image hosting services but some are slow and some refuse to be linked from outside pages (because that avoids the ads that pay for the "free" service).

5814 is just a high-spec 12AU7. Nothing bad about using those, though 12AU7 gain is not large.

Here's an idea. Use 12AU7 volt-amp, 12AU7 Cathodyne, all conventional and possibly already there. Strap the feedback connection at the volt-amp cathode directly to the 8-ohm output tap (no resistor divider). That gives over 12dB NFB, with one bass pole plus the transformer. It also gives input sensitivity of ~14V RMS! So build (or find) another 12AU7 volt-amp on the chassis to give a gain of 7 to 15 (without or with a cathode cap) to bring the line-input up enough to push the output stages around. With any sort of semi-proper bias, and 300+V supply, 12AU7 can make the 20V-30V of swing with low distortion, a couple percent at most, possibly less with local feedback and a little cancellation between stages.

 

[Miles Prower]

But many fine amps use 12AU7. Most of the US-made Williamsons including the mighty HeathKits. Fairly common as the output driver in classic-era pro audio broadcast and recording equipment.

That may be, however it sure didn't work out when designing the 807 amp I mentioned elsewhere here. That's where I kept getting those horrendous THD estimates. Changing to the 6SN7 gave better results, and also fits better with the overall "retro" appearance.

Must be a very odd loadline.

Probably was.

You don't see 12AU7 everywhere because it is an unexciting tube. For sheer Gm/Ik, 12AX7 beats it with ample current for hi-fi needs. For gain, 12AT7 and 12AT7 are the first-choice tubes. When 12AU7 is almost-enough, the several super-12AU7s like 6CG7 and 12BH7 get the call.

As I said before, the "everywhere" I found it included mostly RF applications. The ARRL Handbook mentions the 12AU7A as a transmitter type, and one of the LWCA's earliest DiY projects used it for a longwave Part 15 transmitter.


At what load (what is the next-stage grid resistor)? Even 1Meg is a significant load on a tube with 330K plate resistor and ~120K plate resistance. The hi-fi HK250 uses only 330K resistors on the 6L6 grids, and you sure won't get 136V p-p with 330K plate and 330K next-grid.

Also what is the bandwidth? Take 30pFd grid and stray capacitance unless you have a specific load in mind. (A Cathodyne on the same socket as a volt-amp can be built with less than 30pFd grid-and-stray.)

I guess I should've mentioned that: 1.0M (248.12K AC R(P) ). This was originally going to be included in the design of the 807 amp as a gain block with an A(vol)= 1625 and 46.2db(v) feedback for an A(vcl)= 8.0. I didn't pursue it any farther after deciding not to include it, yet it looks like the open loop BW is around 9.0KHz (-3.0db). Including feedback would increase that.

As of now, I don't have any use for the 6SF5, however, I'm definitely keeping it in mind for any subsequent projects. Price is good as well: $4.50 NOS from a local supplier.

 

[rickl]

Alternatively, if you have webspace (many ISP accounts have some, though finding them is not supposed to be easy) just put the big file on the web and post a link in the message.

Like this: big pic of underside of HK

 

[SY]

Yes, old-type 6SN7 are more linear than the 12AU7 that came from 6SN7. IIRC, by the 1970s you could not tell a 6SN7 from a 12AU7; they probably put the same guts in different bottles.

The samples I have of late '70s 6SN7 (RCA button base) have guts that would not fit into a 12AU7 envelope. And, except for one duff section in one tube, the linearity is about an order of magnitude better than the 12AU7 measurements reported by Morgan Jones.

 

[rickl]

Schematic

Here is my first pass at the HK 6L6GC schematic. Note how the 6L6 tubes are wired. This seems like a problem. My thoughts are to use the circuit from HK schematic link above on the 6L6 cathodes and leave the drivers alone. Once I get the 6L6 bias set, focus on the drivers.

My initial worry over the black soot has subsided. I'm planning to add the bias circuit and check voltages again. Clearly some past 'tweaks' were incomplete.

rick

 

[Miles Prower]

Here is my first pass at the HK 6L6GC schematic. Note how the 6L6 tubes are wired. This seems like a problem.

Actually several.

• No DC return for the 6L6's.
• The 3.3K resistor from the screen supply to the plates. Not only will that make the plate voltage lower than the screen voltage, the sum total of 13K will really pull down the actual applied voltage. Given likely values for plate current (50mA) and screen current (8mA) you're looking at 750V of DC drop right there. You'll need a supply voltage of some 1000V(DC) to get reasonable voltages at the 6L6 electrodes. Why do that?
• The phase splitter is all wrong. You have 270K in the plate circuit, and yet just 24.4K in the cathode circuit. Big time unbalance there. Furthermore, what's this suppose to be? Cathodyne or LTP? One or the other, but not both!
• Why have three gain stages outside the feedback loop? The 6L6 doesn't need that much voltage amplification ahead of it for reasonable preamps and line stages.

 

[PRR]

> No DC return for the 6L6's. ....The 3.3K resistor from the screen supply to the plates.

Either rickl is not reading what is there properly, or someone before screwt-up bad.

The 6L6 need a 250-300 ohm resistor from the common cathode to ground. This may have burned up and been removed.

Surely the OPT CT is tied to the highest B+ around.

Almost certainly, the screens are tied to a 3K3+10K voltage divider to set G2 voltage about 75% of plate voltage AND have enough swamping current to absorb the large changes in 6L6's G2 currents. This is a very common setup.

> The phase splitter is all wrong. You have 270K in the plate circuit, and yet just 24.4K in the cathode circuit.

26K6 in the cathode, which suggests 27K (not 270K) in the plate. I'd say a red stripe is looking orange in its old age, or that some previous hacker mistook orange for red.

With the positive feedback (bottom 2K2 resistor) there almost has to be overall negative feedback. The 100pFd is not in the right place nor the right value.

 

[rickl]

Miles, PRR,

First I'd like to thank everyone for the replies.

>Either rickl is not reading what is there properly, or someone before screwt-up bad.

I spent a lot of time re-tracing the wires/passives before the post. I sketched the schematic three different times to make sure I got it. I just about can dream how the wires are hooked up. I also had someone else look the schematic and all agree the amp is hose up. Funny how the PO only said a cap might be leaky.

>No DC return for the 6L6's.
> The 6L6 need a 250-300 ohm resistor from the common cathode to ground. This may have burned up and been removed.

Yep, this is a problem.

The plan is to add a 270/10w cathode resistor across the 47uF cap for the cathode resistor.

>The 3.3K resistor from the screen supply to the plates.
> Surely the OPT CT is tied to the highest B+ around.
>
>Almost certainly, the screens are tied to a 3K3+10K voltage divider
> to set G2 voltage about 75% of plate voltage AND have enough
>swamping current to absorb the large changes in 6L6's G2
>currents. This is a very common setup.

I may have drawn this wrong. G2 (pins 4) are tied together with the 10k going to B+. Also from one of the pin 4, there 3.3k/2W going to the OPT CT. This too seems wrong. There is also a 22uF cap tied from pin 4 to ground missing from my posted schematic :-(.

Plan to remove the 3.3k. Tie the OPT CT to B+.

The plates are tied to the OPT only.

I’m trying to picture the 3K3+10K divider. 10K from B+ to G2. 3K3 to ground?


>The phase splitter is all wrong. You have 270K in the plate circuit, and yet just 24.4K in the cathode circuit.

> 26K6 in the cathode, which suggests 27K (not 270K) in the plate.
>I'd say a red stripe is looking orange in its old age, or that some
>previous hacker mistook orange for red.

This could be my fault. It is hard to see between orange and red on some. I have not measured any resistors. The location of the 270k plate resistor is under the ‘taped’ caps and is where most of the soot is. I’ll look again tonight.

Is the 210k plate resistor on the same driver tube (pin 6) ok? Using the HK250 schematic as reference again, this isn’t the same.

>With the positive feedback (bottom 2K2 resistor) there almost has
> to be overall negative feedback. The 100pFd is not in the right
> place nor the right value. [/i]

The ceramic cap says 100k. I could never remember in college (nF?) and that was a long time ago. This is consistent with the HK 250 schematic (except for the 4 ohm tap).

Thanks again for every ones help. I really am learning something.

rick

 

[PRR]

> Is the 210k plate resistor on the same driver tube (pin 6) ok?

Probably works. 210K is a decidedly odd value for the 10% series and for this amp's vintage. I'd be thinking 270K.....

> Using the HK250 schematic as reference again, this isn’t the same.

Yes. On third sight, that HK250 driver is odd. The first stage cascodes the cathodyne. It may work, but I don't know why it would work good.

> I’m trying to picture the 3K3+10K divider. 10K from B+ to G2. 3K3 to ground?

What _I_ had in mind is 3K on top, 10K on the bottom to ground. Feed +400V to the top. The junction will be +300V, a good value for 6L6 screens.

However look at the HK250. They make the raw +400V, then through a 34Ω/20uFd R-C filter. That supplies the OPT CT. Then they drop through 6K8 and 8uFd, and feed the Screens. Then they drop through 4K7 and 8uFd, feed the driver.

You have only two screens to drive (the HK250 has 4), but a pile of low-level mike-amps etc. So the "6K8" might be twice as big for just two grids, then both the "6K8" and "4K7" could be 2/3rd the size to allow for added low-level tubes. That makes about 10K and 3K, as you found. Except the 3K should be going to low-level stages, not to OPT CT.

> The ceramic cap says 100k. ... This is consistent with the HK 250 schematic

Well, yeah, but where is the 12K resistor?

> I could never remember in college (nF?)

Here, I think "K" is a type or voltage or tolerance designation, not the units.

> This could be my fault.

ALWAYS blame the guy before you for any stupidity found along the way.

 

[rickl]

Schematic rev 2 ECO

Well, after reading/listening/looking/measuring with DVOM I did find a couple of mistakes in my schematic.

1. OPT center tap is wired to B+
2. wired 3k3 dropping resistor to G2
3. fixed plate resistors (pin 1 is 27k, pin6 270k). Found only 1 210k to tape monitor. This agrees with PRR.
4. added 15k resistor from cathode (pin 8) to 4 ohm speaker tap. note "POC" screw terminal on back of amp.
5. Added B+ and filtering caps.

Still no cathode resistors on 6L6 tubes

I'll post the rev 2 schematic tomorrow.

That makes about 10K and 3K, as you found. Except the 3K should be going to low-level stages, not to OPT CT.

Again you are correct. B+ to OPT CT, 3k3 2W dropping resistor to G2, another 10K dropping resistor for low-level stages. This makes sense to me.

Tomorrow, I might just add my trusty 270 10w cathode resistor, hook up my test speaker and see if any magic smoke comes out.

Thanks again,
rick

 

[rickl]

Here is the updated schematic.