AA-23 Heathkit conversion done, debugging begins :)

[CitizenCain]

Hi guys, I finally got the Heathkit conversion fired up and it works pretty well, but I need to debug a couple things.

As a summary of my last posts, this is an AA-23 Heathkit mono hifi amp that I placed into a Princeton Reverb repro chassis from Marsh for a buddy.

The first question I need to figure is this...is there anything special about the value of the first dropping resistor in the B+ rail, the one after the first filter cap? This amp has a 3k resistor and many other schematics I see have a similarly small resistor or a choke.

My problem is that the voltages after this resistor are all too high. Before the resistor, in the power amp section, everything is fine. After the resistor, in the PI and preamp, the voltages are all high. The drop between sections is OK, they're just all almost uniformly too high. Any problems playing with that resistor value to get a bigger drop leading into the preamp area?

[G-hoppa]

Congrats on the addition of a Heath guitar amp to the world! I am still loving my recent build.

When you write that the values are too high, do you mean higher than you want to have at the plates, or higher than your math indicates they ought to be given a 3K dropping resistor? For the first, what about changing the values of the plate load resistors to get the voltage you want to see there at the plates? Replacing the dropping resistor (with a high watt rating, of course) would change all the following B+ values, maybe not all to what you want to see. You may end up having to swap the plate load resistors anyway-- why not just do that to begin with?

If the second, either the resistor or the math is off, I guess.

Of course, I'm no expert-- a similar set of questions face me with my next conversion-- a Bell hifi. I find the math a bit confusing, honestly.

[CitizenCain]

When I wrote the voltages are high, I meant high compared to the ref voltages given on the schematic. Here's a schematic of the final circuit I'm using with my measured voltages marked in red. I didn't have a ref voltage for V1 since I paralleled it.

As you can see, replacing the dropping resistor between filter caps A and B looks like it would make everything after it fall into line pretty closely.

http://www.visi.com/~sstolle/heath/aa-23-mod-006.JPG

[G-hoppa]

If the math looks good, I don't see why not then.

I do wonder about some of those voltages . . . point D is further off spec (~15%) than B and C (~10%) -- and look at the plate voltages of the two triodes fed by D. V 2A is much further off than V 2B. If D is off by a certain percentage, shouldn't these two be off by the same percentage? If that one dropping resistor is in fact the problem, I'd expect more consistent problems down the line.

-C

[CitizenCain]

V2B might be a little closer to spec because I had to swap out the cathode resistor after a slight "incident" It was originally a 2.2K resistor. Swapping to the 1.5K would bias it a little differently and pull the plate voltage down some from the 2.2K, wouldn't it?

Also, there used to be an "E" node on the supply rail that was used to feed V1. Since I paralleled that tube I just eliminated that node and used "D" to feed both V1 and V2. I think there's enough capacitance there, the whole preamp is pretty heavily filtered since it was a hifi amp previously.

FWIW, here's the original unaltered circuitry. Even tho this schemo is for an AA-181, it's identical to the AA-23. I got a repro assembly manual for the AA-23 and the schemo in there matched this one exactly.

http://www.visi.com/~sstolle/heath/aa-181.gif

[Alex/Tubewonder]

When you connected the input triodes in parallel you should have replaced plate and cathode resistor with 1/2 of their original values.

Are the power tubes originals or new production replacements?

Lets take a walk through original scheme: you expect a 75V drop on the 3k resistor (R36), this is 25mA current flow. Preamp section including phase inverter was originally drawing 15mA (calculated from voltage drops on plate resistors).

Your measurements say preamp current draw (current through R37) is about 11mA, ok, given the input gain stage is not drawing the current it was originally designed for that's fine.

Now, you have a drop of 47V on R36 that means current flow is ca 15mA.

So we have additional 4mA to account for - easy, look at voltage drop on the screen grid resistors and calculate current: two screen grids draw 4ma total.

So in your circuit everything is ok except for one thing:
Original circuit is designed for power tubes that are spot on spec and assumes ca 5mA/tube screen grid current draw from point "B". Your tubes draw about 2mA each.

Things to do:

Replace screen grid resistors with lower value in 200-300 Ohm range
Put 1 Ohm resistors between each 7591 cathode and ground so you 'll be able to measure each tubes idle current.
Adjust bias voltage so you get something in the 35-45mA range to bring to the operating conditions Heathtkit designed it for.
Don't stare at voltages given on original circuit too much. If you get there within 5% you're fine.

[CitizenCain]

Wow, thanks for the detailed explanation, Alex. I'm still very much a noob about this stuff. Calculating current draw never even entered my mind, I'm sorry to say.

On the paralleled triodes, I asked the question here about needing to change the resistor values but the feedback I got was that they could stay the same. It makes sense they should be halved.

The power tubes are JJ 7591S tubes. In fact, all the tubes are JJs except for the 12AU7, that's an old Tung-Sol.

When you say to "Replace screen grid resistors with lower value in 200-300 Ohm range", do you mean the resistors should be 200-300 ohms or should I subtract 200-300 ohms from the current values?

I actually do have a 1 ohm resistor in place between the 7591 cathodes and ground. The layout was such that it was only conveinient to use one shared resistor. When I measure the current, I thought I'd allow for double the value of a single tube and figure from there.

Thanks again!

[CitizenCain]

OK, I changed the resistors on V1 and it actually helped a bit. The voltage on the plate only dropped less than 4v, but the drop at the filter cap went from 317v to 305v.

OK, I'm pretty set with the voltage now I think. It seems to be close enough for now at least. I checked the tube specs for the 7591s and with the voltages I've got, it's almost dead on spec for the 450v plate/400v screen listing.

On to the next issue. This thing is a gain monster. In fact, it's almost unusable. There's no clean headroom at all. Turn the volume past 1-2 and it's all saturated gain. Plus, the more you turn it up, it actually seems to get quieter and sometimes sputter a little bit, all while maintaining max saturation. Is this blocking distortion?

As for the super gain, I think I need to put some interstage attenuation in place. I mean, the plate/cathode resistor combos aren't that radical, are they? How do you figure attenuation resistor values, just a little trial and error? I'm thinking of putting a 1M resistor between V1's coupling cap and the Vol pot, and another between V2B's coupling cap and the grid resistor for V3A.

Here's the latest schematic http://www.visi.com/~sstolle/heath/aa-23-mod-007.JPG

[Alex/Tubewonder]

Yes, you've got a straight gain cascade there. The only inter stage attenuation is provided by tone controls, with both at 12 o'clock you get flat response with ca 15db attenuation. Instead of adding things it's often more fun (and cheaper) to remove stuff. Try removing cathode decoupling capacitors (3.3 and 5uF electrolytics) in preamp one by one starting with input stage see what happens.

[CitizenCain]

Unbypassed cathodes give a lower gain? I thought it was the other way around. I have to go back and read the Aiken papers again

[CitizenCain]

Yep, unbypassed cathodes have lower gain Just reread Aiken's site. Now, to put it ot the test...