Low Power w/ SF Bassman

[hasserl]

I've been working on a SF Bassman project and I've run into a new problem with it that has me scratching my head. Previously I had installed a three prong cord and new filter caps and bias circuit cap, that all went well, everything worked fine. The I added a Master Volume in place of the ground switch, which I removed when I installed the new power cord and rewired the primary side of the power trans to eliminate the switch. I used a Ken Fisher style post PI style MV, using a dual 250k pot. Again, it all worked fine and the amp tested well after that. Then I did just a couple of tweaks to the Normal channel, and still everything was working and sounding great. A nice sounding amp.

Then I rewired the Bass channel to a 5F6A type circuit using the unused triode as a cathode follower. I used 250k resistors to isolate the channels just ahead of the PI input cap. And this is where I ran into trouble. Now the amp has real low power, only about 15-16 watts output. Troubleshooting showed I'm loosing AC voltage at the grids to the power tubes. With a 1k 100mv signal applied to the input, I end up with about 10 volts at the grid of the PI, but only about 16 volts coming out of it (with the MV turned all the way up). Interestingly, if I turn the MV down the volts come back up. If I pull the power tubes I get about 80 volts at the grids! But put the tubes back in and the AC volts drop back down to around 16. I'm just not getting enough voltage swing to drive the power tubes. I checked the DC voltage at the PI plates and it is a little high compared to the schematic (schematic calls for 220v, I've got about 230v). Same with the cathode voltage at the junction of the cathode resistor and the tail resistor (schematic calls for 100v, I've got about 110v). This amp doesn't match any schematic I found for Bassman's, it's kind of like a cross between an AB165 and an AA371.

Since the last modification I did before it started having this low power problem was to the bass channel, I disconnected the output at the Treble pot, to remove that channel from the equation, that did not change anything, I still have the problem. I can't figure out why I'm losing signal voltage from the PI. Voltage is a result of amps x resistance. Resistance hasn't changed, so if voltage is down, current must be also. But why? The DC voltage at the PI seems to be correct, and I changed out the PI tube to see if perhaps I had a failed tube, that did not help. What am I overlooking?

[dai h.]

MV isn't wired wrong? (since turning down brings the output voltage back up)

[Amp Kat]

What is the voltage on the Cathode of the Cathode follower. That is a direct coupled follower so did you just add a follower or just like the 5F6A schemo ?

[hasserl]

MV isn't wired wrong? (since turning down brings the output voltage back up)

Yeah, thanks Dai, that's a good place to look. Though the MV did work just fine after installing it and the problem only appeared after modifying the Bass channel. Still, not a bad place to look for issues. Here is a link to the schematic I've drawn up of the amp and the changes so far. I think it's mostly complete. The MV is wired up as in this schematic.

http://members.dslextreme.com/users/...od%20schem.pdf

[hasserl]

What is the voltage on the Cathode of the Cathode follower. That is a direct coupled follower so did you just add a follower or just like the 5F6A schemo ?

The voltage at the cathode of the CF is 148vdc. I just printed off the schematic and took it out and checked all the voltages, and I was wrong about them being a bit high, they appear to be low across the board. B+ is only 418vdc. That's quite a bit lower than any of the B-Man schem's show.

I used the unused triode of V2 for the CF.

[dai h.]

if it's not a mis-wired pot, and the driver output voltage goes down, maybe some kind of cancellation happening? Perhaps through the power supply? Or how about if the wipers were going to the wrong grids (crossed)? Would that cause what you're describing?

[oc disorder]

I first thought something to do with phase inversion and cancelling out
via audio in ht supply
edit whoops dai h already said that
...well got to think outside the square...
what about pot failure..? Just a coincidental fault developing at the same time?
I presume as the wipers get closer to the middle there is hardly any signal there as the audio is canceled (out of phase).
Weird.. be interested to know the outcome...

[dai h.]

yeah this sounds interesting. Wonder what it is. Is the mid-point of the pot resistance the strongest signal?

[hasserl]

OK, here's an update on this for those who were reading it and helping out.

To make a long story short, two different sets of power tubes later and modifying the preamps to eliminate the bass channel and go straight from the Normal channel .1uf coupling cap to the PI input I was back up to full power, well, about 50 watts max into a 4 ohm resistive load. That's close enough for me at this point. But hooking the channels back up according to the attached schematic causes the drop in power again. Both channels are weak, but the Bass channel is particularly weak, I'm only getting a couple of watts output with it. The AC voltage from that channel, with a 150mv signal at the input gives me just a few signal volts AC at the PI grid even with all controls dimed. The Normal channel is much stronger, but still low.

So the problem definitely lies with the preamp circuits, though I don't see why at this point. The Bass channel is a pretty straight 5F6A type circuit, and appears to be wired up correctly.

[Enzo]

Bass channel has one less gain stage compared to the Normal, so it would be weaker.

Measure resistance to ground from either end of both 250k mixing resistors just for grins.

Lift the 0.02 cap feeding the PI and stuff a signal into the cap there with a signal generator. Can you drive the PA to full?

Lift one 250k, then the other, one at a time to test each channel all by itself. Are they both still weak individually, or do they recover?

Are the 250k resistors actually 250k? As opposed to being mismarked or mistaken 250 ohm installed.

With the amp assembled and in the low power state, measure signal level at pin 2 of the PI or at the 0.02 cap. Now measure signal level at the outputs of th two channels: ie. the channel ends of the two 250k resistors. SHould be about the same. Now lift the two 250k and check the signal level at those two preamp output points: the treble wiper in the upper channel and the 0.1 cap in the lower. Has it increased?

In the low power condition, scope the circuits, is ther any RF oscillation going on? That could sap most of your power, but you wouldn't hear anything.

Point is we want to verify the PA still works, then isolate the problem.

[hasserl]

Thanks for the help Enzo, see the answers below:

Bass channel has one less gain stage compared to the Normal, so it would be weaker.

Measure resistance to ground from either end of both 250k mixing resistors just for grins.

With Tone controls all cranked up full:
Normal ch - ground = 990k
Bass ch - ground = 494k
.02 PI cap to ground = 743k

Lift the 0.02 cap feeding the PI and stuff a signal into the cap there with a signal generator. Can you drive the PA to full?

W/ 6.5 volts (max I can get out of my signal generator) at input of .02 cap I get ~40 watts power out of PA. If I could get more voltage from the generator I think the PA would put out more. Though the signal was squared off and very symmetrical.

Lift one 250k, then the other, one at a time to test each channel all by itself. Are they both still weak individually, or do they recover?

Both remain weak

Are the 250k resistors actually 250k? As opposed to being mismarked or mistaken 250 ohm installed.

248k on one, 246k on the other

[quote] With the amp assembled and in the low power state, measure signal level at pin 2 of the PI or at the 0.02 cap. Now measure signal level at the outputs of th two channels: ie. the channel ends of the two 250k resistors. SHould be about the same.

W/ 150mv applied to Normal input, 6.9vac at pin2 of PI, 87.1vac at channel end (junction of .1 cap & 250k resistor).

W/ 150mv applied to Bass input, 2.4vac at pin 2 of PI, 14.6vac at channel end (Treble pot wiper) .

I'm losing a LOT of voltage here!!!

Now lift the two 250k and check the signal level at those two preamp output points: the treble wiper in the upper channel and the 0.1 cap in the lower. Has it increased?

Treble wiper in Bass channel = 14.6vac
0.1 cap in Normal channel = 106.6vac

In the low power condition, scope the circuits, is ther any RF oscillation going on? That could sap most of your power, but you wouldn't hear anything.

I have to be honest here, I really don't know how to do that.

Point is we want to verify the PA still works, then isolate the problem.

I think the PA works fine, I think I'm losing a lot of voltage between the channel outs and the PI in, and I think that's the problem. But it's got me stumped right now as to why.

[dai h.]

feedback looks weird in the latest schematic. How about if the fdbk. is disconnected?

[Enzo]

Imagine you are looking at a speaker sitting there. The cone appears sharp and in focus as you look. Now imagine a large signal coming through it. The cone will be moving, and it will not appear as sharp or clear to the eye because it is moving in and out fast. Like taking a photograph of something still and then while moving fast. Even if the speaker was moving wildly at too high a freq to hear, you could still see that the cone was in motion, it would appear spread out or fuzzy, like the photo of the moving object.

In your scope, a steady signal, or no signal at all for that matter, will appear as a nice sharp clean line - either waving to the signal or just straight across. If the signal is very high in freq, the trace will be up and down your screen so many times per sweep, it looks like a wide band of light instead of a striaght line. That is what you'd see. (This wide band of light could be resolved into a waveform if you just speed the trace way up. Of course then an audio waveform would stretch so far out it would only look like a barely sloping line) If there is an audio signal trying to pass through, it will make the edges of this band move with the signal.

I assume your signal generator will produce higher freqs than audio? Try this. Connect your scope to the generator, set the gen to 100Hz and the scope so the waveform is at least half the height of the screen - a nice big sine wave. Set the sweep speed on the scope so the trace shows maybe two complete cycles. Pretty and clear, right?

Now, don't touch the scope or the signal level, but switch the gen freq to 100,000Hz. Now instead of 2 cycles across the screen, there will be about 2000 of them. they will be so close together it will just look like a wide band of light across the screen. This is what high freq oscillation in an amp looks like.

You can't hear it, but the amp is cranking it out for all it is worth, making the audio postion very weak sounding. I don't know you have it, but that is what to look for to check.
***********************8

Looks like your PA is indeed OK if you can get the 40 watts from it. But I find it off you can get the 40 watts strong from it with 6,5v from your gen, but with 6.9v from the normal channel at the same point, it is weak.

Between the 0.1 of the normal channel output and pin 2 of the PI, you are losing 80v. Is it droppng across the 250k or across the .02?
To take this reading, don't compare volts to chassis at each end, actually put the meter across each component and read the AC volts. The reason is we tend to think our meters are invisible to the circuit, but they are not. A typical meter might have a 10 meg input impedance, which is pretty high, but when we are looking at relatively high resistances in the circuit, the meter can load the signal down. So measuring across a suspected compnent doesn't load down the signal by becoming part of a voltage divider.

I notice the PI input seems to load the normal channel down about 20v, or in this case about 20%.

The amp used to work, until you changed it. I am less inclined to think we will find a circuit related problem here, than one of mishap to the parts board. When doing whatever changes you made, a wire might have gotten moved to the wrong point somewhere, or a solder bridge is hiding under the eyelet board, or a connection was left open. Or a new part was left with an untrimmed lead under the board. Or even a cap was damaged in soldering something next to it. Like that .02 for example.

[dai h.]

point of fdbk. insertion looks like it'll cancel the more the master is raised (seems to fit with earlier descrip. of increase in volume when turning DOWN). Maybe just simply miswired to the wrong spot?

[hasserl]

feedback looks weird in the latest schematic. How about if the fdbk. is disconnected?

point of fdbk. insertion looks like it'll cancel the more the master is raised (seems to fit with earlier descrip. of increase in volume when turning DOWN). Maybe just simply miswired to the wrong spot?

That is the original fdbk circuit. One of the common mods for these SF Bassman's is to move that over to the other side of the PI, but then it'll be positive feedback, so you have to reverse the OT secondary wires to compensate. I was considering that mod but haven't gotten there yet. Got stuck at this point.

Enzo, I have had the scope on the circuits and haven't noticed anything like you described. But I haven't looked specifically for it either. I have to put this off until tomorrow now as I've got to take my wife somewhere this morning and have a blues jam this afternoon/evening.

Thanks again guys, I'll be back.