New build with a baffling problem

[ldervish]

I am building an Ampeg SB-12 clone bass amplifier (2- 12ax7 + 2 - 6L6GC) on an old Rowe chassis, with a couple differences. First, I'm using full-wave recto rather than tube; and second, the bias supply is provided by a single (yellow) tap w/ half-wave neg recto followed by a 3k3 in series then 100uf electrolytic to ground paralleled by (at the moment) a 200k resistor, with 270k's going to the grids per schematic. I know the dropping resistor seems large but it is how I am able to generate -53.5v on the output grids. It also drives the plate voltage up to near the 475 mark that the schem calls for. So maybe there's a better way to get enough bias voltage? That's issue #1.

The second issue is baffling me, but I think has to do with the phase inverter. With the treble pot almost at minimum or almost at max, the output current drops pretty significantly (say by 12-13ma)and the amp is underbiased but sounds pretty good. With the treble anywhere else in its range the current jumps to about the right level, but the sound is very distorted. I have switched out the baxandall treble caps and the treble pot, verified the V2 cathode resistor and tried to find any leakage in the V1 coupling caps, without joy.

Any suggestions where I could look for the problem?

The schematic is good except I didn't include the "bright" input, and the bias supply is as described above.

[loudthud]

First experiment would be to disconnect one side of the (I think, can't read clearly) 5.6K resistor that goes from the 16 ohm tap of the output transformer back to the phase inverter. That should increase the gain but stop any oscillation. If that clears up the problem with distortion and the treble pot, you need to reverse the output transformer wires at the output tube plates (pin 3). If that is not convienient you can swap the wires going to the grids (pin 5) of the 6L6 tubes.

[ldervish]

First experiment would be to disconnect one side of the (I think, can't read clearly) 5.6K resistor that goes from the 16 ohm tap of the output transformer back to the phase inverter. That should increase the gain but stop any oscillation. If that clears up the problem with distortion and the treble pot, you need to reverse the output transformer wires at the output tube plates (pin 3). If that is not convienient you can swap the wires going to the grids (pin 5) of the 6L6 tubes.

Thx for the reply. Disconnecting the fb (6k8, sorry for the poor schem quality) did not resolve the buzz and signal was now slightly less clear at the max treble position.

[ldervish]

Retracing my steps

Time to go back and check again (for the umpteenth time) my wiring. I must be missing something.

[loudthud]

The symptoms you report strongly suggest an ultrasonic oscillation. Try connecting your DMM across the speaker terminals set to AC Volts. Does the reading jump up as you rotate the Treble control?

[nickb]

Talking of baffling, I really don't get your bias circuit description. What you described would give hundreds of volts on the grids since the series resistance is only 3.3k. If OTOH if it really is 200k, as you allude to, then what is in the ground side of the divider?

I think Loudthud's suggestion of oscillation is good, but since the treble control affects it so much I'm thinking that the loop causing it might be more than just the output section (V2-V4). Does it change if the volume is turned to zero, so removing the first stage from the loop? How about if the instrument is unplugged - that should short the input to ground and take that out of the picture. The cause could be a wiring layout issue where wires with big voltage swings (e.g. 6L6 plates) are close enough to sensitive (low level) areas such as volume and input wiring to mention just a couple.

[ldervish]

Got it

The symptoms you report strongly suggest an ultrasonic oscillation. Try connecting your DMM across the speaker terminals set to AC Volts. Does the reading jump up as you rotate the Treble control?

OK, please bear with me.... I had to redo my reply.

I found the source of the problem, I had bundled the lead running to the Ext. Amp jack (tied to V2 grid and baxandall output) in with the OPT primary leads. Once I separated it, output current is steady, Treble has no effect on it. I found it after reading another of your posts re: lead dress, Loudthud, and nickb's comments re: leads with high voltage swings being near those of low voltage. Thanks so much to you both. I thought I had been pretty careful with lead dress but that one got past me. The amp is very quiet.

Re: the bias supply, there is only the one yellow tap so it must return via the ct. It does not pull from the secondary ht leads as many do. The original Rowe circuit (this was once a PA amp) used the 3k3 series R and a 33k dropping R, with ~ 43vac before the diode. Increasing the dropping R has given me about -52vdc on pin5 and simultaneously driven the PT secondary up for a B+ on the plates of about 460vdc.

[loudthud]

OK, please bear with me.... I had to redo my reply.

I found the source of the problem, I had bundled the lead running to the Ext. Amp jack (tied to V2 grid and baxandall output) in with the OPT primary leads. Once I separated it, output current is steady, Treble has no effect on it. I found it after reading another of your posts re: lead dress, Loudthud, and nickb's comments re: leads with high voltage swings being near those of low voltage. Thanks so much to you both. I thought I had been pretty careful with lead dress but that one got past me. The amp is very quiet.

Re: the bias supply, there is only the one yellow tap so it must return via the ct. It does not pull from the secondary ht leads as many do. The original Rowe circuit (this was once a PA amp) used the 3k3 series R and a 33k dropping R, with ~ 43vac before the diode. Increasing the dropping R has given me about -52vdc on pin5 and simultaneously driven the PT secondary up for a B+ on the plates of about 460vdc.

When an amp breaks into ultrasonic oscillation, there is usually a shift in the bias. You will probably need to reset the bias now that the oscillation is fixed.

[ldervish]

When an amp breaks into ultrasonic oscillation, there is usually a shift in the bias. You will probably need to reset the bias now that the oscillation is fixed.

Correct once again, tubes are too hot now and plate voltage dropped.

I'm contemplating changing the bias supply to the schematic version which pulls from the ht lead, b/c the way it is now it takes a lot of added resistance to bring the bias voltage down to where it needs to be. I feel like I would have more latitude with the schem arrangement, subbing a fixed R and pot for the last 56k. Opinions on that change?

[loudthud]

If the bias current is too high, you have to do something. In the first post you said you had a 3K3 resistor in series with the rectifier. The first thing I would do is knock that down to perhaps 470 to see if that increases the bias enough. If that doesn't work, either try a voltage doubler or the circuit like Ampeg used.

[ldervish]

If that doesn't work, either try a voltage doubler or the circuit like Ampeg used.

I ended up with the Ampeg circuit but put a 100k pot in (temporarily) in place of the 56k bias resistor. It gives me plenty of bias. Once settled I'll use a smaller pot and a fixed resistor.

There are a few other things to sort out, like the baxandall gives me way more bass than I want and zero treble control, maybe something went south with all the U.O. going on, or I have a cold solder joint. On initial power-up both controls worked so a little sleuthing is in order to find whatever changed.

Thanks for the help with the oscillation.