• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

6LR8 push pull amp build

gadget73

gadget73

junk junkie
Subscriber
I've made mention of a project a few times lately, so here is a thread on it. Its not done, but far enough along that its at a point that it makes sense to post about.

I'm bad about taking pics, so you'll have to excuse the lack of step by step, but it looks like this.

Top side, gold hammertone on the transformers, bronze on the chassis

20261017_204830.jpg

Bottom view

20261017_204707.jpg

Chassis is recycled, it already had the meter hole so I just put a meter in it. Its 1mA, but across the 1 ohm cathode resistors its 100mA full scale. Selector switch lets it read bias current on each tube across the 1 ohm cathode resistor. Row of 4 pots closer to the front are output section bias, the other two are AC balance

Bonus points if anyone can tell me why the colors on that selector switch are in the order they are.

Iron topside is Fisher TA800, chokes underneath are from the parts bin. Somewhere I have notes on their particulars, but its not in front of me. Supply is a voltage doubler with an LCRC filter arrangement on the high side. Screens feed from the center point with an LC arrangement.

Bias supply is also a doubler, the stock Fisher setup didn't give me enough to work with.

The extra heater winding is bucking the primary.

Amp schematic

20261018_190835.jpg

Nothing radical here, floating paraphrase inverter with fixed bias pentode output. At some point I may change the screen to get more voltage, or swap this to a split load inverter but this is where it sits now. Also may consider 6GF7, its a triode with the same high mu section and a low mu that shares the pinout, other than not using the screen pin.

And now for the problem that hopefully someone can steer me towards fixing

With the feedback connected, the output of the voltage amp starts to seriously flat top by the time it hits 20v peak to peak. This happens at 1.5 volts peak to peak at the grid. DC at the cathode is 1.9 volts. I don't think its a DC bias problem because I can easily get 60 volts peak to peak if the feedback resistor is unhooked and connected to ground to maintain the same sort of operating conditions. Essentially the same in both channels so I don't think its a single component problem, but quite likely some design or construction problem on my end.

If anyone spots the cathode connection problem in that underside Pic, I already fixed that. Initially the two were tied directly together instead of having the 100R splitting them.

Also pay no attention to the slightly dark resistor in the power supply. Its not overloaded, I slipped with the meter probe and it was in the blast zone.

I also did try a white LED that conducted about 2.9 volts, and it started flattening at an even lower AC voltage.

I walked away from it for the night, hoping someone may have a thought on where to go from this point.

Feedback is just barely over 10dB, plan was for more but I figured this would serve for basic sanity tests.
 
Register to hide this ad
Power supply, got lazy with the bias stuff but there are 4 sets of pots, resistors, and caps after the doubler
 

Attachments

  • 20261018_222514.jpg
    20261018_222514.jpg
    136.7 KB · Views: 21
Really not interested in plate to plate feedback on this. I'd prefer a standard global loop setup. I think I'm doing something wrong that probably ought to be obvious but at some point looking at a problem for too long makes it hide.
 
Does bypassing the LED (>100 uF) help? 1K feedback passes about 15 mA (AC) at full output. There's about 6 mA at idle.
 
I can try that. I know its shunting enough current to dim the LED a bit.

I did confirm that it works when just tied to ground, so the DC conditions at least seem acceptable, it just doesn't like the AC part of it. I was wondering if maybe the AC signal at the cathode + the AC signal at the grid are actually hitting 0 volts, but didn't think of that until I decided to quit poking at it for the night.

not exactly sure what to do with that info though. I guess conventional resistor bias to get more voltage at the cathode, or higher plate voltage and run a higher volt LED to get to the same place.
 
Plate dissipation can be set a little high without issues using these tubes.
Mine runs at 14.7W as per the amp design.
 
I haven't gotten quite that far. Right now just for initial testing its at 20mA per the meter, which looks to be 18mA measured across the 1 ohm resistors so I guess I need to do a bit of tweaking there. Details, not critical just now. Need to get the driver to function correctly first.

the annoying part about the meter thing, I tested it by running my DVM as an ammeter in a circuit off the power supply with a 1 ohm across it and it tracked very well. Now its slightly off if I'm reading millivolts. Makes me wonder if my DVM is off in either amp or volt mode by a small amount.
 
Use a different DMM and see what you get? I had seven of them but I'm down to four due to giving a few away to my sons. I've used several in the past to compare reading with various things like bias, etc.
 
Don't have another that reads millivolts unfortunately. I have a few that can do milliamps. Its an easy enough thing to rig up for testing, 1 ohm resistor, however many meters I want set to current, a load of some sort, and a power supply. See what the milliamps are and then see what the millivolts across the resistor are, something is going to disagree, though my luck all of them will disagree and then I'm still at square one.
 
Tom Bavis said:
Does bypassing the LED (>100 uF) help? 1K feedback passes about 15 mA (AC) at full output. There's about 6 mA at idle.
Click to expand...

That actually fixes the problem. Lets it wind up to a bit over 20 watts and its got some HF oscillation stuff going on but it certainly deals with the drive issue. Thinking its just too much current shunting through the feedback resistor to make the LED an effective low impedance path to ground?

if so, I guess its resistor + bypass cap time. Easy enough fix. I laid this out in such a way that conventional bias on the front end is not difficult.
 
heh 680 ohms with a 220uf / 50v cap (its what I had) per channel both fixes the nasty flat-topping thing, AND the HF oscillation thing in one shot. OK, screw the LED, back to old school.

I was actually thinking that it would be funny if the "modern" part of this amp turned out to be the problem, and sure enough. The one thing that I did that couldn't have been found in any amplifier design circa 1960 and its what didn't work.
 
Got the tuning and all that fun stuff done, need to do a full documentation workup and make a few updates to the schematic but it doesn't suck. Once I do that I'll take the schematic to work and scan it so its actually readable. Didn't realize how bad it would get mangled between downsizing and my phone not being great.

big thanks to @Tom Bavis for pointing me in the right direction. Still amusing, the part of the amp that I was sure would be the easiest, but also the one I'd never used before is what fouled me up.
 
Nice! Those tunes sohnd pretry good. Even if they do roll off at 40hz.
I wonder if the few other drop in replacement tubes, although would have a lower output, would work in your PP amp like they will in my SE amp?
I'm glad you built a PP version. I was wondering how well one would perform. A fairly simple circuit too having voltage gain and output in one bottle.
 
The other one I'd like to try is the 6GF7 twin triode, mostly because its the same pinout other than N/C to the screen. There is one other triode/pentode, forget the number offhand, but its lower dissipation and voltage so not entirely sure how well it would behave. I'm at 400 on the plates, or near enough.

The small triodes are not perfectly identical, but so close that I don't think it would have any problems subbing one for the other.

I'd have to see what the bias requirements are for the output section, possibly would have to make some changes to get enough adjustment range. There is about -65v available, but the actual adjustment range is something like -25 to -35 as it sits. I'm using 5K pots because its what I had 4 of. If it needs more range I can always change the pots to a higher value. I'd prefer a shorter shaft pot anyway but junkbox build. If you look close you may even notice the pots aren't the same. Two AB and two Clarostat pots.
 
Great build thread @gadget73 !

Tom Bavis said:
Does bypassing the LED (>100 uF) help? 1K feedback passes about 15 mA (AC) at full output. There's about 6 mA at idle.
Click to expand...
So if I am understanding this correctly, the LED was not the excellent free "AC short" it's always made out to be- dV/dI is a resistance, and the change in current from idle to high power (6 to 15mA) with this lower mu triode resulted in too high resistance here for the unbypassed LED. Funny that the big value add with this technique is supposedly being able to eliminate the bypass cap. I can see how this might work somewhat better for a very low bias current 12AX7, etc, but I think it's still ugly. I've never been a fan of using a diode for this- As @gadget73 has mentioned before, it's hard to find a good LED that operates well at very low bias currents, and here, where you have a decent current it still does not operate well.

1792525689723.png
 
Last edited:
gadget73 said:
The other one I'd like to try is the 6GF7 twin triode, mostly because its the same pinout other than N/C to the screen. There is one other triode/pentode, forget the number offhand, but its lower dissipation and voltage so not entirely sure how well it would behave. I'm at 400 on the plates, or near enough.

The small triodes are not perfectly identical, but so close that I don't think it would have any problems subbing one for the other.

I'd have to see what the bias requirements are for the output section, possibly would have to make some changes to get enough adjustment range. There is about -65v available, but the actual adjustment range is something like -25 to -35 as it sits. I'm using 5K pots because its what I had 4 of. If it needs more range I can always change the pots to a higher value. I'd prefer a shorter shaft pot anyway but junkbox build. If you look close you may even notice the pots aren't the same. Two AB and two Clarostat pots.
Click to expand...
I used 5K/8 OPT's in my SE amp. There is one other tube thats exactly the same as the 6LR8, but has a plate cap instead of the plate being connected to a pin.
I thought about using them when I built my amp since they were even cheaper than the 6LR8's at the time. Around $2 each compared to $5 each for the 6LR8's.
 
wparks said:
Great build thread @gadget73 !


So if I am understanding this correctly, the LED was not the excellent free "AC short" it's always made out to be- dV/dI is a resistance, and the change in current from idle to high power (6 to 15mA) with this lower mu triode resulted in too high resistance here for the unbypassed LED. Funny that the big value add with this technique is supposedly being able to eliminate the bypass cap. I can see how this might work somewhat better for a very low bias current 12AX7, etc, but I think it's still ugly. I've never been a fan of using a diode for this- As @gadget73 has mentioned before, it's hard to find a good LED that operates well at very low bias currents, and here, where you have a decent current it still does not operate well.

View attachment 3860426
Click to expand...


I suspect the problem was related to the second current path through the feedback resistor. It just didn't have enough current to be sufficiently low impedance with the 1K in place so it didn't end up bypassing correctly. Possibly could have fixed it with a big coupling cap but the size required probably means electrolytic, and I suspect overall that would be a worse situation than just a conventional bypassed resistor.

I actually need to re-check the feedback level now that I think about it. I had it at 10.3 dB but thats with the LED. If that wasn't sufficiently low impedance that value may be wrong.
 
You must log in or register to reply here.
Back
Top Bottom