• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Question about building an amp without a preamp stage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
So I'm starting my Push-Pull 6L6 build (transformers FINALLY arrived... its like christmas :D). I am using a design from Jason Thorpe on AudioKarma (thread can be found here Looking for schematic for nice clean push pull 6l6 monoblock )

I would like to omit the 12AX7 preamp stage as the last thing I need is more gain in my system. My 12AU7 preamp (eventually I'll post a thread about how that turned out) can drive just about anything.

My question is this: How would I go about redesigning the feedback circuit to work with just the 6SN7 phase-inverter as an input? Would I just connect it to the grid of the first triode? I know I'll probably have to rework the resistor/capacitor values in the feedback circuit, but I planned on that anyway since I'm using a different brand of transformers.

I'm also designing an 807 amp (first time designing my own schematic from scratch), and the same issue will apply, though I'm planning to use a single 6J5 for the phase inverter instead of an LTP (Kind of curious to see how it compares). I also plan to use a 10M45 for a CCS on that 6J5. I'll post that schematic when it's done tomorrow.

Thanks for the help, you guys have been very helpful in learning this hobby. BTW, any recommendations for useful books and/or raw information on designing tube amps?
 

Attachments

  • ST60-UL.pdf
    60 KB · Views: 182
Last edited:
Look up Eli Duttman's El Cheapo or Gingertube's Baby Huey for examples of LTP front-ends without preceding voltage amplifier.

You'll probably want to re-jig the LTP to have the grids at ground potential (likely will require a negative rail for the LTP tail).
 
Last edited:
This is ever so slightly off topic of the original post. I suppose I'm hijacking my own thread.

This is an 807 push-pull amp design. This is my first crack at actually designing an amp, rather than just slightly tweaking someone else's design. Of course, this sort of design has been done a million times, but I figured I'd see what I can come up with using cheap tubes (807 and 6J5).

Output transformer is undecided. Once I get my current 6L6 build done with the Edcor transformers I'll decide if I want to use them again. Otherwise I may just give Heyboer a call and see what it would cost for them to wind something up for me. I seem to remember that Dynaco made a transformer with a 6600k primary. Maybe they can wind something like that. I have a feeling that Magnequest might be a bit out of my price range for this.

Power supply design is next, I'm not 100% sure what PT I'll use either.
 

Attachments

  • 20180606_112648.jpg
    20180606_112648.jpg
    723.9 KB · Views: 459
H713,

Comments on your 807 schematic:

I like the idea of a simple 807 push pull amp. But here are a few considerations:

Bias A and Bias B should go on the Left side of R5 and R6. Grid stoppers are supposed to be the only thing connected to the control grids (and right at the socket not further away).

Your 6J5 circuit will not work. I think you were trying to make a concertina with a current source in the cathode.

If you want to use 6J5 tubes, then use 2 of them, with common cathodes and a current source in the cathode circuit. You will need to have a negative power supply to power the current source (as tikiroo said). Use -20V to have some headroom for the IXYS current source, it does not really work well with less than 10V across it. Use identical resistors in the 2 6J5 plates, 27k will work, and connect the coupling caps to the plates. Or use a single 6SN7 (dual 6J5). The phase invertor gain will be a little less than 10. You will not be able to use very much negative feedback. Apply the signal to one grid, and apply the negative feedback to the other grid, like jazbo8 said.

Your negative feedback as drawn will have wild variations as you change the volume control. Not a good idea.

The 807 maximum screen grid rating is 300V in pentode mode. The 807 maximum screen grid rating is 400V in triode mode. Ultra Linear is different.

I would recommend screen voltage of 350V at most in Ultra Linear (the plate voltage will swing 2.5 times faster than the screen voltage, and the screen will draw way too much current then as the control grid voltage heads toward 0V, and the plate voltage swings close to the cathode voltage). To keep the screen voltage low in Ultra Linear, you would need to reduce the B+ to 350V. But that would not give as much power.

You could use pentode mode, and set the screens to 250V or 300V. Then you could use the 410V B+ for the plates. You will need to use negative feedback in pentode mode. There is lots of gain in pentode mode, so adjusting negative feedback will be paramount.

Or you can use triode mode, 400V B+, and tie the screens to the plates with 100 Ohm resistors. You can do this, even if you do not use negative feedback. Triode mode will have less power. But this is just about as simple as push pull gets.
 
Last edited:
  • Like
Reactions: 1 user
What would be the advantage of doing it that way?

The schematic shows direct connection from the voltage amplification stage to the LTP, so LTP bias is set by the DC voltage on the ECC83 anode to 125V, with 130V on the cathodes of the LTP. If you don't have the ECC83, you'll have to bias the LTP another way. You could just set up a voltage divider to supply 125V to the LTP grid but this would require DC blocking capacitors on the input and feedback, and is not how you'd normally go about it. Better to lower the supply voltage to the LTP by 125V, have the grids tied to ground through resistors, and send the tail of the ccs to a negative supply (as discussed by 6A3sUMMER).
 
A few years ago I built a PP amp using the 1625. I used Hammond 1650H (not HA) outputs, in ultra linear mode with 400V to the center taps. Output power was 22 watts, both channels driven. Later I converted that same amp to pentode outputs with a 285V screen supply. Power increased to about 25 watts (both channels driven).

I lean more to pentode outputs these days. I can't see a lot of downside over UL, except a slightly more complex power supply and the requirement of application of a bit more feedback, which not all output transformers may be able to handle, depending on how they are wound.
 

PRR

Member
Joined 2003
Paid Member
...omit the 12AX7....

The amp claims 0.44V input sensitivity.

The 12AX7, here, is giving gain around 25. {Edit: sheet says 33.}

For the same NFB factor, the input sensitivity will now be 0.44*25= 11 Volts. {Edit: 0.44*33= 15V.}

That is a heck of a high level to expect from your control unit. Higher even than Hammond used on lines that ran the long way around large churches.

_I_ would *first* just build it as-drawn. Don't muck-up a working plan until you try it. Just turn-down the input pot. (Note that it is Audio taper- someone expected large turn-down.)

Then.... experiment with the 6.8K. I bet it can be 3.3K. It might go 1K without howling. At some point this hi-gain loop is sure to howl. So you want a proper test-rig with dummy-load and wide-range audio voltmeter or 'scope. If it is stable at say 2K, then gain is 21, and if that is "too much" for your system you could maybe look at the rest of your system and find out why you have a gain-flood.
 
Okay here's a second try at the 807 schematic.

If I were to stick with the cathodyne phase inverter using a 6J5, does this make slightly more sense? I realized I had no grid biasing in the last schematic... probably not ideal.

Feedback circuit was moved to a more logical place. For testing I'll probably use a 100K pot instead of that 10K resistor and adjust it to where it seems happiest. I adjusted the plate voltage of the 6J5 slightly as well.

Those unlabeled grid stoppers are 2k. I was planning to have four bias controls (one per tube) instead of using a bias-balance system. Thoughts?

The 50K volume pot may be omitted. I don't anticipate it seeing a lot of use.

As far as the screen voltage on the 807s, that was the reason for the 1.5K screen resistors. Would this not work?

This may still have plenty of issues with it, so come at me. Also, I hope it's semi-readable. Drawing this up in a schematic editor is on my list of things to do.
 

Attachments

  • 807-2.jpg
    807-2.jpg
    691.3 KB · Views: 160
You need to capacitor couple the signal from the input potentiometer or else the input will be at a high voltage, since the concertina is self biased...

Also- your feedback scheme will not work as shown, you will be leaving high voltage from the top of the cathode load to the transformer secondary, effectively shorting out the cathode load with the feedback resistor, as they will appear in parallel more or less... It may work if you bring feedback into the grid of the concertina, however.

Edit-

Perhaps a better idea would be to ground the 4 ohm tap of the secondary, and run a very-low-value feedback resistor to the cathode of each output tube from either end of the secondary (remember, impedance is squared, so 0, 4, and 16 are the end/center/end that are most balanced) this will only work if the secondary is very symmetrical, or if you have dual secondaries (toroids used as outputs sound like a good place to try this...)
 
Last edited:
What if I were to put a capacitor in series with the feedback loop to block that high voltage from the cathode? Edit: Wouldn't connecting the feedback loop directly to the grid of the 6J5 pull down the grid voltage and change the bias of the tube?
 
Last edited:
Yes, it would need capacitor coupling to work either way, but putting feedback where you have shown will unbalance the blameless/perfect balance of the concertina either way.

The only usable input for feedback in this situation would be from output to input-grid, and it will need to be AC coupled to work, unless you go to a bipolar power supply of some sort... It could work putting feedback into the grid if you were to run a negative supply such that you have 50-60% of the total rail-to-rail voltage across the tube, and the remainder split evenly across the load resistors. something like 290 volts supply, with a -90 volt rail would probably work, with 90-95 volts across each load, and ~200 or so across the 6J5. The trick is to keep each of the plate loads at 20-25% of the total rail to rail voltage for max swing with low THD... Running output to grid feedback would work in this situation, so long as you keep the input large enough to keep input impedance high, although this may require a pretty big input and feedback resistor, which may or may not contribute to noise... If driving this circuit with a strong driver that can push into low impedance it's not an issue, but it really all depends how you want to go...
 
I'm not opposed to a preamp stage. I just want to avoid the issue I have with my Heathkit amp (AA-121) where RF noise is a real issue and I have my preamp at like 1/4 volume...

And it would be fairly easy to design a preamp stage with a 12A_7, whichever seems fit.
 
Do you already have multiple 6J5 to use?

What's the output level of your 12AU7 preamp? If it's hot enough then a simple 6J5 input/gain stage feeding the concertina would be a good way to go, with feedback to its cathode at the input tube. That, or a single 6SN7 per side in place of the 6J5 pair would be a good solution too. Maybe even a 12AT7 LTP standing on a CCS with feedback to the non-driven grid. Plenty of ways to go.
 
I'd have to check. I can say that cut down on the gain significantly, however under testing I remember that I could get it to output like 25-30 volts before clipping. That said, I was using a pretty serious input for that. I can increase the output of that preamp without any trouble by removing a couple resistors that I used to attenuate the input signal by like 50%. I believe it has a gain of 10 as it is, and it could be increased to 20 without any trouble.

As of right now, I do not have any 6J5s, though they are relatively inexpensive. With this design I wanted to stick with older tubes. I was probably going to use the metal bodied 6J5s, paint the chassis grey and add plate current meters to make it look like an old transmitter. Just an idea though. Suggestions for other input tubes are welcome. I was also going to use some old TV dampers as rectifiers.
 
Remember that the split load inverter offers no gain. It actually has a gain factor of slightly less than unity, at about 0.9 or so. Thus to drive your current design to full power (assuming you fix the other gotchas discussed above) will require about 25V input! And that's before you add any global feedback. You really must add a voltage gain stage before the phase inverter, even if your preamp can pump out 25V undistorted.

If you have a second 6J5, that would be great to use for that purpose, or you could use a single glass envelope 6SN7. Either way, you need a voltage gain stage. But it could be made out of pretty much any low mu type tube you want. You can direct couple that voltage gain stage to your split load inverter so as to eliminate the inverter self biasing if you choose.

You can then bring your global feedback to the cathode of the voltage gain stage. This is normally what you'd see in this type of topology. It works extremely well, and will give you a solid performing product if engineered correctly.

Screen resistors seem overly large to me. You can put 100 ohm or 150 ohm resistors in place of your current 1.5K resistors and be just fine.

It's difficult to know the correct value of the phase compensation capacitor in your feedback circuit without running some tests on the actual amp, because the value of that cap is highly dependent on the high frequency characteristics of the output transformer.

I've done the individual bias arrangement for output stages and it works fine. However, my personal preference is a bias/balance type setup because it can sometimes be easier to set up your output stage. For example, with a bias/balance setup per pair of tubes, you can (mostly) set the DC balance between tubes in the push pull pair, and then independent of the DC balance, adjust bias up or down to suit your tastes. This is easier to do this way vs. separate bias adjustments for each tube. Basically to adjust bias this way, you need to adjust only two pots instead of four. But ... it's not that critical if you don't mind the extra effort.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.