It's spring--time to build an amp

[kward]

My son asked me to build him a "compact" integrated amp. Compact to me means no larger than maybe 12 x 14 or equivalent square inches. He will use it mostly to drive some speakers for his computer and TV. I've got some Fisher 400 output transformers, so the output stage will use either 7868 or 7591 tubes.

For the frontend, I've got some nice samples of the 6U8A I'd like to use, so one option will be a straight-up pentode voltage amplifier fed to a split load inverter like Scott and many others have done. This would certainly fit the "compact" requirement.

But another option I'm considering, maybe not so compact, is to run the pentode section of the 6U8 in triode mode for the voltage amplifier, and use the triode section as maybe a poor-mans constant current source or as a cathode follower (gotta use the triode section some how). Then run the signal into a differential pair phase splitter.

The amp must support enough gain so that no preamp is needed. Certainly the "Scott frontend" approach I know will give the gain needed even after 10 dB or so of feedback is applied.

Does anyone have triode curves for the pentode section of the 6U8? I need to know things like mu, transconductance, where it's most linear, and plate impedance.

Comments on the overall approach certainly also welcome.

 

[nerdorama]

Which 400 transformers? The early units have a traditional primary impedance for 7591's while the later units with p/n ending with an X have around a 10k primary. This is all according to Dave Gillespie from his 400 EFB thread. Probably a good read if you haven't already seen it as this issue significantly effects the operating point.

I agree with your title. I'm having the same amp building urge.

Have fun,
John

 

[Scuzzer]

But another option I'm considering, maybe not so compact, is to run the pentode section of the 6U8 in triode mode for the voltage amplifier, and use the triode section as maybe a poor-mans constant current source

I thought a pentode was a better CCS than a triode. Either way it sounds interesting and I'd be interested to see the schematic you come up with. BTW, November is the time I usually the feel itch to build an amp. Weather is turning crappy, snow falling, time to settle in with a project.

I also just "fixed" an amp with 6U8As in it. Although everything is running correctly it sounds kinda spiteful, as in it just doesn't like music. Nothing against the tube here, just the implementation.

 

[zandru]

Bigger! Faster!, and the Bold use of interstage Transformers………..

 

[kward]

T1020-116-1AX. 10k primary, measured and verified. Yes I have read Dave's thread. And if I have room I thought I might throw in some EFB, too. Just looked in the box and remembered I also have the Fisher 400 power transformer! Could do a Fisher 400 clone with that...

 

[kward]

I thought a pentode was a better CCS than a triode.

I think you're right. I was thinking the pentode strapped as a triode may offer more gain though than the other triode section. Purely guessing at this point, but I believe a mu of about 30 to 40 is all I would need even with some feedback thrown in.

 

[Scuzzer]

I've been out of the tube game for a couple of years but I remember (perhaps incorrectly) that the high plate resistance of a pentode was the primary reason it made a better CCS than a triode. I vaguely remember sketching out a schematic with a 7199 as a voltage amp/CCS. I only bring this up to give you a few more options on your build. I have a hard time building stuff to a recipe (schematic) simply because I like to try and figure out new ways to pound the square peg into the round hole.

 

[jazbo8]

Does anyone have triode curves for the pentode section of the 6U8? I need to know things like mu, transconductance, where it's most linear, and plate impedance.

Here you go... let me know if you need to have it plotted differently.

 

[dcgillespie]

K -- I had a strikingly similar project a number of years ago for my son that I had posted about here:

http://www.audiokarma.org/forums/showthread.php?t=553403

It is based on the transformer set from a KX-200, but uses 5881 tubes instead of 7591s. I offer it to give some context to the idea of "compact". Still gets many hours of use daily in my son's home.

You've got a neat project ahead! It's always special when you're building from scratch for family. I know yours will turn out great. Good luck with it!

Dave

 

[kward]

Here you go... let me know if you need to have it plotted differently.

Perfect Jasbo8! That's exactly what I need. As I suspected, mu about 30. I think that will work beautifully for what I'm contemplating.

Bigger! Faster!, and the Bold use of interstage Transformers………..

One day I really would like to try this....

K -- I had a strikingly similar project a number of years ago for my son that I had posted about here:

I do remember seeing that thread a while back. Yes, I'm afraid I'm bit with the same complex--I will end up putting as many gadgets in the thing as possible!

 

[nerdorama]

I've used IT's and really like them. Have also tried grid chokes on the output tube grids and they work a treat as well. They seem to make for a more powerful sounding amp to me and seemingly more dynamic.
John

 

[kward]

I think I have a build plan now. First off, I feel it would be a crime NOT to build a clone of the Fisher 400 power amp section, given that I have both the output transformers and the power transformer, so that's what I will do. I will use Dave's Fisher 400 modifications as presented on his thread in the Fisher forum, including the addition of EFB. However, I will probably add DC balance controls also.

The power amp section has sensitivity of 1.9V AC (according to Dave's thread). That's not enough gain by itself to drive the output stage to full power from standard line sources, so I will need to add an on-board preamp. If I want a sensitivity of .28VAC (that's what Dave's mods offer), then I need the preamp stage to have forward voltage gain of 6.78.

I've decided I'm going to use the 6U8 tube for the preamp section. The pentode section of the 6U8, strapped as a triode, has mu of 30, and used in common cathode topology given the rail voltage and bias point I've chosen, gives open loop stage gain of 15. I need to reduce that stage gain by a factor of 2.21 to get amp sensitivity to .28VAC, so I need 6.89 dB of local feedback in the preamp section.

The preamp will invert signal polarity, so to "right it" again, I will need a unity gain inverter. That's where I will use the triode section of the 6U8. A unity gain inverter built from a triode has a gain of about 0.9 (or about -0.915 dB), so to keep the amp sensitivity at .28VAC, the local feedback factor needs to be reduced to 5.97 dB, or just round that to 6 dB straight up.

I will probably be able to direct couple the preamp stage to the unity gain inverter, then after the unity inverter, place a coupling capacitor, then a volume pot, which will feed into the first 12AX7 stage of the power amp. To keep Miller at bay, I will restrict the volume pot to 100K in size. A 100K pot means a 0.33 uF coupling cap, for a 5 Hz cutoff frequency.

Also little fingers (children) like to flip switches a lot, so I am thinking of adding some sort of high voltage delay circuit that will reset the count down every time the power is cut.

So that's the basic topology.

Next is to put some hard numbers to paper to validate all of these assumptions, then if that works out, draw up the schematic.

Feedback or observations on all of this is certainly welcome.

 

[kward]

Not the build plan I thought

My idea of compact was basically not compact at all, so I needed to rework the design into fewest tubes possible. This is what I came up with. It's got a Stromberg Carlson ASR-433 frontend, a Scott (ish) phase inverter, and a Fisher 400 output section with Dave Gillespie's mods. So there's nothing on this amp that is really my own!

I've not ever used the 6U8 tube before, which is why I thought I would follow the ASR-433 on this one. I'm hoping the Stromberg frontend works well, and it's the piece I'm most concerned about. I'm not sure how much gain that frontend will produce, since I don't have plate curves for exactly the screen grid voltage used. I expect it will be enough to drive the output stage to full power with standard line sources, but I'm not sure.

Also, while the schematic uses Dave's EFB that is almost stock from his Fisher 400 thread, I did tweak it just a wee bit to set the operating point of the bipolar transistors on the EFB board right at 1.5 mA under "average" biasing conditions (-15V).

Okay, any feedback on this? I think it's pretty solid but I might have missed something. Mostly concerned about the gain of the 6U8 pentode.

 

[dcgillespie]

K -- Looks good! -- But the top diode in the bias voltage doubler rectifier circuit is backwards. You should be able to get plenty of sensitivity out of this format, but if you maintain 20 db of NFB, you'll likely need to lower the screen voltage (raise the value of the screen dropping resistor) of the AF amplifier stage, and bypass the pentode's cathode resistor as well. You're using R/C coupling to minimize disturbing the inverter's operating point? With a pentode gain stage, that's probably a good idea, minimizing any sensitivity to tube selection.

Good luck with it!

Dave

 

[peterh]

A few things :
-the G2 could benefit connected in a voltage divider ( i.e. add a resistor to ground)this will stabilize the voltage better.
- the ac balance is moot. remove and use equally sized ( 22k) on anode and cathode.
( a concertine has no choice then split equally, if power tubes have differing Gm, replace them with matched pairs)
-add a 100ohm5w resistor across the OT br/bl , this protects the amp from open output condition

 

[kward]

K -- Looks good! -- But the top diode in the bias voltage doubler rectifier circuit is backwards. You should be able to get plenty of sensitivity out of this format, but if you maintain 20 db of NFB, you'll likely need to lower the screen voltage (raise the value of the screen dropping resistor) of the AF amplifier stage, and bypass the pentode's cathode resistor as well. You're using R/C coupling to minimize disturbing the inverter's operating point? With a pentode gain stage, that's probably a good idea, minimizing any sensitivity to tube selection.

Good luck with it!

Dave

Thanks. I would have caught that backwards diode eventually .

Are you saying lowering the voltage on g2 of the 6U8 pentode will give the stage more gain? I was understanding that raising the voltage on g2 will raise the stage gain (given everything else equal) because higher voltage on g2 increases the tube's transconductance at the bias point, but I might have this understanding backwards somehow.

Yes, I capacitor coupled the stages exactly for that reason, to minimize the differences in 6U8 samples on disturbing the inverter's operating point.

I'm not sure how much feedback I will apply. I tend to like about 12 dB, but I will experiment up to Fisher's recommended 20 dB.

A few things :
-the G2 could benefit connected in a voltage divider ( i.e. add a resistor to ground)this will stabilize the voltage better.
- the ac balance is moot. remove and use equally sized ( 22k) on anode and cathode.
( a concertine has no choice then split equally, if power tubes have differing Gm, replace them with matched pairs)
-add a 100ohm5w resistor across the OT br/bl , this protects the amp from open output condition

Yeah, one of the H.H. Scott schematics I looked at (299D maybe) uses a voltage divider on g2 as you suggest. This is exactly the Stromberg Carlson ASR-433 frontend. Thanks for your ideas. I'll consider them.

 

[kward]

First bit of real progress

I've cleaned up the transformers and painted them. They did have some rust on the end bells bubbling up under the old paint that took a while to sand off and refinish. The color scheme I've chosen for this amp is black and silver as you can now see. I used high temp automotive paint.

Now the harder part is to figure out what to do for the chassis and get that ordered.

 

[marantzfan]

Nice! I like your idea of mixing scott, fisher and stromberg together. Curious to see this build progress. I really enjoyed watching your last one!

 

[dcgillespie]

K -- Sorry I missed your post #16. A couple of points. Looking back at your schematic again, your screen dropping resistor is 1M, which is fine. For some reason, I thought it was originally much lower in value, but I likely needed to look closer......

As for your comment about Gm relating to gain, Gm is a function of plate current, and is therefore very dependent upon the operating point chosen. Within a broader context however, if Gm were the sole determiner of gain, then it would be impossible for a 12AX7 to have more gain than a 12AT7 would, wouldn't it.......

Voltage gain on the other hand is a function of Mu, which is why the 12AX7 trounces the 12AT7 in the gain department. With a pentode, the screen's voltage relationship to plate voltage (within reason) can act to vary the effective Mu of the tube, which is why pentodes are generally not discussed in terms of Mu.

As a practical exercise however, imagine a pentode output stage, where the screen and plate voltage are nearly identical. Take for example, any 6L6 Fender output stage, where both plate and screen operate at (about) 450 volts. Grid bias is (about) -50 vdc, meaning that a peak-to-peak grid drive will be (about) 100 volts to produce full power output. But what happens if the screen is reduced to 350 volts? Now, to produce an appropriate quiescent current draw, the bias voltage only needs to be about -38 volts -- requiring only 76 volts P-P drive. Yes, the plate load would need to be increased slightly, but the point is, the effective gain of the stage -- as measured against the signal at the plate -- has been increased, primarily because the screen voltage was decreased.

With small signal pentode tubes, the effect is exactly the same. Again, within reason, as the screen voltage more approaches the plate voltage, gain goes down, because more bias is required. As the screen voltage is lowered, bias can be reduced, and gain goes up. Taken to an extreme, where the screen voltage is greatly reduced, and the plate load resistor is elevated to many megohms, the gain of the stage can literally move into the thousands, achieving a form of operation know as "starved circuit" operation.

I hope this helps....

Those transformers look great!

Dave

 

[kward]

Yes, that makes sense. The Fender example helps. I tend to think in mathematical terms first before intuitive terms, which in this case has lead to my misunderstanding. For example the voltage gain equation for a pentode is approximately equal to:

Av = (approx) gm x RL

From that equation, I deduced that if gm goes up, then voltage gain must go up given the same RL. Certainly that must be true because the math doesn't lie. What was not intuitive (because it's not part of the equation) is how screen voltage affects gm. From your example, I am deducing that lowering screen voltage causes gm to rise. That's not intuitive at first glance! So the Fender example you quoted has helped me in that understanding. Thank you!