Subminiature Conversion

[drewspriggs]

Howdy,

Love using subminiatures and while I've got a couple of scratch-builds under my belt, I'm looking at emulating a couple of the most classic amp topologies with them. While the power amp situation is obviously different, my focus at the moment is just getting the pre-amp as close to the 'original' as I can.

Pretty much everything I'm looking at copying was built arund 12AX7's, and I've got stacks of 6112 subminiatures. I know these tubes are obviously very different, but I figure with a bit of playing around I can get (somewhat) close. Everything I've seen others do have just been replace one with the other, lower B+ down a little and leave it at that but I think it should be possible to get closer.

Is something like changing grid stoppers and making a lot more effort to match frequency response with Ck worth it - or am I just massively overthinking things?

[katopan]

I went through this with my submini version of an 18 Watt Superlite TMB. Started doing what everyone else had done, just dropping a 6112 into the existing circuit values as you say.

What I found lined up with the plate curves on the datasheet - noticeable lower sensitivity (and therefore gain) than a 12AX7 even in the context of going into the more sensitive 5902 power stage instead of EL84s. I experimented to find the biggest plate resistor value for max gain before it dropped back down again. Then I adjusted the cathode resistors to bias at the same point between clipping and cutoff as the "real" amp. Ended up with vastly different resistor values but a much closer representation of the full size amp. Even then the gain of each stage was still a little less than a 12AX7, but close.

[TUBEDUDE]

Did you socket them, mount the tubes on a strip, or run the tube leads to their respective connections? I have a small pile of pencil tubes I've been meaning to use and am curious about mounting and potential problems. Thanx.

[drewspriggs]

Did you socket them, mount the tubes on a strip, or run the tube leads to their respective connections? I have a small pile of pencil tubes I've been meaning to use and am curious about mounting and potential problems. Thanx.

For both 5 and 8 pin tubes I cut the leads about 1/4" long and solder them into a row of strip or a DIP8 socket

I went through this with my submini version of an 18 Watt Superlite TMB. Started doing what everyone else had done, just dropping a 6112 into the existing circuit values as you say.

What I found lined up with the plate curves on the datasheet - noticeable lower sensitivity (and therefore gain) than a 12AX7 even in the context of going into the more sensitive 5902 power stage instead of EL84s. I experimented to find the biggest plate resistor value for max gain before it dropped back down again. Then I adjusted the cathode resistors to bias at the same point between clipping and cutoff as the "real" amp. Ended up with vastly different resistor values but a much closer representation of the full size amp. Even then the gain of each stage was still a little less than a 12AX7, but close.

Do you happen to have any info on what values you used? I did some modeling and got wild values which I discounted, but maybe they were right after all

[katopan]

I also socketed them. Soldered the submini into a DIP-8 machined pin socket. I made a PCB for my amp, so it also has DIP-8 machined pin sockets for the valve locations. Then the submini socket plugs into the PCB mounted socket matching orientation marker.

My submini build was 16 years ago. I'll dig up the details later and let you know tomorrow what plate and cathode resistor values I ended up with. I wrote up a build thread on the 18Watt forum.

[drewspriggs]

I also socketed them. Soldered the submini into a DIP-8 machined pin socket. I made a PCB for my amp, so it also has DIP-8 machined pin sockets for the valve locations. Then the submini socket plugs into the PCB mounted socket matching orientation marker.

My submini build was 16 years ago. I'll dig up the details later and let you know tomorrow what plate and cathode resistor values I ended up with. I wrote up a build thread on the 18Watt forum.

Ahhh I knew I recognized that username - have seen the thread, but all the images are gone

[katopan]

Through some breadboarding, I found a 220k plate resistor gave me the most gain out of a 6112 preamp stage. Anything bigger and gain dropped off again despite what the plate curves said in theory.

Then I tested bias point between grid clipping and cutoff and got the following values of cathode resistor relevant for the stages in the Superlite TMB preamp:
6112 220K/3K9 = 12AX7 100K/1K5
6112 220K/2K2 = 12AX7 100K/820R

As I said the stage gain relative to the next stage input sensitivity was still a little bit lower than 12AX7s at normal preamp B+ voltage, but it was very acceptable and as close as possible using 6112s.

I also changed the resistor values in the PI to get as close to the Marshall 18 Watt PI operating points and gain.

If you just use the standard values from the full size amp you are copying, the gain structure will be noticeably lower and the bias point between clipping and cutoff also wrong. This is a mistake made by many.

[katopan]

Here are the final schematics including PI/PA VVR.

[katopan]

And some pics. 16 years ago, I still have the amp and it sounds great.

[katopan]

Guts and size comparison.

[drewspriggs]

Through some breadboarding, I found a 220k plate resistor gave me the most gain out of a 6112 preamp stage. Anything bigger and gain dropped off again despite what the plate curves said in theory.

Then I tested bias point between grid clipping and cutoff and got the following values of cathode resistor relevant for the stages in the Superlite TMB preamp:
6112 220K/3K9 = 12AX7 100K/1K5
6112 220K/2K2 = 12AX7 100K/820R

As I said the stage gain relative to the next stage input sensitivity was still a little bit lower than 12AX7s at normal preamp B+ voltage, but it was very acceptable and as close as possible using 6112s.

I also changed the resistor values in the PI to get as close to the Marshall 18 Watt PI operating points and gain.

If you just use the standard values from the full size amp you are copying, the gain structure will be noticeably lower and the bias point between clipping and cutoff also wrong. This is a mistake made by many.

Those are similar to the values I got off of modelling, which is perfect - thankyou so much! The first amp I'm working off has 100k/1k5 on the stages so that works out nicely.

[drewspriggs]

Here are the final schematics including PI/PA VVR.

Thought your idea of a quadrupler was fantastic, so I've come up with a little PCB using a couple of the new MeanWell N78 series switchers - giving me a 12.5v heater rail for power tubes, and a 6.5v rail for the preamp tubes of what I'm building. Incredible little pieces of kit; will take anything up to 30v and give you regulated positive (or negative using the same part) rails up to 2A, requiring no heatsinking, and fully EMI compliant with zero external components.

Absolute genius!

[katopan]

Thanks. Floating the heater supply at half the 170V B+ was inside the submini valve cathode to heater voltage spec and meant I could get B+ and heaters from a single 0-30V transformer.

Keep us posted with your build!