6L6GC Single Ended amplifier with first attempt OPT

[Mars]

Hi All

While it has become quite a busy season, I wanted an amp to listen to while I work I had an chassis with PT and choke in the workshop, so I thought I'd try out the OPT I wound a few months back for a client, who changed his mind and went for the push-pull amp.

I only had a pair of 12AU7 in the workshop, so I mounted them as a SRPP with 250V on the top plate and 10mA running through each triode.

The sound was promising, and my companion of the last few months (SANSUI AU101) is up for sale.

But the 12AU7 was just not delivering the goods. I needed about a gain of 30X to drive the 6L6's properly.

BTW the 6L6GC's are connected ultra linear (43% tap) with cathode feedback from the secondary winding on the OPT.

A good friend (Pieter de Kock) gave me some 6DJ8's. This seemed to be the tube for the job. I scanned the net for a mu-follower example and found this from Tubecad site:



I have a SS rectifier C-L-C-R-C power supply. All caps 100uF 500V. I must say that the sound is great.

The speakers are KEF Q floor standers Wessel lent me; very good sounding boxes!

[handsome]

the 'classic' mu-follower takes the output from the upper tube's cathode resulting in a low output impedance plus a very high load for the lower tube. since the upper tube is a constant current source the load it presents to the lower tube tends toward infinite - giving a near horizontal load line for the lower tube - this very high load drastically reduces distortion on the lower tube. however, taking the load off the lower tube's anode can nullify this; the lower tube is now feeding an input resistance much smaller than it's load. the true load on the lower tube is this input resistance (100k in this case)  in parallel with the active load of the upper tube. thus the load on the lower tube has been reduced from very high to rather low (relatively speaking of course) and the low distortion of the active load has been lost. If one is to use the lower output of a mu-follower one should direct couple it to the next stage. but of course if sounds good.........

[Mars]

I was also wondering about the position of the output, but my curiosity made me try the position in the diagram. After giving it a good listen, I shifted the output lead to the cathode of the top tube. First thing was a noticeable improvement in high frequency response. Nearly too much! It is like hearing detail way up to the limit of my hearing. On electronic music this can be irritating. I would like to experiment with different coupling caps to see if the effect changes. Definitely more "hifi". I'm leaving it here for the mo' Taking the output from the bottom triode's anode has a more old school relaxed sound with less sizzle on the hi Hz side. I guess it is similar to a single triode setup with an anode load resistor.   I would still like to go over the calculations of the mu follower and check the resistor/cap values for the current application.

Another thing I noticed is the bass response of the amp. The amp sounds huge! At conversation level I can hear each instrument loud and clear. It is like I'm sitting close to the stage at the side where the bass amp stands.  I'm listening to "Love and Theft" (Dylan) I guess the pentode effect in the ultra linear circuit is showing up. But with the volume turned up, the bass distorts too early. An increased air gap in the OPT might improve that issue.

[Mars]

Here's a schematic of the circuit in its present state:

[ludo]

Thanks for this Karel, and Handsome. Did some sims and it was pretty curious. I think I've learnt something.

I like somewhat weird circuits like this. In part because it's not so common, and in part because every time I have to replace a pair of 5881s in a guitar amp, there's still a working one left... Re-using it feels like replacing brake pads on the left front wheel only, but on a monowheel like this one could give it a shot

OK so I've got brake pads. And 12AU7s to [not] match Now if only I had a wheel...

[handsome]

the next step with the mu follower would be a solid state or pentode current source getting you closer to infinite resistance. with infinite resistance then - in theory - the lower triode should produce no distortion! look up alan kimmel on the web i recall he had some pentode loaded mu follwer circuits that he believed where vastly superior. i built a mu follower preamp a while back and it did sound very nice especially with solid state power amps (cheap NADs and Rotels). however i have always found single ended amps whether triode or pentode sound amazingly realistic. would love to hear some of your OPTs some time Karel will give you a call when the cash flow increases.

[handsome]

i was talking a bit of toddle there...infinite resistance will not produce zero distortion it will simply remove varying anode resistance as a distortion mechanism.....

[Ampdog]

The first circuit gave a constant current load (well comparatively) to the triode, the cathode follower output type a constant current load followed by a cathode follower, i.e. lower output impedance. The distortion will still be a function of the lower triode's characteristics, but is usually better than with a finite (low) resistance as plate load.

Problem with this arrangement is with the greater effect of Miller capacitance on the lower triode, as governed by the input (source) impedance. In the given case there was also an unbypassed cathode resistance, further increasing the lower triode internal resistance, which will begin affecting the circuit. In that sense this circuit is less desireable as an amplifier input circuit, because of the possibly unknown source impedance - particularly as part of a main amplifier, where the varying Miller effect (input pole) can compromise feedback stability. An interesting comparison (which I have not done) will be between this circuit and using two triodes in a cascode circuit. Such a stage has high linearity but high output impedance, but mainly, the effect of the source impedance is largely cancelled as a result of negligible Miller effect, thus wider bandwidth (though depending on the following stage).

[handsome]

the miller capacitance of the mu follower will increase over a normally resistive loaded triode simply because the active load has maximised the gain of the lower triode. the unbypassed cathode resistor of the lower triode will increase it's anode resistance and thus reduce the circuit's power supply rejection as the resistive divided formed the the (active) load and the lower valves anode resistance is altered. the cascode will give far more gain than the single triode can with very low miller capacitance thus very wide bandwidth, but very high output impedance (effectively the value of the load resistor) thus zero power supply rejection - the circuit's output impedance is so high there is no resistive divider action with the load resistor (john broskie does show a method of using the upper triode's grid as a noise canceling input). ironically using the cascode with it's high output impedance now exposes you to (any) miller capacitance problems of the next stage. so here too would be a circuit that requires a buffer or direct coupling in order to maintain it's considerable advantages. as i see it mu-followers are a great way of creating a low distortion, quiet(ish) gain stage best used with a bypassed cathode resistor and direct coupled to a cathode follower but can be used with the upper cathode output for some driving ability. the cascode will give more bandwidth at expense of noise (more work on the power supply perhaps?), good linearity and a lot of gain but hampered by an extremely high output resistance. it will be a more stable stage but with all that gain it also most probably will the first stage in a design whilst the mu follower with a nod to it's better (but not perfect) driving capabilities would probably come closer to the output of a circuit much like Mars' single stage SE amp. a circuit using a mu-follower like this is (at least i think it is) the CAT SL1 preamplifier however note the anode stopper on the final cathode follower - looks like a bit of instability being quelled there....

[Mars]

Thanks for the info handsome

From what I gather is that cascode's are less affected by the Miller capacitance of the input tube? I've been wanting to check out a 6SN7 in cascode configuration. It is not so difficult to DC couple a cathode follower to the output tube. Then there is still only one cap in the signal chain. Maybe those 12au7's can still be used to drive the SE output tube.

Then if the anode of the cathode follower are directly connected to the screen grid, there is some regulation benefits of negative feedback via the output tube; not so?

[handsome]

the point of the cascode configuration is bandwidth through the elimination of miller capacitance, it is also a very quiet stage with lots of gain. interesting thought to using the screen to feed the cathode follower i would be interested to see how much linearisation would result. ultralinear taps linearise the output tube, your feedback loop should add even more - keep us posted