• 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.

OTL Topologies

In class A the P-P tube rp curvatures would largely cancel out, but in the class B portions (of class AB operation), the rp curvatures make for odd harmonics.

With the difference in gain between top and bottom banks, the gain also sweeps from low to high as the output current polarity and magnitude changes, causing 2nd harmonic too. It doesn't just "average out". Might be a good idea to add a toggle switch to allow flipping between MP mode and balanced Futterman mode drives. Have it both ways, the listener can decide.

N Fdbk to the rescue for OTL.

The B+ transformer is listed as an isolation xfmr. Usually those have a boost in their turns ratio to preserve output V at max. load (compensating for internal resistance). Since this unit seems to be generating +-150V instead of +-170V I'm guessing the isolation xfmr is installed "backwards" to get a V reduction already. But if not, then that would be an option for compensating for todays higher line voltages. (or a small auto-xfmr reducer) I would guess the power supply xfmr is around 90% efficient, plus some bleeder resistors on the caps.

13/15FM7 tubes appear to have 12B4 guts in them for section #2. These however have a 10 Watt rating and 175 mA peak rating versus the 12B4's 5.5 Watts and 105 mA peak. These used to be on the various $1 lists, but no longer. Could be cheaper to purchase in quantity than 12B4, but 15FM7 are Compactron based.

If someone were going to build another OTL from scratch, I would suggest the 38HE7 tube which is still on the $1 list at ESRC. This is a compactron pentode with 800 mA peak and about 15 Watts Pdiss approx., when operated without the damper diode in the same bottle. (pins 12 and 10 allow operation of just the pentode at 21V and 0.45 A heater, an available option on about 80% of the 38HE7 tubes I've seen) This is 3 times more efficient for peak cathode current versus heater power compared to the 12B4. It is capable of low saturation voltage in pentode configuration, and also has excellent triode curves. But has a low grid 2 V rating (150V).

Two of them in parallel are equivalent to a 42KN6 tube (it is half of the early twin pentode _KN6 version). Which is very close to a 6LX6 tube (which is what is in the later single section 42KN6 tubes). (this is a premium tube, like 6LF6) You'll only need like 3 or 4 of 38HE7 to equal the 30 pieces of 12B4. $4 + sockets.
 
Last edited:
Looks like a toggle switch for selecting MP unbalance or Futterman balance would be a good mod. Can hide that easily. Futterman won't be around to collect patent royalties now.

If building one of these from scratch now, 13FM7/15FM7 would be cheap and easier (almost 1/2 the tubes), and one could have an individual driver triode per output triode too. (self scaling for any size)
(maybe better to use the "free" top and bottom small driver triodes in an LTP splitter/driver, like N parallel OTLs, could make modular two tube top&bottom PC boards, that plug into each other like dominoes for any power level needed.)

38HE7 would get one down to 8 total output tubes I think. 26LX6 getting down to 4 output tubes, 6LF6/6MH6 down to 2. (for roughly the same Watts out) (38HE7 $1, 26LX6 $20, 6LF6 $50 )
 
Last edited:
Would expect higher closed loop gain to give higher Zo, yes, since it wouldn't be available in the global N Fdbk loop.

The inverted Futterman should increase the output stage Zo (alone) using the local pos. Fdbk to make both banks into grounded cathode mode, but the greatly increased open loop gain available then, using the global N Fdbk, should bring it right back down.

The normal Futterman should have decreased output stage Zo (alone) using the local N Fdbk to make both banks into cathode followers. But then the reduced loop gain available would reduce the effect of the global N Fdbk. So same net result on Zo?

If all of the schemes: MP, Futterman, and inverted Futterman have the same tubes, and the same final closed loop gain, I would expect them to have similar Zo and dist. With the MP scheme having slightly higher dist. due to the mis-matched gains, and maybe slightly higher Zo too. (some gain wasted with the tube banks fighting each other somewhat, but the increased dist. would be 2nd harmonic, some may like that effect)

Edit: OOPS!! I think I've got the Futterman versions messed up. Too late to think this out.
Edit2: I guess its correct. I'll check again tomorrow.
 
Last edited:
smoking-amp

I`m not sure for which OTL topology you refer that is` Inverted Futterman` ? ,
is that topology from Fig.2(`Anti-Futterman`)?, or that is from Fig.3 known as `Technics OTL variant `? .
 

Attachments

  • OTL topology.jpg
    OTL topology.jpg
    372.5 KB · Views: 486
I'm no expert on the OTL versions, but I think Fig. 2 would be called Inverted Futterman. (anti-Futterman is a typo I think)

Fig. 3 Technics variant is the AC equivalent of Fig. 2, except for a phase inversion for the signal going through the amplifier.

Since one can also use LTP/CCS phase splitters similarly for OTL, Fig. 2 and 3 would then just be using swapped outputs of the LTP.

I think the main distinction really should be whether the output stage is being run as two cathode follower banks, or as two grounded cathode banks. (or then there's the 3rd ODD case: the MP design, with mixed CF and grounded cathode banks. Rosenblit had a patented PL509 amplifier like this too, except the patent was incorrect, was supposed to be equal bank gains.) Well, then there's the Circlotron design too, where both banks are 50% CFB.

Usually one hears about which design has lower output impedance. But I suspect when global N Fdbk is included in the calculations (using the same gain tubes for each) all the cases come out near equivalent on that (just putting some of the gain in the local or global position). No idea if there is any "sound" difference between the designs. Would seem to be equivalent to 1st order, but "sound" can be 2nd or 3rd order subtle distortions. Could be some N Fdbk stability issues that are different.

Oh, using a triode for the splitter does cause some complication to the bootstrapping, due to the internal plate N Fdbk. So that might make for some gain (and Zout) variation between the OTL versions.
 
Last edited:
I guess using a high Mu splitter triode would make the OTL versions near equivalent.

But a low Mu splitter triode would lower the overall gain for the case where the triode splitter plate load has positive (bootstrapped) Fdbk on it. So Fig. 2 would have somewhat reduced overall gain versus Fig. 3. (since the splitter plate has to swing higher V with the positive bootstrapping)

A LTP/CCS triode splitter would have the same issue.
 
Last edited:
Disabled Account
Joined 2013
Futterman secured 2 patterns, Fig 1 referred to 1st one and Fig 2. referred to 2nd one, both pattern increased the gain of top bank (PFB, boostrapped) and in do so reduce the gain of bottom bank. Rosenblit thought the bottom bank gain should not be reduced...

Inverted Futterman Fig3, just reverse Futterman 1st pattern. Techics version only use NFB to reduce the gain of lower bank but doing nothing to top bank.

You can also use completely separately Class A driver one each for top and bottom bank and then apply PFB to top driver and NFB to the bottom, keep the feedback level of either one fixed and the other variable so it is used as AC balance, which seem to be forgotten by most, but it's neccessary due variation of tubes in repairs maintainace.
 
Last edited:
Ahh, makes sense.
What looks like NFB to the splitter bottom load in Fig. 1, just ends up increasing current through the splitter to raise the top side gain.

Fig. 2 does that directly with pos. Fdbk bootstrapping to the top load.

Fig. 3 reduces current through the splitter, to lower the splitter plate output V swing going to the bottom output bank.

Increasing the drive signal to the top bank makes them operate like grounded cathode mode for Figs 1 & 2.

Decreasing the drive signal to the bottom bank makes that operate like a cathode follower for Fig. 3.
But since the total loop gain is then lowered by the same amount, the global N Fdbk is less effective at lowering output Z. With a net result of all the schemes, 1 thru 3, coming out near the same Zout I think.

SO, I guess, Fig 1 would be called Futterman, Fig. 2 would be called inverted Futterman, Fig 3 would be called Technics (and maybe also anti-Futterman? )
 
Last edited:
Disabled Account
Joined 2013
There is a difference between the Fig 1 and Fig2, Fig1 is actually positive feed and meaning that its level is not restricted to output voltage level it can be much higher depends on load and number of tubes. Fig 2 is really just boot strapped, it is limited to output voltage level and not any higher, so sometimes this is not quite sufficient to drive properly, when the output level is incorrect in the first place.

The latest version of MP mods has out performed even Fig1 and others in the SIM test, as the balancing is adjustable and optimised to reduce distortion and the output Z.
 
Last edited:
Disabled Account
Joined 2013
Yes, that is AC balance pot for both banks!

The current plot for different load (say 24, 32, 64..) should stay symmetical, this is the main target, that AC balance comes in handy: say you adjust it for 32 Ohms and then change the load to see whether symmetry is still there.

Remember you still have GNFB, it can now better handle the correction when the correction range has been much narrow down to a more manageable level :).
 
Last edited:
The latest version of MP mods has out performed even Fig1 and others in the SIM test

Remember you still have GNFB, it can now better handle the correction when the correction range has been much narrow down to a more manageable level

OH, I guess I mis-interpreted the earlier remark above. So Fig. 1 was referring to your earlier post on the Sim tests, not the Fig. 1 posted by Banat.

I certainly would expect balanced gains to outperform the original MP design.
But, how does it compare now to the Futterman Fdbk design(s)?

-----------------------------------------------------------------------------

Tubecad articles, for reference on some earlier posts on OTL:

The Tube CAD Journal: E-Mail, Transcendent OTL amplifier

Circlotron Hybrid OTL: Circlotron Amplifiers Once Again
 
Last edited:
Disabled Account
Joined 2013
"But, how does it compare now to the Futterman Fdbk design(s)?"

MP/Fut
THD% 0.202/0.227 @44.8V RMS both
Output Z 0.35/0.4 Ohms
My option:
In Futterman amp, the cathode of the phase splitter is fully sitting on output, this is 100% NFB for bottom and 100% PFB for top bank, this can be overkilled, and hence slight inferior performance.
 
Disabled Account
Joined 2013
Yes, I think so. I change to LTP from Concertina sometimes ago, and I think have gotten ride of the "congestion" in the sound. But my most favorate is seperate Class A driver and feedback to each, of course with LTP phase splitter in the front end. This sounds more SET to me, dispite all, that is why you don't really have to worry but just trust your ear. Btw the 2 class A tube is asymmetry, from triode/strapped from 6H12P. This is to minmise any sonic difference due unequal driver level.
 
IMO by Futterman design (Fig.1) power triode V3 grid also gets local NFB via Rk of V1(is that Shade Fb?) dynamically lowering V3 anode gain close as possible to 1, in the same time V1 anode is dynamically boosted up (by bootstrap action) forcing V2 to come close as possible again to gain of 1.
 
Disabled Account
Joined 2013
Go by the SIM:
Futterman

When it's full output the bottom is 20V, the top is 80, the input is 27. The gain is 20/27(0.74) for bottom and 80/27 (3).

MP mod
Same story, except the top gain is 82.5/27 (3.05)

The output stage has a combined gain of 3, the top would need that because it is cathode output but the bottom is common cathode less gain (0.74) is needed.
 
Last edited: