SPICE Design of 6F5P SET Amplifier

Author: Dmitry Nizhegorodov (dmitrynizh@hotmail.com). My other projects and articles


1.   Abstract

In this article we SPICE several Single-Ended Triode (a.k.a SET) designs all based on 6F5P tube, delivering clean 2W of SET power from only .7 VRMS input. 6F5P is a triode/pentode combo, the pentode section is triode-connected.

2.   Introduction

6F5P is a Russian TV deflection triode/pentode combo tube, somewhat like ECL85/6GV8. If the pentode section is connected in triode mode, the tube provides a great playground for experiments with 2-stage SET designs on a dime. Why a "dime"? First, both stages are inside of one tube just slightly large in diameter than a dime, with the necessary aux parts, excluding the power supply and the output transformer, fitting under the tube socket. Second, the tube is not very expensive.

To start SPICEing designs based on 6f5p, we need SPICE triode models for both sections of the tube. Below are the models I obtained matching Koren Improved Triode Model against plate curves published by E.Karpov [2] using my interactive model parameter finder [3]. See curves at [1].

       .SUBCKT 6F5P_T  1 2 3 ; P G C (Triode Part)
       + PARAMS: MU= 73.16  EX= 1.668  KG1= 522.9  KP=353.07
       + KVB= 519.0  VCT= 0.60
       + CCG=3.5P  CGP=1.8P CCP=.25P RGI=2000 
       E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+(VCT+V(2,3)))
       +  /SQRT(KVB+V(1,3)*V(1,3)))))} 
       RE1 7 0 1G 
       G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1} 
       C1 2 3 {CCG} ; CATHODE-GRID 
       C2 2 1 {CGP} ; GRID=PLATE 
       C3 1 3 {CCP} ; CATHODE-PLATE 
       D3 5 3 DX ; FOR GRID CURRENT 
       R1 2 5 {RGI} ; FOR GRID CURRENT 
       .MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N) 

On this SPICE plot, the red line is the driver section's issipation limit = .5W. Recommended quiescent point biasing is -1V to -1.5V.


6f5p's driver is a tiny triode, yet with a strong bite. Its curves are similar to 6an4, except 6f5p is much more linear. Its Mu and Rp are a close match to 6an4. Of course, it can not work on the same sections of the curves and its Rp at 2.5 mA is a fraction of 6an4's at 10 mA.

The driver section is very interesting, the power section, hooked in triode mode, is even more interesting.

       .SUBCKT 6F5P_PT  1 2 3 ; P G C (Pentode Part in Triode Mode)
       + PARAMS: MU= 7.49  EX= 1.622  KG1=1336.3  KP= 59.91
       + KVB= 120.0  VCT= 0.00 RGI=2000 
       + CCG=11P  CGP=.7P CCP=8P
       E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+V(2,3))
       + /SQRT(KVB+V(1,3)*V(1,3)))))} 
       RE1 7 0 1G 
       G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1} 
       C1 2 3 {CCG} ; CATHODE-GRID 
       C2 2 1 {CGP} ; GRID=PLATE 
       C3 1 3 {CCP} ; CATHODE-PLATE 
       D3 5 3 DX ; FOR GRID CURRENT 
       R1 2 5 {RGI} ; FOR GRID CURRENT 
       .MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N) 

Reminds 45, 2a3 or 300B curves? The comparison may sound silly, but do not laugh out loud yet.

Plate curves of RCA 2a3 aligned with 6f5p:


Almost ideal match, except due to its 40% higher Gm the bias lines are 5.7V of 6f5p for every 10V of 2a3. In other words, the 2.5K loadline presented on the curves will be driven exactly similarly by both tubes (same distortion) yet a driver needs to deliver 40% less swing to drive 6f5p. To be exact, 2a3 needs 87 VPP, as shown; 6f5p needs only 50VPP.

Is this too good to be true? In one sense yes - 2A3 can continuously dissipate 16W, yet 6f5p's power section can handle 11-12W (plate plus screen) at most. Please note, though, that this translates to a very small difference in SET amp's output, and is a fair price to pay for tiny 6f5p delivering the quality of 2a3 at a fraction of cost. 2 well-matched 6f5p running in parallel can outperform a 2a3.

The designs presented here are aimed at use of Russian TV audio transformer TWZ-SH, which we represent with the following SPICE model:

     .SUBCKT TWZ-SH 1 2 3 4 
     +PARAMS: LPRIM=10 LLKG=.016 RPRIM=300 CPRIM=.1nf WRATIO = {114/3000}
     RS1 1 2    1000K     
     RP1 1 11   {RPRIM}
     LPleak 11 12   {LLKG} 
     LP1 12 2   {LPRIM} 
     CS1 12 2    {CPRIM}  
     LP2 4 3    {LPRIM*LRATIO}  
     KALL LP1 LP2 {1-1/(QFCTR)}  ; COUPLING

3.   Condenser coupled, simple

First design is a conventional cap-coupled 2-stage design. For simplicity and to obtain more power out of the driver, we run it with the same voltage that is fed to the power stage.


The following is distortion data with various values of the load resistor of the first stage. Input sensitivity of the amplifier is pretty good - 2W of output corresponds to ~ .7VRMS input.

6f5p-hd-3.gif Driver stage resistor is 35k, which yields some visible 2nd harmonic cancellation at the expense of the 3rd.

6f5p-hd.gif Driver stage resistor is 40k. This offers least THD.

6f5p-hd-2.gif Driver stage resistor is 47k. Sonic signature is more of classic SET, with 2nd harmonic dominating.

Similar variations in distortion patterns can be obtained if biasing resistor R3 changes by +- 10..15 %.

An article describing physical implementation of this design is here: 6f5p-set-impl.htm, The article contains real measurments very much in sync with these SPICE results.

4.   DC coupled

This is a DC coupled design tuned for minimal distortion. Changes of driver stage B+, its load, bias and power stage bias affect this local spot quite greatly, and the value will be different for different tubes due to varying mu, bias and other tube parameters.



5.   DC coupled, "tower"


6.   DRD


7.   Parallel connections

Here's what happens when one can perfectly match several 6f5p tubes. Perfect match is in a sense discussed in [4].

7.1   Dual case

This way a SET built with two parallel 6f5p is much equivalent to a 2a3 SET



7.2   Quad case

This way a SET built with four parallel 6f5p is a lot similar to a 300b SET.



8.   References

[1] http://www.tubes.ru/techinfo/GlassTubes/6f5p.html

[2] http://www.next-power.net/next-tube/DataSheets/tubes/6f5p.djvu

[3] http://www.dmitrynizh.com/tubeparams_image.htm

[4] parallel-triodes.htm SPICE experiments with parallelled tube stages.

Author: Dmitry Nizhegorodov (dmitrynizh@hotmail.com). My other projects and articles