Push Pull Ultra Linear Connection

Triode single end amp’s warm sound but less power  leads Diyers to find other solution. Tetrode push pull amp gives larger power but sounds no good.

Compare tetrode and triode plate curve please see picture:

tet tri out

The output area is much bigger for tetrode as well as output power, but the price pay for that

is reduced sound quality. This is because the plate curves are   much more even distributed for triodes.

Some body uses triode connection ( screen grid connected to plate) to get the  plate   curves  looks like  a triode but with reduced output power, there is not thing new.

I wondered why screen grid connected to plate can get triode style curves  so I study the     tetrode  plate curve and fond something. Please see picture:

curve 100 300

For a specified grid negative bias the plate current remains almost constant throughout the graph. But the plate current increases  as the   screen   grid voltage increases. The relationship between plate current and screen grid voltage is   directly  proportional to each other and quite linear.  As an example , the  -5V  grid bias curve gets Ip 10 ma for SG 100V  and 30ma for SG 300V. On other word the plate current Ip is sensitive to screen voltage but not sensitive to plate voltage.

Now we examine the negative -5V  bias curves see picture:

curve -5

We find the direct proportional relationship , in this case screen grid voltage is a constant throughout one step as the plate voltage increased from 0  to 300V to get the curve. And we get this result from three steps, screen grid voltage  100V, 200V and 300V.

Now think about screen grid connected to plate, plate voltage increase means screen grid voltage increase, see picture :

curve triode -5

The screen grid voltage no longer constant but follow the plate voltage.

When   Vp =100V ;  SG=100V

When      Vp=200V ;  SG=200V

When   Vp=300v ;  SG=300V

The plate curve follows the red line now.

The curve family with different grid bias looks like this, see picture:

curve triode all

Ok, now  we understood screen grid connected to plate will get the plate curves  looks like a  triode.  Yes,  I said  it  looks like . The story is not yet finished.

Please see an  actual  example  of 5881 tetrode curve in triode connection, compare with a real triode tube 245 plate curve:

5881 245 even

We find that the distribution of the curve family is much more even in tube 245 real triode, but not so even from that of 5881.  How come ?

Ok let’s look at the 5881 tetrode connection plate curve:


We can see the curve family is not evenly distributed as a well known shortcoming of tetrodes.

The starting point of triode connection curves  are  not evenly distributed how can we expect excellent results,  see picture:

curve no gd

So that triode connection will improve a lot in  Total Harmonic Distortion (THD) is only a beautiful misunderstood.

Think about the screen grid voltage higher than plate voltage, the plate curves will be brought up more close to vertical  position. This is known as ultra linear connection see picture:

UL connect

According to this circuit I arranged   a curve tracer test ,  a 5881 tube with Tamura 2012 PP output transformer in ultra linear connection and the VR is  5K/4= 1.2Kohm. Adjust the VR can get  0%  to 100% ratio.

100%  means triode connection, 0% means tetrode connection,  some  where  in between is ultra linear connection see picture:

t 2012 testt_2012


The graphics tell the results   of 25% , 30% ,  35% ,  40%  and  42%   UL  connection:

ul 25 30

ul 35 40

ul 42 triode


Compare with triode connection, 42%  UL connection got a more even curves  as well as lower  THD.  One more thing, curves more close to vertical position means  lower plate internal resistance (Ra) as well as better damping factor.  These  advantages plus more output power UL obviously will defeat triode connection.

The relationship between percentage UL and THD see picture :

percentage graph

The result is amazing , the lowest THD comes on   about 40% ultra linear tag ,  it  is better than triode connection in THD measure.

Before I got  the curve  tracer results I always suppose that  triode connection is the best, then ultra linear and tetrode connection is the worst ,  many DIYers  agree as a common practices.

Now I suppose ultra linear is the best then triode connection and tetrode connection still the worst.

This changes  the situation a lot, the question is , is it still  worthwhile to  rearrange a triode connection for those vintage tetrode machines without an ultra linear tag on output transformer ? If the answer is no, may be some high sought vintage machines  coming  down from the heaven.  I will only build PP machines with about 40% UL connection, those ancient and   modern    PP output trasnsformers  without ultra linear tags are wasting time.

I am not saying that ultra linear connection solves all the PP power amp problem, but  at lease we had cleared  one part,  yet the PP cross over distortion is still waiting for us.

Thank you    for visiting.