This is an implementation of Frank McIntosh’s unity coupled output stage. The original work, description and analyse is found in a paper called "Description and Analyse of a New 50-Watt Amplifier Circuit" [1]. Some more theoretical analyses of class AB amplifiers and notch distortion was done by Pen-Tung Sah some years previous to that in his paper "Quasi Transients in class B audio frequency push-pull amplifiers" [2]
The corner stone (or what ever I should call it) in this amplifier is the trifilar wound output transformer. The third winding is for regulated screen grid voltage to the two pairs of EL86.The primary impedance of the transformer is 3 x 720 ohms which represents an anode to anode load for each EL86 of 5.8k. The leakage inductance is about 400uH and the transformer has a overall bandwidth of more than 150kHz.
One problem with bi/trifilar wound transformers is the possibility of insulation brake down between the windings, and according to Brian Sowter I am not allowed to run this transformer in a circuit with higher potential than 300Vdc. That is one of the reasons why I choose EL86. Other candidates could have been sweep tubes like PL504 etc.
Circuit analysis:
One problem in a unity coupled output stage is that you will need a generous portion of voltage swing to drive to output tubes. The demanded voltage swing is about B+/2 or ~130Vp in this case. McIntosh solved this problem by bootstrapping the driver stage. In this circuit I have done the same thing, you will find the bootstrapping realized in the 15k plate resistors in the driver stage which is coupled to the anodes of the opposite side output tubes. This will give at least two effects: positive feedback and increasing the apparent value of the plate resistors.
The apparent value of the 15k plate resistor depends on the amplification in the output stage, if we assume that A=0.85 the apparent value of the resistors will be 15k/(1-0.85)=100k which will flatten out the load line for the drivers. The load (C- resistors) is also bootstrapped to increase the apparent value.
As mentioned earlier there will also be some positive feedback: the gain of the 12BH7 stage with no bootstrapping is about mu*Ra/(Ra+rp) = 16.5*15k/(15+5.3k)=~12x and since mu=16.5 max possible positive feedback will be 16.5/12=1.4x or less than 3dB. The output stage itself will contribute with negative feedback due to its unity coupling. One way to estimate this portion of negative feedback is to look at the reduction of gain compared to the same circuit coupled as common cathode. The gm for two EL86 is about 20mA/v so the gain in a cathode coupled circuit will be gm*Rl = 20mA/v*720R=14x and the reduction of gain in a unity stage is 14x/~0.85x=16x or 24dB. This indicates that the amount of global negative feedback could be kept quite low.
An interesting thought is to use pentodes as drivers and insert some negative feedback into the screen grids, something like this:
But for the moment I will stick to the 12BH7 drivers and see where that will bring me. If nothing unexpected shows up I am planning to have the amplifier up and running before Xmas (this year).
Jan E Veiset
[1] www.veiset.net/tech/Mac1949.pdf
[2] www.veiset.net/tech/Mac1936.pdf
The corner stone (or what ever I should call it) in this amplifier is the trifilar wound output transformer. The third winding is for regulated screen grid voltage to the two pairs of EL86.The primary impedance of the transformer is 3 x 720 ohms which represents an anode to anode load for each EL86 of 5.8k. The leakage inductance is about 400uH and the transformer has a overall bandwidth of more than 150kHz.
One problem with bi/trifilar wound transformers is the possibility of insulation brake down between the windings, and according to Brian Sowter I am not allowed to run this transformer in a circuit with higher potential than 300Vdc. That is one of the reasons why I choose EL86. Other candidates could have been sweep tubes like PL504 etc.
Circuit analysis:
One problem in a unity coupled output stage is that you will need a generous portion of voltage swing to drive to output tubes. The demanded voltage swing is about B+/2 or ~130Vp in this case. McIntosh solved this problem by bootstrapping the driver stage. In this circuit I have done the same thing, you will find the bootstrapping realized in the 15k plate resistors in the driver stage which is coupled to the anodes of the opposite side output tubes. This will give at least two effects: positive feedback and increasing the apparent value of the plate resistors.
An externally hosted image should be here but it was not working when we last tested it.
The apparent value of the 15k plate resistor depends on the amplification in the output stage, if we assume that A=0.85 the apparent value of the resistors will be 15k/(1-0.85)=100k which will flatten out the load line for the drivers. The load (C- resistors) is also bootstrapped to increase the apparent value.
As mentioned earlier there will also be some positive feedback: the gain of the 12BH7 stage with no bootstrapping is about mu*Ra/(Ra+rp) = 16.5*15k/(15+5.3k)=~12x and since mu=16.5 max possible positive feedback will be 16.5/12=1.4x or less than 3dB. The output stage itself will contribute with negative feedback due to its unity coupling. One way to estimate this portion of negative feedback is to look at the reduction of gain compared to the same circuit coupled as common cathode. The gm for two EL86 is about 20mA/v so the gain in a cathode coupled circuit will be gm*Rl = 20mA/v*720R=14x and the reduction of gain in a unity stage is 14x/~0.85x=16x or 24dB. This indicates that the amount of global negative feedback could be kept quite low.
An interesting thought is to use pentodes as drivers and insert some negative feedback into the screen grids, something like this:
An externally hosted image should be here but it was not working when we last tested it.
But for the moment I will stick to the 12BH7 drivers and see where that will bring me. If nothing unexpected shows up I am planning to have the amplifier up and running before Xmas (this year).
Jan E Veiset
[1] www.veiset.net/tech/Mac1949.pdf
[2] www.veiset.net/tech/Mac1936.pdf
Wooooooow 
Very nice.
There's another unity coupled at Jogis
But he doesn't use 3 Windings instead he connects the screen grids to the opposit anodes.
Do you calculate the loadline in the normal way and only separate it to the 2 Windings ?
720Ohms in Anode and Cathode means 7210*4 = 2900Ohms quit.
2900*2 because of Push Pull Class A results in 5,8K is that the way how its going with that split load config ?
Regards, Simon
Very nice.
There's another unity coupled at Jogis
But he doesn't use 3 Windings instead he connects the screen grids to the opposit anodes.
Do you calculate the loadline in the normal way and only separate it to the 2 Windings ?
720Ohms in Anode and Cathode means 7210*4 = 2900Ohms quit.
2900*2 because of Push Pull Class A results in 5,8K is that the way how its going with that split load config ?
Regards, Simon
the_manta said:
Interesting project.
the_manta said:
The reflected load for two EL86 in parallel will be 720 ohm (or 2.88k plate to plate), one EL86 will "see" 1440 ohm so I use that value for the loadline.
Here are the loadlines for pure class B (blue) and 30mA bias class AB (red):
An externally hosted image should be here but it was not working when we last tested it.
Jan E Veiset
Hi Jan,
I've been exploring the possibility of Unity Coupling with 46's and doing a search found this thread. I see from your site that your Unity Coupled amp is long finished and though you've kindly posted your very good square wave results you haven't actually said whether, as a purely musical experience (in contrast with the fun of thinking it through and building it) you find it satisfies your ears or not.
Having only built SE Class A tube amps up to now, I don't know a lot yet about PP and especially Class B. Many say that it is fundamentally not suitable for high quality audio. I'm hoping to find out whether there is any point in investing the time and materials to explore it. Do you have any observations or critique of what you've done and seen/heard so far?
Thanks
I've been exploring the possibility of Unity Coupling with 46's and doing a search found this thread. I see from your site that your Unity Coupled amp is long finished and though you've kindly posted your very good square wave results you haven't actually said whether, as a purely musical experience (in contrast with the fun of thinking it through and building it) you find it satisfies your ears or not.
Having only built SE Class A tube amps up to now, I don't know a lot yet about PP and especially Class B. Many say that it is fundamentally not suitable for high quality audio. I'm hoping to find out whether there is any point in investing the time and materials to explore it. Do you have any observations or critique of what you've done and seen/heard so far?
Thanks
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.