I've been looking at this for a while. The Tamura and Tango magnetics are pretty esoteric and expensive. I sent the interstage transformer specs to Heyboer and had them wind one for me. I'm going to ditch the cathode feedback and UL. I also want to ditch the input stage that is DC coupled to the EL34 grid and just AC couple it to a standard previous gain stage.
The interstage transformer has 2 x 2500 ohm on both the primary and secondary. The primaries are paralleled. My first question is do those windings in parallel present a 1250 ohm load to the EL34 or is it still 2500 ohms, because it's the same turns ratio, with double the current capacity.
Per the drawing, the EL34 has 130v on it's cathode and 120v on it's grid from the DC coupled connection to the previous stage - so I would read that as -10v bias on the EL34 grid. 2nd question - is that correct?
Per the drawing, it looks like the EL34's quiescent current is 26mA. 3rd question, If I want to get rid of the DC coupling to the input stage, do I just replace the EL34's 5Kohm Rk with 10v/26mA = 384 ohm?
Why am I messing with this - I thought it looked cool and might sound interesting with the combination of EL34 SE + PP output....might sound like crap
You do not have the required permissions to view the files attached to this post.
I'm not an expert, these are just hunches, so please take them with the usual grain of salt.
I imagine the interstage transformer works just like an output transformer: the load at the input is a reflection of the load at the output. The thing is, (I think) there really is no load at the output while the output stage is operating in class AB1, because there is little to no AC current passing through the secondary. I'm thinking its only when the output stage is pushed into class AB2, will you see an appreciable load at the EL34's plate.
I would agree that the EL34 appears to be biased at -10VDC.
While I agree with your math (question 3), I'm not sure 384R in the EL34 cathode circuit is advised - but I really don't know. EDIT: Actually, the more I think about it, the more I think you'd be fine with 384R in the cathode circuit: If you could find a 1K, 1W trimpot, you could even make it adjustable (26mA through 1K dissipates 0.676W). But you would need to supply a -10V bias to the EL34 (I'm sure you already know this). Maybe it makes more sense to just used a fixed 384R in the cathode, and make the bias adjustable from, say -20V to -7V, then adjust the bias for 10V at the cathode. ONE MORE EDIT: It just occurred to me that if you lower the cathode voltage from 130V to 10V, but still have 323V on the plate, the EL34 dissipation will rise from 5W to 8.14W - probably not a big deal, but it may affect the amount of bias required to get to 26mA.
EDIT: BTW, when you spec'd the interstage transformer to Heyboer, did you advise them that there is 13mADC flowing through each primary winding? I think it makes a difference.
I sent Heyboer a copy of the schemo, the Tango specs, and a written explanation of what I was up to. It's a 10watt transformer...looks bigger than some of the 20watt OT's I've seen
The interstage is better viewed as just a voltage transformer, where the primary inductance (at the bias DC current) is sufficient for the required voltage swing (to fully drive the KT88's) at the lowest design frequency to not roll off.
The interstage also has to perform at high frequency, where its shunt capacitance combined with the EL34 triode anode resistance causes a roll-off.
If you can get EL34 triode curves, then you know the DC bias current, so can go through the SE textbook design.
You could ac couple - just need to bias the grid with a resistive divider to give the same DC bias current. You then introduce low and high frequency roll offs that you will need to estimate.
Thanks TRobbins,
Regarding AC coupling the grid of the EL34, do you think it's best to keep the Rk at 5kohm and set the grid voltage at 120v with a resistive divider, or just set the bias with a new Rk value?
Mullard Jan 1964 datasheet shows triode curves. For Vak nearly 200V, and Ia about 22mA (a bit less than 26mA anode plus screen), the nominal grid bias is about -15V.
With no KT88 grid conduction, then output loading is very high, so EL34 loadline is effectively horizontal, and appears to be fine and mid-position for large signal swings. If the interstage turns ratio is effectively 1:1, then 60V anode swing on EL34 is easily handled.
Original designer may well have moved the EL34 idle bias point around to identify the lowest distortion operation, so keeping that nominal idle point is a good initial aim if you are interested in minimising distortion.
So perhaps aim to initially keep new cathode bias around 15V when capacitor coupling, and drop supply rail from 323V (=190+130) to about 205V (=190+15).
You could also swap EL34 for 6L6 type, with some bias adjustment to suit, although it has about twice the plate resistance in triode mode.
