Modified RCA BA21 quartet

[blue_luke]

Here is a little progress report.... (ti ti ti diti.. little news music theme )

I have a problem with the power supply section, the IRF730 keeps on blowing! If i bring the voltage up gradually with a variac, no problems! I have about 355Vdc in, and I put a 200V and a 100V zener in series, so differential is about 50 Volts. Things are peachy!. If I turn the power switch ON directly, the mosfet returns to silicon heaven!
I checked the specs for the IRF730 and it is a 400V device. My guess is that I am exceeding the specs at turn on. I just got some IRF840 this afternoon..

I also elected to use DC heater voltage because of the transformer I have. simple affair really, 25Vct on a big diode bridge, 3000uF, a 2N3055 with a TL431 and a trim pot in tha base to get exactly 6,3 V at the output. No problem there.

Also built a first channel with mixed success...
The first stage and the phase splitter works, but I must have done something wrong with the push-pull (V2) the current resistor puff a bit of smoke and I read about 25mA on the power supply.  Hmmmm?
Probably some wiring error. Also I don't have the proper tubes, so I use a 12AX7 as V1 and a 12AU7 for V2. Far from optimum, but enough to poke around with a scope.

To answer to emrr about wether or not a zobel network is required on the input transformers, I placed a 1Khz square wave to the primary, and placed my scope directly at the grid of V1, with power on, so the transformer is seeing its 'natural' environment. Quite a bit of ringing. I will have to investigate further more  this matter.
I have not evaluated the impedance ratio, but I can confirm a 20db gain from prim to sec winding, using the 600ohms input. so its a 1:10 turn ratio... don't remember how to calculate the impedance ratio, something with PI square.... will find out .

Going down to the basement, will keep you posted of my progress

Luc

 

[emrr]

12AU7 will draw more, I think. 

Ringing test needs proper input termination/source, be sure of that.  I find it makes a lot of difference. 

 

[blue_luke]

Ringing test needs proper input termination/source, be sure of that.  I find it makes a lot of difference. 

Using the normal output impedance of the freq gen (50ohms) Beckmann industrial function generator.
I also have a Heathkit IG18 which has a true 600ohms output.
As I said, the transfo is connected exactly as the original RCA schema.
Luc

 

[blue_luke]

Now this is getting more difficult than I thought !  :'( :'( :'(

I am using a heatkit bench power supply for the B+ and a DC lab supply for the heaters.
Right now the minus side for the heater supply is not connected to the chassis ground, so the heaters are effectively floating with no reference.

Now I am using a 12AU7a for V2 and a 12AX7 for V1, understanding that these are not optimum in this design but it should work nevertheless, if not perfect.
Right now at this stage all I want is to input some signal and getting it out not too badly mangled and with somewhat more gain.
RATS!!  :worried: all I get even with nothing connected to the input transformer is some heavy motor boating at about 1Hz.
All I can see is the current meter going between 25mA to about 10mA in sympathy with this motorboating. Now the whole channel should use about 15mA non?

Sooooooooo  I decided to put a signal at the input, 1Khz about 50mV. All I get now is an oscillation at about 400Hz that modulates the input signal. The current meter now stays at a steady 25mA.
In order to limit the damage if any I decided to lower the voltage on the power supply to probe here and there (and everywhere :sing: ) then that's it!! motorboating gone! no more oscillation and the signal is right there with no apparent distortion on the scope, this when I apply about 200 volts to the circuit instead of 300.

I placed a 1200ohms load on the output transformer and measured about 1Voutput... roughly 26db gain, this with NFB loop connected.
So, where do I look from here? I know using a 12AX7 for V1 gives me plenty of gain for the first stage but a 12AX7 is not really good as a phase splitter, still, the 12AU7 requires very little drive current... I guess???

Hmmmm!????

 

[blue_luke]

I GOT IT WORKING!  ;D  ................ Sorta!    :


The first thing I did was to remove the feedback network in an attempt to get each stage working as good as possible before NFB.
Then I changed the value of the resistor from the two cathodes to ground in the PP circuit. Went from 390 Ohms, as per plan, to about 1K that I had on hand.
Of course the current dropped like a bomb!
Bingo, no more motorboating, and now plenty of signal output.
This is getting exciting! 

Then I connected back the feedback network and the gain droped to about 2!! 
So just to experiment, I decided to change the value of the 22k in the NFB network...
Did'nt do anything!!??  Now this is curious... then EUREKA!! I built most of the circuit on a piece of veroboard and found that I forgot to cut the tracks under the feedback components. So basically I was shorting the feedback winding of the output transformer directly at the cathode of the first stage.
100% feedback!!   

Works beautifully now!!.

Now I have started tweeking the whole thing.
In order to get a feel for the frequency response, I completely disconnected the input transformer and connected a sweep generator (sine) directly on the grid of the first stage and placed a 1K load resistor on the output. I had this value on hand!! 

To my amazement, this thing will go roughly between 12Hz to about 70Khz! at 2 volts output to within about .1 volts. I just checked on the scope, but the -6db point (half voltage) is about 10Hz and 95Khz!!
WOW! 

The gain of the preamp is about 30db for the moment, including the input transformer. By playing with the value of the cathode resistor in the first stage(8K3 as per plan) which is the NFB return point, I can vary the gain quite easily.
But now it's late and I need to ponder a few things about tonight's observation, before finalizing the design.

Also I don't have the suggested tubes (12AY7) but it works quite well for the moment with a 12AU7 in the PP section and a 12DW7 for the first stage (like a 12AX7) and the phase splitter (like a 12AU7).
I have not listened to it yet, does'nt seem to produce any hum. I will also have to measure the distortion.
The current draw of the whole channel is about 6-7 mA. Does this seems reasonable? I was expecting something more like 15mA for the channel.

Luc

 

[RedNoise]

I GOT IT WORKING!      ................ Sorta!   

Thumb up !!! ;D

 

[emrr]

Your measurements are off entirely because you are using different tube types.

Max feedback via forgotten short; that would do it! Rather than changing cathode resistance to control feedback, change the series resistor in the feedback path; that's why it's there. 

I'm not amazed by the response you plot at all; the (good) old stuff is always just fine in that regard.  The input transformer will probably have more effect on the response than any other part.  NFB is helping everything else in the circuit. 

The 12AU7 will drive the output more than the spec'd 12AY7, so there's greater push through the NFB path.  This should be a greater effect than the increased gain (perfect world only) of the 12AX7 stage. 

 

[blue_luke]

Your measurements are off entirely because you are using different tube types.

That I understand obviously, but I never realised everything is such a fine balancing act! And also I have read somewhere John(?) Manley saying that tubes have no sound of their own if the circuit around them is perfectly adapted to the tube. So poking different tubes in the socket of a guitar amp may actually change the color if nothing else is changed. But if you change a XYZ7 for 12ABC and tweek the circuitry to adapt them perfectly, apart from gain change, you should not hear any differences. (Dixit Manley)

Max feedback via forgotten short; that would do it! Rather than changing cathode resistance to control feedback, change the series resistor in the feedback path; that's why it's there. 

I played with the cathode resistance just because it's more convenient, but I agree with you and I will tweek the NFB network instead. Won't shift the DC bias of the input stage also.

I'm not amazed by the response you plot at all; the (good) old stuff is always just fine in that regard.  The input transformer will probably have more effect on the response than any other part.  NFB is helping everything else in the circuit.

You are right again! (darn I hate those guys!  )
With a square wave, there is just a hint of overshoot. The input transformer on the other hand will not go as high in frequency (sine) and will exhibit quite a bit more overshoot on a square wave. That's why I included a zobel network in my first preliminary schema. Then you had me question it, since there is none on the original, and those guys in those days really knew what they were doing.

They might also have figured that since this ringing during the overshoot is at about 40Khz, it had no audible effect since it's so out of the hearing band?... maybe

The 12AU7 will drive the output more than the spec'd 12AY7, so there's greater push through the NFB path.  This should be a greater effect than the increased gain (perfect world only) of the 12AX7 stage.

 
Sorry but I don't follow here!  : Can you elaborate?

Meanwhile I am taking a few design decision here...
Instead of using 12AY7's as the original, somebody have recommended using 5687. These tubes were made for the military so they have excellent built, and are available for about 5$ each!!
(right now I am a bit in a bind financially, being in recovery after surgery).
So I will tweak the schema to get these working at there best.
The mu is about 18,5 and the dissipation is 3,75 watts per triode. Close to a 12AU7,at 1/3 the price!
Again, this open to debate!

Hey, thanks anyway for the time you take with my tube education. This is very much appreciated!
Luc

 

[emrr]

The 12AU7 will drive the output more than the spec'd 12AY7, so there's greater push through the NFB path.   This should be a greater effect than the increased gain (perfect world only) of the 12AX7 stage.

 
Sorry but I don't follow here!  : Can you elaborate?

Meanwhile I am taking a few design decision here...
Instead of using 12AY7's as the original, somebody have recommended using 5687. These tubes were made for the military so they have excellent built, and are available for about 5$ each!!
(right now I am a bit in a bind financially, being in recovery after surgery).
So I will tweak the schema to get these working at there best.
The mu is about 18,5 and the dissipation is 3,75 watts per triode. Close to a 12AU7,at 1/3 the price!

Heres the problem as I see it.  The circuit is designed for the 12ay7, which operates at a higher impedance while delivering less current than the options you consider/tried.  Look at the data sheets.  12au7 is 3 times lower plate impedance and 3 times higher current, at least for the quoted standard.  The 12ay7 works with the transformer in question, in the circuit in question, and IMO anything more current oriented is akin to bringing a machine gun to a fist fight.  The circuit is rated to deliver +18 dBm, and generally the transformer can be expected to have relatively little headroom beyond that point.  It may be intellectually interesting to redesign the circuit, but the 12ay7 is the cheapest known fix; a redesign may mean a different output transformer to meet same specs - that gets really expensive.  I'd be worried about cooking one of those unobtainium output transformers with excessive current flow, and there's no advantage to be gained by purposefully limiting the range of a higher current tube like the 12AU7. 

I believe the 12ax7's higher than rated impedance is probably stepped on by the lower than rated impedance of the 12AU7 pushing lower than rated current through the NFB path.  The 12AX7 theoretically higher gain mostly disappears in reality, in addition the NFB already erases most differences there, and the NFB effect is changed by the lower drive impedance.  This is why the gain measurements are all different.

if you change a XYZ7 for 12ABC and tweek the circuitry to adapt them perfectly, apart from gain change, you should not hear any differences. (Dixit Manley)

I think you may erase the majority of differences, but there are innate qualities of certain tube types that you will not erase.  There are geometric issues within identical tube numbers of different bottle sizes that create sound differences, even though the specs (on paper, at least) are the same. 

I may step in something if I attempt to say more; may already have!  ;D