[bartas]

Hi,
I have my Cornet 2 for some two years already.
I`m very interested in choke input supply for it. Did some modeling in PSUD and came to such main psu part sequence: 30H/595R - 47uF - 15H/256 - 47uF. This gives fastest response and smooth voltage slope.
Another thing I do not like too much in my Cornet is a ECC82 buffer stage. I want to try implementing a 6SN7 and maybe get rid of bias transistor. The only thing I`m worried about is lower B+ if LCLC power supply used.
I think I would be getting some 270v for the buffer stage.
Would appreciate comments on 6SN7 operating points for this case.

Bartas 

[poty]

I don't think you should worry about voltage drop on LCLC circuit. You have plenty spare voltage dropping on the serial resistors in B+ circuit. The only thing I can see - it's better to raise thhe capacitance of the last capacitorr in the DC filter.

The transistor is mostly for safety, not bias stabilizing. You can easily drop it without any thoughts, but you will need  to choose the correct cathode resistor.

I think it is rather easy modifications than you think!!

[bartas]

Why would you raise the last capacitor?
The LCLC setup already reduces voltage ripple significantly, the bigger capacitor would just slow down the PS.
I`m going to get rid of B+ resistors, but 270v is the best I will get from 275v transformer.
Do you think I can skip the buffer plate resistor also?
Cathode resistor I`m going to try would be something around 17K. This would give me ~8mA and 135v across the tube.

[GRD]

Are you talking about the CCS transistor on the buffer?  I've tried both a bias resistor and the CCS and much preferred the CCS. 

[bartas]

I`m still going for the resistor grounded cathode.
Would apreciate very much if Jim could bring some light on this bias topic.
Octal cornet does use a resistor on 6SN7 buffer cathode and B+ is 325v.
I wonder what is the current intended?

[poty]

Why would you raise the last capacitor? The LCLC setup already reduces voltage ripple significantly, the bigger capacitor would just slow down the PS.

What do you mean for "slow down"? If it is longer time to charge the capacitor - then it is better protection (the tubes heating would work longer to the point where the plate voltage would be full). But longer charge - longer discharge, and this is your safety magic wand for energy needs for the Cornet's stages. And there could be low frequency oscillation for lower capacitance, but this sentence needs more calculation...
I'd change the LCLC in such a way also where the less inductance goes first. It saves on voltage drop due to mostly AC nature of the current after the rectifier to which the first L would have big resistance.

I`m going to get rid of B+ resistors, but 270v is the best I will get from 275v transformer. Do you think I can skip the buffer plate resistor also?

No, the buffer plate resistor can give you just a couple of volts.
Yes, I agree, I wonder if you get even 275V for LCLC filter. But there can be slightly modified design - CLCLC filter (which gives you 20-30V more) or maybe you should use the transformer with higher voltage.

Cathode resistor I`m going to try would be something around 17K. This would give me ~8mA and 135v across the tube.

Here I should look at the tube's datasheet. I hope I would have time to do this.

Are you talking about the CCS transistor on the buffer?  I've tried both a bias resistor and the CCS and much preferred the CCS.

Then it should be wrong resistor value or something else. The way it is implemented in the Cornet is not the typical CCS, because there is no stable things (6.3 V is not stabilised in any way and there is not even thermo compensation for the transistor). So it is rather for tube rolling and not for current stabilising.
If bartas get the right value for specific tube it would be an improvement, not losing.

[GRD]

To clarify, I've compared using a cathode resistor versus the CCS on the 12AU7 buffer and preferred the CCS.  Agree that the CCS uses the heater voltage as a reference to set the current and the heater voltage is not regulated, so it will "float" with the line voltage - as will the B+.  But it seems to work fine.

I think the CLCLC idea is a good one for increasing the B+.

[bartas]

According PSUD model I`m going to end up with ~250V in LCLC mode. Still I think it is worth trying. I got very good results when converted my linestage PS to LCLC even with lower voltage but successfully rebiased. 
There is a Jim`s post about cornet prototype with choke input PS. But he uses high DCR chokes and I want to try some with lower DCR.
Poty, looking forward for your comments on 6SN7 buffer biasing with ~250V B+.   

Bartas

[poty]

I've looked through the Sylvania 6SN7GTB and GE 6SN7GTB datasheet and found this:
As soon as we get 150V Eg-to-ground from the previous stage we should count around Eb-150V across the tube. The recommended work point for the tubes is Eg=0V for Eb=90V (Ib=10mA) and Eg=-8V for Eb=250V (Ib=8mA). So:

...270v is the best I will get ...
Cathode resistor ... would be ... 17K. This would ... ~8mA and 135v across the tube.

Eb~270-150~120V
Eg=-1.5V
Ib=9,5mA
Rc=16k

Poty, looking forward for your comments on 6SN7 buffer biasing with ~250V B+.   

Eb~250-150~100V
Eg=-0.5V
Ib=10mA
Rc=15k

Actual values may vary, but as soon as we don't have big voltage for output signal, it is not so important. You see, the current 1.5 times more than for AU7.

[bartas]

Thanks, Poty, that`s very kind of you.
I`ve done some further modeling and this is what I came up with so far. 6SN7 OP came out perfectly.
I just hope Jim will not mind me abusing his circuitry sheets in such way.
The EQ values are from Cornet 3 prototype.
Everything works fine on paper but I am no expert.

[GRD]

Now I see what you're working on.  Very interested in your findings. 
Question - why do you have 3 - 220K bleeder resistors on the B+?  I would think they wouldn't be needed on each channel.

[bartas]

My cornet 2 is done one DIY boards and hardwired and it has separate board for PS. Two bleeders are there primarily for safety and they help me tune voltages. These now are approximate values which came from playing with PSUD. They are to be tuned in reallife supply.
I`m more concerned about some other things:

1. I took the RIIA EQ circuit directly from Jims Cornet 3 prototype and R204/10K is missing there. Is it really not needed? I am no expert of EQ but I was reading somewhere about importance of this resistor in EQ.
2. I am not really sure how the 6SN7 will act without 220R resistors on its plate and in front of output capacitor.
3. And finally only real prototype will show if the small film PS capacitors could be enough for gain stages. I`ve done similar with my linestage, but there was no buffer stage dragging 10mA of current.   

[poty]

I'm glad you come to some draft. You use Cornet 3 schematic, so the first 2 stages will work definitely. The last stage may have some tuning after build, but the tubes are very flexible for that and you definitely will bring the device to work!
I still have some thoughts about your layout:
- in high voltage part of power supply I can't understand the R101 and R102. As safety resistors? For what? You have alreay had choke input! I'd drop them fully.
- the 220k resistors everywhere in the high voltage lines - what for do you use they? They just dumping current with no useful meaning.

Two bleeders are there primarily for safety and they help me tune voltages. These now are approximate values which came from playing with PSUD. They are to be tuned in reallife supply.

I think it's better to have serial resistors to achieve the goal

1. ... R204/10K is missing ... Is it really not needed?

It seems that there is less plate voltage, so less gain. To save the overall gain some resistors have removed or lowered. I don't think it is so important.

how the 6SN7 will act without 220R resistors on its plate

I don't see any problem for that.
- I'd not drop the electrolytic caps and use them with bypass capacitors of your liking!

3. ...only real prototype will show if the small film PS capacitors could be enough for gain stages.

Let's see then.
- I'm not shure why you decided that with the same values in the schematic (as in Cornet 3) you'll have 0.75mA instead of 0.5mA for the first two stages. For that you have lowered R18 (according to Cornet 3) to 5k I think? But then you would have not 210V for the first two stages, but more. And the grid resistor will need some tweaking then. The Hagerman's values (I think) is adapted to lower heating voltage of the tubes and you have not twiked anything in this field either.
- You mustn't connect to the ground the center wire of the 6.3V transformer AC output!
- I'd increased the value of the C208 caps instead of lowered it.
I think the device will work! You can definitely try to implement the changes with almost definite shure about it. But you should know that the common plate (last stage) schematic involves some negative feedback and it is not always the right thing (and not bad either).

[bartas]

in high voltage part of power supply I can't understand the R101 and R102. As safety resistors? For what? You have alreay had choke input! I'd drop them fully.

These are for easier rectifier life but yep  they will be dropped as transformer secondaries have decent 470 ohms resistance.

I think it's better to have serial resistors to achieve the goal

I want to try least resistance PS so there is only one series resistor which is also a bit of the separation from last stage.
The bleeders definitely stay! How do you bleed PS caps any other way???
At the time being I think optimum capacitance and not extra is the way to go in the power supply. Modeling these values came out promising, but the ears will hear
 

- I'm not shure why you decided that with the same values in the schematic (as in Cornet 3) you'll have 0.75mA instead of 0.5mA for the first two stages. For that you have lowered R18 (according to Cornet 3) to 5k I think? But then you would have not 210V for the first two stages, but more. And the grid resistor will need some tweaking then. The Hagerman's values (I think) is adapted to lower heating voltage of the tubes and you have not twiked anything in this field either

I let myself recalculate Jims currents according to his voltages and 0.75mA was the result: 210v-120v=90v/120K=0.75; correct me if I am wrong. Anyway I am going for Jims OP and will be tuning them on the spot (the same with heating voltage).

I'd increased the value of the C208 caps instead of lowered it.

The thing with output cap is that I`ve calculated the lowest possible value here which doesn`t yet degrade frequency response looking into mu linestage and further on. This is because I like Audio Note caps in signal coupling very much and they are expensive

Poty, do you know the purpose of 220R resistor in front of C208 cap (you will not see it on my circuit)? I assume it is for buffer stability in some way...

[poty]

I want to try least resistance PS so there is only one series resistor which is also a bit of the separation from last stage.

The least resistance and the choke input is not compatible. You have the resistance of both chokes already and they are not so easy to overcome in total by another series resistor. You may choose not to regulate the high voltage at all, but let it be as is, tuning only bias (cathode) resistor in the last stage. The plate voltage for the previous two stages can be really tuned by only one resistor. It's better, because you don't have to obtain more current from PS.

The bleeders definitely stay! How do you bleed PS caps any other way???

But why do you need to bleede the caps? The C6+33uF caps for the last stage discharged after less than 2 sec by plate current of 10mA (the heater will be still heat ). The combined 4.3uF+4.3uF caps for the first two stages discharged in around the same time. If you follow the "original prototype" Cornet 3 schematic, you will have slightly longer time for the first two stages (about 10 seconds), but it is also not so big time.

I let myself recalculate Jims currents according to his voltages and 0.75mA was the result: 210v-120v=90v/120K=0.75; correct me if I am wrong.

I think you are right. The only doubt I have is about lower heater voltage... Maybe it is here something about such difference...

The thing with output cap is that I`ve calculated the lowest possible value here which doesn`t yet degrade frequency response looking into my linestage and further on. This is because I like Audio Note caps in signal coupling very much and they are expensive

You mean you have already the value and want to use it?

Poty, do you know the purpose of 220R resistor in front of C208 cap (you will not see it on my circuit)? I assume it is for buffer stability in some way...

I think it is just a precaution (like the plate resistor of the same 220 Ohm). If for some reason you short accidentally the output you have no load resistance for signal frequencies (the C208 is chosen in such a way it serves like the wire for signal). So it can be potentially harmful for tube (no restricting plate current resistance).

[GRD]

http://www.tubecad.com/2006/10/blog0083.htm

About half way down the article.

"The added cathode resistor further linearizes the cathode follower and buffers it from excessive capacitance."

I also have read that cathode followers can resonate at high frequencies and a small resistor can tame the rf resonance.

[GRD]

And more from tubecad.  See section 3 in the article.

http://www.glass-ware.com/tubecircuits/Tube_Line_Stage_Amplifier.html

[bartas]

So I have ordered the chokes today. Will be waiting for the delivery.
Let experiments begin! Really like this part

[hagtech]

Yes, in the octal cornet I ran the 6sn7 at 10mA.

jh

[bartas]

Hey, Jim,
glad you joined this topic.
What do you think of this hybrid? Are RIIA EQ values ok?
Could you verify the need of 220 Ohm resistors on 6SN7 plate and cathode?

I would be very glad to hear from you...

Bartas