• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Cathode follower driver

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi guys,

I just wrote down this schematic.
It is a three stage integrated amplifier:
- 1st stage (12AX7) is a CCS loaded SE driver, should be very linear.
- 2nd stage (6N6P) is a cathode follower
- 3rd stage (KT88) is a parafeed output stage, loaded by a CCS (actually a KT88 pentode CCS).

I know first stage can work, already tried it in the past; I have also built the third stage (parafeed output stage, at the moment on my desk under test and development), and drove it with a 1:10 transformer, fed by my PC sound card.

The new stage is the cathode follower in the middle, included to add some juice to the KT88 driver.

The feedback is optional, at the moment there's no feedback at all in the prototype.

My questions:

1. Is the second stage correctly developed? Would it work well?

2. How can I determine the values for R4 and R5? I would like to fix the idle operating point of 6N6P tube to around 120V@20mA.

3. What happen if I substitute the R4 resistor with a CCS@20mA?

4. The alternative to this topology is to use the 6N6P tube as a plate follower, using a CCS to load it: would this topology work worse or better than cathode follower?

5. 12AX7 and 6N6P have two triode each, so I could use one of them for each channel (I'm building a stereo amplifier, of course): is there any problem using the same valve for both stereo channels?

Thanks in advance.

Ciao,
Giovanni Albergoni
 
I second Gluca's opinion. Direct coupling to the power tube by driver as a cathode follower. The Italians have used this topology a lot; ie the Geloso amps. You will need to have a split supply for the driver, so the driver's cathode will feed the proper negative bias voltage to the output's grid. This makes for a powerful driver that is extremely fast; this will sound good.
Using a cathode follower direct coupled to the power tube has many advantages but is harder to implement.
Capacitors are generally bad but a necessary evil and if you can avoid using any, then that is the way to go.
Daniel
 
You don't need R4.
Value of R5 depends on a maximum current you need to drive your output monster from one side, from the other side it is limited by max current 6N6 can sustain (power dissipation) without significant distortions and plate melting.
And I would use some high voltage power PNP instead of pentode for CCS...

Gentlemen who suggest you direct cupling forgot about the 1'st triode that needs some positive voltage on the plate to work properly.
 
croccodillo said:

1. Is the second stage correctly developed? Would it work well?

2. How can I determine the values for R4 and R5? I would like to fix the idle operating point of 6N6P tube to around 120V@20mA.

3. What happen if I substitute the R4 resistor with a CCS@20mA?

4. The alternative to this topology is to use the 6N6P tube as a plate follower, using a CCS to load it: would this topology work worse or better than cathode follower?

5. 12AX7 and 6N6P have two triode each, so I could use one of them for each channel (I'm building a stereo amplifier, of course): is there any problem using the same valve for both stereo channels?

Thanks in advance.

Ciao,
Giovanni Albergoni

1) Looks OK, however, I would add a negative rail, and DC couple the cathode follower to the final. That way, you get the advantages of fixed bias, with less power wasteage, better THD, enough current to charge up Ci + Cmiller + Cstray without overdrive blocking. Move the AC coupling to the first preamp.

2) Loadline. Also, R4 either isn't necessary, or pick one with 1 -- 5V across it and bypass with an electrolytic if stability and/or PSRR is gonna be a problem. With a cathode follower, you want the plate to be at AC ground.

3) You waste a CCS, and invite instability. If you're gonna do that, then the CCS belongs in the cathode lead, not the plate lead.

4) Much, much worse. Do that and your feedback becomes positive instead of negative. All you'll get is a fixed frequency square wave generator.

5) Yes. No.

Furthermore, I would ditch C4, and connect that gNFB at the load on the OPT secondary, That way, the xfmr is included in the error correction. This will help to linearize it, and reduce the Zo for better woofer damping and better bass performance.
 
Hi Wavebourn

I can only use other people arguments, but actually that is not that bad at all! I will use the argument's by Morgan Jones, in his Valve amplifiers book. Hmm...before spitting rubbish here (2:33 in the morning) I will read Morgan Jones again, and understand exactly why. If noone explained why ray moth's approach is better, I will try to do it, maybe tomorrow.

But it has to do with overloading of stages and class of operation. THe cathode follower can't be overloaded, but the output stage can...

I have to go to bed... ray moth, jump in (I know you have experience with the Crystal Palace).

Erik
 
ErikdeBest said:
Hi Wavebourn

I can only use other people arguments, but actually that is not that bad at all! I will use the argument's by Morgan Jones, in his Valve amplifiers book. Hmm...before spitting rubbish here (2:33 in the morning) I will read Morgan Jones again, and understand exactly why. If noone explained why ray moth's approach is better, I will try to do it, maybe tomorrow.

But it has to do with overloading of stages and class of operation. THe cathode follower can't be overloaded, but the output stage can...

I have to go to bed... ray moth, jump in (I know you have experience with the Crystal Palace).

Erik

I mean, what reason in moving the coupling capacitor one stage ahead?

Possible answer is that a grid current of the output toob will charge the cap causing dynamic bias.
 
That's very much it. "Blocking" is the word you're looking for. In most designs, the output stage will clip first, and direct-coupling the cathode follower to the output tube grid minimizes the pain caused by grid current piling up charge on the coupling cap, overbiasing the stage for a time constant or three.
 
Ex-Moderator
Joined 2004
Wavebourn, you were the one who pointed out the problem, I thought you would understand my answers to it. As you said, the first triode must have a positive voltage om the plate (compared to its cathode, that is). The second triode (the cf) needs a negative voltage on its grid if it is directly coupled to the KT88.

One (obvious) way to resolve this would be to have AC coupling between the first triode plate and the second triode grid, so they can operate at their correct DC voltages. This is quite normal where directr coupled cf drivers are used.

Another (possible) way could be to have DC coupling from start to finish but use a negative rail for the cathode resistor of the first triode. I doubt if it would be practical, however, because that negative rail would also have to be the basis for supplying the bias voltage for the KT88 (via the cf) and I think it would be difficult to stabilize.
 
Thanks all,

You're right, DC coupled stages are much better.
My power transformer has some auxiliary windings that can be used for the purpose; it has 4 different secondary windings:

- 250V 300mA
- 250V 300mA
- 100V 100mA
- 100V 100mA

So, now I'm using only the first two windings to have a B+=300+300V=600V (two separate rectifier on thw two windings, two regulators and then the two 300V voltages are connected in serie).
In this way I can have a central voltage of 300V, useful for output capacitor (less stress) and for drivers.

I can use the last two windings to build up a 125+125V, and use those voltages for a negative supply.
With such a negative supply I can connect the 12AX7 cathode to -250V, setting its working point at about 250V, and the 6N6P cathode to the -125V supply, and thus having lot of room for the output tube drive swing.


And I would use some high voltage power PNP instead of pentode for CCS
The CCS for the driver is realised with a mosfet around a schematic I drawn; when I will be sure it works fine I will show you.
The CCS for the output tube is actually a pentode KT88 driven by a SS CCS, realised with another P mosfet; actually the pentode and the mosfet are cascoded.
This is the latest schematic, where you can see also the parafeed CCS (not the SS schematic, however).
Actually the voltage regulator for the upper pentode screen is much more complicated, involving a mosfet, diodes, zeners, in manner to keep as far as I can the pentode screen voltage stable to 300V with respect with its cathode.

My bigger doubt now is about the input: is there any problem driving the amplifier with my CD player, since the input tube (12AX7) grid is at about -250V?

Comments? Could it work? Suggestions?

Ciao,
Giovanni
 
Croccodillo I thought you might be interested in an observation I have made comparing my choke loaded parafeed version compared to my CCS loaded version. Despite the poor efficiency of the CCS loaded version it has much tighter bass than the choke loaded version. Of course there could be any number of reasons for this, but personally I attribute this to the fact that the CCS version behaves more like a PP output with better current drive throughout the wave cycle.
This results in tighter punchier bass overall.
So poor efficiency but better sound. You pays your money and makes your choice.


Shoog
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.