ray_moth said:
I was trying to keep the thread general but, since you called me on itmy specific interest had to do with the possibility of using some Sovtek 6SN7s that I have as cf drivers for EL34s in a PP amp (pentode or triode, haven't decided yet). I wondered if the Sovteks would sound OK as cfs, even if they're not so good in grounded cathode configuration. However, from reading the general trend of replies, it would seem that a higfher gm, higher mu tube (or a MOSFET) would be a better solution anyway. MOSFET's are OK by me, I'm not interested in 'purity' - this is a hobby, not a religion!
How about use the gary pimm CCS as plate loading your 6SN7. pimm's CCS has a mu out (mosfet follower) that will have enough juice to drive the EL34.
The mu out can be couped to the EL34 via a capacitor. Or you can do something like Gary did to his 300B amp. Have the mu out directly couped to the EL34.
ray_moth said:...I'm not interested in 'purity' - this is a hobby, not a religion!
Hey, I'm not interested in a philosophical food fight (well, OK, a little food fight might be fun - we can throw Easter eggs). But isn't this hobby a place where you can strive for perfection without commercial constraints? Also, why even bother with tubes if purity is not important? I see no reason to use tubes if we throw out signal purity as a desirable trait. Transistors make cones vibrate too. Just havin’ fun with you… Don't ever take me too seriously
SY said:"Ideas have consequences."
Yah-well.
Since there seems not to be too much worry regarding mixing ss and glass (also sand basically) .....
let us then go for a complete transistor amplifier! (Ohhhhhhhh...)
I get a quite lower distortion design (no, not thd - give me credit for some grey matter!) on a 100W transistor amplifier with moderate NFB, at 40% of the parts cost than a tube amplifier ... and both these designs are top-of-the class!
You did say the best of both worlds ......
Seriously (not that the above was not), where does one stop? It is not normative, but to me there is somehow some hidden attitude in the mind that suggests a sort of ... er ... capitulation by mixing the technologies. Pure "man's vanity" I am sure, still ...... Imagine whether you will ever fit a 1959 Chev Impala or Rolls Silver Shadow with a computerised engine "organiser". (Now you have done it, JP!!)
Technically, what about the input C of a MOSFET however well proportioned? CFs are also used for low loading on the previous stage, say a high gain pentode, RL=220K. Lower than 15 pF? That starts to cut at 48KHz, just sliding into NFB stability criteria.
Regards!
Johan Potgieter said:
Not sure. How about this one:
http://db.audioasylum.com/cgi/m.mpl?forum=magnequest&n=5167&highlight=transconductance&r=&session=
current amp. SS driver, Pentode output stage.
@ ray_moth:
I'm using Sovtek 6SN7s as CF drivers for an Class AB1, 807 (pentode mode) amp. They work just great, and can take the finals part way into Class AB2 until the OPTs core saturate. Sounds great.
@ Brian Beck:
"What's a "Kimmel"? As in Alan Kimmel? You must mean a mu-follower with low output Z - that certainly helps . Then why use a FET follower at all if you already have a mu-follower?"
Yes, Kimmel Mu-Stage. This design is for Class A2, 812 monoblocks, which won't work unless you take the control grids positive. The Kimmel all by itself can't handle the control grid current demand. A MOSFET source follower can do so with ease, and with an even lower Zo. So far as the voltage requirement, this runs off +/- 125Vdc rails, and swings 80Vp-p. The idle current is around 50mA, so that will dissipate about 12W. You wouldn't do any better with a xfmr coupled, VT driver in that department. As for parametric frequency multiplication in SS devices, there isn't anything you can do about that. It's still a problem with total SS implementations, only moreso since there are more SS devices in the signal path. No perfect technologies, and there are always design trade-offs. We can only do the best we can with what we've got, and I do believe that, overall, it won't be such a problem, especially with judicious application of gNFB.
"You got me. I don't know what a "NBG" is either." No Big Deal.
I'm using Sovtek 6SN7s as CF drivers for an Class AB1, 807 (pentode mode) amp. They work just great, and can take the finals part way into Class AB2 until the OPTs core saturate. Sounds great.
@ Brian Beck:
"What's a "Kimmel"? As in Alan Kimmel? You must mean a mu-follower with low output Z - that certainly helps . Then why use a FET follower at all if you already have a mu-follower?"
Yes, Kimmel Mu-Stage. This design is for Class A2, 812 monoblocks, which won't work unless you take the control grids positive. The Kimmel all by itself can't handle the control grid current demand. A MOSFET source follower can do so with ease, and with an even lower Zo. So far as the voltage requirement, this runs off +/- 125Vdc rails, and swings 80Vp-p. The idle current is around 50mA, so that will dissipate about 12W. You wouldn't do any better with a xfmr coupled, VT driver in that department. As for parametric frequency multiplication in SS devices, there isn't anything you can do about that. It's still a problem with total SS implementations, only moreso since there are more SS devices in the signal path. No perfect technologies, and there are always design trade-offs. We can only do the best we can with what we've got, and I do believe that, overall, it won't be such a problem, especially with judicious application of gNFB.
"You got me. I don't know what a "NBG" is either." No Big Deal.
Hey, I wear a shirt when I go out of the house (most of the time).
The tube VS mosfet follower debate has some strong supporters on both sides. There is valid engineering support for both sides. There is quite a bit of emotion on each side also. When I first published the PowerDrive circuit, I got lotts of hate mail. I was even told to change my name to Transistorlab.
I have seen data where a high quality mosfet is tested against a high GM tube. The slight edge went to the mosfet. In this testing the load was a fixed resistance. I find that the best place to use a mosfet follower is to drive the grid of a power output tube. Especially a tube that draws grid current. In this situation the load is not constant and go from megohms to ohms when the grid transitions positive. If you are going to use a CF, use one that has a high peak current capability. The prevoiusly mentioned vertical output tubes are best here.
It would be interesting to test a tube CF against a mosfet follower, with the grid of a tube in A2 for a load. I have this set up on my bench, but haven't had the time to do it yet.
Yes the capacitance of the gate - drain terminals varies with the pplied voltage. This change effect is mush worse at lower voltages. The cure is (as mentioned) to keep the voltage across the fet high. Yes, I run the fet in my 845 amp at 20 watts. Some of the newer fets have now capacitance, and the capacitances are low iv the voltage across the fet is over 25 volts or so.
I will post the data whwen I can get it done.
The Whits cf is something for which I have never found any use. If you read what John Broskie has to say, it seems unable to source much current. I found it in the tube CAD online journal.
Tweeker said:
Hi Tweeker,
I've been tinkling with the follower as well, played with 5965, 6N1p, 6922... and it does what it advertised, very quiet... & I think it's neutral too.
with 2 tubes in series, I feel that it has more fuel to feed the power grid current demand... but it's only an hunch...
The 6SN7 will give Zo of between 450~850ohm depending on operating point.
My question is how low is low enough? or should we be looking at how much plate current available in the CF output to overwhelm the power grid instead??
Cheers
If your really trying to provide significant current and not just low Z.O, you need more than most signal tubes can provide. As mentioned earlier here, it is advised that the follower idle >5x more current than its asked to provide. Not always practical here, but you can do better. If using 6SN7 Id use GTA/GTB types, parallel the sections and run them at 7w/envelope.
Don't underestimate the ability of the grid in an output tube to draw serious current as it is driven positive. I had previously measured grid currents of over 100 mA in large transmitting tubes. This current is drawn only as the grid transitions positive and rises sharply and returns to zero over a fraction of the audio cycle. During the remainder of the audio cycle the grid is a very high impedance. This will cause some distortion in any driver circuit. The amount of distortion is dependent on the drivers output impedance. The output impedance is a function of the tube's characteristics, as well as the circuit topology. A tube with low Mu and Gm will have distortion problems in any circuit.
I was testing the feasibility of using A2 on a 6L6GC so I hooked up a PowerDrive circuit and cranked things up. I put the scope probes on either side of the 100 ohm grid stopper and made a differential measurement. I was seeing pulses of 10 volts on signal peaks. This implies 100 mA of grid current on a 6L6GC. The grid stopper turned out to be a source of distortion. I replaced it with a wire and picked up more power and reduced distortion. This implies that your driver needs to have an output impedance below 100 ohms for intended A2 operation. The elimination of overload may not impose such severe requirements. That testing was done while setting up for some measurements in this thread (near the end):
http://www.diyaudio.com/forums/showthread.php?s=&threadid=97384&highlight=
I have not had a chance to finish any of these measurements. This experiment has opened several new questions in my mind about drivers, cathode VS mosfet followers and A2 (SE) / AB2 (P-P) operation. I tried the same experiments with a KT88 and did not discover as much of a power output improvement.
I am not advocating the use of mosfet followers in every circuit, but if you are going to drive an output tubes grid positive, they are hard to beat. Clearly a lot more experimentation is in order. I will get to it as I have time.
I was testing the feasibility of using A2 on a 6L6GC so I hooked up a PowerDrive circuit and cranked things up. I put the scope probes on either side of the 100 ohm grid stopper and made a differential measurement. I was seeing pulses of 10 volts on signal peaks. This implies 100 mA of grid current on a 6L6GC. The grid stopper turned out to be a source of distortion. I replaced it with a wire and picked up more power and reduced distortion. This implies that your driver needs to have an output impedance below 100 ohms for intended A2 operation. The elimination of overload may not impose such severe requirements. That testing was done while setting up for some measurements in this thread (near the end):
http://www.diyaudio.com/forums/showthread.php?s=&threadid=97384&highlight=
I have not had a chance to finish any of these measurements. This experiment has opened several new questions in my mind about drivers, cathode VS mosfet followers and A2 (SE) / AB2 (P-P) operation. I tried the same experiments with a KT88 and did not discover as much of a power output improvement.
I am not advocating the use of mosfet followers in every circuit, but if you are going to drive an output tubes grid positive, they are hard to beat. Clearly a lot more experimentation is in order. I will get to it as I have time.
Hi tubecad,
Great of you to share your experience.
Zo<100ohm, 100mA transcient peak.... There aren't many common signal tube which can cope. More reasonable to use small power tubes with low low rp and high peak capability, maybe parallel some TV deflection tubes? 6SN7, 12B4, EL509, EL519? or Maybe some regulator tubes such as 6AS7, 6C41C, or better still 6C33C... But is linearlity of CF tube going to effected distortion?
Yes, the idling current is huge in comparison...
Cheers
Great of you to share your experience.
Zo<100ohm, 100mA transcient peak.... There aren't many common signal tube which can cope. More reasonable to use small power tubes with low low rp and high peak capability, maybe parallel some TV deflection tubes? 6SN7, 12B4, EL509, EL519? or Maybe some regulator tubes such as 6AS7, 6C41C, or better still 6C33C... But is linearlity of CF tube going to effected distortion?
Yes, the idling current is huge in comparison...
Cheers
Both RDH and in particular Valley and Wallman say that pentodes make superior (to triode) cathode followers. Kimmer says that MOSFETS are superior to pentodes (and easier to implement since you don't have to worry about keeping Vg2-k voltage constant).
Valley and Wallman say to use an active load (CCS) on the triode cathode follower to get the best out of them.The White cathode follower is just a minor variation of a triode CCS loaded triode cathode follower by using some feedback into the CCS.
Ray,
I've been looking at the CF for exactly the same duty as your intended use - that is driving outut tubes in Class AB. The problem is the eliptical load line. You can draw this up by using the usual load line, then calculating the capacitive reactance of the output stage Miller capacitance at the highest frequency of interest (MJ used 20kHz in Crystal Palace calcs but I'd suggest using at least double that). That reactance and the required peak to peak voltage swing gives you the peak to peak current swing which will be + and minus vertically about the idle point on the load line. You can then sketch your elipse to go through those points. That current swing needs to be reasonably low w.r.t. the idle current.
By the way, you said "The White cf is something for which I have never found any use"
Would you believe - Over the Easter break I completed repair of an AMPEG SVT3 PRO 300 Watt Bass Guitar Amp. It uses a 12AU7 White Cathode Follower to drive the 8 output MOSFETS (4 push pull Class AB pairs).
Cheers,
Ian
Valley and Wallman say to use an active load (CCS) on the triode cathode follower to get the best out of them.The White cathode follower is just a minor variation of a triode CCS loaded triode cathode follower by using some feedback into the CCS.
Ray,
I've been looking at the CF for exactly the same duty as your intended use - that is driving outut tubes in Class AB. The problem is the eliptical load line. You can draw this up by using the usual load line, then calculating the capacitive reactance of the output stage Miller capacitance at the highest frequency of interest (MJ used 20kHz in Crystal Palace calcs but I'd suggest using at least double that). That reactance and the required peak to peak voltage swing gives you the peak to peak current swing which will be + and minus vertically about the idle point on the load line. You can then sketch your elipse to go through those points. That current swing needs to be reasonably low w.r.t. the idle current.
By the way, you said "The White cf is something for which I have never found any use"
Would you believe - Over the Easter break I completed repair of an AMPEG SVT3 PRO 300 Watt Bass Guitar Amp. It uses a 12AU7 White Cathode Follower to drive the 8 output MOSFETS (4 push pull Class AB pairs).
Cheers,
Ian
Ian,
To broaden my horizons ... I presume we are still basically talking of hi-fi amplifiers, so where does class AB2 really come in? As you gather I have never used them as domestic amplifiers; I would accept that as theater units they have a place. On the other hand, class AB can still give several hundred watt. A GEC class AB2 amplifier of 700W is giving a maximum of 5% distortion. Is this acceptable for the guitar scene - otherwise?
Regards.
To broaden my horizons ... I presume we are still basically talking of hi-fi amplifiers, so where does class AB2 really come in? As you gather I have never used them as domestic amplifiers; I would accept that as theater units they have a place. On the other hand, class AB can still give several hundred watt. A GEC class AB2 amplifier of 700W is giving a maximum of 5% distortion. Is this acceptable for the guitar scene - otherwise?
Regards.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.