I have it running pretty well now. I’ve parallel some 1k resistors with the 680R cathode resistor (404R) to get the dissipation up to about 18 watts. Plate voltage is 388 above the cathode. Screen is regulated at 255 (235 above the cathode). The transformers measure about 26:1 for 5200 at 8R load. Cathode resistor current is 46ma at idle. I’m operating real close to the max operating line but I don’t see any signs of distress, however I think 480R cathode resistors would be a better choice.
Here‘s how it looks using the 6L6G curves as a proxy for the 6P3S. The curve gradient is 5 volts.
Here‘s how it looks using the 6L6G curves as a proxy for the 6P3S. The curve gradient is 5 volts.
BTW
The Tun-Sol data sheet shows conditions for UL for the 7591. Plate and screen are show as 400V max.
Like you I am not familiar with a commercial product that used the 7591 in UL operation.
Be interested in know how much output power it delivers before it clips.
I am not sure but would be concerned the lower screen voltage above the cathode (235V) may limit the max. cathode current to less than required for full power output.
However if you raise the screen voltage then you have to increase the cathode resistor and that will cause power loss as well. No free lunch.
Cathode bias is a balancing act between screen voltage and cathode resistance to get the most power from any one B+ voltage and transformer impedance.
BTW
Having a easily changeable screen voltage that is regulated makes setting the bias current easy with cathode bias, just dial up or down the screen voltage a bit to tune the bias current for the cathode resistor used. Gives you some range in bias current without having to constantally change the resistor value.
I saw that TungSol datasheet yesterday and thought of my remark, haha.
Good point. have been tempted to add a pot to the screen supply. Right now it‘s binary, 250 or 300v.
It’s making just shy of 11 volts RMS on the 8 ohm load. Let’s say close to 15 watts.
I noticed that load line on post #21 doesn’t look right. The transformers are 30:1 measured 300 across the anodes and 10 at the 8 ohm tap. 30^2*8=7200. Shouldn’t the class A slope be 7200/2= 3600? That slope looks like 5600. GIGO I suppose.
Good point. have been tempted to add a pot to the screen supply. Right now it‘s binary, 250 or 300v.
It’s making just shy of 11 volts RMS on the 8 ohm load. Let’s say close to 15 watts.
I noticed that load line on post #21 doesn’t look right. The transformers are 30:1 measured 300 across the anodes and 10 at the 8 ohm tap. 30^2*8=7200. Shouldn’t the class A slope be 7200/2= 3600? That slope looks like 5600. GIGO I suppose.
Last edited:
This might be correct. I’m measuring about 300 rms across both anodes just before clipping. 150v per tube, 212 pk. That load line green shaded area is 212 v. Do I have that right? I struggle with push pull load lines.
Last edited:
That is a leisurely number.
You seem to have 388 - 212V = 126V anode to cathode at the bottom of the SINE wave.
Most well driven pentodes can pull the anode down to between 60V~90V from the cathode so some voltage is being given up here some where.
What we do not know is what the cathode is doing when the amp is driven. Cathode bias can wander up in voltage when tested with continuous SINE waves eating some of the power output available. Part of the price paid for the stability and ease of cathode bias.
Maybe have a look at the cathode voltage during power testing.
If bypassed by a good size capacitor with music this is less of a issue as the transient nature of music makes movements in cathode voltage smaller and less of a issue.
BYW What is the cathode resistor bypassed with?
Did you measure B+ during power testing. It may well have drooped a fair bit depending on the power supply costing you some power output.
Again like the cathode voltage rise the B+ droop is less of a issue with music.
With 388V plate to cathode and the ability to pull the cathode to anode voltage down to 80 volts or so then you should be seeing 388-80 or 308 volts peak of swing on the each plate.
With a 6600 P-P primary impedance (standard 7591 transformer) and as each tube sees 1/4 of the P-P impedance, peak anode current for each tube will be 308V/ (6600/4) = 187mA
308V peak * 187mA peak = 57.6 watts peak.
57.6 watts peak / 1.414^2 = 28.8W RMS
Take off 10% transformer loss and you get 28 * .9 =~ 25 watts at the speaker
25 watts at the speaker is about the power level I have gotten in my rebuilds of 7591 units with the 6P3S.
Mind you they were done with fixed bias and a B+ of about 425V.
Last edited:
I believe you’ve surrounded the problem right here. It doesn’t have quite enough drive. I suspect this amp was intended to be a class A design from the start. The first and only AF stage is a 12ax7 with a 100k plate resistor and 3.3k bias resistor. Changing the bias resistor to 2.2K helped improve the OP some. It’s now 1.5v and .65ma. It should swing 100v pk. Next is a cathodyne phase splitter with 100k resistors and a 3.3k bias resistor. I changed that one to 2.2k also. It can only handle about 50v pk on paper and less in practice. This is where I need to follow up.
Also, it will do about 325vrms across the primary at clipping. The 300v number I cited was at the end of class A.
Last edited:
Yours looks a bit like a Fisher x100c. Nothing wrong with the old Fishers however the total drive gain is limited for a 6P3S.
Here is what I have used to replace 7591 power sections with 6P3S.
Note diagram shows 5881 outputs but I always used the 6P3S or 6P3S-e tubes with this circuit.
B+ is about 425V.
The out transformer (not shown) is whatever came from the 7591 doaner.
I found Heathkit AA-100 transformers (power and output) worked nicely and were at one time pretty cheep.
The Hammond 1650H also work well but are pricey.
It produces about 25 watts at the speaker with pretty low distortion and decent damping.
It is a fixed bias design for maximum power output.
Note the grid resistors on the 6P3S are only 100K to insure bias stability when driven hard.
The 100K value does add extra demands on the driver stage.
Bias voltage is around -40V volts (regulated), setting idle current to about 30mA for tube dissipation of about 13 watts.
You could set the current higher but I like long tube life and the distortion is pretty good at 30mA bias.
The screens run around 360V-380V regulated. Yes this is over the data sheet however it has proven very reliable in HiFi use.
With lower screen voltages I found maximum power started to drop off so I settled on 360V to 380V.
Some time I will try the 6P3S tubes with a lower screen voltage and AB2 drive (like a 807) and see what they will do.
No idea if G1 on a 6p3S will like the grid current of AB2 opperation.
I find the drive stage gain important with 6p3S if there is to be enough global feedback to get a low damping factor with pentode output tubes.
The drive stage (6F12P) can supply before any feedback about 47dB gain with less than 1% distortion @ 40VRMS per grid.
I find a lot of gain and drive needed for the 6p3S much like a 6L6.
Total gain to each plate of the 6p3S is 64dB.
Total gain to the 8 ohm speaker tap before any feedback is about 41dB.
Here is what I have used to replace 7591 power sections with 6P3S.
Note diagram shows 5881 outputs but I always used the 6P3S or 6P3S-e tubes with this circuit.
B+ is about 425V.
The out transformer (not shown) is whatever came from the 7591 doaner.
I found Heathkit AA-100 transformers (power and output) worked nicely and were at one time pretty cheep.
The Hammond 1650H also work well but are pricey.
It produces about 25 watts at the speaker with pretty low distortion and decent damping.
It is a fixed bias design for maximum power output.
Note the grid resistors on the 6P3S are only 100K to insure bias stability when driven hard.
The 100K value does add extra demands on the driver stage.
Bias voltage is around -40V volts (regulated), setting idle current to about 30mA for tube dissipation of about 13 watts.
You could set the current higher but I like long tube life and the distortion is pretty good at 30mA bias.
The screens run around 360V-380V regulated. Yes this is over the data sheet however it has proven very reliable in HiFi use.
With lower screen voltages I found maximum power started to drop off so I settled on 360V to 380V.
Some time I will try the 6P3S tubes with a lower screen voltage and AB2 drive (like a 807) and see what they will do.
No idea if G1 on a 6p3S will like the grid current of AB2 opperation.
I find the drive stage gain important with 6p3S if there is to be enough global feedback to get a low damping factor with pentode output tubes.
The drive stage (6F12P) can supply before any feedback about 47dB gain with less than 1% distortion @ 40VRMS per grid.
I find a lot of gain and drive needed for the 6p3S much like a 6L6.
Total gain to each plate of the 6p3S is 64dB.
Total gain to the 8 ohm speaker tap before any feedback is about 41dB.
It’s a stereo amp and truthfully, it wasn’t really “planned”. I picked it up locally third-hand for small money because I recognized the Fisher 7591A transformers. That and a pair of good telefunkin 12ax7! Did I mention the almost new quad of JJ 6l6GC?
It’s a bit ugly and it sounded just horrible. The guy said he got it at an estate sale. I suspect the builder never quite finished it. The biggest issue was the phase splitter. It was biased wrong and was clipping the cathode/non-inverted signal. It sounds just fine now but it’s probably not reaching its potential.
It’s a bit ugly and it sounded just horrible. The guy said he got it at an estate sale. I suspect the builder never quite finished it. The biggest issue was the phase splitter. It was biased wrong and was clipping the cathode/non-inverted signal. It sounds just fine now but it’s probably not reaching its potential.
For stereo I'd replace one of the 12AX7 for a 12AU7, using the *U7 as phase splitter (one triode per channel), and the *X7 for the input stage.
Blue systems, I like the DC coupled driver with the elevated AF stage. I had been considering that because I have a preamp that can put out 6 or 8 volts and I like the DC coupled circuit feeding a concertina. I just makes sense.
I see your feedback circuit is interesting. The drawing doesn’t indicate the transformer taps but I see you have the 4 at ground. That’s because you need the feedback out of phase I assume. I’ve seen that on Scott designs and the rationale never clicked until now. Which taps do you use then for your 8 ohm load?
Is the 6BL8 close enough to the 6F12P to substitute? I have a few on hand.
I’m running Polk SDA-2B in my main rig. They require a common ground. Something to consider.
The Amp currently has no global feedback.
I see your feedback circuit is interesting. The drawing doesn’t indicate the transformer taps but I see you have the 4 at ground. That’s because you need the feedback out of phase I assume. I’ve seen that on Scott designs and the rationale never clicked until now. Which taps do you use then for your 8 ohm load?
Is the 6BL8 close enough to the 6F12P to substitute? I have a few on hand.
I’m running Polk SDA-2B in my main rig. They require a common ground. Something to consider.
The Amp currently has no global feedback.
Last edited:
Yes, your point is well taken. The 12DW7 was made for this application. I’ve always been a little superstitious about running both channels together on one tube. It’s probably just that, but are there separation issues or crosstalk to consider? My reluctance is based on something I read years ago and have long forgotten, but it lingers among the cobwebs in the attic.
Last edited:
I get your point. I would expect crosstalk to be virtually nonexistent at audio frequencies for an indirect heated double triode. It is quite common for amplifiers to share a double triode for the first stage. But I have never done real measurements.
The transconductance (S) of the pentode in the 6BL8 is a little low compared to a good 6F12P but at 6.2 mA/V very workable.
Consider the 7199 has a pentode S of 7 mA/V, a tube widely used in the past for the input stage.
Lower transconductance generally means lower gain for the same bandwidth in the first gain stage.
So transconductance, gain and bandwidth are a three way compromise.
The triode section of the 6BL8 is great as it can draw lots of current at low (50v) plate to cathode voltage.
This is important if you want a wide voltage swing in the phase inverter at low enough impedance to keep the bandwidth wide and have the ability to drive low value G1 resistors in the output tubes.
This is one reason to not use a 12AX7 as a phase inverter.
The issue I have is with the 6BL8 and all pentode/triode tubes that have the triode plate (pin 1) and the pentode G1 (pin 2) on adjacent pins is this creates a small amount of stray capacitance between the input grid of the high gain pentode and the in phase audio output of the triode's plate.
So at high frequencies this sets up positive feedback from the triode plate output to the pentode G1 grid input. The capacitance is very small but due to the very high gain any stray capacitance is multiplied by the gain.
For 46dB gain this increases the effective stray capacitance by 200 and this positive feedback impairs the stability and transient response of the amplifier.
I discovered this effect years ago working on replacements for the 7199 tube.
Amplifiers that had nice square wave response with the 7199 would have nasty ringing with any tube that featured pin 1 pin 2 pairing.
Tubes like the 7199, 6AN8, 6AW8A or 6F12P do not have these pins beside each other and have no stability issues at high gains.
The 6AW8A can work however the triode is low in current ability at low plate/cathode voltages and so marginal in driving large signals levels unless you keep the B+ above 340 volts.
The 6BA8, 6BH8 looks good on paper with good pentode transconductance of 8 and 7 mA/V respectively.
Also both have 7mA or more of triode plate current with only 50V plate to cathode.
I have not tried either however.
Be warned that a lot of sellers of the 6F12P are selling defective tubes with low emissions and transconductance.
Sadly for me I learned this the hard way. However a good 6F12P is a excellent tube.
The common ground can easly be moved from the 4 ohm tap to the speaker negative and then the feed back taken from the 4, 8 or 16 ohm tap as wanted.
All one must do is reverse the plate connectons to the transformer to switch the phase the feedback to keep it neagative.
That way a common ground can be used.