-
Feeling lucky? Check out the AK 24th Anniversary Giveaway thread and join in the fun!
Caution on using 6GH8, 7687, 6U8 tubes to replace 7199 or 6AN8 in phase splitter circuits
- Thread starter Bob@Blue5
- Start date
<<Looks like the HK A300 and A500 use positive feedback between the volt amp and the PI cathodes. Probably why is oscillates with the 6GH8.>>
They do, but they use 12AX7, not 7199. If changed to a triode / pentode, there should be plenty of gain without positive feedback.
They do, but they use 12AX7, not 7199. If changed to a triode / pentode, there should be plenty of gain without positive feedback.
Sorry, I missed that.
The factory HK design of the A500 exhibits a voltage gain of approximately 340. That can't be equalled with a pentode-triode tube in this position. Even if it could, the sonic quality of the A500 is excellent, and I would be extremely hesitant to change the basic topology of the driver.
Jack
Wizard -- Your question is relevant to something I touch on briefly in my last thread on getting the most out of the Dynaco SCA-35 Amplifier. In it I discuss some of the measures needed to properly replace the 7199 tube used in that amplifier with more common 6GH8A/6U8A tubes. In that amplifier, if 7199 tube is simply replaced by using the more common tube types with an adapter, most often it will result in poor performance and various forms of oscillation generated.
Relevant to your question, one of the problems generated is the development of parasitic oscillations in the output stage at moderate and elevated output levels when substitute tubes are used without any appropriate compensation. For this particular problem, the cause can be found in a significant difference between the triode section of a 7199 tube, and that of a 6GH8A/6U8A tube, relative to the service the 7199 tube was specifically designed for.
The 7199 was one of the last receiving type vacuum tubes to be developed, so it has all the knowledge and experience of previous tube manufacturing and circuit design baked into it. But for all that experience, the triode section only has a rather low performing published Gm (transconductance) value of 2100 µmhos. That's no better than each section of a 12AU7, which was designed about 20 years prior to the 7199. But this designated performance level was intentional.
One of the greatest intended design uses of the 7199 was for it to be used as is so often seen: The pentode section is (often) direct coupled to the triode section which is used as a cathodyne phase splitter, which then directly drives an output stage without the use of any intermediary driver stage between the inverter and output stage. Countless amplifier designs have used this topology, because when used with high Gm output tubes, it can then make for a very economical circuit without as many stability issues that designs employing additional stages can have. But high Gm output tubes can bring their own stability issues as well if not properly accounted for. For this discussion, the concern is one of having a relatively similar drive impedance presented to the output tubes when high Gm tubes are used.
As is generally well known however, the two outputs from a cathodyne type phase inverter are quite different in terms of the impedance they represent, with the output at the plate being much higher than that at the cathode. This is an inherent characteristic of the design, resulting from the cathode output taking the form of a cathode follower configuration, which by way of the significant NFB that configuration generates, reduces the output impedance at the cathode to a much lower value than that produced at the plate. This is the reason that the plate and cathode resistors for the best designs using this type of inverter are rather low in value, to help mask the otherwise significant impedance differential it presents at its outputs.
But the characteristics of the triode section used for the inverter also plays heavily in this equation when it is used to directly drive a high Gm output stage. The salient point here is that with all else being equal, the higher the Gm is for any tube used in a cathode follower configuration, the lower its output impedance will be at the cathode. So now look at the published Gm value for the triode section of a 6GH8A -- 8500, or over 4X that of the 7199's triode section.
The bottom line is that the Gm of the 7199's triode section was intentionally kept low so that it could directly drive high Gm output stages without causing instability. It's "low performance" triode section worked to minimize the output impedance differential when used as a cathodyne inverter in this type of application. Replacing it then with a triode of much higher Gm then can invite output stage instability, and particularly so when that stage is of the UL type. This was definitely the case at hand with the design of the SCA-35 amplifier.
The cure then is to add a little series resistance in the cathode leg of such designs. In this way, it raises the drive impedance that the "bottom" output tube sees to be much more in line with that with the top output tube sees, and in that way, it eliminates the potential instability that the use of otherwise higher performance tri-pent driver tubes can produce under these conditions. As a result, when substituting a 6GH8A for the 7199 in the SCA-35, the inclusion of a 10KΩ stopper resistor in the gird circuit of the bottom output tube stopped the generation of parasitic oscillations on output waveforms dead in its tracks, that remedy being one part of the steps taken to properly substitute the 6GH8A in the SCA-35 circuit.
I hope this helps!
Dave
Relevant to your question, one of the problems generated is the development of parasitic oscillations in the output stage at moderate and elevated output levels when substitute tubes are used without any appropriate compensation. For this particular problem, the cause can be found in a significant difference between the triode section of a 7199 tube, and that of a 6GH8A/6U8A tube, relative to the service the 7199 tube was specifically designed for.
The 7199 was one of the last receiving type vacuum tubes to be developed, so it has all the knowledge and experience of previous tube manufacturing and circuit design baked into it. But for all that experience, the triode section only has a rather low performing published Gm (transconductance) value of 2100 µmhos. That's no better than each section of a 12AU7, which was designed about 20 years prior to the 7199. But this designated performance level was intentional.
One of the greatest intended design uses of the 7199 was for it to be used as is so often seen: The pentode section is (often) direct coupled to the triode section which is used as a cathodyne phase splitter, which then directly drives an output stage without the use of any intermediary driver stage between the inverter and output stage. Countless amplifier designs have used this topology, because when used with high Gm output tubes, it can then make for a very economical circuit without as many stability issues that designs employing additional stages can have. But high Gm output tubes can bring their own stability issues as well if not properly accounted for. For this discussion, the concern is one of having a relatively similar drive impedance presented to the output tubes when high Gm tubes are used.
As is generally well known however, the two outputs from a cathodyne type phase inverter are quite different in terms of the impedance they represent, with the output at the plate being much higher than that at the cathode. This is an inherent characteristic of the design, resulting from the cathode output taking the form of a cathode follower configuration, which by way of the significant NFB that configuration generates, reduces the output impedance at the cathode to a much lower value than that produced at the plate. This is the reason that the plate and cathode resistors for the best designs using this type of inverter are rather low in value, to help mask the otherwise significant impedance differential it presents at its outputs.
But the characteristics of the triode section used for the inverter also plays heavily in this equation when it is used to directly drive a high Gm output stage. The salient point here is that with all else being equal, the higher the Gm is for any tube used in a cathode follower configuration, the lower its output impedance will be at the cathode. So now look at the published Gm value for the triode section of a 6GH8A -- 8500, or over 4X that of the 7199's triode section.
The bottom line is that the Gm of the 7199's triode section was intentionally kept low so that it could directly drive high Gm output stages without causing instability. It's "low performance" triode section worked to minimize the output impedance differential when used as a cathodyne inverter in this type of application. Replacing it then with a triode of much higher Gm then can invite output stage instability, and particularly so when that stage is of the UL type. This was definitely the case at hand with the design of the SCA-35 amplifier.
The cure then is to add a little series resistance in the cathode leg of such designs. In this way, it raises the drive impedance that the "bottom" output tube sees to be much more in line with that with the top output tube sees, and in that way, it eliminates the potential instability that the use of otherwise higher performance tri-pent driver tubes can produce under these conditions. As a result, when substituting a 6GH8A for the 7199 in the SCA-35, the inclusion of a 10KΩ stopper resistor in the gird circuit of the bottom output tube stopped the generation of parasitic oscillations on output waveforms dead in its tracks, that remedy being one part of the steps taken to properly substitute the 6GH8A in the SCA-35 circuit.
I hope this helps!
Dave
WizardWG4
Well-Known Member
Thanks Dave! That definitely helps and I appreciate the detailed response.
I'm getting ready to work on my SCA-35 with your PC-10A and EFB boards. Would there be any issue with using a 10K grid stopper for the top output or should I stick with 1K top, 10K bottom?
Wiz -- Keep the upper tube at 1KΩ. If you make it a 10KΩ as well, then you're defeating the purpose of using dissimilar values.
Dave
Dave
WizardWG4
Well-Known Member
Will do. Thanks Dave!
Here's a rather random question: stretching beyond same-socket substitutions, I know that Triode Electronics has made an ST-70 driver board using two EF86 pentodes and a 12Au7 for the phase splitters, which incidentally has the benefit of drawing less heater current than two 7199s or 6GH8s (.7A for the EF86-12AU7-EF86 setup vs. .9A for the 7199/6GH8 x2). For retrofitting an amplifier where there is no place to centrally mount a third 9-pin socket, is there any reason that the same idea wouldn't work with a small triode like the 6C4 as a phase splitter in each channel? The 6C4 triode has similar characteristics to the triode in the 7199, and only pulls .15A for the heaters, half of a 12AU7, so it would still have the total .7A load on the 6.3V circuit. On something like my Scott LK-150, where the power supply capacitors and recitifer tubes take up the whole central section of the chassis, it would be rather trivial to either punch a hole for a 7-pin socket on each side next to the 9-pin sockets, or make an adapter board that plugs into the 9-pin socket with 7 and 9 pin PC mount sockets, with standoffs to hold it off the chassis on existing screw holes.
I guess the second half of that question would be “are circuit changes necessary when dropping an EF86 or similar tube into a circuit designed for the pentode section of a 7199 or 6U8/6GH8, likewise with 1/2 of a 12AU7 or a 6C4?” I’ve looked over the datasheets, and the only answer I’ve come up with is “I don’t know.”
There are ruggedized equivalents that may behave better.You may have better luck than I have, but finding a non-microphonic 6C4 has always proven difficult for me. I used to have to screen through dozens just to find a good one.
I hate them
There are no issues substituting the EF86 in the 7199 pentode circuit. I have quite a bit of data on the EF86, including charts showing gain and maximum signal swing for a wide range of voltages on the plate and screen. Some of this information came from tube books that I haven't seen online, and data sheets that may or may not still be available. I would be glad to share, but it might take some time to copy and upload. The EF86 is one of my favorite tubes, in part because it provides low distortion in both pentode and triode modes. In triode, it's one of the lowest distortion tubes ever made, and it works exceedingly well driving a long tail pair for push pull applications. I prefer that approach myself over the pentode/cathodyne configuration, because the first stage (which is primarily responsible for the distortion signature) operates over a much smaller portion of its total voltage capability. Another reason to love the EF86 os that the inexpensive Russian 6J32P is a direct sub and also performs extremely well.
Jack
The basic topology works, you may need to tweak resistor values to get the voltages correct though. Probably the screen resistor is the one to twiddle with though, it will affect bias points on both pentode and triode. The pentode won't care hugely but the triode does want to be reasonably biased or it will affect the output voltage swing.
Where can I find the data to try to work that out?
I usually work this stuff out experimentally, basically build it on a breadboard with stock values and see how it shakes out. If its funny, sub the screen resistor with a pot and mess with it until it seems sane. Normal voltage split on a triode used like this will be somewhere in the range of 1/3 to 1/2 the total supply voltage measured from plate to cathode, with the rest across the plate and cathode resistors. Mess with the screen resistor value until you get that.
I guess if you're cool and Spice doesn't make you want to smash things it could be done in a simulator.
I guess if you're cool and Spice doesn't make you want to smash things it could be done in a simulator.
Scott LK-150 rebuild/modification thread
Bump for opinions. I would have replied to the compliments then added the opinions part....but that's just me. Good luck with it.
audiokarma.org
Post 91 has the latest version of the schematic that dcgillespie and I worked up before life threw me a whole bunch of curveballs. I can build it up with the stock values, buy the new tubes and start playing with the measurements.
Last edited: