I believe you are OK then because the SCA-35 pentode section in the 7199 is not biased by Grid Leak - it is conventionally biased by R31 on the cathode and the grid is referenced to ground by the 4.7M resistor. (According to the voltage chart in the manual the control grid should be at 0V wrt ground so it is not Grld Leak Biased). C17 is there to keep DC out of the 7199, since it is not getting it’s input from a known source like the ST-70 is pre-amplified by a PAS with requisite output capacitors.
You may not agree with it, but as @rayma stated the shelf network is essential for the proper functioning of the ST-70. Likewise, using a correctly functioning 7199 is essential for the standard Dynaco SCA-35 implementation of PC-10. If you use something else you are on your own to perform the redesign and testing.
The fact that you are measuring a plate voltage on the pentode of 83V in stead of 55V per the voltage chart probably means that you are not passing enough current to drop the required number of volts through R30, probably because the pentode is not biased correctly. You are also getting a bias of -11V on the triode section (83-94) so the phase inverter is incorrectly biased also. The chart is of course based on 7199, so I don’t know if you have reengineered your component values for 6BL8.
You originally asked for a “mod suggestion”, not a redesign for 6BL8. I understood you wanted a mod to change to something other than 7199 and Dave’s PC-10A boards that are designed and optimized for 6GH8a or 6U8 tubes (not 6BL8) were suggested. If you want to change from 7199 to something else, Dave’s boards are still my “mod suggestion”. It will save you a lot of redesign work, testing and perhaps headaches.
Last edited:
Today, I attempted to check how 6GH8 fits in my circuit.
I found the stability issue of the positive feedback by accident. It sometimes got stuck at low frequency oscillation, below 50Hz at full swing at the output.
The reason I never find this problem with the speakers is that, I guess as long as you put the load to the transformer, the oscillation is gone. Even though, I don't like it. For me, an amp should be stable even without speaker connected. The oscillation happened to 6BL8. It also happened to 6GH8, but with less extend.
The culprit is the positive feedback. In the original circuit, the positive feedback is through R33, 120K. It regenerates some or all the current that the plate load resistor 270K needed. The lower value the R33 is, the more current you get regenerated. The positive feedback, if you over do it, it may oscillate.
This guy wrote up this in very details. I highly recommend to read through his article below regarding the positive feedback. https://www.angelfire.com/electronic/funwithtubes/pentode-triode.html
The minimum value of the regenerate resistor is 150K based on the article above. That is with 220K plate load resistor.
In the SCA-35, the plate resistor is 270K, I figured 180K should be the minimum for the regenerate resistor. Yep, this did fix the oscillation for me.
The question is, why SCA-35 uses only 120K. The optimal regenerate resistor value also depends on the transconductance(gm) of the pentode. I guess the gm of the pentode in 7199 is pretty low comparing to 6GH8 and 6BL8.
BTW, all my test above was with the RC shunt network added in front of the phase splitter as the Dave's PC-10A design. The high frequency response is always well behaved even with 10K square wave. The oscillation always happens (if it happens) at very low frequency below 50Hz.
Dave did mention his board doesn't work with 6BL8. Besides it has different bias point (less plate current comparing to 6GH8), I found, it is more prone to oscillate because of the positive feedback.
I found the stability issue of the positive feedback by accident. It sometimes got stuck at low frequency oscillation, below 50Hz at full swing at the output.
The reason I never find this problem with the speakers is that, I guess as long as you put the load to the transformer, the oscillation is gone. Even though, I don't like it. For me, an amp should be stable even without speaker connected. The oscillation happened to 6BL8. It also happened to 6GH8, but with less extend.
The culprit is the positive feedback. In the original circuit, the positive feedback is through R33, 120K. It regenerates some or all the current that the plate load resistor 270K needed. The lower value the R33 is, the more current you get regenerated. The positive feedback, if you over do it, it may oscillate.
This guy wrote up this in very details. I highly recommend to read through his article below regarding the positive feedback. https://www.angelfire.com/electronic/funwithtubes/pentode-triode.html
The minimum value of the regenerate resistor is 150K based on the article above. That is with 220K plate load resistor.
In the SCA-35, the plate resistor is 270K, I figured 180K should be the minimum for the regenerate resistor. Yep, this did fix the oscillation for me.
The question is, why SCA-35 uses only 120K. The optimal regenerate resistor value also depends on the transconductance(gm) of the pentode. I guess the gm of the pentode in 7199 is pretty low comparing to 6GH8 and 6BL8.
BTW, all my test above was with the RC shunt network added in front of the phase splitter as the Dave's PC-10A design. The high frequency response is always well behaved even with 10K square wave. The oscillation always happens (if it happens) at very low frequency below 50Hz.
Dave did mention his board doesn't work with 6BL8. Besides it has different bias point (less plate current comparing to 6GH8), I found, it is more prone to oscillate because of the positive feedback.
Last edited:
Maybe we don't really need the positive feedback. A simple bootstrap can get you very high gain without worrying about stability issue.
I don't have tube model. I just use high voltage mosfet in the position of the tubes. You should get the idea.
I don't have tube model. I just use high voltage mosfet in the position of the tubes. You should get the idea.
As the voltage gain of the cathode follower is always less than 1. It has less chance to oscillate.
This might be my final version. The pentode is loaded with the bootstrap. Don't oversize the bootstrap cap, 0.1u in my circuit. The corner frequency of the bootstrap network should be higher than the lowest frequency that OPT transformer can produce. Otherwise, it might motorboat.
It is stable. Just as I predicted, because of no positive feedback, it is stable even without the RC shunt network in front of the phase splitter.
Below is 10KHz, 2Vp response without speaker connected. The signal is injected right before the power amp PC-10 board. Thus, the tone controls are bypassed.
It would be more stable after the load is connected.
Other stuffs that you can mod it farther.
1. Change the negative feedback point from 16 Ohm tab to 8 Ohm tab. If I am right, the new nfb resistor should be 56K
2. For 6BQ5 tubes, replace the cathode resistor with a shunt regulator, or two (one per channel). You can make adjustable shunt regulators with TL431. Like below.
Or you can just purchase D. Gillespie's SCA-35 EFB mod.
PS: Here are the voltages around the pentode-triode tube.
HT voltage on the PC-10 board is about 320v. It is aligned with Dynaco's manual.
Case 1: RCA 6BL8
Pentode:
Plate: 110v
Grid No.2: 33v
Cathode: 0.50v
Triode:
Plate: 209v
Cathode: 116v
Case 2: RCA 6GH8
Pentode:
Plate: 73v
Grid No.2: 32v
Cathode: 0.58v
Triode:
Plate: 236v
Cathode: 83v
I would say I like 6GH8 in my circuit more, as the triode portion has more headroom.
It is stable. Just as I predicted, because of no positive feedback, it is stable even without the RC shunt network in front of the phase splitter.
Below is 10KHz, 2Vp response without speaker connected. The signal is injected right before the power amp PC-10 board. Thus, the tone controls are bypassed.
It would be more stable after the load is connected.
Other stuffs that you can mod it farther.
1. Change the negative feedback point from 16 Ohm tab to 8 Ohm tab. If I am right, the new nfb resistor should be 56K
2. For 6BQ5 tubes, replace the cathode resistor with a shunt regulator, or two (one per channel). You can make adjustable shunt regulators with TL431. Like below.
Or you can just purchase D. Gillespie's SCA-35 EFB mod.
PS: Here are the voltages around the pentode-triode tube.
HT voltage on the PC-10 board is about 320v. It is aligned with Dynaco's manual.
Case 1: RCA 6BL8
Pentode:
Plate: 110v
Grid No.2: 33v
Cathode: 0.50v
Triode:
Plate: 209v
Cathode: 116v
Case 2: RCA 6GH8
Pentode:
Plate: 73v
Grid No.2: 32v
Cathode: 0.58v
Triode:
Plate: 236v
Cathode: 83v
I would say I like 6GH8 in my circuit more, as the triode portion has more headroom.
Last edited:
I did test with a real speaker. As it may disturb other ones in the house, the test was short and I did not take the photo. At this moment, I only want to make sure it is stable. The speaker load just makes the amp more stable. The square wave will be rounded off, no ringing.
Because of laziness, I didn’t pull out my dummy load this time.
Usually, for full power test, I use 110v 1500W space heater as the dummy load. As speakers all have inductance at high frequency. Some inductance shouldn’t matter much.
What’s the best tube to replace 7199?
The candidates I could think of are 6BL8, 6U8, 6EA8, 6GH8.
Among them, 6BL8 is kind of special. From what I read, 6BL8 is the only one constructed with a true pentode. The rest of them are designed with a beam tetrode as a pseudo pentode. As those tubes have similar specs. It is possible that manufacturers might relabel the original 6BL8 to others or vice versa.
The 6GH8 seems to come later in the timeline. It also has very high gain with both pentode and triode. 6GH8 is kind of the best one based on the specs.
One thing that we need to watch for is the hum that coupled from the heating filament. It is especially critical for SCA-35 as the amp board has very high voltage gain over 140x. I only have 2 6GH8 tubes, and one of them exhibited extensively filament hum.
The candidates I could think of are 6BL8, 6U8, 6EA8, 6GH8.
Among them, 6BL8 is kind of special. From what I read, 6BL8 is the only one constructed with a true pentode. The rest of them are designed with a beam tetrode as a pseudo pentode. As those tubes have similar specs. It is possible that manufacturers might relabel the original 6BL8 to others or vice versa.
The 6GH8 seems to come later in the timeline. It also has very high gain with both pentode and triode. 6GH8 is kind of the best one based on the specs.
One thing that we need to watch for is the hum that coupled from the heating filament. It is especially critical for SCA-35 as the amp board has very high voltage gain over 140x. I only have 2 6GH8 tubes, and one of them exhibited extensively filament hum.
Tried one before, it hums. Might be just bad luck.
Maybe the brand matters more than the tube type. The two with hum (6GH8 and 6U8) are GE branded.
My variants RCA and Sylvania ones don’t hum that much.
Those tubes above are almost NOS if the seller was honest.
Below, I would like to add the measurement of a used 7199 tube. The circuit is still the one posted in #26. As it was measured at the different day and the line voltage may have changed, but it is still useful.
7199 (used)
Pentode:
Plate: 93v
Grid No.2: 26v
Cathode: 0.58v
Triode:
Plate: 235v
Cathode: 101v
You can see the bias point is more similiar to 6GH8.
The datasheet mentioned 7199 should have lower noise/hum level than other tubes. However, I don't notice there is any noise advantage comparing to my RCA 6BL8. In the past, I found 6GH8 and 6U8 had large variance. Some of them suffer hums, some don't. That's one of the reason I lean towards 6BL8 tubes.
If you are seeking the best tube of this type you may consider 7687, designed by Sylvania to excel in low hum and noise performance.
https://bms.isjtr.ro/sheets/168/7/7687.pdf
https://bms.isjtr.ro/sheets/168/7/7687.pdf
Yep, that one is promissing. I might pick a pair when the price is right.
I notice the left channal is always noiser than the right channel. I believe something is causing the low level hum other than the tubes. I swapped the tube from channel to channel, same result. Better tubes won't solve this issue. Someone mentioned it is simply because the left channel is closer to the power transformer. If that is true, how to address this?
Some amplifiers have the power transformer on rubber grommets. I suppose it could be vibrating and transmitting that hum microphonically in the amplifier. Or you could try some metal tube sleeves to screen the tubes. I think a lot depends if it is 60hz or 120hz. If it is 60hz it is more likely to be being propagated mechanically.
I scope the 8 Ohm tap. The hum is 60Hz, about 5mVp.
17W max over 8 Ohm tap, the Vp is about 16V. Thus, the signal to noise ratio is only about 70dB. The random noise floor is about 1mV rms. Ideally, you want to see the random noise cover the hum entirely.
Another interesting finding is that the ground(chassis) is carrying some portion of 60 hum when I scope the ground. My SCA-35 is with the original 2-prong plug. Thus, the ground is just floating there. By design, the oscillascope is grounded through its 3-prong plug. Is that an issue?
17W max over 8 Ohm tap, the Vp is about 16V. Thus, the signal to noise ratio is only about 70dB. The random noise floor is about 1mV rms. Ideally, you want to see the random noise cover the hum entirely.
Another interesting finding is that the ground(chassis) is carrying some portion of 60 hum when I scope the ground. My SCA-35 is with the original 2-prong plug. Thus, the ground is just floating there. By design, the oscillascope is grounded through its 3-prong plug. Is that an issue?