To 6A3sUMMER: The matched plate loads will probably go close to balancing the drive but does not allow for differences in each side of the tube. Also there will be a drift in characteristics over time. Using a CRO to balance the drives overcomes drift etc. The balance network in the cathodes of the output tubes eliminates residual magnetism in the core of the output transformer. Again the characteristics of the output tubes drift over time.
peteki,
Consider a cathode coupled phase splitter, with the grid of one side driven, and zero signal volts to the other grid. Then use matched plate loads; and matched coupling caps and matched grid resistors in the next stage:
Then with the cathodes directly connected together, and a perfect CCS (nearly perfect with extremely high impedance versus the cathode impedance), When the cathode current of one side changes, the other cathode current has to change exactly as much, but in the opposite direction (as long as either side does not have grid current). If the change in cathode currents is exactly the same except for one going up and the other going down, then it must be that the plate currents also change exactly the same except for the direction (phase). With equal plate current change and equal plate resistances, the voltage change has to be the same.
The use of a metering resistance, or balance pot between the cathodes is what can disturb the balance of this circuit, because no CCS is perfect. In that case, you would have to use un-equal plate loads to restore the balance.
Of course to make up for an unbalanced output stage, you may want to un-balance the phase invertor in the opposite direction, in order to cancel the output stage's less than perfect balance.
Consider a cathode coupled phase splitter, with the grid of one side driven, and zero signal volts to the other grid. Then use matched plate loads; and matched coupling caps and matched grid resistors in the next stage:
Then with the cathodes directly connected together, and a perfect CCS (nearly perfect with extremely high impedance versus the cathode impedance), When the cathode current of one side changes, the other cathode current has to change exactly as much, but in the opposite direction (as long as either side does not have grid current). If the change in cathode currents is exactly the same except for one going up and the other going down, then it must be that the plate currents also change exactly the same except for the direction (phase). With equal plate current change and equal plate resistances, the voltage change has to be the same.
The use of a metering resistance, or balance pot between the cathodes is what can disturb the balance of this circuit, because no CCS is perfect. In that case, you would have to use un-equal plate loads to restore the balance.
Of course to make up for an unbalanced output stage, you may want to un-balance the phase invertor in the opposite direction, in order to cancel the output stage's less than perfect balance.
kward, well, too far from Europe to buy
Can anybody advice on the choice of the KT88's? I prefer mostly rock, pop and classic (I presume that matters). Surprisingly I have found many good reviews on the Shuguang KT88Z and Psvane KT88-TII quads, but they seem to be most expensive. Alternatively I have at hand the Genalex, EH, Svetlana choice. Can you say some words on the most and less preferred for this type of amp?
Or may it be an option to build it and test it with something simple like Svetlana | EH and afterwards invest more expensive set of tubes?
Can anybody advice on the choice of the KT88's? I prefer mostly rock, pop and classic (I presume that matters). Surprisingly I have found many good reviews on the Shuguang KT88Z and Psvane KT88-TII quads, but they seem to be most expensive. Alternatively I have at hand the Genalex, EH, Svetlana choice. Can you say some words on the most and less preferred for this type of amp?
Or may it be an option to build it and test it with something simple like Svetlana | EH and afterwards invest more expensive set of tubes?
Last edited:
Some KT88's had not been de-gassed properly - Russian Sovtek, whilst most modern ones do not use cathode material that lasts if run near maximum recommended dissipation. My experience with current Genelex has been reasonable (they seem to have a good vacuum).
However none of the modern ones last.
However none of the modern ones last.
I do have the schemas, but I don't have permission to share them. However, your initial schema is correct. I have built the ST-70, 120, and M-125. Also the Curcio M-100. To my ears, the ST-120 sounds the best. The 120 also reproduces 1K and 10K square waves better than the others too...virtually perfect at 1K and very good at 10K. I built mine as mono-blocks because the 120 destroys rectifier tubes almost as fast as you can install them, LOL. Now my rectifiers are hardly stressed. I use 5U4BG's, I can afford the voltage drop. Sylvania chrome dome 6SN7's and KT-120 outputs. I have tried the Lundahl 1679 PP transformer, its a good transformer but the ST-120 trans is just as good as far as I can tell.
Theres not much to tune with the new CCS phase splitter driver board, except the bias. The older VTA driver boards had a balancing pot, but not these. It's my understanding that with the CCS, the splitter is inherently self balancing. Now that you ask, I may scope the output of the splitter though, if I invert one output and add the two I should be able to see any discrepancy. BTW, I ended up converting my M-100s to VTA's octal driver circuit, the regulated screen supplies in the M-100's went up in smoke. Pretty much all I kept was the iron.
Troy Madden,
Good! You checked the phase splitter balance. With a (very good) CCS in the coupled cathodes (and no balance pot), and balanced plate resistors, and balanced grid circuits of the next stage, the balance is essentially intrinsic. And that is what is in this amp.
The same could be said of a Concertina phase splitter.
But a 'serial' phase splitter where the output of the first stage drives a series resistive divider of the two plates, and the 'center' of the resistive divider drives the grid of the second stage . . . does not have intrinsic balance. This is where the scope channels and probes must be carefully adjusted and checked.
Differential probes are expensive; especially if they have very good balance of amplitude and phase, and especially at high frequencies. That means most scope users have to use 2 probes, 2 channels, and then set up with Invert, and Math (no-invert and Math minus (-), Or Invert one channel and Math plus (+). This is a Pseudo Differential test.
It is good to increase the accuracy of a Pseudo Differential test. There is an Oscilloscope and Probe trick, that many scope users are not aware of:
1. Adjust the probe compensation of channel 1 and its probe on the scopes calibration signal.
2. Adjust the probe compensation of channel 2 and its probe on the scopes calibration signal.
3. With both probes on the scopes calibration signal, invert One of the channels (i.e. channel 2 invert).
4. Use the Math 'Add' function and look at the Error signal (mismatch between Ch 1 / probe 1, versus Ch 2 / probe 2. Or do not invert, and use Math Minus (-). If the match is good, the Math trace will be zero signal. Any signal is the Error signal (imperfect match).
5. Now you can accurately check the tube phase invertor amplitude balance and phase balance of the invertor's 2 signals.
If you fail to perform steps 1 - 4 above, then the accuracy of step 5 is totally unknown.
Note: there is an additional in-accuracy, it is when you change the Volts / Division on the two channels. The Amplitude and / or Phase may no longer match . . .
Therefore, it is better to put a square wave into the amplifier and put both probes on one phase of the invertor signal at the same time. Turn Math off, and Adjust the Volts / Division according to the signal amplitude, Then compensate each channel probe on its channel trace. Now turn Math on and check the channels/probes balance. There should be no signal on the math trace (any signal is the error signal). Once the balance on that (single) signal is verified, put one of the probes on the opposite phase. Now check the tube phase inverter balance of the amplitude and phase on the math trace.
If you want to do this without a square wave, you can use sine waves. I sometimes do spot frequency testing: I use a test CD that has 21.5 Hz, 1,000 Hz, and 19,999 Hz. Check the scope and probe balance at each frequency with both probes on one phase; and then move one probe to the opposite phase to see any un-balance. You may see more un-balance at 21.5 Hz and 19,999 Hz than at 1,000 Hz.
Happy Measuring! Happy Listening!
Good! You checked the phase splitter balance. With a (very good) CCS in the coupled cathodes (and no balance pot), and balanced plate resistors, and balanced grid circuits of the next stage, the balance is essentially intrinsic. And that is what is in this amp.
The same could be said of a Concertina phase splitter.
But a 'serial' phase splitter where the output of the first stage drives a series resistive divider of the two plates, and the 'center' of the resistive divider drives the grid of the second stage . . . does not have intrinsic balance. This is where the scope channels and probes must be carefully adjusted and checked.
Differential probes are expensive; especially if they have very good balance of amplitude and phase, and especially at high frequencies. That means most scope users have to use 2 probes, 2 channels, and then set up with Invert, and Math (no-invert and Math minus (-), Or Invert one channel and Math plus (+). This is a Pseudo Differential test.
It is good to increase the accuracy of a Pseudo Differential test. There is an Oscilloscope and Probe trick, that many scope users are not aware of:
1. Adjust the probe compensation of channel 1 and its probe on the scopes calibration signal.
2. Adjust the probe compensation of channel 2 and its probe on the scopes calibration signal.
3. With both probes on the scopes calibration signal, invert One of the channels (i.e. channel 2 invert).
4. Use the Math 'Add' function and look at the Error signal (mismatch between Ch 1 / probe 1, versus Ch 2 / probe 2. Or do not invert, and use Math Minus (-). If the match is good, the Math trace will be zero signal. Any signal is the Error signal (imperfect match).
5. Now you can accurately check the tube phase invertor amplitude balance and phase balance of the invertor's 2 signals.
If you fail to perform steps 1 - 4 above, then the accuracy of step 5 is totally unknown.
Note: there is an additional in-accuracy, it is when you change the Volts / Division on the two channels. The Amplitude and / or Phase may no longer match . . .
Therefore, it is better to put a square wave into the amplifier and put both probes on one phase of the invertor signal at the same time. Turn Math off, and Adjust the Volts / Division according to the signal amplitude, Then compensate each channel probe on its channel trace. Now turn Math on and check the channels/probes balance. There should be no signal on the math trace (any signal is the error signal). Once the balance on that (single) signal is verified, put one of the probes on the opposite phase. Now check the tube phase inverter balance of the amplitude and phase on the math trace.
If you want to do this without a square wave, you can use sine waves. I sometimes do spot frequency testing: I use a test CD that has 21.5 Hz, 1,000 Hz, and 19,999 Hz. Check the scope and probe balance at each frequency with both probes on one phase; and then move one probe to the opposite phase to see any un-balance. You may see more un-balance at 21.5 Hz and 19,999 Hz than at 1,000 Hz.
Happy Measuring! Happy Listening!
Last edited:
Looks like pretty the same driver circuit here, but with noval tubes....http://www.diytube.com/unidriver/poseidon.pdf
I agree the 6SN7 is preferable. I used Nichicon poly film for audio coupling, a couple of .47uf poly Orange Drops across the main power supply caps, Nichicon 47uf low ESR for filtering on the driver board, silver mica in the feedback loop. Nothing too fancy, just good quality caps. I had my fling with boutique caps in the past, don't get too carried away theses days!
Well, some long time has passed, we are back in business.
Has anyone used one of these capacitors for audio coupling? They should be for 600 V in case something goes wrong:
1. Mundorf MCap Supreme 1200 VDC 0,33 и 0,1?
2. Mundorf MCap EVO SilverGold.Oil axial 650 VDC 0.33 и 0,1?
3. Capacitor Jantzen Silver Z-Cap MKP 1200 VDC?
The power supply is slightly bellow 600 V. I can't find suitabele 100 mF / 600 V capacitors. Any hints I can use?
Has anyone used one of these capacitors for audio coupling? They should be for 600 V in case something goes wrong:
1. Mundorf MCap Supreme 1200 VDC 0,33 и 0,1?
2. Mundorf MCap EVO SilverGold.Oil axial 650 VDC 0.33 и 0,1?
3. Capacitor Jantzen Silver Z-Cap MKP 1200 VDC?
The power supply is slightly bellow 600 V. I can't find suitabele 100 mF / 600 V capacitors. Any hints I can use?
As for power supply caps, the most reasonable is serially connected 400 - 450Volt electrolytics with suitable leveling resistors.
All those are coupling caps are so good that I can’t hear much of a difference between the first two, with perhaps a slight advantage to the EVO SilverGold, but I have not used the third. Some people with good hearing apparently hear bigger difference than I. I’m currently very interested in Miflex from Poland as a highly rated, but still affordable caps. If you want to read about a ranking of capacitors you can look here:
Humble Homemade Hifi - Cap Test
Are you still working from the schematic you initially posted? Sorry, if I missed an update. I remember you said B+ was 509V. Could you post you revised schematic with the higher supply voltage?
I found that film capacitors in the power supply enhance performance over electrolytics. Several companies now offer box type film caps in voltage ranges over 600V. They are more expensive than ‘lytics, but sound really good and last forever. See for example:
EZP-V60117MTC Panasonic | Mouser
The Lundahl LL1679 transformer is really good and 4.5k impedance will be perfect for two PP KT88. Or are you now leaning to KT120? The LL1679 AM with amourphous core is even better, if you can afford it.
Schematic in Post # 1 violates KT88 maximum operating conditions:
KT88 with adjustable fixed bias, 100k maximum grid resistor.
But the schematic shows 150k grid resistor. Ouch!
150k is easier for the driver tubes, but can cause the KT88 to go into Thermal Run-away.
And the 6550 with adjustable fixed bias, 50k maximum grid resistor. Don't even think about tube-rolling with a 6550 in your amplifier.
KT88 with adjustable fixed bias, 100k maximum grid resistor.
But the schematic shows 150k grid resistor. Ouch!
150k is easier for the driver tubes, but can cause the KT88 to go into Thermal Run-away.
And the 6550 with adjustable fixed bias, 50k maximum grid resistor. Don't even think about tube-rolling with a 6550 in your amplifier.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.