The following link refers to just this on a 45 DHT. Would like to bump this to find if anyone has implemented use of this tube?
Application Note 05. The white spark problem.
Obvious question is this a solution looking for a problem, or is it an elegant solution for a problem that has required a huge amount of effort to address previously? Thinking if there are benefits, then DHT Preamp is an obvious application.
Regards
Grantn
Application Note 05. The white spark problem.
Obvious question is this a solution looking for a problem, or is it an elegant solution for a problem that has required a huge amount of effort to address previously? Thinking if there are benefits, then DHT Preamp is an obvious application.
Regards
Grantn
4П1Л (4P1L) cannot really be used on ac anyway, since it is more than usually prone to hum.
Even eliminating hum does not cure the real problem of ac-heating - bad sound due to intermodulation distortion. Yes, clumsy application of dc-heating is even worse still, but there's no need to do that.
With properly-implemented current-driven dc heating, there is no hum at all. Expect 100/120Hz component in the low microampere region, to achieve no hum even if the centre-tap is not used, and IMD all but vanished.
Kevin is using my regulator solution for the 4P1L.
Even eliminating hum does not cure the real problem of ac-heating - bad sound due to intermodulation distortion. Yes, clumsy application of dc-heating is even worse still, but there's no need to do that.
With properly-implemented current-driven dc heating, there is no hum at all. Expect 100/120Hz component in the low microampere region, to achieve no hum even if the centre-tap is not used, and IMD all but vanished.
Kevin is using my regulator solution for the 4P1L.
In the interests of full disclosure, I'm not really into the vacuum state devices anymore. But having said that, I'd think the ideal solution for any DHT would be a regulated DC supply with a slow start feature to minimize shocking the filament with the cold start up current. Or the other possibility is that I have no idea what I'm talking about. 
Mike
Mike
In the interests of full disclosure, I'm not really into the vacuum state devices anymore. But having said that, I'd think the ideal solution for any DHT would be a regulated DC supply with a slow start feature to minimize shocking the filament with the cold start up current. Or the other possibility is that I have no idea what I'm talking about.
Mike
Yes, the modern-production DHTs are somewhat prone to filament breakage. Running on ac (or raw dc, or voltage-driven dc) certainly does not help, since the resistance of the filament when cold is ⅓ of the value when hot, or even less. The turn-ON current surges accordingly. With current-driven heating, there is no surge, and the heating power rises gently.
I have build a fully balanced linestage withthe 4p1l in LTP, so the two cathodes need to be connected together and go together to a ccs.
I can see now a couple of different connection/setups and ask myself which one would be the best in terms of consistent electron flow:
A. 2.1V setup - Both coleman regs positive side go to the ccs:
- One pos. reg goes to the middle-tap of the 4p1l (pin 8),
- one pos. reg goes to the other tube to the paralleled pins 1,7
---> Pin 8 of one tube is connected to pin 1,7 of the other tube.
B. 2.1V setup - Both coleman pos. reg's side go to the ccs, but:
- Both pos. reg goes to the middle tap (pin 8) of each tube
---> Pin 8 of both tubes are connected together.
C. same as B., but pos. regs and ccs connected to both Pin1,7 of each tube
D. 4.2V setup:
- Both positive reg outputs go in Pin 1 of each tube, both neg. go into Pin 7
- CCS is connected to Pin 8 of each tube, so the middle-tap forms the cathode like in an IDHT see here Application Note 05. The white spark problem.
E. Same as D, but
- pos. reg out in tube 1 goes to Pin 1 and pos. reg out goes onto tube 2 on pin 7
My suspicion, but not sure if this makes sense:
- In a LTP, both triodes should be in an ideal world absolutely the same in terms of current flow.
- The CCS is automatically biasing the tubes and this works if tunes are heavily matched plus used in the same way/physical current flow between CCS, cathode/filament and anode
- We have to evaluate if the DC loading of the cathode leads to a consistent grid distortion as over the physical cathode the Voltage between Grid (grounded) and cathode changes by 2.1V which is at a bias of -10V more than 20% !!!
my 0.00000005ct:
- In scenario A: the electrons flow to anode from the CCS to the pins defined above, so one tube gets its electrons from the centertab, the other from the two connectors at the end of the filament. Electron-flow/distribution is not perfectly the same between the two tubes.
- In B.: Both tubes behave the same way with electrons coming from one point in the center of the cathode. Like in A, this center point of current flow is more positive than the endpoints of the filament, forcing the electrons to the outside
- In C.: Both tubes behave the same with electrons coming from two points (pin1,7), with Pin8 more negative, so forcing electron to the center
- In D: Both tubes behave the same, but middle point is as well electrical middle point regarding Cathode-grid, overall grid distortion might be higher as -2.1-0-2.1V potential of the heater needs to be distributed between grid and cathode. But cathode-grid is in the middle at least...
- In E: nearly lik D, but the relative Grid to Cathode potential in the tube is flipped around
I would not be surprised if you can hear the differences between these options...I can see that this might be not so bad when dealing with a 300b, were 5V vs. a bias of -70V needs to be considered, but with 2.1V(4.2V) vs. -10V grid voltage, the correct setup might matter....
I can see now a couple of different connection/setups and ask myself which one would be the best in terms of consistent electron flow:
A. 2.1V setup - Both coleman regs positive side go to the ccs:
- One pos. reg goes to the middle-tap of the 4p1l (pin 8),
- one pos. reg goes to the other tube to the paralleled pins 1,7
---> Pin 8 of one tube is connected to pin 1,7 of the other tube.
B. 2.1V setup - Both coleman pos. reg's side go to the ccs, but:
- Both pos. reg goes to the middle tap (pin 8) of each tube
---> Pin 8 of both tubes are connected together.
C. same as B., but pos. regs and ccs connected to both Pin1,7 of each tube
D. 4.2V setup:
- Both positive reg outputs go in Pin 1 of each tube, both neg. go into Pin 7
- CCS is connected to Pin 8 of each tube, so the middle-tap forms the cathode like in an IDHT see here Application Note 05. The white spark problem.
E. Same as D, but
- pos. reg out in tube 1 goes to Pin 1 and pos. reg out goes onto tube 2 on pin 7
My suspicion, but not sure if this makes sense:
- In a LTP, both triodes should be in an ideal world absolutely the same in terms of current flow.
- The CCS is automatically biasing the tubes and this works if tunes are heavily matched plus used in the same way/physical current flow between CCS, cathode/filament and anode
- We have to evaluate if the DC loading of the cathode leads to a consistent grid distortion as over the physical cathode the Voltage between Grid (grounded) and cathode changes by 2.1V which is at a bias of -10V more than 20% !!!
my 0.00000005ct:
- In scenario A: the electrons flow to anode from the CCS to the pins defined above, so one tube gets its electrons from the centertab, the other from the two connectors at the end of the filament. Electron-flow/distribution is not perfectly the same between the two tubes.
- In B.: Both tubes behave the same way with electrons coming from one point in the center of the cathode. Like in A, this center point of current flow is more positive than the endpoints of the filament, forcing the electrons to the outside
- In C.: Both tubes behave the same with electrons coming from two points (pin1,7), with Pin8 more negative, so forcing electron to the center
- In D: Both tubes behave the same, but middle point is as well electrical middle point regarding Cathode-grid, overall grid distortion might be higher as -2.1-0-2.1V potential of the heater needs to be distributed between grid and cathode. But cathode-grid is in the middle at least...
- In E: nearly lik D, but the relative Grid to Cathode potential in the tube is flipped around
I would not be surprised if you can hear the differences between these options...I can see that this might be not so bad when dealing with a 300b, were 5V vs. a bias of -70V needs to be considered, but with 2.1V(4.2V) vs. -10V grid voltage, the correct setup might matter....
Last edited:
Hi Frank,
The connexion method in B. or C. are close to the method used in other amplifiers - and should give excellent results. This is the « known-good » connexion.
The 2.1V mode is best for the 4P1L, to minimise the bias-skew across the filament.
With separate regulators for each 4P1L, there is no crosstalk of anode-current between the two filaments, which is an important consideration for the best possible sound.
The regulators have high ac-impedance looking out from the filament, to prevent the short-circuit or dissipation of the differential signal-voltage, that appears across the filament at all audio frequencies. Naturally, there must be no other components or wiring connected to the filaments: only the regulators and the cathode's CCS.
The connexion method in B. or C. are close to the method used in other amplifiers - and should give excellent results. This is the « known-good » connexion.
The 2.1V mode is best for the 4P1L, to minimise the bias-skew across the filament.
With separate regulators for each 4P1L, there is no crosstalk of anode-current between the two filaments, which is an important consideration for the best possible sound.
The regulators have high ac-impedance looking out from the filament, to prevent the short-circuit or dissipation of the differential signal-voltage, that appears across the filament at all audio frequencies. Naturally, there must be no other components or wiring connected to the filaments: only the regulators and the cathode's CCS.
Hi Rod,
Thx for the clarification...one question: Do I have to use the positive leg of each filament reg in any case attached to the ccs/cathodes-connection ? Even if the ccs's negative connection does not go to ground, but has its own negative supply voltage ? Currently the negative leg of the ccs sees -20v, resulting with each 4p1l at 30mA and about 170V in -10V grid voltage...as the reg is flowing and the ccs as well, I would have thought that I could use either way the legs of you regs...so, attaching the negative side to pin 8 connected together with the ccs would create the most negative single point e.g. at tge physical middle point of the cathode.
Basically option B., but with negative legs of the reags attached to the ccs..
Thx for the clarification...one question: Do I have to use the positive leg of each filament reg in any case attached to the ccs/cathodes-connection ? Even if the ccs's negative connection does not go to ground, but has its own negative supply voltage ? Currently the negative leg of the ccs sees -20v, resulting with each 4p1l at 30mA and about 170V in -10V grid voltage...as the reg is flowing and the ccs as well, I would have thought that I could use either way the legs of you regs...so, attaching the negative side to pin 8 connected together with the ccs would create the most negative single point e.g. at tge physical middle point of the cathode.
Basically option B., but with negative legs of the reags attached to the ccs..
Hi Frank,
You can try negative-return instead of positive-return, and the function will be appear to work equally well..... but using the positive-side has proved to sound better in most cases. I don't have an objective reason for it, but can only suggest that you design the system to be able to test it in both ways, to be sure of getting the best results.
You can try negative-return instead of positive-return, and the function will be appear to work equally well..... but using the positive-side has proved to sound better in most cases. I don't have an objective reason for it, but can only suggest that you design the system to be able to test it in both ways, to be sure of getting the best results.
Hi Rod,
I have just finished my experiments and listening session. You can hear quiet clearly a significant difference between the different setups.
Initially I started with A. and I had a bit strange, fussy holographic experience, I could no longer pin-point the artists or instruments, it felt like you are wearing a big headphone sometimes, like as if something is mixing up the phase.
I asked Rod and Kevin Carter for their advise and Kevin told me that it is not very likely that ghe LL1692a is the cause, while Rod pointed me to try different cathode-setups.
So, back at home I now changed it to both pin 8(mid-tap) got connected together and the ccs. Additionally, pos. legs of the two heater regs per channel to pin 8. Result: The feeling of listening to a wrong ohase was gone. 3D and holografic presentation got much better. Still though, dynamics where less now than before, even worse than A. Felt a bit like driving with a brake on.
So, I turned the negative reg legs now to connected cathodes/CCS instead. Bingo !!! Now we have it: The image is there, the phase is very natural and we have micro and macro dynamics. The magnitude of the difference I would rate in the same category like changing from an OK cap to a real good one.
So, i will for the moment not touch it anymore and continue to optimize other details, maybe I will test at some point of time all other options above as well, but currently it feels just extremely right. THX again for pointing me into the right direction ! Great support !
I have just finished my experiments and listening session. You can hear quiet clearly a significant difference between the different setups.
Initially I started with A. and I had a bit strange, fussy holographic experience, I could no longer pin-point the artists or instruments, it felt like you are wearing a big headphone sometimes, like as if something is mixing up the phase.
I asked Rod and Kevin Carter for their advise and Kevin told me that it is not very likely that ghe LL1692a is the cause, while Rod pointed me to try different cathode-setups.
So, back at home I now changed it to both pin 8(mid-tap) got connected together and the ccs. Additionally, pos. legs of the two heater regs per channel to pin 8. Result: The feeling of listening to a wrong ohase was gone. 3D and holografic presentation got much better. Still though, dynamics where less now than before, even worse than A. Felt a bit like driving with a brake on.
So, I turned the negative reg legs now to connected cathodes/CCS instead. Bingo !!! Now we have it: The image is there, the phase is very natural and we have micro and macro dynamics. The magnitude of the difference I would rate in the same category like changing from an OK cap to a real good one.
So, i will for the moment not touch it anymore and continue to optimize other details, maybe I will test at some point of time all other options above as well, but currently it feels just extremely right. THX again for pointing me into the right direction ! Great support !
Hi Frank,
That is a very useful data-point - thank you for feeding it back to us. I can only say that it reinforces the point that one should design new amplifiers to be able to experiment with the routing of the cathode circuit, where DHTs are involved.
If we have CCS 'tail', it means allowing a little extra negative supply voltage. And if we have grid bias, it mean that the bias circuit must be easily adjusted across a range of voltages greater than the filament voltage.
Good work, and good luck with the rest of your design!
That is a very useful data-point - thank you for feeding it back to us. I can only say that it reinforces the point that one should design new amplifiers to be able to experiment with the routing of the cathode circuit, where DHTs are involved.
If we have CCS 'tail', it means allowing a little extra negative supply voltage. And if we have grid bias, it mean that the bias circuit must be easily adjusted across a range of voltages greater than the filament voltage.
Good work, and good luck with the rest of your design!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.