Not sure if any of this is really needed. As I said earlier I also don't believe that it's the source of your sparking rectifier. We don't know any details about your power supply and this is likely where the problem is. What is the total B+ voltage, choke input or cap input, any inrush limiting? what is the size of the input filter cap, CLC, CRC ....
We do not know the value of the negative voltage source fed to the CCS, but 50 Volts has been assumed.
Lets look at the LTP circuit. There is a 56 mA CCS pulling current through the tube cathodes. The grids are essentially grounded. To get 28 mA (per tube) with 300 volts on the plate, the bias voltage needs to be around -12 volts between cathode and grid (curves from 1940 RCA data sheet). The CCS will suck 28 mA through the cathodes resulting in a POSITIVE VOLTAGE on the cathodes of around 12 volts. This should be measurable in normal operation. The reason for the negative voltage source is for CCS headroom with signal swing, and to limit it's VVC effects (discussed later).
On power up the negative voltage hits the CCS and it tries to suck 56 mA through the cathodes of the tubes, but the plate voltage is not there yet. It will pull the cathodes negative until is gets 56 mA, or saturates trying to do so. The 801's are warmed up, so they do what they are supposed to do, conduct. You now have a variable negative voltage source on the cathodes which is current limited to 56 mA, zero volts on the grid, and an unknown resistance (bleeder resistor?) in series with whatever plate load is on the tube connected to the plate. The tube will conduct and cathode voltage will either hit the negative supply rail or settle at whatever voltage the tube wants for 28 mA. Most of this will be grid current, but some plate current is flowing, or trying to flow since the B+ voltage goes negative.
I see no absolute grid current spec for the 801, but 15 mA is seen as the normal DC grid current and the curves extend to 80 mA. The 15 mA of DC current does not include the RF current flowing in RF amp duty, so my guess is that 28 mA for a few seconds won't hurt anything, and you can insure that more of that is plate current.
Add the diode that I talked about earlier. This will provide a path for the plate current to flow, and limit the negative voltage on the B+ supply to a safe value. I tried to link the schematic for my current driver board in a previous post, but I picked a file that was too large. The proper one is here.
I hesitate to use a silicon device for limiting the maximum negative voltage on the cathodes since all silicon devices are Voltage Variable Capacitors in their reverse biased regions, but much worse at low voltages. We choose mosfets used in tube amps based on their capacitance specs VS voltage curves to minimize this, and run them at higher voltages where the capacitance curves are nearly constant.
Many believe that this parameter is responsible for the TIM and PIM distortion effects that feed the "all silicon is evil" crowd. You will see a common NE2 neon lamp wired from the cathodes to ground.
Many of my boards use +/- 160 volt supplies for the mosfet drivers to reduce VVC. This could break down the HK interface when the board is powered up and all tubes are cold. The NE2 will light up limiting the voltage to about 75 volts. Once the tubes are conducting, the lamp goes out and becomes a fixed 1 pF cap. Unless the negative supply rail is higher than 75 volts the NE2 is not needed here.
If your rectifier still sparks after the diode is added, verify that the input filter cap is not too high for the tube being used. I see a Tesla spec where 60 uF is the maximum, but the curves of current VS secondary resistance are all done with 16 uF. The usual cause of sparking rectifiers is too much capacitance, to low of secondary resistance, or both. There are several ways to fix the sparking rectifier tube. I have added two of them to my SSE boards since it runs the rectifier hard and many current production tubes are not exactly up to spec.
You can add a silicon diode in series with each plate of the tube. A 1N4007 or UF4007 will increase the PIV rating of the tube / silicon combination by 1000 volts. This eliminates the sparking caused by excessive reverse voltage on a tube that's not fully heated yet. The VI characteristics of the tube are still highly dominant over the silicon, so that no "sandy sound" is added.
You can add in inrush current limiter. A low current device like the CL140 can be added in series with the HV winding's CT lead. This slows the current inrush of the high voltage somewhat blunting the hit on the rectifier's cathode as it tries to conduct the full 56 mA when it's cathode is still warming up.
If your power supply id CRC or CLC, reduce the value of the first cap to the lowest possible value that doesn't drop the voltage too much and increase the second cap to kill hum. I have used values from 4.7 uF to 47 uF for the first cap, with 200 to 500 uF for the second cap. The large second cap can be a mix of electrolytic and film cap to get the low ESR / ESL attributes of the film cap at high frequencies.
We do not know the value of the negative voltage source fed to the CCS, but 50 Volts has been assumed.
Lets look at the LTP circuit. There is a 56 mA CCS pulling current through the tube cathodes. The grids are essentially grounded. To get 28 mA (per tube) with 300 volts on the plate, the bias voltage needs to be around -12 volts between cathode and grid (curves from 1940 RCA data sheet). The CCS will suck 28 mA through the cathodes resulting in a POSITIVE VOLTAGE on the cathodes of around 12 volts. This should be measurable in normal operation. The reason for the negative voltage source is for CCS headroom with signal swing, and to limit it's VVC effects (discussed later).
On power up the negative voltage hits the CCS and it tries to suck 56 mA through the cathodes of the tubes, but the plate voltage is not there yet. It will pull the cathodes negative until is gets 56 mA, or saturates trying to do so. The 801's are warmed up, so they do what they are supposed to do, conduct. You now have a variable negative voltage source on the cathodes which is current limited to 56 mA, zero volts on the grid, and an unknown resistance (bleeder resistor?) in series with whatever plate load is on the tube connected to the plate. The tube will conduct and cathode voltage will either hit the negative supply rail or settle at whatever voltage the tube wants for 28 mA. Most of this will be grid current, but some plate current is flowing, or trying to flow since the B+ voltage goes negative.
I see no absolute grid current spec for the 801, but 15 mA is seen as the normal DC grid current and the curves extend to 80 mA. The 15 mA of DC current does not include the RF current flowing in RF amp duty, so my guess is that 28 mA for a few seconds won't hurt anything, and you can insure that more of that is plate current.
Add the diode that I talked about earlier. This will provide a path for the plate current to flow, and limit the negative voltage on the B+ supply to a safe value. I tried to link the schematic for my current driver board in a previous post, but I picked a file that was too large. The proper one is here.
I hesitate to use a silicon device for limiting the maximum negative voltage on the cathodes since all silicon devices are Voltage Variable Capacitors in their reverse biased regions, but much worse at low voltages. We choose mosfets used in tube amps based on their capacitance specs VS voltage curves to minimize this, and run them at higher voltages where the capacitance curves are nearly constant.
Many believe that this parameter is responsible for the TIM and PIM distortion effects that feed the "all silicon is evil" crowd. You will see a common NE2 neon lamp wired from the cathodes to ground.
Many of my boards use +/- 160 volt supplies for the mosfet drivers to reduce VVC. This could break down the HK interface when the board is powered up and all tubes are cold. The NE2 will light up limiting the voltage to about 75 volts. Once the tubes are conducting, the lamp goes out and becomes a fixed 1 pF cap. Unless the negative supply rail is higher than 75 volts the NE2 is not needed here.
If your rectifier still sparks after the diode is added, verify that the input filter cap is not too high for the tube being used. I see a Tesla spec where 60 uF is the maximum, but the curves of current VS secondary resistance are all done with 16 uF. The usual cause of sparking rectifiers is too much capacitance, to low of secondary resistance, or both. There are several ways to fix the sparking rectifier tube. I have added two of them to my SSE boards since it runs the rectifier hard and many current production tubes are not exactly up to spec.
You can add a silicon diode in series with each plate of the tube. A 1N4007 or UF4007 will increase the PIV rating of the tube / silicon combination by 1000 volts. This eliminates the sparking caused by excessive reverse voltage on a tube that's not fully heated yet. The VI characteristics of the tube are still highly dominant over the silicon, so that no "sandy sound" is added.
You can add in inrush current limiter. A low current device like the CL140 can be added in series with the HV winding's CT lead. This slows the current inrush of the high voltage somewhat blunting the hit on the rectifier's cathode as it tries to conduct the full 56 mA when it's cathode is still warming up.
If your power supply id CRC or CLC, reduce the value of the first cap to the lowest possible value that doesn't drop the voltage too much and increase the second cap to kill hum. I have used values from 4.7 uF to 47 uF for the first cap, with 200 to 500 uF for the second cap. The large second cap can be a mix of electrolytic and film cap to get the low ESR / ESL attributes of the film cap at high frequencies.
Wow.
Guys I am really impressed and very thankful for all the thoughts and willingness to help and share.
Thanks a lot !!!
Some more data and thoughts:
On the sparkling rectifier:
- I have this behavior in both occasions (LTP in the DAC as well as in the driver of the 300B) where I use SS-CCS with SS-negative-supply.
- In both sections I use two rgn1064/AZ1, one per channel.
- the filtering is always the same: LCLC with first L 10H(40ohm) - 50uF (ASC x386PP in pil)- L 15H (100 ohm)- 100uF (ASC).
- There is a bleeder resistor across the final cap of approx. 30K
- the negative voltage going into the CCS is about -24V
To cure the flashing, I integrated a nice high-wattage 5K WW resistor into the mid-tap of each transformer/ earth) with a time relais which switches the resistor out of the system after 7 sec. So, the caps can be filled slowly, the rectifier filament can warm up etc. This does work.
Nevertheless, it felt like only a workaround and not really curing the initial issue that something weird goes on at the start in the amplifier to make them spark. But maybe it is just the case that the rgn as a mesh rectifier is very sensitive when not fully warmed up yet...while the CCS is trying to force it as hard as possible to give 58mA
I am not sure for the root cause and found the negative voltage on the HV-caps as an error / unexpected result which give the RGN1064 a negative loaded cathode when heating up...which might or might not do crazy stuff with it.
Tubelabs...thanks a lot for sharing...I will definetly try your CCS-setup...
Rundmaus, would you be so kind to share your tube-based CCS as well ? I always was wondering how a tube CCS would sound different from a SS-CCS. I will as well simply measure what is going on in my case if I switch the CCS not on and have only the HV applied to the warm tube and see if it is like a big resistance or not...and if it is still sparkling or not...
Guys I am really impressed and very thankful for all the thoughts and willingness to help and share.
Thanks a lot !!!
Some more data and thoughts:
On the sparkling rectifier:
- I have this behavior in both occasions (LTP in the DAC as well as in the driver of the 300B) where I use SS-CCS with SS-negative-supply.
- In both sections I use two rgn1064/AZ1, one per channel.
- the filtering is always the same: LCLC with first L 10H(40ohm) - 50uF (ASC x386PP in pil)- L 15H (100 ohm)- 100uF (ASC).
- There is a bleeder resistor across the final cap of approx. 30K
- the negative voltage going into the CCS is about -24V
To cure the flashing, I integrated a nice high-wattage 5K WW resistor into the mid-tap of each transformer/ earth) with a time relais which switches the resistor out of the system after 7 sec. So, the caps can be filled slowly, the rectifier filament can warm up etc. This does work.
Nevertheless, it felt like only a workaround and not really curing the initial issue that something weird goes on at the start in the amplifier to make them spark. But maybe it is just the case that the rgn as a mesh rectifier is very sensitive when not fully warmed up yet...while the CCS is trying to force it as hard as possible to give 58mA
I am not sure for the root cause and found the negative voltage on the HV-caps as an error / unexpected result which give the RGN1064 a negative loaded cathode when heating up...which might or might not do crazy stuff with it.
Tubelabs...thanks a lot for sharing...I will definetly try your CCS-setup...
Rundmaus, would you be so kind to share your tube-based CCS as well ? I always was wondering how a tube CCS would sound different from a SS-CCS. I will as well simply measure what is going on in my case if I switch the CCS not on and have only the HV applied to the warm tube and see if it is like a big resistance or not...and if it is still sparkling or not...
Last edited:
Hi there,
there's not much to share, I just adapted the standard application circuit for a pentode CCS from the Morgan Jones book. I did not expect any audible improvements over a solid state CCS. The reason for using a tube CCS was simply the fact that I had the idea in mind to build the whole amp without any solid state stuff. So only tubes, passive components and relais.
Rundmaus at work - PP1C without sand
Regards,
Rundmaus
Blitz,
What is the value of the high voltage secondary? i.e. 350V-CT-350V
What is the rectifier tube?
What are the voltage ratings of the B+ capacitors? A choke input supply with only a 30k Ohm bleeder, will rise to 1.4X the rms voltage until the output tubes warm up and draw their quiescent current. (30k is not enough to make a 10H choke operate in the critical inductance range, and not enough current to have much voltage drop in the rectifier tube). After the output tubes are warm, then the choke operates in the critical inductance range.
What is the value of the high voltage secondary? i.e. 350V-CT-350V
What is the rectifier tube?
What are the voltage ratings of the B+ capacitors? A choke input supply with only a 30k Ohm bleeder, will rise to 1.4X the rms voltage until the output tubes warm up and draw their quiescent current. (30k is not enough to make a 10H choke operate in the critical inductance range, and not enough current to have much voltage drop in the rectifier tube). After the output tubes are warm, then the choke operates in the critical inductance range.
The secondaries are indeed 350-0-350. The rectifier is a RGN1064 Mesh as said earlier. the caps are 440 VAC Motorrun cap ASC x386 running well within their specs, even in clc...the 30k are there to ensure the caps are fully unloaded within 20sec, not to make the choke-input work as everything is very fast up and running with DHT 801a...not sure if the CLC scenario even exists as both, rectifier and tube are DH and very quickly heated within 2-3 sec simultanously.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.