• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

CCS in LTP drive HV-Cap negative...help...pls

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am a bit confused on the following and would appreciate your help:

I am using a pretty straight forward setting with a LTP/DifferentialAmp im my DAC and in my driver stage. Actually two 801A DHT work in push-pull into a Line-OPT/IT at the anode. Grid is geounded and the grid to ground resistor is at the same time the i/v-converter for the dac (ess9018 dual mono).

The two cathodes are connected to a CCS from Kevin Carter (MOSfet).

OK, what surprises me: When I am not applying any HV, let the tubes just heat and have the CCS running (with a negative supply): I can measure a negative Voltage of approx 19V on the HV-Caps of the Anode-HV-Supply.

When I switch the HV on, I get on my Mesh-Rectifiers(rgn1064/Az1) a big spark and I suspect it has to do with this negative voltage in the HV-Caps.

Has any one an explanation for this ? I am not even sure how the negative voltage gets there as the raw supply of the CCS is grounded to the circuit, but the negative side goes to the cathodes of the tubes only...but this negative potential is now even at the Anodes ??

Any suggestions how to build a setup where the HV-Rectifiers are not stressed/do not see the CCS ?
 

PRR

Member
Joined 2003
Paid Member
....not applying any HV, let the tubes just heat and have the CCS running (with a negative supply): I can measure a negative Voltage of approx 19V on the HV-Caps...

So grids are grounded with small resistor, and a negative supply connected to the cathodes, with no power "to" the plate supply.

The cathodes are pulled Negative until grid current is approximately the CCS current. (This may be too much for these grids.) If there is no bleeder resistor on the HV supply, very little current is needed to pull it negative also. Electrolytic HV caps reverse-biased "should" conduct past a few Volts reverse; if you have pulled them to 19V the wrong way they may be damaged.
 

PRR

Member
Joined 2003
Paid Member
Hasty arbitrary sim. JFET is a limited current, fed from -50V source. "Dead HV" modeled with 10K (I did not see your 30K). Tube is turned-ON. Grid current is large for a small tube (the 801 may be fine). Because tube is on, plate pulls-down, causing your negative rail symptom.

Have they invented crystal diodes yet? A diode from HV to zero would clamp the negative voltage. However I do not believe a mere 19V is causing your "flash".
 

Attachments

  • Blitz-42.gif
    Blitz-42.gif
    4.4 KB · Views: 264
Hmmm...I will test this by not heating the tube...in theory than no neg. voltage should be there, right ? If that is the case the solution might be to first heat the rectifier, switch on HV and a few seconds later apply heater and neg. supply to the ccs...and see what happens.
 
PRR,

The 2N3819 is only a 25V rated part. VDS, VDG, and VGS breakdown are all only rated for 25V. "Your mileage may vary".

A -50V power supply with the 2N3819 will cause Cathode to Grid Current, and the JFET will be in a high current condition, not a constant current mode (it is in voltage breakdown).

Is the JFET really a 2N3819, and is the negative supply really -50Volts? If that is the case, re think what JFET to use, and how much negative voltage to use.

I do not have any idea about the Kevin Carter Mosfet circuit, and that takes us back to the need for a schematic from Blitz to give us the real parts and voltages.
 
Last edited:
Without knowing much about the circuit details, I offer a simple solution to the negative voltage buildup on the B+ supply caps.

Put an ordinary silicon diode like a 1N4007 across the cap so that it does not conduct when there is normal B+ on the cap. It will conduct when the negative voltage appears, limiting it to about 0.7 volts.

This will solve the negative voltage issue. I doubt that a few negative volts on the cap is the reason for the sparking rectifier tube unless it's running near its max PIV rating already. Make sure that the cap value is within the maximum capacitance value for the rectifier tube.
 
I am a bit confused on the following and would appreciate your help:

I am using a pretty straight forward setting with a LTP/DifferentialAmp im my DAC and in my driver stage. Actually two 801A DHT work in push-pull into a Line-OPT/IT at the anode. Grid is geounded and the grid to ground resistor is at the same time the i/v-converter for the dac (ess9018 dual mono).

The two cathodes are connected to a CCS from Kevin Carter (MOSfet).
I don't know if this has to do with your problem... But, i asked Kev about connecting 2 cathodes of a splitter to 1 CCS and he said that doesn't work. That you need a CCS for both cathodes. I was just researching my circuit so i did not follow through.
 

PRR

Member
Joined 2003
Paid Member
PRR, .... The 2N3819 is only a 25V rated part. VDS, VDG, and VGS breakdown .... A -50V power supply with the 2N3819 will cause Cathode to Grid Current, and the JFET will be in a high current condition,... Is the JFET really a 2N3819, ... re think ......

I am very sad that you missed the point about a "Hasty arbitrary sim". And that you typed your fingers to the bone demolishing a straw-man.

Blitz reported an odd result. I could picture the mechanism in my mind, but a GIF picture is easier to share. I did not have the parts he used, but parts-is-parts. Similar parts will give similar results.

No, this is NOT a real plan, nor real parts. Just parts available in my 1995 simulator. This is not my first rodeo with it. I know that it does not enforce breakdown limits on most parts. I did however verify that the JFET's drain current varied little with Vds.

While this sim-result is not "RIGHT", it gives numbers echoing Blitz's observation: negative B+ with grid-cathode energized and B+ "dead". That was all I was trying to show.

Again, sad that you missed the point.
 
But, i asked Kev about connecting 2 cathodes of a splitter to 1 CCS and he said that doesn't work.

I don't know who Kev is or anything about his particular circuit, but for an LTP to operate properly you tie BOTH CATHODES TOGETHER and feed them with ONE CCS. I make life easy and use a 10M45 chip CCS unless the circuit uses low enough power for an LM334.

This circuit uses one LTP direct coupled to another one. I have been building amps of a few watts to 250 watts using this design for several years. it was designed here:

6L6GC AB2 Amp

The rectangular boxes on the schematic connected to the cathode pairs are IXYS 10M45 CCS chips. Each CCS feeds TWO cathodes. If you have a separate CCS for each cathode, it's NOT a true LTP. Note some LTP's are made with a large valued "tail" resistor approximating an LTP. It will work with slightly degraded performance.
 
Blitz,

How many mA is your current source set for? It is sinking current from the cathodes.

Tubelab_com had a good idea to use a diode clamp to keep reverse voltage off of the electrolytics.

When there is no plate supply voltage, much of the current that drags the cathode to ground is going to go into the 801 grids. Suppose the 801 grid resistor is a 100 Ohm load to ground, and is fed by the DAC output; and suppose the other 801 grid resistor is returned to ground through a second 100 Ohm resistor. Suppose your current source is set to 20mA. The worst case is if there is no plate current. Then 10mA will go to each grid. That means -1V will be fed back to your DAC output stage. Just check to see if this might damage your DAC.

You can easily measure the voltage across the 100 Ohm resistor, and calculate the grid current. Any current that does not go to the grids will go to the plates, then to the plate loads, and then to the diode clamp to ground.
 
Last edited:
That is very helpful and you made me think....some more data:

- the CCS is sinking 56mA per channel, so each 801A is running at 28mA at 300V two cathodes connected in LTP as desceibed by tubelab)
- the grid to ground resistor is currently indeed 100 ohm...good idea to measure the voltage across it in the different stages of operation
- the Dac is a ess9018 in dual mono directly coupled to the grid...which itself has an impedance of approx 90 ohms if I am not wrong delivering 32mA current...this setup is running now for many years and the DAC is still alive...
- I have two different CCS of Kevin Carter (based on two IXTP08N100D2, see K&K Audio | Other kits )...the type of CCS does not matter, the effect is always the same.
 
Last edited:
I expected such a behaviour when I adapted a tube-based ccs to the driver stage of my PP-1C amp. That's the reason why I enforce a startup sequence in which the negative rail comes up only after the +HV is present. No issues with this in the last 1-2 years of operation.

Regards,
Rundmaus
 
Rumdmaus, I had the same thought...but how do you bias the tubes in the first seconds ? The ccs controls / produces the negative gridvoltage with the help of the negative rail...no ? So how becomes the grid neagative if no negative supply is active anymore ?
 
With the -HV not present, the CCS is not active in my case. So the cathodes practically float and there is no significant plate current even without a negative bias present.

As soon as the CCS comes up and current flows, the negative grid voltage develops across the cathode resistor, correctly biasing the tube.

It should be noted that I am talking about small signal tubes (ECC88) and only low plate currents (approx. 5mA per tube). Don't know if the trick still works with power tubes and higher bias currents.

Regards,
Rundmaus
 
Last edited:
But you described that you power up HV first and the CCS a couple of seconds later...no ? So the tube is fully active in theory...the question I guess how does the CCS looks like for the tube when its not working...as a high resistor, so the grid becomes very nehative as no current can flow theough the CCS ? Did you measure your grid voltage in that seconds ?
 
But you described that you power up HV first and the CCS a couple of seconds later...no ?
Yes, that's right.

So the tube is fully active in theory...
Yes. The tube would be fully driven without the negative grid. But its cathode is only connected to the plate of a cold CCS tube. So there is no plate circuit for a plate current to flow.

the question I guess how does the CCS looks like for the tube when its not working...
I guess that depends on the type of CCS and could be very different from my situation in the case of a solid-state CCS.

as a high resistor, so the grid becomes very nehative as no current can flow theough the CCS ? Did you measure your grid voltage in that seconds ?
For the tube CCS I use, the CCS becomes a practically open connection in the unpowered condition. So no negative grid voltage is created, but no current can flow in the plate circuit anyway.

Have to admit that I never measured grid voltage as the circuit worked as intended from the beginning...

Regards,
Rundmaus
 

PRR

Member
Joined 2003
Paid Member
One of you smart guys check this wild plan.

Added PNP and divider. If B+ is way low, PNP is saturated, absorbs all the CCS current. When B+ comes up high, PNP is turned-off, out of circuit.

PNP must, as 6A3sUMMER says, be rated for significantly more than the negative supply. It will work mostly off/on so low power, but in the transistion it could be over 1W (for some values of supplies) so a TO220 part.

As shown the B-E junction is reverse-biased perhaps 8V, so will break-down. Base current is limited so not a blow-up. Still a good idea to add a diode B-E. (An LED would give a sign that B+ has come up and the PNP should be off.)

I have not thought-through all the ways this could leak on large signal. It may be that I've missed some key flaw.

I still do not think this "negative B+" is your "rectifier flash". And while grid current is hefty, your proposed triode is designed for larger grid currents.
 

Attachments

  • Blitz-43.gif
    Blitz-43.gif
    4 KB · Views: 126
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.