tube CCS below cathodes, startup issues?

[Rundmaus]

Hi there,

just some thoughts that came up last night...

I am planning to put a pentode ccs powered by a negative rail below the cathodes of a LTP. What I am now wondering about is what will happen if negative HT comes up earlier than B+ on startup.

Without B+, the cathodes of the LTP will allow no current, so if the ccs tries to sink the set current, will it pull them far into negative potential? This might cause problems with max cathode/heater voltage in the LTP.

Hope you understand my point, if not I will attach a simple drawing later...

Greetings,
Andreas

 

[Doz]

Andreas,

Is it likely the B+ will come up after the tubes are warm?
A cold tube won't conduct anything anyway.

 

[Rundmaus]

Hi Doz,

both rails will use tube rectification, so it depends on the characteristics of the GZ34 (B+) and EZ81 (HT-) which rail is up first...

Greetings,
Andreas

 

[atmasphere]

I am planning to put a pentode ccs powered by a negative rail below the cathodes of a LTP. What I am now wondering about is what will happen if negative HT comes up earlier than B+ on startup.

Without B+, the cathodes of the LTP will allow no current, so if the ccs tries to sink the set current, will it pull them far into negative potential? This might cause problems with max cathode/heater voltage in the LTP.

You won't have problems with the LTP, IME, as those tubes will not be able to conduct until the CCS can. OTOH you may well have issues with startup with the CCS itself.

A pentode might manage CCS duties quite well. Can we see a schematic?

 

[piano3]

I use pentode CCS on my balanced linestage; assuming you have the pentode screen directly connected to ground then if the pentode conducted before the LTP the screen would pass the entire programmed current which might destroy it.

 

[tomchr]

If the CCS starts up first, it will pull the cathodes of the LTP down as far as it can. If this violates the Vhk spec of the LTP tubes, you should probably deal with it.

If the LTP inputs are expected to swing a few volts about ground, you could conceivably connect a zener diode between the cathodes of the LTP tubes and ground. This zener diode would need to be off during normal operation but turn on when the Vhk approaches the breakdown voltage. With this setup, the zener will conduct the current from the CCS during start-up. Then as the LTP "wakes up" they'll start taking over that current. You may have to use a diode or a second zener in series with the first to allow the cathode to swing positive with respect to ground.

Makes sense?

~Tom

 

[Alastair E]

You could use a Neon-bulb...

Some strike at around 60-70V, so will keep Vh/k under control....

 

[Rundmaus]

If the CCS starts up first, it will pull the cathodes of the LTP down as far as it can. If this violates the Vhk spec of the LTP tubes, you should probably deal with it.
~Tom

Thanks Tom,

you confirmed what I already suspected. Luckily, discovering errors and problems in the design phase is cheaper than finding them by sparcs and smoke *g*

Right now, I do not see a problem if the situation is the other way round: If the CCS comes up late, the LTP cathodes are floating and unable to conduct until the CCS sets in - but this should not lead to any undesired effects so far?

Am I right in that?

Greetings,
Andreas

 

[Bandersnatch]

hey-Hey!!!,
The EZ81 will beat the GZ34 to conduction. There is another issue you have not yet seen; if the LTP conducts first it could take your main B+ filter negative. Tie a zener anode to ground and its cathode to the LTP cathodes. Make its Zener voltage about double what you expect the LTP bias voltage to be. IFF you've got film caps to filter B+ then this efect does not matter one bit.
cheers,
Douglas

 

[Rundmaus]

Oh oh,

the issue seems not to be solved yet - if I summarize the above answers, it comes to the following:

a) If negative rail and CCS is up before B+, it will pull the LTP cathodes negative, probably violating V_hk of the ECC88

b) If, in addition, the (preheated) LTP is able to conduct before the B+ is up, it will probably pull even the B+ rail negative, blowing my electrolytic filter caps

Argh.

Solutions:

The Zener diode seems the simplest and best solution, but it violates my design rule to do completely without semis in this project.

Ideas:

1) Turn on negative rail manually after B+ is up.
+ avoids situations a) and b) if used correctly
- leaves the LTP cathodes floating for long times (Is this a problem?)
- no protection against B+ failure during operation -> catastrophic failure possible

2) ground LTP cathodes via resistor, open ground connection after B+ is up (Relais)

+ if CCS is up before B+, programmed current can be pulled from grounded resistor, pulling cathodes only slightly negative
+ protects against a) and b) if B+ fails unter operation
- ?

Question is I) will the LTP be in a safe state with the resistor to ground and II) will the thump wenn opening the ground connection kill my ears, speakers or both?

Desperately hoping for comments,
Andreas

 

[kevinkr]

You could use a Neon-bulb...

Some strike at around 60-70V, so will keep Vh/k under control....

This is what I do in my designs and it works extremely well, best of all is that when not conducting it is just some number of pF of capacitance at that circuit node. In the case where the cathode connections of an LTP will be well above ground in operation a diode oriented correctly in series with the neon will prevent it from turning on during normal operation.

 

[tomchr]

How about using a voltage regulator tube (neon bulb) in place of the zener? You'd need to find one that has a low enough breakdown voltage to save your tubes. 0A3, for example, breaks down around 75 V as far as I recall.

You could also argue that the zener is not in the signal path and only on during start-up. Hence, your amp would be semiconductor FREE*.
* = certain limitations and restrictions apply.

~Tom

 

[Rundmaus]

Thanks Kevin and Tom,

I forgot to mention the neon bulb / VR tube idea - in principle I like this very much. Problem is that most neon bulbs I found in electronic stores have ignition voltages of 90-100 Volts. Same for VR tubes, as they rely on the same principles...

Vhk for the ECC88 is only 50V in the relevant polarity, so this might still be a problem.

Looking at the schematic again,

I wonder if the LTP stage of the amp does not simply become cathode-biased when I put a resistor to ground where the CCS connects - so this would answer the question if this is a safe operation mode with "yes". Am I right?

Remains the question whether the opening of the ground connection after B+ is up will cause a signal surge that might damage speakers?

Greetings,
Andreas

 

[kevinkr]

Thanks Kevin and Tom,


I wonder if the LTP stage of the amp does not simply become cathode-biased when I put a resistor to ground where the CCS connects - so this would answer the question if this is a safe operation mode with "yes". Am I right?

Remains the question whether the opening of the ground connection after B+ is up will cause a signal surge that might damage speakers?

Greetings,
Andreas

Hi Andreas,
I think in this case it would be best to clamp the ECC88 cathodes to something less than their rated breakdown voltage during warm up and before cathode current starts to flow. Or better still use a couple of IXYs mosfets in a really high performance cascode CCS circuit that is powered by something like a -35V rail - no clamping would then be necessary. A really good cascode CCS should be inaudible I'm told. (Ten + years ago my transistorized ring of two measured good, but sounded worse than a resistor IMHO in the differential input stage of my line stage - this might have had more to do with the harmonic spectra that resulted in that high cmrr diff stage than the CCS in fact.)

Your idea to use a resistor to hold the cathodes within a safe voltage range until the tubes warm up has two potential issues, the first is that doing this will probably generate a pretty large common mode transient when it is disconnected which may result in anything up to a nasty thump or worse. Second if this is manually operated eventually you will forget at some point and potentially roach the input tube in question.

Another option is to delay the filament supplies to the rectifiers until all the other tubes have heated filaments, you can do this with a single amperite relay or somewhat more sophisticated electronics.

You could also use a set of interlocking power switches both of which have to be flipped to provide B+/C-, usually this works well enough. Flipping one heats the amplifier filaments, flipping the other would normally power up the plate and in some instances like yours the bias power supplies.. This approach can be adopted to your scheme with some thought.