drift in direct coupled tubes

[andrew_whitham]

hello all.

I'm looking at a direct coupled input stage which runs something like a CCS loaded ECC88 into an equally CCS loaded 12BH7 cathode follower. Picture the front end of a Willamson amplifier but the concertina is only driving from the CF output

so far no drama.

my query is then, how does the amp deal with drift of operating points over the life of the tubes? My concern is that even a small change in the steady state anode voltage messes up the operating point of the second tube.

Can anyone shed some light on this? is the drift actually very small? or is some other process going on?

Apologies if this is actually self evident - clearly direct coupled amps work (well) but I'd rather like to understand why...

cheers

Andy

 

[EC8010]

The anode voltage of the first stage may well drift, but it doesn't matter because the cathode follower isn't really very bothered about its Vak and it has lots of feedback wrapped around it. Imagine that second stage (before feedback) as having a flat loadline due to the CCS, then all that's happening is that drift from the first stage erodes a little headroom at one end of the other of the cathode follower. Now, if the second stage had gain, drift of that first stage would be far more important...

 

[AndreasS]

Hi Andrew,

an effective method for reducing the gm for dc of the input tube is a higher cathode resistor. The anode load reduces the gm by the factor 1+ Ra/rp, the cathode load 1+ gm*Rk. Adjust the bias by a positive voltage on the grid.

Regards Andreas

 

[Wavebourn]

Use a CCS with a servo like I do in such cases.

 

[andrew_whitham]

Thanks

EC8010,

thanks for that, I think I see the feedback in the cathode follower is reducing the error (as it would) the cathode is tending to follow the grid anyway so its almost self compensating. Is that about the size of it?

generally though,

If the second stage has gain then, what are the likely problems? I've seen this done with a couple of *sn7 sections, and I guess it works?

I still feel like I'm missing something, If the anode voltage of the first stage changes the second stage operating point is supposed to move... Is the basic rule to assume that the operating point of the second triode is going to be 'in the general area' of -n volts? i.e. realistically you won't get full swing from the second triode?

And if so. what sort of variation is common during the life of the tube (now thats the million dollar question isn't it?!)

Does any one know of any resources that cover tube aging? although to be honest I'm not sure how I'd relate this to the above...

cheers

Andy

 

[andrew_whitham]

waveborne, I'm fully intending to use a CCS's I already tested this out on a previous amp.

If it isn't such a daft question then how are you feeding the servo into the CCS - I assume this is what you are up to?

My scratchings this afternoon involved level shifting from the anode of the first tube and setting the grid bias to maintain a "constant" VA - no, it didnt look so pretty then either.

Cheers

Andy

 

[Wavebourn]

Here you go:

In order to shift levels I use bipolar B+ source.

 

[jnb]

Re: Thanks

If the anode voltage of the first stage changes the second stage operating point is supposed to move... Is the basic rule to assume that the operating point of the second triode is going to be 'in the general area' of -n volts? i.e. realistically you won't get full swing from the second triode?

If you direct couple to a gain stage in the most common way, you'll first need to bring the cathode up to the grid. Using a large value of cathode resistor (like you'd use for a cathode follower) will cause the gain to drop just like a cathode follower. If you bypass that resisotr you get your gain back and your cathode voltage is fixed.

Basically, the first stage DC drift will translate to a second stage bias drift only where a volt or three makes little difference to the first stage, it makes a lot of difference at the grid of the second stage.

 

[ray_moth]

In the case of a mu stage, it can easily be coupled directly to a following stage, with no risk of drift, if the CCS of the mu stage consists of a MOSFET and the output is taken from the MOSFET's source. This is based on the fact that biasing the MOSFET's gate to a fixed voltage will also force its source to be at that voltage.

The bias-point can be fixed using a (capacitor-bypassed) potential divider between B+ and ground or, if you don't trust the stability of your PS, you can use a (capacitor-bypassed) zener. Connect the bias-point to the MOSFET's gate using a high value (several megohms) resistor.

 

[EC8010]

Andrew; yes, you've pretty much "got it." Of course, the second stage can't distinguish between a wanted anode movement and an unwanted, so your "general area" supposition is correct. Now, if you go to a differential pair following another, a common drift in anode currents is rejected (although differences are amplified). As to how to minimise drift, regulated heaters and HTs would be a good starting point; have a look at oscilloscope circuits. Nevertheless, I'm fairly sure there's an old valve book with a chapter entitled, "DC amplifiers, and why to avoid them." Pretty well sums it up, really.

 

[ilimzn]

Here you go:

In order to shift levels I use bipolar B+ source.

Doesn't the high dynamic load resistance roughly equal R3? Obviously 'high' gets to be a relative term, then.

In general, if you have a ccs loaded gain stage, the cathode to plate voltage is the voltage drop on the tube's Rp at that working point. Changing the power supply voltage would not produce drift, but an aging tube will (towards higher plate voltage). If this is directly coupled to the follower, which is also ccs loaded, then it's cathode will go that much higher. If you are direct-coupling to a gain stage that follows, then you have a problem unless the whole thing is differential.
It;'s not terribly difficult to 'servo' this, by 'adjusting' the first tube bias voltage, from the follower output DC voltage. A bit easyer than servoing the plate ccs of the first stage as both input and output of such servo are ground referenced.

 

[andrew_whitham]

Again thanks all,

Wavebourne, didn't that circuit come up when I was last asking about using CCS's in a cathodyne phase splitter ? (answer - yes, and it was nice then too) Anyone interested - http://www.diyaudio.com/forums/showthread.php?s=&threadid=95152 is where its explained... (going to build this one)

ilimzn - the Servo bit was what I was hoping to avoid. I know folks do this without it. Are you saying they just 'put up' with the drift? i.e. its mostly acceptable but there's a better way?

EC8010, great, and I'm convinced enough to just go build the thing with the cathode follower. I must admit I wasnt going to bother with a regulated heater supply, but I might now. HT regulation is part of the reason for building the amp though. testing, testing, testing...

Naturally results to follow, but genuine thanks for the help.

Cheers all

Andy