Calculating cathode resistor

[vampyretim]

Hi,

My friend is trying to build a phono preamp using valves, the design comes from an electronics article in a magazine from the 90's. He wants to change the design from using ecc88's to 6n1p-vi's.

How would we go about calculating the various bias resistors?

I'm a little confused about working out the cathode bias voltage
Many thanks.

[Radio Wrangler]

Method 1)

You just want to set the valve up for the manufacturer's published normal operating current. Their datasheet gives you an anode current and a grid bias voltage. So if the grid is at 0v DC and the cathode is positive by the published grid bias voltage (that is if the data says -8v grid bias, you want the grid at 0v and the cathode at plus 8v) then the valve should pass the published current. The anode current equals the cathode current so use Ohms law. Take the published bias voltage, divide by the published anode current and Voila! you have the required cathode resistor.

Method 2)

So you want to go off-piste and pick your own value of anode current.

Easy! You just need to find a plot of anode current versus grid voltage, pin the tail on the donkey for the anode current you would like, read off the associated grid bias voltage and use those numbers in method 1.

David

[G8HQP Dave]

Then all you have to do is adjust the component values in the RIAA network to take account of the fact that the valve anode impedance will be different from that assumed in the design. Or don't bother and don't worry.

[julie_m]

Method 3)

(For masochists only) Rig up a suitable test circuit using your intended PSU, so you can apply a variable voltage to the grid and measure the cathode current. Plot your own graphs of I\V. Select an operating point as near as you can get it to the middle of the longest, straightest part of the curve.

I'm sure a true masochist would eschew the assistance of a computer equipped with an "experimenter board" having a DAC output and ADC inputs, and tabulate and plot all measurements by hand. I, however, cheated when I actually did this!

[Argus25]

If you assume that the anode resistors remain unchanged and that you want about the same average anode current using your replacement valves (which places the anode voltage at about the same DC level) then this also represents the cathode Current Ik, and you would simply choose a cathode resistorR , that generated the correct bias voltage Ik.R for the valve that corresponded to that anode current value for your new valve...... And everything would seem just dandy or would it ?

One issue you might confront is that depending on the design of the actual valve you try, to be operating at the same anode current value as the original valve (assuming you don't want to change the anode resistor and anode current value) you may find it has a substantially different bias voltage or cathode voltage value (with the grid returned to ground).

Most of the time this has little effect, but in some circuits, but if the bias voltage is on the low side of 5 volts or less, the grid can be driven by signal positive with respect to the cathode and your signal will be asymmetrically clipped. So this is one thing to watch out for trying to swap and re-bias different valves into audio circuits.

[Diabolical Artificer]

This article - http://www.valvewizard.co.uk/Common_Gain_Stage.pdf goes into a lot of detail about the common gain stage, drawing load lines and choosing a cathode R, not had a look at the datasheet for a 6N1P but it's given as an equivalent for the 88 so they shouldn't be any need to change Rk, however the interelctrode capacitance might be different which might through out the RIAA stage a tad.

Andy.

[Radio Wrangler]

Julie's brought up a subtle issue.

The advantage to measuring your own valve is that you therefore get the bias exactly right for the very valve you intend to use. Fine if you're only building one.

The disadvantage to measuring your own valve is that you therefore know the figures for only the valve you have, and it gives you no idea of the statistical distribution of others of the same type.

If you want to manufacture your design, or publish it, you'll have a better chance of the duplicated units being successful if you use the manufacturer's data and trust that they quote the mean across a number of batches. As an alternative, measure a number of devices yourself and try to get different makes and batches.

If you're just building one unit for yourself you can enjoy a number of freedoms. You might get bitten if you need to fit a new valve, or publish the design.

On the whole, this issue isn't too much trouble with valves because of their relatively small variations in parameters. This leads many people to be comfortable with valves.

Transistors on the other hand have relatively enormous variations in parameters, and doing a design to accept all in-spec parts is serious work. Not understanding this has put a number of almost irreproducible designs in magazines, and a number where the builder has to select devices to get it to go. As aversion therapy, it's scared a lot of people off.

So there you have it; a solid engineering foundation for why many people like valves and dislike transistors. And we didn't need Hari Seldon!

David

[GW4FRX]

Originally Posted by julie_m

Method 3)(For masochists only) Rig up a suitable test circuit using your intended PSU, so you can apply a variable voltage to the grid and measure the cathode current. Plot your own graphs of I\V. Select an operating point as near as you can get it to the middle of the longest, straightest part of the curve.I'm sure a true masochist would eschew the assistance of a computer equipped with an "experimenter board" having a DAC output and ADC inputs, and tabulate and plot all measurements by hand. I, however, cheated when I actually did this!

Attention all audiophools. Are you entirely sure all your valves are matched carefully enough? Shouldn't you be doing precisely this?

I suspect Julie has just opened up a whole new market (or possibly a can of worms).

[Radio Wrangler]

It's alright. They match their valves by appearance. Length and smoothness of anode, colour of markings, Maker's logo.

They know they mustn't make any numerical measurements. Even on an AVO VCM, they are not allowed to read from the scale, they are only allowed to say 'strong'.

David

[GW4FRX]

Originally Posted by Radio Wrangler

They know they mustn't make any numerical measurements.

Oh yes, of course, silly me. I had completely forgotten that numerics of any kind are eschewed in audiophool circles.

[kalee20]

Originally Posted by Radio Wrangler

Method 1)

You just want to set the valve up for the manufacturer's published normal operating current. Their datasheet gives you an anode current and a grid bias voltage. So if the grid is at 0v DC and the cathode is positive by the published grid bias voltage (that is if the data says -8v grid bias, you want the grid at 0v and the cathode at plus 8v) then the valve should pass the published current. The anode current equals the cathode current so use Ohms law. Take the published bias voltage, divide by the published anode current and Voila! you have the required cathode resistor.

Method 2)

So you want to go off-piste and pick your own value of anode current.

Easy! You just need to find a plot of anode current versus grid voltage, pin the tail on the donkey for the anode current you would like, read off the associated grid bias voltage and use those numbers in method 1

Method 1 works, but isn't necessarily optimum. I well remember sketching out the values for an ECC82, in teenage years... The data book gave Ia = 10.5mA for Vg = -8.5V, at Va = 250V. So, Rk = 820Ω and I must use well over 250V of HT... So if I have 325V available, I must drop to 258.5V on tha anode... That's 66.5V drop in the load, at 10.5mA, isn't it?

Yes, of course it is. But it's very wasteful of HT current and the valve runs pretty hot! Much better to use Method 2, get the complete set of characteristic curves, set the operating conditions to suit the signal levels you will see, and take it from there.

The data book just gives one set of figures (or occasionally two), which might just suit one particular use, but are more likely just a set of numbers which work THD valve fairly hard, for use with a valve tester to assess goodness. So, except for rectifier valves and output valves, you probably wouldn't see these conditions in actual use.

[julie_m]

My original trick with the computer could be extended, and the bias constantly tweaked to ensure the valve was always at the correct operating point. But, I fear it would be a case of "the more watches you have, the less sure you are of the right time".

Think about it: If the settings are perfect now, and they were perfect at some previous point in time they were also perfect but were not the same then as they are now, then according to the Intermediate Value Theorem, there must have been at least one moment between then and now when the settings were wrong.

Just the fact that something can be adjusted at all is basically an admission that a device is imperfect -- if it was perfect, then there would never be any need to alter the settings. And if it is so complicated that it needs to be adjusted automatically, then there are obviously more ways for it to go wrong than right. Which is bad, because you would be forever listening out trying to catch it sounding wrong .....

So I don't think the hi-fi bores would be up for such an invention. Mind, there would be bound to be some who liked it. And they would be praising the valves to high heavens, while all the time it was the fiendish negative feedback loop doing the lion's share of the work in spite of the valve!