• Please note that Audiokarma will be offline briefly for updates early on Monday morning, September 28th.

Output Tube Bias Using a Micro-controller

tcdriver

tcdriver

AK Subscriber
Subscriber
In a thread in the AudioKarma Manufacturers Forums > Fisher a discussion was started on tube bias.

Link: http://audiokarma.org/forums/index.php?threads/fisher-500-c-bias-a-case-history.843474/

This discussion turned to using an Arduino or something similar automatically set output tube bias. This subject deserves a thread of its own.

Here is my idea of the parameters needed in a micro-controller auto-bias system. This is open to discussion.

1. current sensing: The micro-controller should be able to measure the voltage drop across a 10 ohm resistor on each of four cathode bias resistors. Typically the voltage measurement would be between 0 to 500mV.

2. grid voltage adjustment: The micro-controller must be able to provide grid voltage to each of four output tubes: -15V to -25V.

3. micro-controller current draw: < 25mA

4. small size: The micro-controller should be small enough to fit into a crowded Fisher 500-C or similar chassis.

This is a start. Let us see how far we can get with this. Comments welcome.
 
Last edited:
Register to hide this ad
When music is playing, measured bias current goes up substantially. If the controller is actively regulating bias the controller needs to know that somehow.. maybe sense ac voltage at the input and turn off active regulation, when signal is applied?
 
Schematic of an IBAM added to show the basic idea (Thanks Drew). The controller would adjust 4 10K digital resistors individually to maintain a voltage drop across the four cathode bias resistors.

The bias supply can fluctuate a bit as AC wall voltage fluctuates, so it would need to not care about voltage coming in to the bias supply (within reason).

This bias adjust could be a routine that is run on start up only, but it could monitor and and provide a warning or even shut off routine if a tube redplates.

IBAM for 400.jpg
 
trainbuftony said:
When music is playing, measured bias current goes up substantially. If the controller is actively regulating bias the controller needs to know that somehow.. maybe sense ac voltage at the input and turn off active regulation, when signal is applied?
Click to expand...
Yes, absolutely correct. The system must sense when no music is playing and then make the bias adjustment.
 
Also to consider are startup and shutdown. A reasonable bias voltage must be present on startup. One possible startup would be to set the grid bias to maximum negative voltage for the first 20 to 30 seconds and then slowly ramp to the operating value, determined by measurement. I do not foresee any special requirements for power shut-off.
 
I think it should have a "set" button, so it only activates when the amp is not in use when setting bias, and a storage function to remember the last settings when unplugged
 
I wonder if it is better to have a button you press where at that moment, it tests and sets and retains those settings until the next prompt to set. I feel that the settings should not migrate too much (I just set the bias of my tube amp after eight months and while it changed, it did not change all that much). With this you could easily set it once a day, once a week, or however frequently you want.

It could still monitor for redplating and such and act if such things are needed.

I admit my arduino knowledge is somewhat limited, so I am not sure how easy it is to set an input to run a certain script. (also, I was told to wait a bit after turning an amp on to check the bias, giving it proper time to warm up and settle down).

edit: @s-petersen was too fast for me it seems.
 
rufleruf said:
Schematic of an IBAM added to show the basic idea (Thanks Drew). The controller would adjust 4 10K digital resistors individually to maintain a voltage drop across the four cathode bias resistors.

The bias supply can fluctuate a bit as AC wall voltage fluctuates, so it would need to not care about voltage coming in to the bias supply (within reason).

This bias adjust could be a routine that is run on start up only, but it could monitor and and provide a warning or even shut off routine if a tube redplates.

View attachment 1323363
Click to expand...
Thanks for providing the schematic although I would use something more along the lines of the one I used in my Fisher 500-C.
Bias Circuit.png
 
Requirement: Add settings for minimum_bias and maximum_bias that are used by the circuit to determine when to make adjustments. If the bias measurement falls outside these hi-lo marks, the circuit kicks in to make adjustments. Or if not worth making it a user interface, then the range should be hard-coded in the circuit software.
 
ADC inputs on the processor could easily detect intervals of zero input signal, when it's safe to make bias updates. Why not use an ADC input for B+ as well, to implement Enhanced Fixed Bias (EFB) in amps that can benefit from it?
 
Well, since this is a wish-list discussion, here's what I'd like to see:

Why stop with just auto setting the bias under no signal conditions? Why not continuously sample at say twice the rate of the highest frequency signal the amp will reproduce, and adjust for optimum bias setting relative to plate and screen so you have perfect voltage relationships at all times, including idle, and all the way up to full power output? You might even be able to detect an overdriven condition (clipping) and adjust bias accordingly so you have better behaved response under those situations. You could also make it switchable for different tube types. I'm guessing this is more-or-less what the "Adaptive AutoBias" circuit from PrimaLuna does, but that's just a guess.

So I think what I'm musing about is a digital EFB controller. (a la Dave Gillespie's analog EFB controller, but done in the digital domain). Dave's EFB controller assumes a linear relationship between screen and bias voltages--in practice I think that relationship is close, but not exactly linear, so in the digital domain you'd have the opportunity to correct for those small irregularities.
 
This just strikes me as an overly complicated way to make things less reliable for no real benefit.

yes, I'm good at raining on people's parades, but reliability and benefit have to be primary concerns for any sort of modification. Unless you can improve both, it just seems to me that its a solution in search of a problem.
 
gadg - this is one of those because you can sort of things I think.

I had this idea to use an Arduino for bias monitor and set from a project where a guy in France wrote code that allowed an Arduino to replace the processor in a Sequential Circuits Prophet 600 synthesizer. There were lots of improvements to the original design possible due to the improved processor speed.
 
judsonw said:
I wonder if it is better to have a button you press where at that moment, it tests and sets and retains those settings until the next prompt to set. I feel that the settings should not migrate too much (I just set the bias of my tube amp after eight months and while it changed, it did not change all that much). With this you could easily set it once a day, once a week, or however frequently you want.
...
Click to expand...
You bring up a fundamental question. How often do you need to adjust bias? The answer may not be so simple. The reason this tread started is that I was trying to bias my Fisher 500-C and realized that the bias drifting and wandering all over like a drunken sailor. I had modified my Fisher to allow for individual bias on each of the four output tubes. What I realized is that for this amplifier, with these tubes, a simple set and forget would not be possible. It now amazes me that the original design ever worked well at all. An added problem is that modern 7591A tubes have somewhat different characteristics. In particular, the bias voltage required by the new tubes is greater, more negative, than the original tubes. If one just inserts new manufactured tubes in the original circuit they are very likely to run too hot, which could lead to early failure.

Making things worse is the original Fisher design that uses grid resistors that are larger in value than the recommended maximum in the tube data sheets. With the too high value resistors, grid leakage can become an issue. If the grid is leaky or starts leaking it will cause an increase in bias current. An increase in bias current will cause the tube to run hotter. As the tube temperature rises there will be more grid leakage.
 
That's one of the reasons I tend to go with cathode bias, no muss, no fuss. The end user never needs to adjust anything.
 
BinaryMike said:
ADC inputs on the processor could easily detect intervals of zero input signal, when it's safe to make bias updates. Why not use an ADC input for B+ as well, to implement Enhanced Fixed Bias (EFB) in amps that can benefit from it?
Click to expand...

Sounds like a suggestion to incorporate the microcontroller into the EFB? I don't know enough about how the EFB operates yet to respond. (Probably Dave might see this and chime in. Maybe the EFB is self-bias'ing or stable enough not to need automation. I just don't know.)

For the OP who seemed to be indicating automation of the basic IBAM, a further complication is that when adjusting the bias on one tube with an IBAM it affects the current in the other 3 tubes. Kind of like nailing jello to the wall - the other 3 tubes bias keep moving around as the bias on one tube is changed. It may be possible to do a simple automation and apply it independently to each tube with the Arduino (a one tube algorithm applied separately to all 4 tubes if you are lucky), or it might require a bit more thought to ensure that the automation isn't hunting all over the place trying to find equilibrium across all 4 tubes because of the interactions between the individual tubes. This is what happens with my IBAM on my Fisher KX-100 tube amp where the bias for each tube is adjusted manually. I end up going through the adjustments a few times before all the tubes current flows match pretty closely. Perhaps having the algorithm make baby steps instead of trying to fully compensate each time it checks and adjusts would tend to stabilize the bias on all 4 tubes. [This could be a great college project in control stability - and not atypical for college Analog/digital Control courses either.]

There is an alternative to the IBAM called the IBBA (Individual Bias and Balance Adjustment, I think) which might lend itself to a more stable automated solution, but I'm just guessing.

Is any of this necessary or needed? I don't know, but it is a cool concept.
 
You must log in or register to reply here.
Back
Top Bottom