Advice on building lane switching circuit

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Hello, I am new to electronic design but I am learning as much as I can.
I need to make a circuit that has 2 lanes, lets say 1 lane is a -30v DC and the other lane is 150v DC with an AC sine wave attached.
I want to make a circuit that will switch the output to the -30v DC lane when there is no AC signal and then switch back when the AC is applied.
Maintaining signal purity is important for the 150vDC/AC lane so having it go though anything other than a lowish value resistor is not preferable.


Can this be done?
 

#12

Joined Nov 30, 2010
18,224
I think so. We're going to need the amounts of current and the frequency of the sine wave. We need to know how quickly the circuit needs to recognize that a change has happened.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Current should be 20ma for the 150v lane and probably a low value for the -30v lane.
As for how fast, as fast as possible, any delay at all is not good. Signal purity is most important, can't have it being cut off. It is for a hi-fi audio circuit.
Frequency range is 20hz to 20khz.
 

#12

Joined Nov 30, 2010
18,224
I'm sure you can see that 20Hz defines the slowest wave that needs to be recognized. That's 50 milliseconds if the circuit can recognize the change in only one cycle.

"As fast as possible" brings to mind switching speeds in the micro-second range, and that isn't going to happen here. We can switch in microseconds, we just can't recognize that it needs to be switched for at least 50 milliseconds.

Please excuse if I don't come back tonight. It's Christmas and family beckons.
 

#12

Joined Nov 30, 2010
18,224
recognize and switch
You just named the technical basis of the difficulty.
If you will back up from what you want to what you need, a better answer might be forthcoming.
Nerds can brilliant at simplifying if you start with the most basic way to state the problem and the goals.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
If you will back up from what you want to what you need
Sadly what I want and what I need are sort of the same. I can't have any signal cutoff at high frequencies. I may be able to lower the requirement to 17kz but that is still .056 ms.
Essentially this will be used for a very high fidelity hybrid amplifier as a tube biasing method. When the signal stops, the bias kicks in, when the signal starts, the bias stops. So basically the intention is to keep the tube from melting via a negative bias to the grid while still utilizing a positive DC signal source.
 

sailorjoe

Joined Jun 4, 2013
364
I'm thinking we need a mosfet transistor pair configured like a 2-in-1-out multiplexer, with a detection circuit driving the control line. Concentrating on the detector circuit first, we need to know what it means in this application to say "when a signal is present". Does the DC component go from 0 to 150 V in a microsecond, a millisecond, a second? Is it always at 150 VDC, and we want to detect the audio component on top of it. If it's the latter, what does that signal look like in terms of peak voltages? Assuming it's detectable very quickly, does it need to be switched on at the beginning of the very first cycle, or can we catch the start of the next cycle? Do you need to prevent sharp transients in the audio signal as it switches? That implies the need to switch on/off at the zero crossing points of the signal.

I'm curious about the -30 V signal, too. Where did that number come from?
 

sailorjoe

Joined Jun 4, 2013
364
Ok, I see where the -30 V comes from. So far, this seems doable, but complexity will depend on the requirements details. Something to think about while sipping a hot toddy.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
we need to know what it means in this application to say "when a signal is present"
As in, when any AC signal within 20hz-20khz is active
Does the DC component go from 0 to 150 V in a microsecond, a millisecond, a second? Is it always at 150 VDC,
The 150v is always 150v. It is sourced from the input stage of the amplifier. Actually now that I think about it it will probably be more like 160v.
If it's the latter, what does that signal look like in terms of peak voltages?
I do not know actually
does it need to be switched on a the beginning of the very first cycle
Probably the the very beginning of the first cycle, I do not know the audible effects of missing a cycle but the amplifier and listening equipment is very high resolution.
Do you need to prevent sharp transients in the audio signal as it switches?
Basically as far as the AC signal is concerned, what goes in needs to come out exactly the same.
I'm curious about the -30 V signal, too. Where did that number come from?
The -30v will actually be from a DC offset servo so it will actually be a varying voltage but it will be around -30v.
It is to bias the tube, the cathode of the tube expels electrons to the plate and the grid (where the signal is applied) modulates its voltage to act as a variable impedance.
Without a grid bias to set an idle operating point the current will essentially melt the tube.
 

#12

Joined Nov 30, 2010
18,224
As a person who knows about 10% of the state-of-the-art in audio control and modification circuits I believe that you can not have arrived at an unsolvable problem.

I think you have mistaken beliefs. For instance, "I must use a tube design that will melt within a few microseconds if it doesn't have an audio signal, but I can't show you a schematic of the circuit." "There is no such thing as a circuit that responds correctly without a switch in it." "There is only one way to make this work: start where I tell you to start." "A DC signal as high as a thousand volts can be blocked with nothing more than a $1 capacitor, for decades, while allowing the AC components through with no distortion at all, but I can't use that." "Tens of thousands of audio professionals have developed and built audio switchers that cost hundreds of thousands of dollars, but I can't learn from them or use the methods or circuits they use." "I know exactly what I want, but I refuse to explain from the beginning, what I need to accomplish with the audio signal, because dozens of well educated electronics engineers and audio specialists, with practical experience that can be measured in centuries are waiting to help me."

I don't believe you have arrived at the first insurmountable problem in audio processing, but I can't work with the information you are willing to provide.
 

AnalogKid

Joined Aug 1, 2013
10,986
The peak amplitude of the audio signal riding on the 160 Vdc is critical. If you don't know that or are unwilling to share it, why did you ask for help in the first place? Another missing piece is the noise on the 160 Vdc when the audio is absent. This sets the minimum detection level. I hope you understand that random audio buried in noise is undetectable.

Separate from the problem of detecting when the audio starts is detecting when it ends. This probably is why #12 is talking about the maximum possible 1/2 cycle signal period. This kind of "squelch tail" has been a common aspect of automated audio switching for, like, ever. If you think about it, you will see that there is no way for a circuit to predict the future.

But more important that any of that is your circuit environment. Are you really saying that the difference between 20 milliseconds and 50 milliseconds turn-off time is oh-so-critical? Turn-on time is different, and you need to face the fact that some of your audio will be clipped. How much depends on you, but again, a capacitor cannot predict the future.

ak
 

sailorjoe

Joined Jun 4, 2013
364
Coinmaster, I think we can still make progress. I get what you're trying to do to protect the amplifier tube.
As a beginner, there are details you may not know about that are critical to being able make a workable circuit. AnalogKid touched on some very important ones. Let's talk about detecting when audio is present.
Ever empty a bottle of milk? How do you know it's empty? The drips slow down, but there always seems to be one more drop in there. An engineer would create a specification like this: "The bottle is empty when the time between two drops is more than 5 seconds." In the same vein, how do you know if the bottle is full as you're pouring milk into it? A specification might read, "The bottle is full when the top of the liquid is between 1 and 2 mm of the top lip." On your 150 VDC line, there could be anything from zero to many millivolts of just noise, unrelated to the audio signal. Suppose we detect a 5 millivolts sine wave that lasts for one cycle. Is that good audio, or just a little noise? A detector can't decide between noise and signal without a discriminating feature.

Just occurred to me...is there another signal perhaps that turns the audio on and off, and which perhaps could be used to make the switch?

Do you know how long you can keep the 150 VDC on the grid until it melts the amplifier tube? I get that you want "as soon as possible", but there has to be an outside limit, otherwise what's the point.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
A DC signal as high as a thousand volts can be blocked with nothing more than a $1 capacitor, for decades, while allowing the AC components through with no distortion at all
Anyone who's ever build or modded an amplifier will disagree with you. Including myself.

I don't believe you have arrived at the first insurmountable problem in audio processing, but I can't work with the information you are willing to provide.
The peak amplitude of the audio signal riding on the 160 Vdc is critical. If you don't know that or are unwilling to share it, why did you ask for help in the first place?
Because even if I did know I wouldn't know what to do with it, I need to find out.
Another missing piece is the noise on the 160 Vdc when the audio is absent.
The DC supply and regulated should be very filtered, another thing I need to measure though.
If you think about it, you will see that there is no way for a circuit to predict the future.
Well I was originally thinking the the AC signal could be the on/off switch itself, or the circuit could switch after a circuit amount of "no AC"
The switch on timing is not as critical as the switch off.
you need to face the fact that some of your audio will be clipped
I can practically hear the heartbeat of the singer though my equipment(not really), in order for this to work the clipping would have to be as minimal as possible. I don't know the thresh hold at which it becomes audible but I seriously doubt that much clipping on what might as well be an audio microscope is a good thing.
Suppose we detect a 5 millivolts sine wave that lasts for one cycle. Is that good audio, or just a little noise? A detector can't decide between noise and signal without a discriminating feature.
I plan to filter the crap out of the supply so hopefully it won't be an issue.
Just occurred to me...is there another signal perhaps that turns the audio on and off, and which perhaps could be used to make the switch?
The only thing I can think of is the binary before it goes into the DAC. If the 0/1s could be used to make the switch maybe there could be a sort of "delayed output" to the dac as well if needed for better precision.
 

sailorjoe

Joined Jun 4, 2013
364
Woah, you have a DAC feeding the audio. Great! We can tap the digital signal and if it's not zero, turn on the audio to the grid. Completely unambiguous. Alternatively, we can use the quantization of the DAC, plus it's precision and noise figures parameters, to determine the min audio value to expect. How many bits feed the input of the DAC? Is there a part number?
Am I correct in assuming that you have an existing high power amp design that shows how to connect the parts, but doesn't tell you what all the signals are? No one could design an amplifier unless they knew the DC and audio voltages coming into it. If the audio voltage is too small you won't get anything useful out of the amp. If it's too large, you'll hear clipping distortion. So the answer to "what would I do with it" is complete the design of the amp and begin the design of the detector.
Lastly, do you have some reference information to indicate the need for this switch to protect the tube? I've never come across that requirement before.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Alternatively, we can use the quantization of the DAC, plus it's precision and noise figures parameters, to determine the min audio value to expect. How many bits feed the input of the DAC? Is there a part number?
This is the dac http://www.audio-gd.com/Pro/dac/NFB12015/NFB12015EN.htm
Am I correct in assuming that you have an existing high power amp design that shows how to connect the parts, but doesn't tell you what all the signals are? No one could design an amplifier unless they knew the DC and audio voltages coming into it
Essentially my learning experience thus far has been modding my amp with greatly increased quality parts and then the dismantlement and reverse engineering of my amp and asking lots of questions on the internet. I think the input voltage to the amp is 5 volts. Is it supposed to say somewhere in the spec sheet?
Lastly, do you have some reference information to indicate the need for this switch to protect the tube? I've never come across that requirement before.
There is no reference information, I am trying to bias the tube without the need for a DC buffer or other signal altering methods of biasing. In theory, as long as the grid is modulated (depending on the tube) the tube will not go passed its dissipation thresh hold. Normally biasing is done by setting a default operating voltage to the grid by setting it x amount of volts lower than the cathode, or by using a cathode resistor to adjust how positive the cathode is relative to the grid.
The problem is in order to use grid bias you need a DC blocking capacitor to separate it from the input stage, capacitors have long been the bane of audio, and cathode resistors have their own imprint on the way the output sounds (albeit greatly reduced compared to capacitors), not counting the fact that a cathode resistor is usually bypassed by a capacitor for the AC signals to pass through anyway.
 

sailorjoe

Joined Jun 4, 2013
364
That information helps a lot in understanding what you want to do. Thanks for that.
First, I noticed that the DAC is taking in an audio bit stream, not a parallel bit pattern, which makes sense, but essentially eliminates the ability to use it to detect the presence or absence of audio. OK, so let's take that option off the table for now.
Yes, the specs for the DAC show the output levels, and that's what I needed. Its not clear to me whether or not those levels are enough to directly drive your amplifier, but save that question for later. Is the DAC currently wired into the amp, and it works? Does the DAC directly the drive the grid, or is there something in between?
If you take the XLR output of 5Vp-p of audio, and divide that by 32 bits of precision, you get the difference between audio and no audio as 0.5 nanovolts. I don't know if you realize how infinitesimally small that is, but without extraordinarily complex circuitry, you can't detect that reliably. Anyway, that's the lower bound, so we just have to go up from there.
Since one of your goals seems to be the elimination of capacitors, have you considered rearranging how the tube is powered? For example, let's say that the DAC ground is the system ground. It puts out a 5Vp-p audio signal that is 2.5 VDC quiescent, for example (I'm not exactly certain of the signal). You feed that directly to your grid. Your cathode wants to be X volts higher than the grid, so you make a power supply to set that voltage. Avoids the problems of self bias. Then you set a different power supply for the Anode. These power supplies can be high quality low noise solid state power sources, and can be made adjustable. One of the reasons for things like self biasing cathodes and other tricks is that power supplies are expensive, so any method that eliminates one is good for business. But if sound quality is your primary goal, then added power supplies might be better for you.
I'm very concerned that if you find yourself switching the grid from150 V to -30 V back and forth, you'll be introducing a large audio component into your grid that will swamp all other audio sources.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Is the DAC currently wired into the amp, and it works?
The dac works, the amp or more or less a scrap heap at the moment, I'm waiting on a batch of parts to arrive to officially rebuild it.
Does the DAC directly the drive the grid, or is there something in between?
It goes into the driver tube grid which has an amplification factor of 20. The there's no need for compromises in the driver stage usually.
The next stage (the output stage) is the difficult one where the problems and compromises arise. This is the stage where an alternative method of biasing would be beneficial in order to reduce compromise.
Since one of your goals seems to be the elimination of capacitors, have you considered rearranging how the tube is powered? For example, let's say that the DAC ground is the system ground. It puts out a 5Vp-p audio signal that is 2.5 VDC quiescent, for example (I'm not exactly certain of the signal). You feed that directly to your grid. Your cathode wants to be X volts higher than the grid, so you make a power supply to set that voltage. Avoids the problems of self bias. Then you set a different power supply for the Anode. These power supplies can be high quality low noise solid state power sources, and can be made adjustable. One of the reasons for things like self biasing cathodes and other tricks is that power supplies are expensive, so any method that eliminates one is good for business. But if sound quality is your primary goal, then added power supplies might be better for you.
I have applied the method you speak of to the input stage via a gyrator on the plate and a CCS on the cathode supplies by a filtered and regulated power supply.

The problem is, the output stage is the tube equivalent of a a unity gain source follower or whatever they are called in solid state, basically the cathode of one triode is connected to the plate of the other, a servo adjust the bias of the input triode so the impedances of the tubes are matched to the point where there is 0VDC on the output, this removes the need for an output capacitor or transformer.
So essentially the method you speak of would work fine but I would simply be moving the capacitor from the interstage to the output.

I'm very concerned that if you find yourself switching the grid from150 V to -30 V back and forth, you'll be introducing a large audio component into your grid that will swamp all other audio sources.
Yeah I've been thinking about that, you may be right. It may be a futile effort after all.
 
Top