Sensing load impedance

sailorjoe

Joined Jun 4, 2013
364
The VCO is part of the full circuit that measures the instantaneous frequency. The other main part is a PLL, a phase locked loop. Both of these are very hard to design and implement properly.
The purpose of the function generator is to take the signal from the instantaneous frequency measurer and convert it to the proper control voltage for the plate resistor, based on the shape of the headphone impedance curve. Remember the headphone impedance is a function of frequency, and the function generator takes the frequency in and outputs what the headphone impedance is at that frequency. Then that signal is used to control the plate resistance.
The band analyzer is less like a spectrum analyzer and more like a band equalizer. It's a few selected bands each filtered out independently. A spectrum analyzer uses one of several derivatives of the Fast Fourier Transform to analyze a time domain signal. Very different process, and in my opinion, the wrong math for what you want to do.
 

sailorjoe

Joined Jun 4, 2013
364
Theoretically, the instantaneous frequency methods has almost infinite granularity in terms of matching frequency to impedance. But it's also the most complex circuitry to build, even for an expert. Also, it's not clear to me yet how much delay and responsiveness this new circuitry will have. In other words, it may not be able to keep up with the music feed.
The band analysis method has far less granularity, but the granularity is theoretically adjustable by increasing the number of bands, at least up to a point. It's circuitry is far simpler to implement, but my no means trivial. Same issue with inherent delay and responsiveness. Also, it's not clear whether we need to use the average or the maximum value from the band filters to drive the function generator (same one as above), so that will need to be discovered.
Those are the top level pros and cons. There are a lot more details to consider within each option. For example, active or passive filters, induced phase delays and their effect, if any, on the outcome, op amp characteristics and performance, test and verification capabilities (does it work they way you thought it would), control signal filtering, and finally, how to control the plate resistor with a FET in such a way that it doesn't create any distortion in the final output.
 

sailorjoe

Joined Jun 4, 2013
364
I'm not sure if JFETs are the most linear transistors or not. I haven't tried every transistor in the world to be sure. I doubt the two you provided in the links are good choices. The first one is designed for high speed high current switching. It's clearly not very linear. You can see this in the graph of drain current versus gate current or voltage. The second one is much more linear, but it's designed for high currents, much higher than I believe you need. It's also quite sensitive to temperature. No need to worry about the FET at this point, though. There are some other steps to do first. Once we get to where we have the function generator ready to generate a control signal, we can find the right FET for the job.

First step: establish a baseline. Do you have an o'scope? Doing this without one is virtually impossible.
Do you have a signal generator? You will need to generate high quality sine waves into your amplifier for a number of different tests that will be needed now, and later. If you have both of these things, then run the first test, which is to find out if the distortion of an unmatched plate resistor is detectable. Set the plate resistor to the match the impedance of the headphones at a particular frequency. Make sure the load signal is a pure sine wave. Now vary the plate resistor. Since the headphones have 350-650 ohm impedance, use that range. Put the o'scope on the output signal, and see if you detect distortion. If not, there are other options, but more complicated.
 

sailorjoe

Joined Jun 4, 2013
364
You know, I just had a thought. What if the plate resistor was converted to a plate impedance, with an impedance versus frequency function that matched the headphones? Then the plate resistor would just naturally follow the signal frequency and self adjust for instantaneous frequency. Has anyone tried that before?
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Do you have an o'scope?
Do you have a signal generator?
I have a https://www.picotech.com/oscilloscope/2200/picoscope-2200-portable-oscilloscopes so that counts as both.
then run the first test
I deconstructed my amp months ago when I reverse engineered it. The PCB is pretty busted so I'm trying to figure out a way to rebuild that isn't a flimsy dangerous mess. I'm planning on point to point and hotglue, could be a month or so before I officially have it rebuilt.

You know, I just had a thought. What if the plate resistor was converted to a plate impedance, with an impedance versus frequency function that matched the headphones? Then the plate resistor would just naturally follow the signal frequency and self adjust for instantaneous frequency. Has anyone tried that before?
Wait what? Can you explain further?
 

sailorjoe

Joined Jun 4, 2013
364
Sure. In simple terms, an impedance is something that varies with frequency. It varies in two ways, both amplitude and phase change. A resistor doesn't change with frequency, at least in theory. It should be possible to build a small circuit that replaces the plate resistor with an equivalent impedance that just happens to vary exactly like the headphone impedance does. You would be matching both amplitude and phase shift of the headphones. To do that, we'd need to add at least one reactive component to the plate resistor. Still interested? I'll be back tomorrow.

By the way, nice scope.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Hmmm, I've never heard of such a device. The phase of the frequencies at the plate resistor would be 180 degrees out of phase with the output. You have to be careful what you throw onto the plate to act as a resistor, even different resistors have been known to affect tone, I'm not sure how you would build this device but it would have to be linear. By reactive component you mean capacitor or inductor right?
 

sailorjoe

Joined Jun 4, 2013
364
To be clear, impedance is not a device, it's a feature of all physical electronic devices.

Yes, reactive devices are capacitors and inductors. But combined with resistors they don't throw the signal 180 degrees out of phase. The thing is, the impedance of the headphones is a reactive effect, too. It's not purely resistive. If you want to match the impedance of the headphones, perhaps a reactive match is better.

I believe you that different types of resistors can change the tone. All physical devices are imperfect representations of their theoretical counterparts. Resistors have a small amount of capacitance by virtue of their physical implementation. They may also have some molecular effects that are not ideal. The best that anyone can do is find the type that works best in a specific application.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
I thought impedance was just the simple term for AC resistance, how are we going to create this thing you speak of that matches the curves of the headphones?
But combined with resistors they don't throw the signal 180 degrees out of phase.
The tube plate itself is 180 degrees out of phase.
 

sailorjoe

Joined Jun 4, 2013
364
Correct me if I'm wrong, but isn't the plate voltage 180 deg out of phase with the plate current, but the plate current is in phase with the load current. In other words, if the plate current goes up, the load current goes up. No?
 

sailorjoe

Joined Jun 4, 2013
364
Coinmaster, I've been thinking about your situation, and I've come to the conclusion that the original two options I suggested would simply not work. The basic problem is that at any instant of time there are lots and lots of frequencies in the signal simultaneously. Sure, there is the special case where only one note is played, but that's the exception. So if we could actually vary the plate resistor based on the signal, it would only be a good match for one frequency and all the others would be mismatched. Make sense? Even sensing the instantaneous frequency is sort of bogus, because it only lasts for an instant, like a microsecond. The next instant we'd get something else. But the down side is that now the tube quiescent operating point is being disturbed constantly, and that can't be good for the purity of the amplifier. The amplifier would be generating its own noise signal!
So in the end the only option is to replace the plate resistor with a plate impedance that matches the headphone impedance. That way, every single frequency would see a matched impedance simultaneously and all the time, yet the quiescent operating point wouldn't be jumping around.
I need some time to work up a circuit, so look back periodically and see what I come up with to try.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
That is a good point. The problem with that is the impedance at each frequency would have to work out to the result of the equation laid out in the cavalli paper. Unfortunately I don't completely understand it and I've been tied up lately and haven't had time to further analyze it.
My question is how does such a thing affect the operating point? I also wonder if a shifting operating point is the basis of the problem at all. Based on LT spice simulation the DC offset servo in my amplifier keeps the operating point stable with a changing load impedance, at least of a single value. Any alteration from 0v at the output will cause the servo to adjust the impedance of the valve accordingly maintaining 0v at the output.
Seeing as the bottom triodes operating remains stable then the top triode will always be at 75ma/75v respectively in order to maintain a 0v output.
There's also the fact that conventional knowledge of deciding plate resistor value is not determined by load, this way of thinking was determined by John Broskie and the author of the cavalli paper. So the true nature of the plate resistor and the load escapes me. These guys are highly respected so they must know what they are talking about.
Have you gotten around to taking a look at the cavalli paper yet?
 

KL7AJ

Joined Nov 4, 2008
2,229
At audio frequencies you can sense current with a very small series resistor or with a hall effect sensor. The hall sensor does not place a measurable impedance in series with the output, but there is some phase delay between the incident signal and the sensor output. A series resistor has no such phase delay, but it is a small resistance. You might be able to place the resistor inside the amp's overall feedback loop, reducing its effect on the system output impedance.

Separate from that, do you want to track the instantaneous load impedance, or it's RMS value? Or, more correctly, it's impedance based on the instantaneous or RMS value of the current through it? Is this for a test setup or something to be operating while playing music?

Finally, why?

ak
Yep.
 

Sensacell

Joined Jun 19, 2012
3,432
Let's design a system that servo's the quiescent point, maintaining it at a constant level, regardless of the phase or frequency of the input signal.

I think it might end up looking like this:

PSU.jpg
 
Top