|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.223
In Reply to: RE: Big Oops! posted by danlaudionut on September 07, 2014 at 17:34:02
Dan, I ran the simulation at 5 kHz. There is sufficient drive at the 211 to create grid current peaks of more than 30 mA. I examined AC (short-term) and DC (long-term) voltages both with and without grid current being drawn. I also ran the sim with various values for transformer inductance, bypass capacitance and bias supply resistance. This was by no means a static test. :)OK, with that out of the way, you know how people say "I made a mistake once - it was when I thought I had made a mistake"? Well, I haven't been paying enough attention to the polarity of currents in this sim. Now I think my first statement - that A2 grid current doesn't increase bias supply current - was correct. After studying the sim at length this evening, I now know what happened to make me think otherwise.
The phenomenon I missed is that grid current causes the bypass cap under L2 to gradually charge to a more negative DC value. I hadn't noticed that before, because I was so focused on the current flow. As the cap becomes more negative, it creates a DC current through the bias supply that I misinterpreted as current draw. What I was actually seeing was the negative "overvoltage" at the cap draining off into the bias supply through the -48V low-impedance voltage source. I have remedied this by installing a diode between the voltage source and resistor R2 (see D1 below). With the diode in place, the sim now demonstrates conclusively that there is NO ADDITIONAL CURRENT drawn from the bias supply when the 211 is in A2 mode. That holds true regardless of how long the amplifier remains in that condition. So, now I can say "I made a mistake once, but it was only because I thought I had made a mistake." :)
Ironicaly, all this additional simulation and analysis has pointed out a bias problem you didn't ask about. As I said above, the bypass cap becomes more negative when the 211 draws grid current. Over time, it becomes so negative in fact that the amplifier is no longer drawing grid current. Instead of -48V, the cap will rise to -60V or -70V (depends on drive level). This causes the negative peaks of the signal at the grid to extend beyond cutoff, so the output distorts. Worse, when signal levels subside, the cap holds that charge. It can't discharge into either the bias supply or the grid.
I'll give this situation more thought, but my initial reaction is that maybe a regulated bias supply is needed after all. It doesn't need to provide much current (uA only), but clearly, some form of shunt regulation is needed to maintain the correct voltage at the bypass cap.
--------------------------
Buy Chinese. Bury freedom.
Edits: 09/08/14 09/08/14 09/08/14 09/08/14Follow Ups:
I've built, rather than simulated, several iterations of 211s being driven into A2 and the grid current always causes the bias to slide.
Mark Kelly
Mark, were the amplifiers you measured transformer coupled like the one under discussion? Also, tell me what you mean by "slide" and how you measured it. SPICE is showing the exact action I would expect from this topology.
--------------------------
Buy Chinese. Bury freedom.
I shouldn't have responded, I have no desire to go into the minutiae of this.I gave up on grid driven A2 a long time ago, I think it's a fundamentally flawed topology. My experience from when I was working on it about two decades ago supports what Paul has said.
Mark Kelly
Edits: 09/08/14
Clearly, different builders have different experiences. My SE 211s can be driven to 30W in A2. The transition from A1 is virtually seamless, and I'm extremely pleased with the way they sound. Difficult for me to imagine ever building a high power DHT any other way. Different strokes and all that... :)
--------------------------
Buy Chinese. Bury freedom.
Agreed, I love A2 once the power gets high. My 833C SET monos do 40W in A1 and then transition to A2 up to 200W. Sound is incredible. Spice sims show typical harmonics in A1, with second dominating, then transitioning to more third and higher harmonics as the amp goes progressively into A2. At that point it's pretty loud (he he).
"At that point it's pretty loud"
What speakers?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Infinity RSIIb. They should be up near 100dB at 40W. Of course the peaks need more, hence the 200W capability. They get as loud as I can stand without breaking a sweat.
Very nice.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
"My 833C SET monos ... transition to A2 up to 200W."
Those would make dandy preamps for these little Westinghouse units...
--------------------------
Buy Chinese. Bury freedom.
Those would make a great calibration standard for WAF, to set the zero point.
Yeah, but just think... what if you got away with it??!!
--------------------------
Buy Chinese. Bury freedom.
Eliminate the diode and install an appropriately sized resistor in parallel with C3?
Make that resistor a zener (as suggested below) and you are there :^)
I couldn't bring myself to recommend a zener
;-)
Jim, the diode is strictly for simulation purposes, to prove that bias supply current doesn't increase when the amplifier enters A2. That's really the point of the entire thread. FWIW, I probably should have had the diode in the circuit all along. A typical half-wave bias rectifier would look just like this - a source (transformer) feeding a series diode. The diode would be oriented the same direction, and just like this, it would prevent the cap from discharging.
Yes, a resistor across the cap would discharge it and help to stabilize the voltage. However, based on the simulation, the resistor would need to be a fairly low value, maybe only a few thousand ohms. That would require much more bias current than Dan has available.
--------------------------
Buy Chinese. Bury freedom.
Thanks for the explanation, TK.Off the top of my head, I was guesstimating a 2k7 5 Watter would work (which would draw approx 18 mA per channel) - so, you'd probably need a bias supply that could supply 50 mA (stereo).
Edits: 09/08/14
then I would eliminate the cap altogether and let the bias supply feed the grid. Job done and full A2 capability.
Naz
"I would eliminate the cap altogether and let the bias supply feed the grid."
Don't forget that the capacitor serves to decouple audio-frequency AC at the bottom of the secondary. It has to provide a low-Z path to ground at the lowest frequency of operation in order to be effective. Any bias supply capable of substituting for that would need a similarly low-Z output capability.
--------------------------
Buy Chinese. Bury freedom.
Sorry, I should have been more specific. Eliminate the cap and feed the grid through the secondary. In fact I would do it this way even with the Tent Labs Bias supply. Yes, the bias supply must handle all the A2 grid current but as I said, I have had extensive experience with 845s and know that it will get him further into A2 than a Sim will indicate.
I don't know enough about the Tent Labs supply but I would strongly suspect that it will handle that reservoir cap I mentioned and that would certainly help transients go all the way into A2.
Naz
I'm not being critical, just trying to understand - given the overall expense of this undertaking:a) is DanL attempting to limit "full A2 capability?"
b) does a 10 mA (stereo) bias supply vs a 50 mA supply constitute serious design limitations/concerns?
Edits: 09/08/14 09/08/14
The limiting factor is the Tent Labs Bias supply so I see basically two options, ditch it for something with higher current capability for full A2 operation or use it a find the best compromise. I think we are working with the latter.
From experience, I know that it can be made to work reasonably well but more current would be a better solution. That's my only point.
Naz
The capacitor is charging because the diode in the bias supply prevents it from providing the current needed to discharge it. Which is equal to the net average DC grid current.
"The capacitor is charging because the diode in the bias supply prevents it from providing the current needed to discharge it."
Yes, but that doesn't mean grid current is coming from the bias supply. Just the opposite, in fact. When the capacitor charges due to grid current, it becomes more negative than the bias supply. No current flows from the supply when this occurs. If the diode is removed, current will actually flow *into the supply* as the capacitor discharges. In other words, when grid current is drawn, the bias supply is not required to deliver current, rather it is *fed* current from the bottom of the transformer winding. This means the power consumed by the grid - of which one component is grid current - is supplied entirely from the driver tube, not the bias supply.
--------------------------
Buy Chinese. Bury freedom.
My apologies, the words we are using are confusing. I should have said the grid current "flows through" the grid bias supply, not that it "comes from" the supply.
Given what you are saying, the grid bias supply could be just a zener diode across the capacitor - all it has to do is absorb the grid current!
That was supposed to be funny, not a serious worked-out design. :^)
" the grid bias supply could be just a zener diode across the capacitor "
It still needs a negative supply to provide bias when grid current isn't flowing. I like the idea of a Zener for setting a ceiling on the voltage across the cap. Not sure how to adjust it, but it's a low parts count!
--------------------------
Buy Chinese. Bury freedom.
As I said, not a worked-out design. :^) Lots of more complicated ways to do it; ultimately it's a shunt regulator. I think. Not a worked out design ...
TK
"I Thought I was Wrong Once…But I was Mistaken"
said by Oscar Wilde
In my situation ...
Using the Tent Labs Bias Supply would
give me the regulation needed.
It can source or sink 1.5mA.
DanL
"...the Tent Labs Bias Supply ... can source or sink 1.5mA."
According to the simulation, source current is very small, less than 100 uA. However, sink current might be more problematic. I added a simple shunt regulator to get an idea of what's required. In the circuit below, if the grid is driven hard, Q2 must sink 3-4 mA in order to maintain a constant DC voltage level at the bypass capacitor.
This doesn't mean the Tent supply won't work, but its effectiveness will depend on how hard the 845 is driven. It appears that if you input a sine wave at full A2 power, bias voltage will increase if the sink is limited to 1.5mA. However, if the amplifier only experiences brief A2 operation on transient peaks, the 1.5 mA sink should be satisfactory. Also, the larger the value of the bypass cap (or reservoir cap, as Naz referred to it), the longer it takes to charge up when grid current is drawn. That provides a larger window over which grid current can be drawn without significantly disturbing the amplifier's operating conditions.
I could try to simulate this with a 845 if you're interested. However, I've found that many tube models don't draw grid current properly. The 211 I'm using now was the third one I tried.
--------------------------
Buy Chinese. Bury freedom.
From the pic on the website ...
It looks like a PP complimentary transistor pair
directly to the output with no cap.
So I guess a IRF9610 CCS loaded follower
would be the answer to my needs.
Source to Ground
Drain to CCS to -300V
Tent Labs to Gate
Output from Drain
Modest 10mA on CCS
Peaks would draw more from IRF9610
DanL
I need an N channel not a P channel FET.
I have 2SK2700s on hand that are overkill.
Same FET the Power Drive uses.
Same circuit as described above.
DanL
There are a number of ways to get this done. I like the simplicity of bipolars for this sort of thing, so I'd probably use the Tent to drive a NPN high voltage Darlington, maybe the NTE2540 or similar. Whatever works!
--------------------------
Buy Chinese. Bury freedom.
Edits: 09/08/14
Dan, are you married to the Tent labs Bias supply or could you get a little more current by using something else? I'm not familiar with what you are using but my thought is that it would open up your circuit configuration options.
Naz
I have some questions on the simulation:
1) why is R2 such high value?
2) the grid current for the GL211 model seems to be very different from the datasheet, so how does it match up to the real circuit?
Also, just how do you define "A2 only on the peaks"? Is it 5V, 20V, or 100V... over 0V?
"1) why is R2 such high value?"
It was my intent to model the circuit in a way that would generally represent a real-world, high impedance bias supply. Many use half-wave rectification with a large value resistor like this connecting the supply to the grid. This seems appropriate considering Dan's mention of limited current capacity.
"2) the grid current for the GL211 model seems to be very different from the datasheet, so how does it match up to the real circuit?"
The GL211 model biases to approximately the same anode current relative to bias voltage as the real thing. It also draws grid current when excitation voltage swings above zero (some models don't do that). Whether the model draws the same amount of grid current relative to drive voltage as an actual 211 probably isn't too important. I used the 211 only to flush out basic circuit operation. Dan will use the 845, so values will be different.
"Also, just how do you define 'A2 only on the peaks'?
What I mean by this is that the amplifier is essentially biased for Class A1 operation. It operates in A2 only during musical peaks that draw grid current.
--------------------------
Buy Chinese. Bury freedom.
Thanks for the clarifications, so it's very mild A2 we are talking about - then the likely grid current should be pretty low. But the Tenlabs bias supply seems to be designed with a very low output impedance, so how does it fit into the equation here?
Edits: 09/08/14
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: