|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
128.194.160.53
In Reply to: RE: another update.... posted by vinnie2 on December 06, 2016 at 10:33:09
An interesting trace. It looks like the burst of high-frequency 35V peak-to-peak output occurs for quite a small fraction of the mains cycle. And it seems to be either on or off, with no half measures. That's quite surprising, because it means the true rms voltage must be really quite low.
By the way, is that with your 100uF filter capacitor still in place, or did you remove it?
One other thing, concerning the sweep speed on your scope. It would be worth looking at the mains frequency itself, on the 5V winding of the transformer you were talking about the other day. Since that is a reasonably precise 60Hz sinewave, it would provide a useful frequency standard for checking the accuracy of the sweep speeds in the mS range.
Chris
Follow Ups:
I think I am using my fluke 179 properly now for measuring voltage at higher frequencies and it is giving what it says is a true rms reading of 9.65 volts.
Yes, the 100uf cap was still in place during these traces. I wanted to do some listening tests with them in place right after running the traces, so I didn't take it off. Does that make all the traces bogus?
No, the traces are presumably perfectly valid. They might look rather different with the 100uF capacitor removed, and it could be interesting to see what happens to that latest trace you posted if the capacitor is removed. But what you posted is presumably a true depiction of what the output voltage looks like in the present configuration.It is striking that the burst of high-frequency output voltage occurs for only a small fraction of the time; maybe about 1/6 of the time, estimating from your scope trace. If we assume that the waveform during the burst is an exact square wave, of say 36V peak-to-peak, then that would mean the rms voltage would be 18V during the "on" period of the bursts, and 0V during the "off" periods between the bursts.
Thus, when we do the time-averaging to calculate the true rms, we would get, I think, 18/(Sqrt6)V, which is about 7.4V rms. Since the square wave is not perfect, the actual true rms voltage would be a bit lower even than this.
I'm not sure how much trust one can put in the ability of a "true rms" meter to handle a case like this, where the output voltage is composed of periodic short bursts of a distorted high-frequency square wave. I don't know how these meters work, but I could easily imagine that such a peculiar type of waveform was not really being catered for when it was designed. Maybe I am being too skeptical though ...
Chris
Edits: 12/06/16
all I know is the Deafbkh said I should try different ranges for voltages when using high frequencies, that it would bounce a lot until I hit the right range and then it would lock on. That is exactly what happened. It locked on to 9.65 and held it very steady. Other ranges had it boucning all over the place. It is capable of measuring high frequencies, so I guess they maybe planned on measuring voltages at higher frequencies too?
We appear to both have the same supplies/boards
If you haven't removed any windings, you should net about 13v with a 220uf cap.
I did this with a 1.3amp load, here's a scope shot of what mine looked like.
Cant remember if this shot was after I removed some secondary windings.
I can give some better shots of this later in the week.
"I can give some better shots of this later in the week."
Is that a stored single-sweep trace in that shot?
While you are getting some better shots, I wonder if you could also get one done at a much slower sweep rate, so that the modulation of the high-frequency output by the rectified 60Hz mains would be visible?
I realise that you have a 220uF filter capacitor connected at the moment. If it were not too much trouble, it would also be very interesting to see the mains-modulated output in the case when the capacitor is not connected.
Chris
The Fluke seems to somewhat agree with the scope.
Without the cap is 12v, with the cap is 14.7
I believe removing about 3 windings from the toroidal will net about 10.5 if I recall.
Edits: 12/07/16
Now this gets interesting, this is on my variac with 60vac in.
Output freq goes up and voltage is about half.
I wonder if a 220-230v electronic transformer would do the same with 120vac input. Might work well on 801's, 300b's, etc...
That is very interesting. Thanks for the idea for the lower filament voltage tubes.
ray
I think the only differences between the 120V and 240v units might be the component values in the startup circuit and turns ratio of the output transformer. I would want the startup circuit to be designed for the voltage actually in use, so my preference would be to use the 120V unit and rewind the toroidal output transformer.
What I find most interesting about all this is the rise in oscillator frequency at reduced voltage. I don't see a good way to make use of that, though. If I'm correct, B+ is dependent on the mains voltage, not whether the unit is designed for 120V or 240V.
--------------------------
Buy Chinese. Bury freedom.
The best I could reduce the voltage before the transformer quits working is about 8-9 turns off or about 5volts.
I ordered some 220v supplies and will post results once I start testing.
My thought was it may allow a further reduction, since the input voltage can be cut in half.
"The best I could reduce the voltage before the transformer quits working is about 8-9 turns off or about 5volts."
I've been thinking about that for some time. I suspect the cessation of oscillation isn't directly related to the "front end" of the supply. It's very possible that reducing the number of turns on the secondary is simply creating insufficient inductance, and the winding is acting like a shunt. You could prove or disprove this by leaving the secondary turns as-is (operational) and rewinding the primary with more turns. That will have the same effect on output voltage, but will maintain critical inductance. I would do the test myself, but the toroidal versions I just ordered won't be here for two to three weeks.
--------------------------
Buy Chinese. Bury freedom.
OK, I spent a few minutes to throw something together on the bench.The first photo below is a modified 60W lighting transformer driving a 211. The cap hanging off the bottom of the PCB is 47u/250V. The original fusible resistor has been replaced with a 3.3 ohm 3W carbon film (lower right). Windings have been removed from the primary of the oscillator transformer in order to raise the frequency.
Here's a screenshot of the square wave at the pins of the tube. I'm using resistors in series with the transformer that reduce the voltage to about 8.5 VAC (values that were handy for the shot). Frequency is about 55 kHz. Note that the unloaded frequency is about 66 kHz. I'm not very happy about this shift, and I don't remember it occurring in earlier units from 10 years ago. This is another reason I want to try the standard SMPS for this (their frequency isn't load-dependent).
The shot below shows the 120 Hz modulation envelope. This is really very clean, especially compared to a standard 60 Hz filament transformer. Used with a hum balance pot at the filament, this level of hum would be completely inaudible.
Finally, this is the 120 Hz modulation envelope with the 47uF filter cap removed. As remarked by member cpotl, RMS output is much lower. The 211 lights much more dimly than with the filter cap in place. This raises the issue that even in the filtered version of the supply, 120 Hz ripple can reduce actual RMS. It's important to reduce ripple as much as possible if RMS will be calculated solely from the square wave amplitude.
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/07/16 12/07/16 12/07/16
Nice shots! That switcher seems to be behaving in exactly the way one might have hoped!In the final example, without the filter capacitor, the envelope of the HF oscillations is amazingly close to the ideal for a rectified sinusoidal supply. Now it is really quite easy, in both the unfiltered case and the filtered cases, to estimate the true rms voltage. Since the HF voltage, in your first shot, is a pretty good approximation to a square wave, one simply has to use the envelope seen in the low-frequency traces in order to compute the rms value. A slightly more refined calculation could also be done, where one first allows for the not quite square wave nature of the HF waveform and does an rms averaging over that, and then afterwards a low-frequency rms averaging.
It is interesting, by the way, that one can see how the HF oscillations fire up once some critical voltage is reached in the rectified mains sinewave. And then on the downward portion of the sinewave segments the HF oscillations continue almost until the sinewave gets to zero.
Nice!
Chris
Edits: 12/07/16
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: