|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.22
In Reply to: RE: update posted by cpotl on December 04, 2016 at 11:11:01
I agree with cpotl. Like anything else, filament transformers have winding resistance. They're rated for output voltage at full load. Your measurement is essentially correct.About the lighting transformer, yours has a toroidal output transformer. Based on the photo, the output winding has 9 turns, which means approximately 3.89V pk-pk/turn (35V pk-pk total). So, if you remove three turns (11.67V), the output should decrease to 23.33V pk-pk, or (for a perfect square wave) 11.67V RMS. Because your output wave isn't as square as we would like, I would deduct perhaps 10-15% from that, which leaves 9.92-10.5V RMS. Under load, it will probably decrease somewhat, but in any event, this should be safe for the 813.
When you remove the turns, try to spread out the remaining turns so they're fairly evenly spaced around the core. That will aid primary-secondary power transfer.
Can you provide a link to the units you purchased? That oscillator transformer is also different than the ones I've been buying. I suspect it's a lower permeability, which would account for the higher switching frequency. Also on this point, if you wouldn't mind, could you confirm the measurement regarding frequency that you made earlier? Don't need a photo, just confirmation. Thanks!
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/04/16Follow Ups:
Those are some pretty good guesses there TK. Whilst you and cpotl were responding to my last email, I was out in the shop slinging solder. I had decided to just go ahead and take a couple of turns off the winding just to get something rolling. I left the 12sn7 hooked up to it so it would have a load and I got about 30 volts p to p, which gave me roughly 10.6 volts, but probably less since my square wave must have been rubbing up against a sine wave and some of rubbed off. A long ways from perfect. Any way, I figured with the 813 on there it would drop a bit more as well. So decided to try my luck with an 813. I got 27.5 volts for about 9.8 rms so decided to go for it and do a visual check at the moment I turned it on. The second 813 in the right channel is still on a switcher unit, and it is right on 10 volts dc. I figured if the ac one glowed too much brighter than that one at turn on I would shut it down instantly and hope for the best. As it turned out it glowed darned close to the same brightness, just a tad less, so I think my voltage estimates are pretty close. I will have to do this all again when the caps have been installed as they will boost the voltage, but it is a fairly simple process.
I went ahead and played some music for a while as the hum was not as bad as thought it might be. Probably because it's a hybrid right now with half dc and half ac on the filaments. It didn't sound too bad at all. I think if I can get the noise whipped we might be on to something here. Tomorrow I will move this halogen to the left channel and also add the second halogen to that channel so that I can compare the sound of the ac halogen channel to the dc switcher channel. I have the 560 uf caps ordered, but they will probably be a few days getting here. I hope they will knock out most of the hum.
Almost forgot.... I looked up that symbol and it is the Greek letter mu and it is used for micro as coptl suggested. All of my measurements were done with the sec/div on either 1 or 2 microseconds.
I have attached a pic showing the halo in the front and the switcher in the back. It takes up a lot less real estate. I will look up the link you wanted and post it soon.
Found it right away. Here it is...
Edits: 12/04/16 12/04/16 12/04/16 12/04/16 12/04/16
I tried using an infrared thermometer in my early tests but did not have too much success since I was only testing a resistor on a sheet of aluminum. I could not get much consistency.
However it might be interesting to use such a thermometer to check the temperature of DC heating compared to AC heating or high frequency AC heating if you could devise some accurate repeatable set up. Hand held is not good enough I suspect.
ray
Why not attach a thermistor to the resistors and read the resistance hot? You'd need to calibrate it using the DC supply, say at 9.5, 10 and 10.5 VDC. The reading with AC could then be compared and adjusted to obtain the same RMS power.
--------------------------
Buy Chinese. Bury freedom.
Good idea.
ray
Great, sounds like good progress! Is the 560u cap for the purpose of smoothing the DC in the lighting transformer? If so, you can get by with much less. Probably as lttle as 47u/250V will render the hum inaudible. I'm just sayin' if you have something like that lying around, you might not have to wait. Congrats on getting this together, I'm anxious to know whether you hear an improvement using AC with the 813!
--------------------------
Buy Chinese. Bury freedom.
560 is what Ray P used used in his, so I thought it was needed. However, I sure will try it with less in the meantime, like tomorrow, as I have some 47uf/450v units in my parts bins. Thanks for the tip. Let's see what happens.
Wait a minute, it just dawned on me you said reduce the ripple in the DC.... where is the dc in the halogen unit, I thought it was all ac?
Edits: 12/04/16
The halogen switchers are configured like the schematic below. If you don't add the capacitor in dotted lines for filtering, the output will be a square wave modulated from zero to peak voltage at a 120 Hz rate. IOW, probably just as bad from the standpoint of audible hum as a standard transformer.
Look at the output of the supply again with your scope. Reduce the sweep speed so you can see low frequencies, and the 120Hz modulation envelope will be clear.
--------------------------
Buy Chinese. Bury freedom.
"Reduce the sweep speed so you can see low frequencies, and the 120Hz modulation envelope will be clear."
I hadn't realised that these switchers were operating on unfiltered rectified AC. That could perhaps explain why the scope trace that vinnie got was so shaky.
But more significantly, it could also mean that the true rms output voltage may be substantially less than might at first appear. If that 35V peak-to-peak is only what is achieved at the peak of the mains voltage cycle, and it is then dropping down to essentially zero during the part of the mains cycle where the switching transistors are receiving only a small voltage from the rectified mains, then the true rms output voltage from the transformer, with the averaging now being done over the entire cycle of the mains frequency, will be much less than simply the rms calculated under the previous assumption of a continuous somewhat distorted 35V peak-to-peak square wave.
Maybe this could indeed bring the true rms voltage down close to the advertised 12V. I can imagine the relation between the instantaneous rectified mains voltage and the output square wave voltage on the transformer could be a rather non-linear one, and so estimating how to do the further rms averaging over the 120Hz cycle might be tricky.
It would be very interesting indeed to see a scope trace of the modulated output voltage with the timebase synced to the mains frequency. This could allow a fairly useful estimate of the true rms output voltage to be made.
Chris
Tell me how to set it up and I will take and post a picture.
"Tell me how to set it up and I will take and post a picture."Try turning the vertical sweep rate down to something like 2mS per division (2 milliseconds), or so. With luck, you might be able to adjust the sync so that it locks onto the mains frequency that will be modulating the high frequency square-wave oscillations.
You should hopefully be able to get a trace that shows as a sort of solid block but with an amplitude that is modulated in sync with the mains. (Imagine first a rectangular filled-in strip running across the screen, and then imagine that the top and bottom edges of the strip are themselves wobbling up and down in some sort of vaguely sinusoidal fashion.)
If you can't make it sync on the mains frequency that is modulating the amplitude of the high-frequency output, then assuming you have a dual-trace scope there is another way you could achieve it. First hook up that 5V mains transformer you were talking about yesterday to the one of the scope inputs, and get the sinewave nicely displaying and synced. Then, connect the output of the halogen switcher to the second scope input. You will still be getting the syncing from the 5V transformer, but the second display, which is the one you really want to see, should be nice and steady too. (Of course, the 5V transformer should be connected to the mains supply for this procedure to work.)
Chris
Edits: 12/05/16
I will see if I can do this.
Is there any reason a fluke RMS meter would be far off. Vinnie has one of these and now with the combined issues on the scope would RMS meter be less quess work?
In regards to the sine square wave combination would that be better or worst than a pure sine or square wave for filament supplies?
This is on my to do list now. Would be nice to get rid of 10,000 to 20,000 electro caps on current filament supplies.
Thanks Tom
Someone posted that the the fluke gets flaky on voltage past a frequency of 1k. Apparently that is all it is designed for.
I found on my Fluke 87 that if I use a certain range, its pretty close.
Automatic range will bounce back all sorts of voltages depending on which range scale the reading falls on. I can back check this and tell you which scale. You'll see it right away if you try it, results will be around 12v on one scale and jump 20 volts on another.
Whats weird, is the the freq (Hz) on the 87 will give the correct frequency.
Edits: 12/05/16
thanks for the suggestion. I will have to give that a try.
Good catch Vinnie!!! Looked up my 87 specs and it is not happy from 5k to 20k. Keep us posted this is what DIY is all about.
Thanks Tom
No need to post the link to the switcher. I searched eBay, and there's only one listing for a 120W lighting transformer. I'm guessing only the high wattage units have toroid outputs. The smaller ones can use conventional transformers without regard for their reduced efficiency, plus they're cheaper.
I would still appreciate it if you could confirm the frequency when you have time.
--------------------------
Buy Chinese. Bury freedom.
I have the 60 and 80w units
The 60W lighting transformers I've purchased do not have toroidal output transformers.
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/05/16
So how would you modify them? No wraps of wire to unwind.
I don't modify the output transformers. Early models that I purchased when I was designing my SETs would output a 12V square wave. I reduce it to 10V with a resistor on each side of the filament. The resistors provide an easy start for both the lighting transformer and the tubes. Newer models of these are different for some reason. The output is higher than 12V, and the wave shape isn't so square. It's not clear to me at this time whether I can still use them. That's why I'm looking into the standard DC-output switchers as a source of high-frequency AC. They're regulated, they're adjustable, they're filtered (the DC for the oscillator), and they operate at higher frequencies right out of the box. There's also very little difference in price between the two types, and the DC supplies can be used in the case they're designed for (no need to add the large external filter capacitor). All things considered, I think this approach holds more promise going forward.
--------------------------
Buy Chinese. Bury freedom.
I think from what I heard this morning listening to the halo's I am inclined in that direction myself. I will know for sure after I build the left channel with switcher units today and do some side by side listening.
Vinnie, I want to stress that the listening tests are for naught if the switching frequency is only around 32 kHz. That can mix with audio signals or their harmonics and create audible intermodulation products.
--------------------------
Buy Chinese. Bury freedom.
I am waiting to hear back from "deaf" about a method for kicking that up to about 65 Khz, will that be high enough? I plan to give that a try and see what happens. Actually, I just got the left channel working with all DC about an hour ago and have been doing some more listening and the differences don't seem as great when I can switch back and forth between them in a few seconds with the volume knobs. I can't quite decide if the "edginess" in the Halo channel might not just be a cleaner sound. Obviously I have a lot of tweaking and listening to do before I ever decide which way to go. I also just discovered that the 26 in the halo channel is still on dc, so I have to switch that over to ac for a true comparison. Still running a hybrid at the moment.
I thought I had. How do I confirm the frequency?
Sorry, I was just asking you to check it one more time for me.
--------------------------
Buy Chinese. Bury freedom.
I am still unclear on how I am supposed to check it. All I did was check the sweep speed. What is it exactly that I need to do to check the frequency?
Vinnie, I'm not asking you to do anything different. Just to do it again and confirm that you get the same result. No matter what the level of experience, people make mistakes. It's always good to confirm things like this, especially if initial results don't seem to fit the norm.
--------------------------
Buy Chinese. Bury freedom.
TK, WHAT is it you want me to do again? I do not know how to check the frequency is what I have been trying to tell you....... If I have already done it I don't know what it was I did. Please spell it out for me what you want me to do.
Edits: 12/05/16
Ahh, miscommunication, sorry!! I was just asking you to put the trace on your scope again and note the sweep speed (1uS, 2uS, 10mS, or whatever it happens to be) and note the number of divisions for a full wavelength. That would be the distance between two consecutive rising edges, or between two consecutive falling edges.
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/05/16
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: