|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.22
Here's the schematic of the lighting transformer (basis for ultrasonic filament supply) discussed in the threads below. This is the 60W version, but I'm sure they're all similar. Component designations are those on the board, except for T1 and T2, which are unmarked. I also located an online writeup of this type switcher at the link below. I can't vouch for the accuracy of the entire theory discussion, but it looks interesting.
--------------------------
Buy Chinese. Bury freedom.
Follow Ups:
Near as I can tell the unit that gives you AC as an end product is the electronic transformer and the unit that gives you DC is the switching unit. That is the one I have now. The attached pic is of that unit and I am wondering if anyone can see from the photo what mods I might need to make to use it as a noise free dc filament supply for an 813. I have ordered two of the 120w halogen electronic transformers (AC output), but don't expect to get them for awhile, so I thought I would see what I might be able to do to get this one dead quiet and sounding good. All suggestions welcome. Thanks.
Edits: 11/20/16
Vinnie, the photograph you took shows a 16-pin IC above the oscillator transformer. Can you tell me if there's a part number on it?
--------------------------
Buy Chinese. Bury freedom.
There are two numbers on that part that I can see ..... one is KA7500B and the other is HYC 576 H.
Thanks Vinnie! The KA7500B is the SMPS controller for that supply. I'm thinking now there might be a better way to do all this. Here's what I have in mind...
The lighting transformers are essentially high power, square wave oscillators with a step-down transformer. They're small and cheap, and they were my first effort at ultrasonic heating nearly 15 years ago. However, there are two drawbacks. First, they oscillate at a fairly low frequency. The frequency can be raised, but not as high as I would like. The limit is created by the oscillator transformer core itself, which saturates during oscillation. I've tried other cores, but haven't been able to push the frequency above 65 kHz or so. Worse, the frequency is load sensitive, and the newer models I bought want to run at 45-50 kHz when loaded near their rated limit of 5 amps. The second downside to these is that they're not regulated. The square wave amplitude varies as a function of AC mains voltage.
The DC switchers you bought are somewhat similar, but they add a rectifier and filter at the output. In addition, because they utilize a separate controller, there's no oscillator transformer to limit the frequency. The controllers I've seen (including the KA7500B) have one or more pins for setting oscillator frequency with a resistor and capacitor. The datasheet for the KA7500B states a 300 kHz upper limit, and while I think that might not be practical from the standpoint of efficiency, 150 kHz is probably a reasonable limit. That would be a huge improvement over 65 kHz! The controller also provides good voltage regulation, usually accomplished with an opto-isolator driven from the DC output.
So, with all this in mind, I'm thinking it might be better to use the DC supplies for ultrasonic heating. It should be a simple matter to cut the PCB traces at the (+) and (-) output terminals, and to rewire those terminals directly to the secondary of the internal transformer. That would provide a regulated, high frequency square wave for the filament. The DC circuitry could simply be left unused, in parallel with the transformer secondary, in order to provide regulation feedback to the controller.
If I get time tonight, I'll dissect one of the little 3A switchers I bought to see if they're similar. This could open up a whole new method for ultrasonic heating!
--------------------------
Buy Chinese. Bury freedom.
Well, it turns out that the "3A switcher" is actually only a 1.3A switcher. I remember now that I bought these just to power a few 12SL7s and the like. It's similar in operation to the large unit you purchased, but with less sophistication. Primarily, the oscillator is discrete, so there's no controller IC. On the other hand, it oscillates like yours using a RC time constant (no oscillator transformer). In theory, it can be easily modified to run at higher frequencies. I also found the frequency doesn't change as a function of load, and that the circuit uses an opto-isolator to regulate the output voltage. Unlike yours, it has no output filter at all, and the input filter is not overly effective. So, it spews a lot of high frequency garbage out both ends.
Here's the switcher next to a thumb drive for scale:
The internal view indicates only two inductors - the common mode input choke on the left, and the output transformer near the center. The opto-isolator is just to the upper-right of the output transformer, and the transistor on the rear wall is used to chop the transformer primary.
Here's the output with a 1.2A load (10 ohms). The main wave is 66 kHz at roughly 10 mV. Notice how much dirtier this is than the larger units you bought with their nice, built-in output filters.
This is a delayed sweep view of the switching spikes in the photo above. I'm pretty sure this is the output transformer ringing. This energy is 25 MHz at about 40 mV. When the switcher is running, this junk floats onto everything, including the line cord.
--------------------------
Buy Chinese. Bury freedom.
Well, I wish I had waited to read this before I ordered two of them to use on the filaments of the 26's. Oh, well we'll see what happens. Maybe I can tack one of my filters on it and clean it up a bit. I wonder at what rating they put the other goodies back in?
Edits: 11/22/16
As I explained before, your filter won't clean it up. There's some improvement due to the additional shunt capacitance, but a common mode choke does virtually nothing in this situation. All this differential output noise will just pass right through.
--------------------------
Buy Chinese. Bury freedom.
I understand what you are saying, but it sounds better with my filter than without, and until I hear something I don't like I think I will settle for that. I can always do more later.
OK, I just wanted to be sure I was getting the point across. Maybe tonight I'll go into the small switcher and see how high I can raise the frequency. I have no idea what the response of the various components might be. The limiting factor is probably the output transformer.
--------------------------
Buy Chinese. Bury freedom.
Well, the joke's on me. The "transistor" on the back wall is actually a 5-leg PWM controller. There are three versions of the part; "L," "M," and "H." Mine is "M," which means (what else?) the frequency is fixed at 66 kHz. I could buy the 100 kHz "H" version (actually a KA5H0265R) on eBay from Hong Kong or Malaysia, but that doubles the cost of this little switcher. Guess I'll just have to resign it to my original intended use of DC for my 12SL7s. Proving that higher frequency ultrasonic heating is feasible will have to wait for another day.
--------------------------
Buy Chinese. Bury freedom.
I will be really curious to see how the 3 amp model works on the 26. Mine are due to get here today or tomorrow. Thanks for the post.
As with the 813 supply, the primary concerns will be that neither output terminal is grounded and the output voltage is reasonably clean. Maybe when you get them, you could open the case and tell me which controller they use? I'd like to order the 5A versions for some experiments, but it's a crapshoot regarding the ability to modify the frequency. Too bad they don't include the controller type in the auction descriptions.
--------------------------
Buy Chinese. Bury freedom.
It didn't make it in today's mail, but I am hopeful for tomorrow. I will let you know when they get here.
That all sounds over my head TK. Have fun figuring it out and let us know what you find.
Have you done any of the things that have been suggested regarding the use of these switchers?
The only thing of significance not labeled in the photo below is the output rectifier. I just can't be sure where it is, might be the device clamped to the wall.
--------------------------
Buy Chinese. Bury freedom.
I am pretty sure the negative input is not grounded, but i don't think I am going to be able to see anything else the way this thing is put together. Not much room between the board and the bottom. Since it appears there is now not much difference between the two channels sound wise now, I am going to assume for the time being that the switchers are ok and try to get the 26 and the preamp squared away and then listen again. I have attached two scope shots. The first was taken at the outputs of the switcher with a 1x probe, volts/div 50m, sec/div 10us. The second shot was taken with at the outputs of my filter, volts/div 20m, sec/div 10us, 1x probe.
Edits: 11/20/16 11/20/16 11/20/16
It would be very rare to find an OEM power supply with either the negative or positive grounded. The reason is the manufacture has no idea of how the power supply is being used. You may want a positive ground.
Consider a bipolar +12,G,-12 power supply made from two identical units. One unit would have it's negative grounded and the other it's positive grounded.
The filament that the supply is going to power will be grounded through a resistive CT (or hum pot) and cathode resistor.One wouldn't want the supply to be otherwise grounded.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 11/21/16
"I am pretty sure the negative input is not grounded..."
Do you mean the negative output? All you need is an ohmmeter to see if the 12V negative output is connected to the case.
"...but i don't think I am going to be able to see anything else the way this thing is put together. Not much room between the board and the bottom."
Please don't try to make live measurements anywhere inside this unit. Most of it is tied directly to the AC mains. This is an electrolution hazard.
"I have attached two scope shots. The first was taken at the outputs of the switcher with a 1x probe, volts/div 50m, sec/div 10us."
The wave there is 50kHz, and amplitude 50mV. Just to be sure that's all there is, you should reduce the sweep speed to see if there's any 60/120Hz energy also. The residual 50kHz can be mostly filtered out, but you should wait to see if you're actually going to use this supply.
"The second shot was taken with at the outputs of my filter, volts/div 20m, sec/div 10us, 1x probe."
What filter is this? That's a 25kHz wave.
--------------------------
Buy Chinese. Bury freedom.
I just checked and there is no continuity between the negative output and the case, so it must not be connected. Don't worry, I won't be messing around in there when it's hot.
The filter is my home made one I was using with the 10 volt trans. It hs a .2 ohm resistor between the + terminals of two 10k caps and the - terminals tied together, and has a dual winding common mode choke with positive going to one winding and negative to the other. The outputs of the windings go to the + and - of the filament.
Does it matter if it drops the frequency to 25kHz in this application?. It's not my design, I borrowed it from a design for an 845 filament supply that someone had posted on the asylum several years ago.
PS Just remembered the output is adjustable.... boy it is ever nice to be able to adjust the voltage to the nearest 1/100 volt by turning a screw. Now set at 10.00 vdc, if my fluke isn't lying to me.
Edits: 11/21/16 11/21/16 11/21/16
I'm not sure how 25kHz is being created from what appears to be a 50kHz triangle. Maybe the filter is ringing. The second trace does look more like a sine wave.That aside, you're not using the right type of filter for this. The 50kHz wave at the output of the switcher isn't common mode. It's differential between the (+) and (-) terminals. In fact, the power supply already has two devices rejecting high frequency common mode energy between the AC mains and the 12V output. One is the input noise filter, the other is the output transformer. The latter isolates common mode audio signals at the filament from the active circuitry in the supply, and it precludes the need for anything else in that regard.
To reduce the remaining 50kHz energy, you'll need to add two chokes, one in series with each output terminal. The chokes should have a value of about 50uH and be rated for 5A or more (813 filament). The (+) and (-) voltage at the output side of the chokes should be shunted with a 220uF cap. This design will attenuate the 50kHz energy about 60dB, or 1/1000. Based on your measurement, the result should be about 50uV of HF noise.
Incidentally, did you have the supply loaded when you measured the 50kHz energy at the output? That's a necessity in order to obtain an accurate reading. Output noise will be much less when the supply is loaded, due to attenuation by the chokes in the internal filter.
--------------------------
Buy Chinese. Bury freedom.
Edits: 11/21/16
Ok, before we go any further I need to bring you up to date on what I got done this morning. I would have posted it sooner, but my honey do list got enforced priority about mid morning. : )
Anyway, I was working at tweaking things a bit and noticed that the hum pots were no longer doing anything when I had the meter on the wings of the pot and set to mV. I got the lowest reading when I had it cranked hard to port or starboard, and I was still getting about 25 mV. Not good. So I decided to try taking the pots out of from 813's and replacing them with a couple of 50 ohm resistors tied to the filament pins and each other with a wire going to the cathode cap and resistor. Well how 'bout that sports fans, the ripple dropped to 1.7mV at the filament pins! So then I change the filter on the 26 to match the filters on the 813's, and danged if that ripple doesn't drop to 1.7 mV too. Haven't removed the hum pot on the 26 yet but will try that next. I was feeling pretty cocky by now, so I checked the ripple at the speaker terminals and get slapped in the back of the head. It was 105 mV. Phooey says I, and decide to put a shorting plug in the amp input, and when I check it this time at the terminals I get 0.5mV! Couldn't hardly believe it. I double checked everything to make sure it was correct, and it was. So that means I have a problem with my preamp, but I will work on that later. Just to be sure, I turned the volume up and put my ear right next to the speaker with the shorting plug still in and it was dead quiet.
So to get back to the rest of your post, I am not sure why, but my filter seems to be working pretty well, at least as far as hum goes. What would be the symptoms of the 50kHz if it is still there? Right now it sounds as if all my troubles are gone with this right channel and it sounds good to boot.
Edits: 11/21/16 11/21/16 11/21/16 11/21/16
Great, sounds like you've made a lot of progress!
"What would be the symptoms of the 50kHz if it is still there?"
You'll have a 50 kHz wave at the speaker terminals. Check for this with your scope. It can be nulled with the hum pot, just like 60 Hz.
--------------------------
Buy Chinese. Bury freedom.
Where would the hum pot need to be placed?
Wire it just like you normally would. The only thing different here is that the AC isn't 60 Hz. On this point, you may not be getting any appreciable high frequency "bleedthrough" at the secondary of the ampifier's OPT. Take a look with your scope - if it's only a few millivolts, I'd let it go for now.
--------------------------
Buy Chinese. Bury freedom.
But then why is the speaker dead quiet as it is now?
Because you can't hear 50 kHz. It might mix with something else though, so it's important to attenuate it as much as possible.
--------------------------
Buy Chinese. Bury freedom.
Hate to dog this, but would the mix be audible, and if so in what way? If not, I am not sure why we need to try and get rid of it. Right now I don't hear anything I think I need to get rid of.
Edits: 11/21/16
The potential issue is IMD. When two frequencies mix in a not entirely linear device (like a tube), sum and difference frequencies are created. Those frequencies can mix again in never ending combinations with harmonics of the original frequencies. These spurious frequencies are known as intermodulation distortion products. Their audibility can vary considerably with different source materials, speakers, specific output tube, etc. Regardless of whether you hear it now, it's good practice to minimize the potential for this effect.
--------------------------
Buy Chinese. Bury freedom.
;
I remember contentious discussions on this exact topic years ago having to do with the ultrasonic response of piezo tweeters. I'm hoping for several reasons that no one is still using those. :)
--------------------------
Buy Chinese. Bury freedom.
Very interesting link. Thanks for the info.
ray
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: