|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.22
The discussion about these transformers is slowly getting buried, so I decided to open a new thread. Here are the basic steps I use to rework the oscillator transformer to produce a higher frequency. I want to stress again that this area of the circuit is not isolated whatsoever from the AC mains. Under no circumstances should it be touched or contacted while power is applied!Here's the 60W lighting transformer as it comes from the factory:
Looking underneath, the oscillator transformer is connected to both power transistors. One leg is in series with a resistor. I pull the transformer from the PCB using solder wick to remove solder from the pins and board. These boards aren't through-plated, so once the solder is removed from the backside, the transformer should wiggle free.
Here's a top view of the transformer surrounded by a schematic of the windings. Note that a toroid is considered to have one complete turn by simply passing a wire through the hole. The wire does not need to come back around or do anything else. Looking at the photo, the oscillator coils are on the left and right. Each has four turns. The single-turn coil at the bottom is a feedback winding.
Viewed from underneath, it can be seen how each wire is wrapped around and soldered to its post.
To remove turns from the transformer, first snip the two main windings above the base.
Next, carefully heat each attachment point and slide the coil off its pin. Exercise care here not to overheat the pin and soften the plastic base. This job is easiest if you have a rubber-jawed bench vise that can hold the transformer while you work, but it can be done without that if care is used.
When you're done, it should look like this...
Now unwrap all but two turns of the two main windings. I use tweezers for this.
The next step is to tin the ends of the wires. This requires a very hot iron, such as the 100W Weller, or a gun. I use Weller 100/140W on the high setting. You might need to puddle a little solder in the crook of the tip and immerse the wire in it. It takes 10-20 seconds for the solder to break through the enamel coating. Once it does, you'll be able to tin the lead quickly by sliding it through the puddle of molten solder.
Now wrap each lead around its pin one or two turns and snip off the extra length. Remove any excess solder by quickly touching the joints with solder wick. Again, be careful not to overheat the pins.
This is the transformer after being modified and reinstalled on the board. Note that in the transformer pictured above, I brought the wires down along the sides of the base. The transformer below has them positioned down the front. Either method will work fine.
I did try this with only one turn on each main winding, but the oscillator wouldn't run. I suspect a single turn doesn't create enough flux in the core to saturate the transformer. With two windings as shown, the no-load frequency is about 66 kHz.
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/07/16Follow Ups:
Same method as above just the newer boards you can find on Ebay.
Just search for "Electronic transformer" and find one that looks like the attached picture. You can find them up to about 130watt and 120 or 240 input.
I found that you can remove about 8-9 turns of output transformer to adjust voltage before it gives up.
Also found that 60vac in almost gives you the same result so if you purchase one with 240vac input it should be able to do lower voltage DHT tubes like the 300b, 801, etc. May not work on 2.5v unless you use a dropping resistor.
Edits: 12/07/16
Thanks for posting that! There are apparently at least two versions now of the 60W electronic transformer. I'll order a couple of those so I have them when I need them.
--------------------------
Buy Chinese. Bury freedom.
Last time I checked; the square / rectangular / puled DC are the enemy in any Audio Chain be it SS or Tubes. Why Square / Rectangular / Pulse DC heating your filament which will ultimately gets to the OPT and then can cause unintended Oscillations at much higher frequency and will be feed back into the GND bus SET amps?
.
.
.Thou shall not stand where I type for I carry a bottle of Certified Audiophile Air and a Pure Silver Whip.
This is a differential signal applied across the filament. It doesn't produce the effects you describe.
--------------------------
Buy Chinese. Bury freedom.
How does the square wave heating affect the bias? With DC, you can shift the bias by the filament voltage depending which end of the filament you reference to ground. With AC, you simply take 1/2 the filament voltage and add it to the bias. I would guess you would do the same with the HFac but then making it a square wave makes me scratch my head a bit.
The reason I bring this up is I have a µTracer that operates off of a 20V laptop supply. In order to do the filaments it uses PWM off the 20V supply and alters the duty cycle to give the desired heating. What is interesting about this is that it samples each data point for 1ms and times it to sample when the PWM is in an off cycle. This means that the tube is actually traced with 0v on the filament and no filament gradient whatsoever which is very different that the gradient given from DC or AC sine heating. Moving this to HF square waves makes me see the bias voltage abruptly shifting the full filament voltage and wondering if that is a good thing.
thoughts?
dave
The technique I'm using applies ultrasonic AC to the filament in a manner that is both differential and balanced WRT ground. It does not affect bias. The circuit below is a simplified version. I've only used this with DHT output stages, so I apply bias to the grid. Bias could alternately be applied to the cathode, for instance by inserting a resistor or parallel resistor/cap combination in series with the wiper of the hum pot. The ultrasonic heater supply does not cause bias to vary in any case.
--------------------------
Buy Chinese. Bury freedom.
Let me try to put this in a different manner. What you describe is the AC nulling out to zero but I am talking about the "static" DC bias position.
To take it a bit to the extreme for my illustration lets take an 841 and assume a 500V P-k and -10V of bias for 4ma of current.
Everything is all nice and tidy until we realize that there is a voltage gradient of 7.5V across the tube which means one end of the filament is at -10V and the other end at -17.5V What happens if we suddenly feed the filament with a -7.5V which is done by simply referencing the + side of the filament to ground? Suddenly the tube will draw in the realm of 13ma of current and if we switch to 7.5Vac the bias reference goes to the midpoint of the filament and we get 8ma. What has always interested me about AC heating is that the voltage gradient covers the entire range of + and - DC options plus a factor of 1.4 and at some point has a voltage gradient of 0 across the filament.
When we look at DC square wave heating from a DC bias viewpoint we are establishing an average bias point at 1/2 the heating voltage but is it any different conceptually that switching the bias reference point from the + to the - side of the filament at a given rate of time?
dave
"Suddenly the tube will draw in the realm of 13ma of current and if we switch to 7.5Vac the bias reference goes to the midpoint of the filament and we get 8ma."Maybe we have a miscommunication, but I think you're not grasping the concept of a floating, differential heater supply. The only ground reference is the tap on the hum pot. If that's electrically centered, bias voltage and current are unaffected by the AC heater supply, even when analyzed for instantaneous changes. Put this into SPICE, insert a 0.1 ohm resistor in series with the hum pot wiper connection, and you'll see no HF current flowing to ground. Yes, there's a constantly changing gradient across the filament. However, it's balanced with respect to ground, always sums to zero, and has no effect relative to bias voltage or the common mode signal path.
--------------------------
Buy Chinese. Bury freedom.
Edits: 12/08/16
I see what you are saying, so if you place a similar reference for a DC supply filament polarity will not matter so reversing it at a high rate of speed is not an issue. I was caught up on one end or the other being a reference for DC and not considering what happens if you reference the middle.
dave
In Dave's cathode supply where one leg is tied to ground without a humpot, I can see where that may be an issue. I might experiment with this on a GM70 once I get the 120w units. Attached a picture of typical hum pot circuit.
Edits: 12/08/16
How do you apply a differential signal to the 2 pins filament configuration?
Or do you meant a POT approach to fold the current in the filament?
.
.
.Thou shall not stand where I type for I carry a bottle of Certified Audiophile Air and a Pure Silver Whip.
Thanks for the post TK. I don't think mine is quite the same as yours though, per the attached pics. Mine has no plastic base, just 6 wires going through the board and soldered to it. The red and blue wires are the ones with the coils around the doughnut, and there is a third green wire that just goes through the doughnut without wrapping around it and the two ends of the wire are soldered to the board as well. As I mentioned in my earlier post, I took one end each of the blue loose and took off one wrap of wire off and soldered it back on. The result was that the Khz reading went down by about 50 %. I then took one end of the red wire loose and took a wrap off and it went down a bunch more after I soldered it back on. I don't have a clue what the green wire does. If you look at the pic of the bottom of the board you can see 6 spots of solder in the upper center that hold the 6 wires. If you look close you can see where I removed one end of both the blue and the red. I did not touch the green.
Got any idea about how I should approach this one? Why did the Khz go down with one loop of red and one loop of blue off? Thanks! Nice post!
Edits: 12/07/16 12/07/16
It sounds as though you did the right thing. How are you measuring frequency? We never did reconcile that issue regarding the scope trace.
--------------------------
Buy Chinese. Bury freedom.
Well, my fluke read it as 32.5Khz without modifications, and that is what deaf said he got on his and it is the same transformer I have according to him. So I have been using that as the scope was giving me readings that were over 120Khz, and I did not think that was likely. I do not understand why the readings would go down when I took off windings though.
Edits: 12/07/16 12/07/16
Triode, what if he reversed the winding direction? Does it matter?
Reversing either winding driving the transistors should prevent oscillation. Reversing the winding that's in series with the output transformer will prevent the oscillator from self-starting. Incidentally, I previously referred to that as a feedback winding, but I believe it's actually the primary of T1.
--------------------------
Buy Chinese. Bury freedom.
I am pretty sure I put it back just like it was because I only took one loop off each wire. Also, when I checked it after putting it back together it was reading about 30 volts. However, since we know yours is working properly, I will make mine look just like yours tomorrow.
Edits: 12/07/16 12/07/16 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: