• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Strip board project completed.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've got some miniature tubes that will run on 90V, it would be nice to make an efficient switcher that could power them.

I may try winding a lower voltage switcher this time.

I've got Mylar tape, Kapton Tape, and believe masking tape can be used as margin tape if one can cut it narrow enough. It does not take any voltage stress and is only there for mechanical spacing.

I'll probably get a dozen of these transformers and cook half for cores and break the cores out of the remaining six for plastic forms. I think the original design was for a 200KHz switcher, so I should be able to design a Royer to run at that frequency.

When I break one open for the plastic form, I'll slowly dissect it to see what the construction is like. Although they are varnish impregnated so it will be a PIA.
 
Disabled Account
Joined 2010
I've got some miniature tubes that will run on 90V, it would be nice to make an efficient switcher that could power them.

I may try winding a lower voltage switcher this time.

I've got Mylar tape, Kapton Tape, and believe masking tape can be used as margin tape if one can cut it narrow enough. It does not take any voltage stress and is only there for mechanical spacing.

I'll probably get a dozen of these transformers and cook half for cores and break the cores out of the remaining six for plastic forms. I think the original design was for a 200KHz switcher, so I should be able to design a Royer to run at that frequency.

When I break one open for the plastic form, I'll slowly dissect it to see what the construction is like. Although they are varnish impregnated so it will be a PIA.

I think the trick is to make a higher voltage than needed and regulate to chop the ripple off the DC then filter..It will be interesting to see what you find.. :)

Regards
M. Gregg
 
Ecpos B82731M2182A30

Strip off the windings but leave one attached at one end. Use it to wind the primary of 8T, twist a loop as Center tap, and wind 8T more. Connect the end to the other end posts.

Wind a secondary on the remaining two form sections. I used about 30ft of 28AWG wire. I lost count of the number of turns.

A shuttle is needed to wind the secondary.

So far I'm getting 200V at 25mA at 55% eff.

I suspect the primary inductance is too high as the only way to shift the frequency is changing the input inductor. Changing the snubber value has little effect. Adding an additional cap across the primary has little effect.

Transformer inductance values in the schematic are measured values, iirc leakage inductance was 1/100 of primary inductance.
 

Attachments

  • Epcos 001.jpg
    Epcos 001.jpg
    112.6 KB · Views: 149
  • Epcos 002.jpg
    Epcos 002.jpg
    240.3 KB · Views: 137
  • Epcos Common Mode Choke.JPG
    Epcos Common Mode Choke.JPG
    8.3 KB · Views: 140
  • Royer.JPG
    Royer.JPG
    80.2 KB · Views: 147
Disabled Account
Joined 2010
Ecpos B82731M2182A30

Strip off the windings but leave one attached at one end. Use it to wind the primary of 8T, twist a loop as Center tap, and wind 8T more. Connect the end to the other end posts.

Wind a secondary on the remaining two form sections. I used about 30ft of 28AWG wire. I lost count of the number of turns.

A shuttle is needed to wind the secondary.

So far I'm getting 200V at 25mA at 55% eff.

I suspect the primary inductance is too high as the only way to shift the frequency is changing the input inductor. Changing the snubber value has little effect. Adding an additional cap across the primary has little effect.

Transformer inductance values in the schematic are measured values, iirc leakage inductance was 1/100 of primary inductance.

Looks good,

Do you get any "singing" (High pitch whistle).. I guess it should be OK with each winding having a different former...I found the way I got best results was to wind and count the turns onto a former of equal size then use some type of very low speed drive to wind the wire onto the bead and guide it. I did have one test piece "sing" so I potted it..guess what it made no difference..if it sings rewind potting does not stop it (I was surprised I thought it would..probably because no vacuum impregnation)..:)

Looks like your having fun..efficiency looks good!
Have you tried UF diodes?


Regards
M. Gregg
 
Last edited:
No singing, although I probably couldn't hear it at 18KHz any way. The dogs didn't notice so I suspect it is not singing.

I've got some UF4007s I can try, for now I am concentrating on the waveforms. Snubbers tame the ringing on the leading edge.

The overshoot without snubbers is 100% the switched waveform height, and ringing proceeds for over 50% of the pulse width. With snubbers it is a slight overshoot (maybe 1V) which is not quite critically dampened.

I was hoping to get 85% eff, but it looks like it will take a lot of work to get there.

I rewound the primary using 5 turns bifilar wound and measured the transformer this morning at work.

Measured values are now:
Lp1.......=142mH.........Rp1........=0.018R
Lp2.......=142mH.........Rp2........=0.018R
LP1-2....=615mH.........Rp1-2....=0.036R
Ls.........=99mH...........Rs.........=1.44R
Cp-s......=0.006nF
Leakage inductance = 0.0036mH

I'll try it tonight.
 
Last edited:
Pretty much the same results. 50-55% eff depending on output voltage.

I switched to UF4007s and the output went up but eff remained the same.

I tried snubbers on the 1N4007 and was able to tune the overshoot to critically dampened, but never could get rid of it.

Interesting exercise, but I think I've about run my course.

It is running at 26KHz so I managed to move it out of the audio band by reducing primary inductance although even out of the audio band artifacts can still be an issue.

The power supply works best if tuned for power out, so the best approach would probably be to follow it with a simple series pass regulator with a shunt regulator to keep the load constant.
 
Disabled Account
Joined 2010
Pretty much the same results. 50-55% eff depending on output voltage.

I switched to UF4007s and the output went up but eff remained the same.

I tried snubbers on the 1N4007 and was able to tune the overshoot to critically dampened, but never could get rid of it.

Interesting exercise, but I think I've about run my course.

It is running at 26KHz so I managed to move it out of the audio band by reducing primary inductance although even out of the audio band artifacts can still be an issue.

The power supply works best if tuned for power out, so the best approach would probably be to follow it with a simple series pass regulator with a shunt regulator to keep the load constant.

Are you going to try it driving a tube stage?
(Which is probably the most interesting)

Or calling it a day? :)
And never know the results... :)

Regards
M. Gregg
 
I don't know, guess I must say something..LOL

If your circuit and layout is correct then you may have some cap coupling because of layout. Possible problems could be no input from the 4093. ie no switching.
To check this make a simple probe with a 1K resistor and a LED test for input to pin 14 of 4017.

Other cause from coupling can be the balance set by the 18K an the 0.1 cap on the reset out of balance. So to test this reduce the 18K to 10K and test. if the value gets to low the auto reset will not work. If its to high the counter will remain in reset and will not count.

The only other thing could be the enable which should be at Gnd.
I will watch to see if you are OK. I can't leave you to struggle.

Regards
M. Gregg

Ok, I have tried all of the above and it still won't switch between the inputs!!! I did get a small result when I shorted the header pins on the main board rather than using the push switch, but only between the first and second input. Could this be caused by the switch and led PCB being a 200mm cable legnth away from the main PCB? I have also used BC639 rather than the spec'd BC107b as I had these to hand, seem close to me?

I'm using a Philips Type FX2240 pot core + bobbin on my inverter and am getting a small amount of "singing". I was going to pot it so thanks for the heads up.

Cheers
Dagwood
 
Dagwood,

I think your builds look great, the first link you used did not work but the second shows your Ipod dock..Nice

Regards
M. Gregg


Thanks, this was my first project, even the timber frame was shaped and cut by hand. It looks even better when it's powered up with blue led's under the sockets and the front switch has a blue ring when on.

Will add pic to album.

Cheers
 
Disabled Account
Joined 2010
Ok, I have tried all of the above and it still won't switch between the inputs!!! I did get a small result when I shorted the header pins on the main board rather than using the push switch, but only between the first and second input. Could this be caused by the switch and led PCB being a 200mm cable legnth away from the main PCB? I have also used BC639 rather than the spec'd BC107b as I had these to hand, seem close to me?

I'm using a Philips Type FX2240 pot core + bobbin on my inverter and am getting a small amount of "singing". I was going to pot it so thanks for the heads up.

Cheers
Dagwood

Try putting a switch straight on the driver...If it works try putting the driver by the switch and drive the cable with the Schmitt Trig. I think you need to get the circuit working as in the breadboard.

Is the first LED on and relay powered..If it is then the transistors should be OK.

You need to create the same situation as in the breadboard..Do you still have the breadboard..the working version?

Have you tried Just the selector with no other circuits powered IE inverter?

Just for interest my switch is about 150mm away and works fine..:)

Sometimes pick up on the power rails can cause a problem so if they are smooth try putting a 0.1 across the power rails right on the chip connections under the PCB. (Small disc Ceramic.)

I'll catch you laters :)

Regards
M. Gregg
 
Last edited:
Disabled Account
Joined 2010
Yeah I think your right, I will re-build it on the breadboard and simulate the cable run

Running the input selector directly from my bench supply at this stage.

Just tried switching from the driver, it worked but only inputs 1 & 2 and occasionally 4 ???

Cheers

It sounds strange,

It could be two faults..The De-bounce should stop it skipping across the count. possibly the length of cable to the switch but I would be surprised if it was. If you have 1-2 and 4 working then it could be the gain of the transistors 1-2-4 being higher than the others..ie what type did you use on the bread board. I had this many years ago with a home made computer I/o Card and it was the transistor gain across the batch being different. try twisting the cables from the switch if it is a distance problem..or switch a transistor before the Schmitt Trigger..(High resistance contacts is a possible but grasping at straws..You will get their :) ) I would try the 0.1 across the chip connections (direct on) for the supply under the board it is common practice but you should not need it in this application. (Keep the leads short)

Just add your working link showing your Dock again... http://s1155.photobucket.com/albums/p551/Dagwood72/


Regards
M. Gregg
 
Last edited:
I had trouble reading the schematic. Is the switch debounce 4.7K and 0.1uF?

If so, increase either the resistor value or the capacitor value.

The time constant of the debounce is 4.7ms. Relays and switches typically switch around 3ms bounce, however it can vary considerably and this may be insufficient.
 
Disabled Account
Joined 2010
I had trouble reading the schematic. Is the switch debounce 4.7K and 0.1uF?

If so, increase either the resistor value or the capacitor value.

The time constant of the debounce is 4.7ms. Relays and switches typically switch around 3ms bounce, however it can vary considerably and this may be insufficient.

4.7K & 0.1..Yes..

I think its the gain of the transistors 1/2 are OK the other two are suspect..LOL

Will be interesting to see... :)

Regards
M. Gregg
 
I tried one more experiment tonight. I wrapped a crude feedback loop around the converter.

Crude, poor regulation but it works. So I can set the output voltage level high enough under expected load conditions to allow for a series pass regulator without dissipating lots of excess power.

I'll come back to this later and wind a smaller core to see if I can get the eff up.

Right now I have other things that need attention.
 

Attachments

  • Royer_W_Feedbacl.JPG
    Royer_W_Feedbacl.JPG
    89.1 KB · Views: 118
Thanks M Gregg, I was supposed to be resting my damaged ankle last night but could not resist going back into the garage and firing up the solder station, lucky because adding the .1uf ceramic cap worked a treat!

I have added a few pics to my photo album http://s1155.photobucket.com/albums/p551/Dagwood72/

Starting to think that a second EM84 might be a good idea, I do like symmetry :)

OK, here's my odd idea
I have a couple of power amps and a couple of headphone amps that I use, I'm going to use a second input selector as a output selector....is that mad??? Seems like a solution to me as it will save me swapping cables all the time. I do understand that this will introduce more noise into the system but plan to be very tidy with the cable runs, your thoughts?

Cheers
Dagwood

P.S
M Gregg, please hang around to see the final result :)
 
Disabled Account
Joined 2010
Thanks M Gregg, I was supposed to be resting my damaged ankle last night but could not resist going back into the garage and firing up the solder station, lucky because adding the .1uf ceramic cap worked a treat!

I have added a few pics to my photo album http://s1155.photobucket.com/albums/p551/Dagwood72/

Starting to think that a second EM84 might be a good idea, I do like symmetry :)

OK, here's my odd idea
I have a couple of power amps and a couple of headphone amps that I use, I'm going to use a second input selector as a output selector....is that mad??? Seems like a solution to me as it will save me swapping cables all the time. I do understand that this will introduce more noise into the system but plan to be very tidy with the cable runs, your thoughts?

Cheers
Dagwood

P.S
M Gregg, please hang around to see the final result :)

The output selector should be OK as long as you use the common contact on the relay as the output so when its not energised the normaly closed is used to Gnd the output so you dont have an open input to an amplifier.. :) (You could use a low ohm value resistor in series with each relay Gnd just to prevent any pop's at switch over) <<it might be OK without.

Thats great news about the decoupling on the power rails with the 0.1 ceramic. :)

I guess you have the EM84 up and running?
It would be a good Idea to get everything connected to the project PSU and get all grounding "good" so you know that you will not get any more coupling problems before assembly.

I will be interested to see the completed project. :)

Just a thought..don't forget if you change speakers over not to leave a power amp with no load on the speaker terminals..

Regards
M. Gregg
 
Last edited:
Thanks M Gregg

I will add a resistor to the new PCB design, can always use a link if not needed.

Yes, very happy to have it up and running! :)

Yes the EM84 is up and running, just need to make another one to cover both channels. Have connected all elements and no hums or pops.

Cheers for the heads up on not leaving an amp with no load :)

One question, earlier you said to have the output connected through a relay that is feed directly from the main power supply ( no caps ) but the problem is my power supply is putting 15.1 v into the inverter and my relay's are 12v (max 13.4) should I run it through a L7812 Votage regulator or through a few diodes to get the voltage drop?? I guess I'm asking which will be quicker to drop the relay out??

Cheers
Dagwood
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.