auto radio solid state vibrator for vintage tube radio

[PM3295]

I believe that the old tube radio transformers required the drive to have some dead-time for maximum efficiency. It was related to the time needed for the vibrator contacts to switch over. Maybe we can limit the max duty as to have some minimum dead-time.

I found some servicing notes (attachment), about the shape of the switching waveforms.

[sphinx]

sounds good to me, tubes are not exactly my area. logic stuff is what ilike

[sphinx]

this is now in last stages of designing not much wil be changed after this documantation/info is added to schematic

there might be a resistor added to limit the duty cycle. the circuit can be used as is if one knows that 100% might not work properly

i also modified the ground trace to be on bottom for better current handling

i dont know what more i could add

if you have followd this thread you know what this is supposed to work as there could be other ways to use this but i guess the imagination is the limit

pdf's and gerbers are included

latest changes are that now it is a sngle layer boards since it was only one trace that was on top layer now it is possible to make this even at home.

this is also a 12 volt only design since ic 8volts and up

/sphinx

[sphinx]

this is sort the last big edit on the schematics and pcb

nothing i can really think of to modify or any other suggestions to the design

on the schematic i added a resistor (R12) that could be used to set limit of duty sycle if wanted other wise shorted
info is in the schematic sheet

on the pcb i moved around compoents to make the ground trace more compact short efficient, i figured out the word it is optimize
previously it was long and all over the board

thanx PM3295 for for this new design and additive suggestions for extra features

 i made this a single layer board, pdf included and so are gerber files

do as you wish with it

/sphinx

[Sideshow Bob]

Would it be an idea to enable the internal soft-start. To do this you only need a 5uF cap connected to pin 8

[sphinx]

i will have a look at it i havent got a cap with a smallish footprint with that value i guess a 4.7 would work too.

i will have to do some searching for one

hopefully wont be much trouble since it is only connected ground

[Sideshow Bob]

i will have a look at it i havent got a cap with a smallish footprint with that value i guess a 4.7 would work too.

i will have to do some searching for one

hopefully wont be much trouble since it is only connected ground

I would think any 5uF ish cap value will do. A larger value like 6.8 or 10 uF will most probably create a slightly longer soft-start period but nothing more. The track to ground can also be quite long. Or if it is more easy use a low ohm resistor(like one ohm) as a jumper

[sphinx]

a 4.7uf cap has been now been added for soft-start support which can be omitted as well is this function is not wanted

info is in the schematics sheet as well and so are pdf and gerbers files

do as you wish with it and modify it to your likings

i was sorta lucky and it could be placed with a short track to ground not that it was necessary but the space was sorta easy to make for it

all the extra features can be disable/(not used) info is provided in the schamtics sheet and pdf make it as simple/advanced as you want it to be.

/sphinx

[PM3295]

It appears that the switching frequencies used are 100 Hz or 115 Hz. The optimum drive needs about 1ms of dead time. There is also a resonating capacitor over the primary winding to improve efficiency. This YT video gives a good indication what drive waveform is needed.
https://www.youtube.com/watch?v=7EWTa1jQmng&t=1s

[sphinx]

interesting video i saw all of it from start to end

this guy is using a MCU for the control and his design is made for 2 6 12 24 and 48 volts

quite nice design

[PM3295]

The MCU is an easy and cheap solution, as long as you don't need to make the duty cycle adjustable. It also serves to protect their product design from being blatantly copied, cloned and sold on ebay etc. This may be a problem if it breaks down, and you can't replace the processor if blown, and need to buy another complete unit. At least with the switching-regulator type designs, most people will be able to repair it.

I think the 3525 circuit can be modified to work down to 6 V (overriding the UV cutout) without much trouble. You will need about 9 V at 14 mA on the Vin pin, which can be generated by a charge pump circuit, synchronized to the main clock (200 - 230 Hz). The Vc pin can be tied to the 6 V supply rail. This will reduce  inter-mods  you may have when running the pump asynchronous at higher frequencies, with harmonics showing up on the longer wave bands.  The main internal clock still runs happily at 6 V (see scope pic), and only the internal output drive is locked out. You most probably have to use logic-level gate threshold MOSFETs for proper switching.

Included, the schematic and prj file in Proteus. The circuit does not simulate correctly since the 3525 model shuts down completely under 8 V supply on Vin. Also the OSCOUT pulse is much wider than with the actual hardware, preventing the 555 from working correctly. I added a 20 uS wide pulse to represent the expected pulse input. At least you can get it running and play with it. The free running 555 should be set slightly slower than the sync frequency.  

[sphinx]

i was wondering about adding the 555 circuit to the design

i got 555's but not 3525's so i can't buid it on breadboard and try it, i will try to order some in near future and add some modfets as well to have try with

youre right about the mcu solution it is all about not making it simple to copy design but i wouldn't think it is that hard to write software for it.

[PM3295]

The SW would be easy, but not every user will be capable of doing their own code. Some older hobbyists avoid coding or learning to do so.

[PM3295]

The 555 part seems to work well. I made one modification by returning the discharge resistor to the discharge pin, and not the output. Switching seems more stable this way. I only had 1N914 diodes at hand, so the output voltage (8.6 Vavg) with appropriate diodes should increase at least another 500 mV. I loaded the output with 14 mA. First pic shows the operation with sync off, and then with 20 us wide pulse.

[sphinx]

this a version of the "vibrator" with a charge pump added to board

this is still a work in progress trying to make board a bit smaller if possible

sadly not single layer at this moment

p.s. perhaps time to go smd what you guys/gals say size of the board is getting bigger ?

[PM3295]

It looks good! As long as it can  fit into the old vibrator standard enclosures, it should be ok. I have seen some-odd  ones that are smaller, but that is not common. I will try and breadboard one fully to verify all works as it should.

Maybe we should limit the frequency of operation to around 100 -115 Hz (200 - 230 Hz) for the clock, as the charge pump will not sync correctly over a very wide frequency range. As it is intended for tube radios, it should only be required to work near the two (200 and 230 Hz) standard frequencies. It may be an idea to have a selection jumper/pads.

[sphinx]

thru  hole size is 1400x3100 mils, i thought it was getting a bit biggish

the whole design is sort of modular if a function is not wanted compoents can be deleted or jumpers placed
info is in the text

[PM3295]

I was thinking about three selection pads to pick operation between the two frequencies. The MCU board design may have something similar.

[sphinx]

ah ok thats understandable

i am on the way of tryng to shrink the board with smd's

my thought was to leave the mosfets as thru-hole since they a bit easier to replace than smd and most common to break is my guess

i was wonder if using tantalum caps would be ok to use on this board, since i am not sure about temps inside a radio or in the place where this board would be ?

[PM3295]

Ok, I managed to string together the two circuit blocks. It appears to be working as expected, with good startup and sync. I have to experiment with the values to determine the set limits for guaranteed sync action.
From the pics, you can see that the main clock starts running at around 5.2 V, and the output enables when the charge voltage reaches the 8 V threshold.

Tantalum's may be necessary to shrink the design.  

[sphinx]

ok sounds good i will put pcb'ing on ice so i can work with a complete circuit instead of modifying for every change

i will make this version smd since pcb is getting a bit big for my taste and i think it wouldnt fit in a vibrator can
i will have the mosfets thru-hole for easier reapir

[PM3295]

Here is the updated diagram with values that worked well for me.

1) I changed R14 so that the charge pump will only start once the sync pulse comes from the 3525 (see pic).
2) I found out that the OSCOUT pin has very limited current drive capability and caused strange pulse doubling of the output when loaded too much. The solution was to use a small darlington to prevent excessive loading of this pin. (MMBTA13,MMBTA14)
3) Adding D3 helps the charge pump to come up more quickly, and if the pump circuit is not needed, all the pump components can be left out.
4) R1(15k) now gives good sync over the stated frequency range.

It will be good if you can perhaps get one going as well so we can see if you get similar results.  

[sphinx]

this is a prerelease of the smd board with the latest changes

i feel like the routing and placement is a bit of mess not really satisfied with it.

i am trying to make the board small so it could be easy to fit (2000x1200 mil)

i need to have a look at it for a bit to "see" what changes i can make to make it cleaner
some traces need to be thicker as well but that will come later when i figure out a better placement and routing

[PM3295]

Maybe move C8 on the diagram, so it will stay in for filtering on Vin if the pump is not needed or populated. You already have it close on the pcb layout.

[sphinx]

i moved the cap on schematic as you can see i also did shuffling around with some compoents and moved C8 a tiny bit closer to IC
not much work with that

i moved some other compoents and modified some traces as well.

i take a day off to clear my head and come back with a freshish mind.

the top layer looks prety ok but the bottom layer not so much i will see if i can clear some of it up.

some power traces needs to get wider  but that comes after the other stuff is beautyid up.