Audio modulation by BUCK power supply

[Acid Byte]

Hey guys and girls!

Been a while since i've posted.
i've been working on a Halfbridge SSTC powered by a buck converter(Class D type).
The SSTC driver circuit is based on steve wards SSTC5 (and very similar except for some added caps for noise protection)

The buck modulator gets fet with a audio signal wich gets converted to PWM to drive the buck converters gate.

One thing i still have to work on is a better gate drive for the buck converter.
As the fet gets really really warm. (70C+ for runs upto 1 minute on 150vdc busvoltage).

This print is the buck modulator,
it takes bluetooth audio and transforms it into PWM
(now running on a UCC37321 @ 15vdc to drive the gate but this does not seem as enough)


Here you can see the actual BUCK converter.


Bridge and buck



As the secundairy coil also gets to hot for the coil form i 3d printed a mount for a fan so i could cool the secundairy from underneath the coil








This is her running on 150vdc bus voltage before finding out i had to do something about the heat of the buck converter.

/>
The PDF attachments are somewhat the schematic for the device. (there are some recent changes i have not put into there yet.

Things on the list to do for the apparatus are:
#1 get better gatedrive
#2 change the modulator to have active feedback
#3 raise the busvoltage to 320vdc

Let me know what you guys and girls think!
Any help is greatly appreciated!

Best regards,
Jeroen van Dijk

[Mads Barnkob]

Great project, I absolutely love to see new ideas mixed with good old proven designs!

Forced air cooling on the secondary! Yay! How about the primary coil, does it heat up that much as well, it has to be worse around there for sure!

I like the simple audio implementation with the LTC6992, that seems like a great little IC, thanks for bringing that around

What frequency is the buck running at? And what other interesting parameters is there to tell about it? A quick calculations on the required gate drive from a few guessed parameters also show that you are properly going to need some 5-10W of gate drive there, so it heats up from being run almost linear all the time. The 44N80 also has a pretty large gate charge.

[Acid Byte]

Great project, I absolutely love to see new ideas mixed with good old proven designs!

Thank you Mads!
The idea is not completely mine though i have to credit Jerome Miche for sharing some of his stuff hehehe
And yeah why go full blown on ud while that old circuit runs fine for this application (wich also is like your sstc1 right?)

Forced air cooling on the secondary! Yay! How about the primary coil, does it heat up that much as well, it has to be worse around there for sure!

Yeah that secundairy gets HOT haha to have any chance of survival for the coil form it just had to be cooled.
as for the primary this gets hot too. and think i have to go for maybe water cooled copper brake liner. but for the short runs its fine now.

I like the simple audio implementation with the LTC6992, that seems like a great little IC, thanks for bringing that around

The LTC6992 is a pretty ok chip i think for this application not the best.
when scoping the wave form on the bridgecap the sometimes are small overshoots.
going to upgrade the modulator board more similar to the qcw modulator board,
With the active feedback the overshoot will be less i think.

What frequency is the buck running at? And what other interesting parameters is there to tell about it? A quick calculations on the required gate drive from a few guessed parameters also show that you are properly going to need some 5-10W of gate drive there, so it heats up from being run almost linear all the time. The 44N80 also has a pretty large gate charge.

The buck converter is running at only 86khz at the moment.
Expirienced some issues trying to run it higher on the LTC6992 but that could just be not having enough knowledge of it all.

Yesterday i made a new gatedrive board with a FOD3184 supplied by a 12v and a -12 supply. (the gate drive from the qcw modulator board)
This solved the heat issue quite a bit.
Comming back from 70 in a minute to 26.5 in 10 minutes.
so thats a big plus. (however i dont think im there yet?)
Any recommendation on gatedrive circuits?

[Mads Barnkob]

Yes, I think all my SSTCs are really based on the Steve Ward SSTC5, in either half- or full-bridge.

Regarding the break line, I have actually been wanting some for a while, but there never turned anything up at the scrap yard, so forgot about it multiply times, this seems to be the cheapest in Europe, 10£ for 7 meters of 5 mm tubing, thats okay. http://www.ebay.co.uk/itm/25ft-Copper-Brake-Line-Pipe-3-16-OD-x-0-131-ID-5-male-5-female-nut/362133334637

I ordered a few LTC6992 just to get some hands on experience with it.

The estimate I did for gate drive power was only at 70kHz, so at 86kHz it requiers even more. I would use +/-15V, since the maximum rating is +/-20V, which you could also go up to if you really wanted to punish it. But as you lowered the temperature by that much and it can now run so long, I would not invest more time in it, unless the gate waveform suggests that you should adress some issues.

For gate drive I almost exclusively use GDT driven by MOSFET driver ICs and if its brick, a MOSFET output stage like on Steve Wards DRSSTC UD1.3/2.1, it is proven, sturdy and a easy to build design.

[Acid Byte]

Yes, I think all my SSTCs are really based on the Steve Ward SSTC5, in either half- or full-bridge.

I think thats just because its the perfect circuit for the job

Regarding the break line, I have actually been wanting some for a while, but there never turned anything up at the scrap yard, so forgot about it multiply times, this seems to be the cheapest in Europe, 10£ for 7 meters of 5 mm tubing, thats okay. http://www.ebay.co.uk/itm/25ft-Copper-Brake-Line-Pipe-3-16-OD-x-0-131-ID-5-male-5-female-nut/362133334637

I orderred some of the tubing here at a webshop
There it was 10 meters 5mm for 30 Euros. (made a nice qcw primary with that hehehe)
Thanks for the link could still be usefull for new coils =]

I ordered a few LTC6992 just to get some hands on experience with it.

Nice!! Pretty easy circuits for there application!

The estimate I did for gate drive power was only at 70kHz, so at 86kHz it requiers even more. I would use +/-15V, since the maximum rating is +/-20V, which you could also go up to if you really wanted to punish it. But as you lowered the temperature by that much and it can now run so long, I would not invest more time in it, unless the gate waveform suggests that you should adress some issues.

I will invest a little more time into the build as i still want to change the pwm its driven on now to a active feedback system (so with sensing).
Worsed case i have to do better heatsinking on the 44n80.

at the moment this is my gatewave on the 44n80 with the FOD3481 driver chip on 12v+ and 12v- rails. still with the 10v zeners on the gate. (note that the 0v was not set to the centerline but 1 div below)

For gate drive I almost exclusively use GDT driven by MOSFET driver ICs and if its brick, a MOSFET output stage like on Steve Wards DRSSTC UD1.3/2.1, it is proven, sturdy and a easy to build design.

The only problem with that ive been told they are a no go on changing PW's?

EDIT:
i forgot to mention.
i was testing the new gate drive the other day.
for this test i replaced my 10v zener protection for a 15v zener protection.
this worked till some degree.
but after 20/30 seconds the waveform on the gate start decreasing.
any idea where that could have gone wrong?
for the gatedrive i now use the same circuit as used in ggy's qcw.

[Mads Barnkob]

If you want analyse more on the gate drive waveforms, Richie Burnett got this nice little guide, which was never really publicly available as far as I remember, but can be found from google: http://www.richieburnett.co.uk/temp/gdt/gdt2.html

You are right about the UD not being able to change the pulse width, as it only works on adjusting on-time for controlling the power output.

About your new gate drive test, did the drive decrease as you turned up the bus voltage? If so, see the link I gave at first and read about the miller effect.

[Acid Byte]

If you want analyse more on the gate drive waveforms, Richie Burnett got this nice little guide, which was never really publicly available as far as I remember, but can be found from google: http://www.richieburnett.co.uk/temp/gdt/gdt2.html

You are right about the UD not being able to change the pulse width, as it only works on adjusting on-time for controlling the power output.

About your new gate drive test, did the drive decrease as you turned up the bus voltage? If so, see the link I gave at first and read about the miller effect.

The gatewave was decreasing with the tesla coil not even on.
This was only with the 15v zeners.
Not with the 10v zeners,

Ill read that page! Thank you!

Edit:

i found the problem.
it was withing the psu.
waiting for new transfomers and then ill do an update.
i think this might be it....

[Mads Barnkob]

If you want analyse more on the gate drive waveforms, Richie Burnett got this nice little guide, which was never really publicly available as far as I remember, but can be found from google: http://www.richieburnett.co.uk/temp/gdt/gdt2.html

You are right about the UD not being able to change the pulse width, as it only works on adjusting on-time for controlling the power output.

About your new gate drive test, did the drive decrease as you turned up the bus voltage? If so, see the link I gave at first and read about the miller effect.

The gatewave was decreasing with the tesla coil not even on.
This was only with the 15v zeners.
Not with the 10v zeners,

Ill read that page! Thank you!

Edit:

i found the problem.
it was withing the psu.
waiting for new transfomers and then ill do an update.
i think this might be it....

Power supply design is often the most neglected discipline, because everyone just wants to get busy with the interesting parts I have done this myself too many times. Was the power supply just too weak to supply the current needed or was it actually a malfunctioning unit?

Looking forward to a new run

[Acid Byte]

Power supply design is often the most neglected discipline, because everyone just wants to get busy with the interesting parts I have done this myself too many times. Was the power supply just too weak to supply the current needed or was it actually a malfunctioning unit?

Looking forward to a new run

While other seem to get it working on a 1 coil transformer without center tap.
i needed to get myself a dual coil transformer and use the centertap as 0v floating ground.

Got the unit to work but to find out the FOD3184 IC's all die after a minute run.
Looks like them chinese chips aint no good.

End of the week ill put out a order on authentic FOD3184's and see if this solves my problem.

One thing for sure.
Adding that centertap stopped the rails from floating and decreasing in amplitude so thats a big plus.

Ill keep you updated!

Almost finished the new active modulator board too so could very well be possible that ill change out the audio modulator too.
This new modulator board is basicly a replica of the qcw modulator only difference is that instead of the fiber input and pwm filter.
Theres an small audio pre amp (tl072) wich makes you able to bias and amplify the audio into the range the tl3016 can understand.



Thats where im at and im gonna test out pretty soon once i have the authentic fod ic's

Greeting Jeroen