IRS27951 / IRS27952

[nigelwright7557]

I had a look at running the circuit on 30 volts tonight.
The output of the LO and HO are wrong.
I think I am having layout problems with my pcb again.
I put on a SOIC8 padset and also a PDIP padset.
When I use the soic8 the HO and LO run at 50% duty cycle.
When I use the PDIP it runs at about 5% and 95% which is wrong.
I think this is confusing matters.

I also had trouble getting 27952's to solder in the pcb, some would stop working and others would work ok. I guess I need a proper SMD solder heat gun.

I think I would need to get some more pcb's made.
By the time I have paid for more pcbs and bought a SMD heat-gun I don't think it is worth the hassle.

I might just go back and revisit my fly-back SMPS and finish sorting out problems with that.

[Reactance]

Some more educational resource(s) for LLC resonance converters.

Microchip: LLC Resonant Converter Reference Design using the dsPIC DSC - YouTube

http://www.ti.com/lit/ml/slup085/slup085.pdf

http://www.onsemi.com/pub_link/Collateral/AND8255-D.PDF

Personally, I don't feel I am capable yet to construct this, its way out of my experience range, I will stick to a simple control like SG2535

@lorylaci

Thanks for the tips lead me to the above.

@nigelwright7557

Dont give up, tip do try calculate all the critical parameters as lorylaci suggested he seems very tolerant. (a quality many senior members on this forum lack "must be the age" )

Another tip why don't you make a lower power version say 20Vin and step it down to 10v and see what's the success factor. its way safer for you and the power supply also its less destructible.

Instead of building a PCB every time the design changes why not build your LLC control board on a breadboard and all high current stuff on a PCB, thus allowing easy modular design, even better a daughter.. (I see you have an expensive double-sided PCB, that must have cost more money to develop. why not go for a single side with no silk screen to reduce costs.

At the end of the day the math determines the success of this project coupled with diligence and effort.

[5th element]

Just a standard paint stripping hot air gun is fine for soldering SMD parts that are thermally enhanced with an exposed pad on the underside of the IC. I use a really old Black and Decker one, along with solder paste from ebay, for just that and it works great. I've soldered high power LEDs with it too and it even worked well for those given the temperature sensitivity of the parts.

[Reactance]

Originally Posted by 5th element

Just a standard paint stripping hot air gun is fine for soldering SMD parts that are thermally enhanced with an exposed pad on the underside of the IC. I use a really old Black and Decker one, along with solder paste from ebay, for just that and it works great. I've soldered high power LEDs with it too and it even worked well for those given the temperature sensitivity of the parts.

You don't need a hot air gun to solder smd parts a simple tempretaure control iron is sufficient enough.

I use a hakko fx-888 for SMD work

[nigelwright7557]

Originally Posted by Reactance

@nigelwright7557

Dont give up, tip do try calculate all the critical parameters as lorylaci suggested he seems very tolerant. (a quality many senior members on this forum lack "must be the age" )

Another tip why don't you make a lower power version say 20Vin and step it down to 10v and see what's the success factor. its way safer for you and the power supply also its less destructible.

Instead of building a PCB every time the design changes why not build your LLC control board on a breadboard and all high current stuff on a PCB, thus allowing easy modular design, even better a daughter.. (I see you have an expensive double-sided PCB, that must have cost more money to develop. why not go for a single side with no silk screen to reduce costs.

At the end of the day the math determines the success of this project coupled with diligence and effort.

I powered up the circuit with 30 volts to start with. Checking for a 50% signal on HO and LO is a good indicator the circuit is doing something.
My first mistake was not using star grounding on the pcb and this caused the HO and LO to be wildly off.

I thought I could scale the application project by scaling the transformer but that didnt work.
My pcb's are fairly cheap as they come from China.

I have ordered a SMD heat gun so will start again with a new chip when that comes.

The frustration stepped in with the transformer as I couldn't get its inductance correct.

[nigelwright7557]

Originally Posted by Reactance

You don't need a hot air gun to solder smd parts a simple tempretaure control iron is sufficient enough.

I use a hakko fx-888 for SMD work

I am finding sometimes my IC's blow up after being soldered in.
I guess I am either frying them or blowing them up with static.
Either that or my soic 8 footprint is shorting on the pads ?

The hot air gun will also help remove any bad IC's.
I ripped off a pad last time I removed an IC.

[lorylaci]

Originally Posted by nigelwright7557

I am finding sometimes my IC's blow up after being soldered in.
I guess I am either frying them or blowing them up with static.
Either that or my soic 8 footprint is shorting on the pads ?

The hot air gun will also help remove any bad IC's.
I ripped off a pad last time I removed an IC.

Nigel, attached you will find my controller board. As you can see its pretty small, however I use a simple temperature controlled solder station (a cheap Solomon SL-10), and usually I only fry about 1 out 30 from SOIC (for TSSOP my statictics are not so good).

Do not use PDIP converters, it os no question why this ICs are made only in small SMD packages. At freqs over 100kHz layout is important.

Would you like to have the layout and schematic plan of my simplified controller board? I can send you if you'd like.

On what inuctors you tested you meter? Some meters work pretty bad on ferrites, however work well on low-freq inductors and transformers and iron-powder cores.

Nigel, look at the datasheet of your ferite cores, it says Al=5000nH for thed Ferroxcube ETD54 3C90 ungapped. So if you do the math: L=Al*turn^2=845uh ungapped. So I am sure there is some problem.

An ETD39 has an al about 2700nH ungapped, so 13 turns should get 456uH. I am do using an ETD39 for >1kW with 13 primary turns (working 100-300kHz freq range), and I do have 50uH total primary inductance but WITH "BIG" GAP. So this is why I thinking about your LC meter.

I also send you a scope picture of how the outputs should like. Here I am using a BJT MMBT3904/3906 pair to boost the output of the IRS27951, to drive FETs with high gate charge (this is only necessary over 1kW).

[5th element]

Originally Posted by Reactance

You don't need a hot air gun to solder smd parts a simple tempretaure control iron is sufficient enough.

Indeed, I usually just use my Metcal, but for any thermally enhanced packages, like LEDs or SOIC/SSOP/MSOP with an exposed pad on the underside, one can't easily solder what's not accessible. If you're using SOIC parts you can drill a large enough hole under the device to give you enough access from the rear, to solder the pad to the reverse side copper, but with smaller parts, this isn't practical and here the heatgun with solder pasted works a treat.

As it stands though, the parts Nigel is using aren't thermally enhanced and I've never destroyed a SMD chip with too much heat or from ESD. I don't know how robust these chips from National are though, they could be prone to destruction.

[nigelwright7557]

Originally Posted by 5th element

As it stands though, the parts Nigel is using aren't thermally enhanced and I've never destroyed a SMD chip with too much heat or from ESD. I don't know how robust these chips from National are though, they could be prone to destruction.

It could be that it is my pcb at fault for the problems and some chips are working and others aren't.
I didn't use a star ground so noise will get into the timing circuit.
I did hack the pcb to get the VB loop in a star connection.

Once I get my heat gun I will try again to get the circuit working.

[lorylaci]

Originally Posted by nigelwright7557

It could be that it is my pcb at fault for the problems and some chips are working and others aren't.

Again I can offer you to send my controller's PCB design. Its pretty easiy, does not use compontents smaller than 1206, so can be soldered by hand easily. Also its on a separate board, so you can check the controller working separately, which is pretty useful.
Do you need my PCB design?