Induction brass annealer redux

[dskogman]

“I want to ask about your 50A model, what voltage do you run it at? “
I have a 36v x 1000W power supply. It runs at a higher amperage since it’s a lower voltage board. I have heat sinks on the bottom and bigger fans on top that run full time. The fan on the board only runs when it’s on so is pretty useless. The heat generated dissipates fairly quickly thru the PCB board which has a copper mesh embedded in it.
I looked at water cooled heat sinks like some of the new PC’s are using but figured I was far enough down the rabbit hole.

 

[SGK]

Bill, if this is the model you are contemplating https://www.ebay.com/itm/50A-1000W-...wer-Supply-Board-DC12V-40V-Y5N5-/311655795345 note that the fan is really sitting in the wrong place - it's not pushing air over the caps. I suspect mounting the coil piping would be a bit more challenging also. (It is likely very easy to rewire the fan so that it runs all the time the unit, rather than the board, is powered.)

This board has fans in the right place

 

[SGK]

And here is an interesting link (assuming your are interested in how these things work and how to make a better version) because it discusses principles of operation, tuning via PLL etc.

 

[GrocMax]

So easy to miss the little bit circled in red. The Sestos timer manual implies that it is already held at 0V internally. (As if pins 1 and 9 are connected when powered by DC.)

Took me about 2 hours to figure out why I couldn't trip the timer when inserting a case. And I did it the hard way - I only looked at your schematic after I had figured it out. FFS.

Yeah the switches need the external 0v ref.

 

[BillK55]

For brass tubing the ideal frequency seems to be in the 90K-120K Hz range. With the 1 1/8" OD 8 turn 1/8" tubing coil on the 1000w induction board listed on page one I measured ~110 KHz.

I finally found the reference I was looking for back on pg 28. It was GrocMax who measured his operating at 110 Khz. That's near the middle of the optimum range for brass. Ok Groc, do you remember what voltage you were operating at?
I'm asking these questions because I had already read up on theory and watched the guys on youtube that are doing experiments with different coils and documenting the effects on frequency. I like dskogman's approach of using the beefed up ZVS and PS, but now wondering what's the impact of his PS's 36V max output.

I know that ebay seller ouyou2010 replied back it had no fan, but this morning I found what appears to be the same one on banggood with more pics including the back, and theirs DOES have a fan. https://www.banggood.com/AC110VAC22...ter-p-1264650.html?rmmds=buy&cur_warehouse=CN I think ouyou2010 is just too busy to answer questions and it was simple to say no since the listing doesn't address it, and his has a fan as well. Its just silly to build one with closed case and no fan.

The 50A ZVS's have more cap's, divide the work up between 4 MOSFETs instead of 2 as on the 20A version, so its logical if I only push the 50A rated version upwards of 20A, then its going to be less likely to overheat. Yes still make sure I have fans and add the $10 heatsinks on the bottom. Sure wish the MOSFET heatsinks fins sat inward to catch the fan's airflow. Outward layout is logical before thoughts of a fan I bet.

If I'm not going to push the ZVS anywhere near the 50A, then why pick a 1000W PS rated at 36V (27A capacity) if the 48V 1000W (20A capacity) will let me run at a voltage closer to what gives me the optimum frequency for annealing brass and still give me comfortable ampacity?

 

[BillK55]

The 'root cause' is a natural product of the circuit.

I meant root cause of heat related failures, not the cause of the heat itself. I understand that the components of the oscillator get hot. Maybe I should have asked it differently. Do we think the overheating of the pcb under the cap's is from high resistance of the copper paths, bad solder joints, or from heat radiating out of the capacitors?

 

[BillK55]

Just realized in my previous discussion about ZVS operating voltage and oscillator frequency I made an assumption that isn't necessarily true. I assumed that if you take two of the same 1-1/8" GinaEric design coils and place one on the 50A rated ZVS, the other on a 20A rated ZVS, operate them both at the same voltage, they run at the same frequency. (oops, too many rabbit holes around here) No really, inquiring minds want to know. Anyone out there made the comparison, either measured the frequency, or have empirical data (anneal time by case size) to support it?

 

[BillK55]

I suspect mounting the coil piping would be a bit more challenging also.

Yes the fans blowing down do complicate the ZVS output connections. A while back I did have an idea for making the layout of the ZVS a little less restrictive in regard to it needing to be mounted close to the front of the unit. There is a design for flexible water cooled conductors that could be used here to run between the ZVS and the annealing coil. Its used for the power lead on TIG (GTAW) welding torches. The standard TIG ones are too long and have larger size pipe thread type fitting crimped on the ends, so you'd have to make your own, but it wouldn't be that difficult and not prohibitively expensive. For a 200A rated TIG torch what you see on the outside is a reinforced rubber hose on the order of about 3/8" size (don't think its as big as 1/2"). That hose serves as both the electrical insulator and containment for the cooling water. The next layer inside is a tinned copper hollow braid cable, similar to what you'd see as the braided shield on coax cable. You then have a hollow core through which they circulate water out to a radiator in a closed loop. The braid isn't bonded to the outer hose, so its very flexible and there is intimate contact between the water and the strands of the braid. Combination of high surface are of the many, many strands of the conductor and the fact you're cooling the conductor directly allows for a much lighter and flexible power lead than compared to a 200A stick welder. Much easier to route inside your case than standard copper tubing, and its actually flexible. Something like this https://www.ebay.com/itm/302172765975 @ $0.99/ft plus shipping inside some 3/8" tygon tubing and handmade transitions on the ends. Just an idea that might be useful for someone.

 

[GrocMax]

From what I remember when I measured freq at the induction coil DC power voltage into it didn't matter. My understanding of the tank circuit resonant frequency is the coil itself and the capacitor values are the primary components that change the frequency, particularly the capacitors, and they are a special type for this application.

The caps for this app are shielded and encased, not sure how convection cooling is gonna work when they have a suit and overcoat on. Majority of the heat transfer out of the capacitors goes out the wire leads into the PC board.

 

[BillK55]

I always like an excuse to buy a new tool. How about this frequency counter https://www.ebay.com/itm/201919128979
Its made to hold near a transmitter, no direct connections required. Think it'd work for checking our ZVS frequency? For $10 its hard to go wrong
This one looks fancier, but then I'd have to connect into the circuit via its antenna connection
https://www.ebay.com/itm/181860091360 Still just $10

 

[GrocMax]

https://www.allaboutcircuits.com/te...rrent/chpt-6/parallel-tank-circuit-resonance/

Formula for determining tank circuit frequency.

 

[BillK55]

The caps have a suit and overcoat (love the analogy) on ...… Majority of the heat transfer out of the capacitors goes out the wire leads into the PC board.

That makes sense then why the board gets hot under the caps. If that's the case then your heat sinks attached with thermal epoxy are probably best bang for buck, that and make sure those heat sink fins get some air across them.

From what I remember when I measured freq at the induction coil DC power voltage into it didn't matter.

That's good to learn. Thanks. Helps work through the possibilities

Formula for determining tank circuit frequency.

Ok that was readable (lol for me), thanks. Voltage has no affect on frequency. I did see something scary in there though, at resonance, the circuit draws NO POWER FROM THE AC POWER SOURCE. Nobody warned me we were dealing with perpetual motion machines …. shh its ok now, I calmed down.

So the 50A ZVS has 8 caps vs 6 on the 20A model. Could it be a network with the same capacitance as the 20A model? (worried now will the same coil design work same as before if they've changed the tuning of the LC circuit)

Ok I think I just slipped into the deep end of the pool. Someone toss me the life ring …. please.

 

[Gina1]

Blub... Blub... Blubb.... Long way from the basic GinaEric design.

 

[GrocMax]

Doesn't matter how many caps are on it, just total capacitance, then need to know coil inductance. It can be calculated, however since capacitors have a wide tolerance (-/+ 5% and often more) measurement is the last word.

The lower the frequency the deeper the penetration into the workpiece, for thru heating of large pieces of steel they use low frequencies and gobs of power. 100KHz is about ideal for brass, around .010" penetration.

 

[SGK]

Blub... Blub... Blubb.... Long way from the basic GinaEric design.

True. But I think a beat that's been missing from this thread is how collectively we can design a truly professional, yet affordable, build with all the bells and whistles many would like but individually many don't have the time, inclination or skill to implement. Auto case detection, for example, could be a part of everyone's build if designs, boards, code and BoMs are shared. Same with PWM controlled fans and temperature monitoring. I have about 14 spare PCBs for my controller board sitting here. I'm happy to sell them at cost, share the code, PCB design files and BoM etc. (I think the parts cost for this PCB is about $20 or less with the 2oz copper PCB about $8. The IR switch and 4 NTC add about another $7.50. All if buying at prices for an order size of 1. Not bad for 4 NTC temp monitoring, PWM control of 2 fans, auto-case detection and autofeeder drive etc.) Something missing from my design? We could think of what a Rev2 might include, prioritise features, implement and produce. I already know what I would change if I did another PCB run. If I don't have the time nor inclination to extend it further someone can pick up where I left off with all the detail, files, code etc backing what I had done. Or maybe we back an Arduino platform. A reasonably-priced autofeeder that would work across a large number of GinaErick annealer enclosures? Bring it on - work in progress.

The original circuit and design advice is/was awesome. Thank you so much for sharing. Collectively we can and should continue to add to it. In the last few days a few things have been kicked around including current limiting that doesn't require the expensive MW PSU, a better induction board, various 'control circuit' typologies, whether the liquid cooling circuit could be improved upon, a more generic GinaEric auto-feeder...I'm for all of them so long as they are "open source" with disclosed BoM, schematics and designs and made from readily available, or available via 'group buys' parts/PCBs. Products like the EZ Anneal should really only exist for those who want a boxed, warrantied and completely 'done' product. After all they are just commercialising the talent broadcast on threads like this.

 

[SGK]

If someone wants to do a little free improvement on the no-fan 1kW ZVS board may I suggest the following. Unscrew the standoffs which support the copper tubes and place them (or replace them with shorter ones) on the underside of the board. At least then the heat rising from the caps isn't heating your copper and coolant as they pass over the top. I wish I had thought of it sooner.

 

[#40Fan]

Do you really think there is that much heat being transferred to the copper piping? I sure don't. But, I do have a 120mm fan above the board.

 

[SGK]

If you don't then you don't need the the coolant, pump etc. Copper is a great conductor. I think heat is transferred via the ambient temperature inside the enclosure, the piping passing over hot capacitors and cases getting momentarily hot inside the coil. So...blow cool fresh air into the enclosure, minimise direct heating of the tubing as it passes over the hot caps, run some coolant through the piping and eject the cases swiftly from the coil. I'm much less worried about the temp of the coil even with the minimal liquid cooling we are implementing. But the suggestion costs $0 to implement. Just 30 seconds with a screwdriver. I don't have a fan running directly over my induction board - a fan blowing hot air doesn't achieve much; you need to get cooler air into the enclosure. That and/or assist the caps to dissipate heat by improving their thermal coefficient.

 

[#40Fan]

I do see the need for coolant running through the piping, but not protecting the piping from the heat coming from the board.

Give it a try though and report what happens. Do a before and after temp reading of the coolant reservoir water with two sets of identical cases ran at the same length of time.

 

[SGK]

I do see the need for coolant running through the piping, but not protecting the piping from the heat coming from the board.

Give it a try though and report what happens. Do a before and after temp reading of the coolant reservoir water with two sets of identical cases ran at the same length of time.

My enclosure is already set up to take the tubing via the stand-offs "as is". So I won't be doing the test. Were I designing a new enclosure I would definitely make this change as it costs nothing, even for an as-yet unquantified, but IMHO definitely positive, gain. Instead I will be adding some Kapton tape to the tubing where it passes over the caps. I already have the tape, don't need it for anything else and so the cost of deploying that measure is $0. I used Insultherm on my coil(s). That cost about $5 for two coils.

If you see the need for coolant then you would have already asked yourself "what is the source of this heat, how is it transferring to the thing I think needs to stay cool and how can I minimise that transfer?" You can't change the basic properties of the induction circuit and the heat it necessarily expels. If you followed the latest typical BoM you felt it necessary to buy a radiator ($18), a fan to go with it ($13), an aquatic pump ($22), tubing ($10-14) and coolant ($10-15?). (Rough prices for new items.) If it costs nothing to reduce the load on those items it's probably worth doing.