Induction brass annealer redux

[BlackICE]

As I and many others have pointed out it isn't wise to trust cheap no name SSR from brought from vendors that are not distributors of electronic parts as their main business. As SGK and McFred pointed out you can easily control the ZVS by controlling the 2 MOSFET already on the PCB for less than $1 in parts.

 

[hdmunger]

Hi, all. Finally finished the rebuild of my induction annealer!

Here is a pic of the original build

Annealer1.jpg

drive.google.com

Rebuilt it into a full-tower computer case so that I would have room above the ZVS board and coolant system for the automatic case feeder.

Some things I've added since the original build:

- case feeder
- counter
- dual temperature display (coolant tank and ZVS connection to coil
- flame sensor (can be used as primary case drop activator with the timer set to a little longer than the case needs, or as a backup to the timer in the event a case reaches flame sensor threshold before timer runs out...can also be switched off so that the timer is the only control)
- also, I replaced the 48V to 12V buck converter and used a computer power supply to provide all the 12V, 5V, and 3.3V power to run components not directly associated with the ZVS board, which also reduces the load on the 48V power supply

Annealer 1 .jpg

drive.google.com

Here is a view with the case feeder cover closed and open (sorry its kind of dark in there...).

Annealer 2 - TopFront and FeederOpen views.jpg

drive.google.com

Tried to keep it as clean as possible for the view through the glass side of the computer case.

Annealer 3 - guts.jpg

drive.google.com

Not so clean looking on the other side, but I shouldn't have to look at it very often.

Annealer 4 - the ugly parts.jpg

drive.google.com

Here is a 'close-up' of the area around the ZVS board...hard to see anything with that cooling fan in the way.

Annealer 5 - ZVS 'closeup'.jpg

drive.google.com

Here is a video of me rambling while I run a 5 case demo (I think the counter double-counted one of the cases...hmm).

Annealer Video.mp4

drive.google.com

Thanks to all who provided direct help and to those that provided inspiration...I appreciate it.

PS - for those interested, here is a schematic of my build at this point...

Annealer Circuit Diagram (3-21-22).bmp

drive.google.com

 

[davexre]

As I and many others have pointed out it isn't wise to trust cheap no name SSR from brought from vendors that are not distributors of electronic parts as their main business. As SGK and McFred pointed out you can easily control the ZVS by controlling the 2 MOSFET already on the PCB for less than $1 in parts.

i’ll have to look that up - I suspect I built mine before you guys had that discussion… LOL

 

[SGK]

Hi, all. Finally finished the rebuild of my induction annealer!

Nice. Man, those cases get bright red! They're way over 950F...

 

[SGK]

As I and many others have pointed out it isn't wise to trust cheap no name SSR from brought from vendors that are not distributors of electronic parts as their main business.

Agree re getting quality parts from reputable suppliers. I've not had any problems with the automotive relay I am using.

As SGK and McFred pointed out you can easily control the ZVS by controlling the 2 MOSFET already on the PCB for less than $1 in parts.

Maybe a little more than $1... To high-side switch 12V to the gates you will need either a relay (mechanical switch) or some sort of MOSFET driver, eg a photovoltaic one, because the downstream side of this switch 'stays high'. I've used VOM1271 before for this sort of application. One would have to check if they can switch fast enough to do the anneal 'stutter' @oliverpsmile is doing. Or you have to look at pulling the gates low when you want the ZVS board off. Whatever pulls them low needs to be able to manage the current for long periods.

 

[oliverpsmile]

I've used VOM1271 before for this sort of application. One would have to check if they can switch fast enough to do the anneal 'stutter' @oliverpsmile is doing.

Just kidding - may be another SSR?

 

[SGK]

To high-side switch 12V to the gates you will need either a relay (mechanical switch) or ...

Just kidding - may be another SSR?

You can't use a simple "SSR" ie MOSFET alone. The source of the SSR/FET isn't at ground but rather 12v so you need a driver. (Why I prefer pulling the gates low to stop annealing.)

But always better to be switching tiny-current-low-voltage than high-current-higher-voltage.

 

[oliverpsmile]

You can't use a simple "SSR" ie MOSFET alone. The source of the SSR/FET isn't at ground but rather 12v so you need a driver. (Why I prefer pulling the gates low to stop annealing.)

But always better to be switching tiny-current-low-voltage than high-current-higher-voltage.

As I said - just kidding.

 

[SGK]

Well given 48V is available one could use that, switched by an Arduino or PIC combined with a simple FET, to the gate of the 12V SSR/FET switch. But one would have to make sure (perhaps with a voltage divider) that the Vgs(max) of the SS switch wasn't exceeded. There's a few ways to skin the cat. I prefer less 'switches' as possible.

 

[hdmunger]

Nice. Man, those cases get bright red! They're way over 950F...

Yeah, those cases are some oddball 308 cases that I have probably run through the annealer a half dozen times or more, and I need to fine tune the flame sensor to trigger just a little sooner (not much).

 

[SGK]

Basic options for switching 12V to the gates of the ZVS board HEXFETs.

 

[itchyTF]

Induction coil ??

@VenatusDominus provided a formula for Henries given coil dimensions. If I recall correctly the Henry value for inductors in series is additive. I believe it was suggested for a multilayer coil to average the diameter. So, if I have a 3 layer 3-1/2 turn coil with the length being .625" - averaging the diameter would be .87". The value would be .228 uH. If I took each layer separately they would be .1062uH, .228uH and .3737uH. If I add them together it would be .7079uH. If I took the average and multiplied by 3 it comes out .684uH. Close to the additive value. If each layer had a different number of turns I don't see how the averaging method would work.
I believe the goal is about 1uH so the above example is no good (unless I want the frequency to be higher).

What is the correct method to determine uH?

 

[SGK]

You can find online calculators for a simple, one layer air-gapped coil. (I used one to approximate my own work coil and 'lo and behold its result when combined with the amount of capacitance on the induction board yielded about the same frequency as I measured with my oscilloscope.)

The situation with a multilayer work coil is more complex. Good luck chewing on this!

An introduction to the air cored coil

 

[CGB]

You may be familiar with this site but it has some good basic design information regarding coils/inductance.

https://www.eeweb.com/tools/coil-inductance/

 

[SGK]

As SGK and @McFred pointed out you can easily control the ZVS by controlling the 2 MOSFET already on the PCB for less than $1 in parts.

So I've been playing around with this. The gate capacitance of the two induction board HEXFETs need to be charged and discharged very rapidly (the frequency of oscillation). If only 12V is supplied, the values of resistors R1 and R2 (470R on the unmodified board) need to be very low. (Closer to 10 ohms.) This places a big demand on the 12V power supply. (Close to 1 amp in simulation with 10 ohm resistors.) That's not an issue if the 12v supply can supply it and your switch can handle the current. (Even if my 12V PSU had no issues supplying the current the tiny switch on my control board won't like the current it has to sink to turn the board off.)

On the other hand, with the unmodified board the current demand on the 48V supply is modest. (Circa 176mA in simulation.) A lot of energy is dropped across the resistors and that's why they are big 5W ones. To have a low current, low voltage point of switching the switches (in whatever form they take) have to be on the downstream (right) side of R1 and R2. That may make a slightly more complicated board mod for someone want to switch that supply connection on/off. An alternative is to pull the gates to GND when you want the board OFF. A connection at each gate (or downstream side of R1 and R2) and a switch to GND should achieve this. When the switch pulls these two points to GND it needs to handle 204mA at a drop of 12V. I'm going to test this now.

 

[BlackICE]

Thanks SGK useful information. I plan to only pull the gates low, so only about 204mA is needed. A saturated NPN TO-92 transistor of the proper voltage rating should be able to do that using a micro-controller pin to switch it on or off. I don't need direct control of the oscillations, but only to turn the circuit on or off at no more than 500 / sec max, probably much less. I guessing an limiting resistor (~50 ohms) should be used between the gate and transitor's collector to limit the current from bleding the charge from the gate. I can either use two transistors one for each gate, or use 2 diodes to isolate the 2 gates an only one transistor.

 

[SGK]

A word of caution. I've not been able to get it to work. Either I have damaged my board (I doubt it) or the board doesn't like not having power to the gates hit at the same time as elsewhere in the circuit. Grounding the gates turns it off ok but it won't restart properly when that grounding is removed. More to follow.

 

[BlackICE]

I was wondering would that happened myself. I have read did people in the past have had problems if their power supply didn't have sufficient current at start up, the board would not oscillate. Maybe it's best to control both gates separately starting one side 1st and wait about 10us and then start the other one.

 

[SGK]

@BlackICE and @itchyTF

Hmm. So I MIGHT have found the problem but I'm not sure I can be bothered going back and retrying things with the change!

I started out by supplying 12V from my 12V PSU connected to the induction board via 470 ohm resistors. I was pulling the gates low via an SMD FET switch that connected the gates to GND on the 12V PSU when I wanted the induction board off. I had been worried that there might appear voltage differential between the ground on the 12V PSU (against which the 12V is referenced) and the ground on the induction board (against which the Vgs is referenced since the sources of the HEXFETs are tied to ground). So I wired the two PSU grounds together.

Later I realized that rather than supplying 12V to the induction board via a 470 ohm resistor I was better off continuing to use the board's 48V input. So I reassembled the original 5W resistors thereby reconnecting the 48V traces. I was still turning the board off by pulling the gates low. But I couldn't get the board to 'turn on' - to begin oscillating when I removed the ground connection at the gates. My amp meter would show very high current draw but the case wouldn't heat.

So today I reconnected my automotive relay and reprogrammed by control board to drive the relay rather than driving the gates of the induction board FETs. (That is, switching the 48V supply to the board.) The induction board worked, cases got hot etc so my inability to get earlier configurations to work was not because I had damaged the board. However, I noticed my amp meter was showing very high amps 20+ and cases got very hot, very quickly. Given my 48V supply has a limiter at circa 15A something was wrong. Given the only change now (except my control board was managing timing rather than the Sestos timer) was the GND wire between the two PSUs I decided to remove this. Everything went back to normal.

So... @BlackICE you might want to try switching the induction board off by pulling the gates low (keeping the board unmodified except a wire from each gate, or the downstream side of R1/R2 solder pads, to your switch to GND) - but without the GND wire between PSUs that I added. Given I have rewired everything back to use the automotive relay I'm not sure I have the enthusiasm to try again until my relay fails (it's worked flawlessly for a couple of years).

 

[SGK]

@BlackICE and @itchyTF

I now realize my mistake. (In addition to the potential issue noted above.)

It's a really stupid one. Doof!

I had run two separate wires, one to each gate. (There are plenty of entry points on the bottom of the board - the downstream side of R1 and R2, the legs of R3/R4 or D1/D2 that aren't at GND.) The idea was to ground both gates. HOWEVER, I only had one free switch available on my controller board. I joined these together at the controller. Of course the problem with this is when they're not grounded they're still tied together!! The board won't work. Duh.

Someone else will have to test grounding/ungrounding both independently but concurrently. Let me know how you get on!