Induction Heater problem 2

[itsmoked]

I've got to go look at an industrial induction heater tomorrow.

What I know:
It's an EASYHEAT 7590LI solid state induction heater.
It's less than 2 years old.

It worked great for heating a tube up for a subsequent insertion of some piece into it for shrink fitting.
I'd do ~150 pieces a day typically.
One day it started not heating as well, taking longer and longer to heat the workpiece. No material/process changes.
They sent it to the manufacturer to check. Manufacturer said there were some minor issues, certainly no show stoppers, and that they took care of them. Returned it with a clean bill of health.


Still not working.

That's all I know so far. Besides the usual suspects of bad power, damaged cables, etc., is there anything typical of induction machines that's problematic?

Skogs I know you've worked on one.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[btrueblood]

They sent the control unit/power supply only, or the induction coil too? Is the coil water-cooled, and is it getting good flow. Can't remember if there are conductivity limits on the coolant...

 

[VEBill]

It's presumably all about driving (AC/RF) current into the coil. Google says 9kw and 150 to 400 kHz (hmmm... be careful).

If the "transmitter" has been checked out, then concentrate on the feed line and coil. Perhaps just a bad connection.

Sad Google results: "Your search - easyheat 7590LI diagnostics - did not match any documents." You'd think it would provide some operator feedback and system self diagnostics about the output power and load.

 

[itsmoked]

Thanks guys.

Today's visit was.. odd. One of those "we can't run it because they're still using the press part of it".

They've replaced it with a bank of toaster ovens. They gun the work pieces and when they're at temp they grab them out and stuff them into the press part and run it.

btrue; It has a chiller and is water cooled. It has flow sensors that seem happy.
Conductivity limits.. I'll check that.

VE1BLL: Yes it's 9kW.
The unit has the power unit - a big 19" rack unit - and the "Transformer" (shoe box) which has The Capacitor in it and a crazy football shaped RF transformer with a fist-full of taps.

As you run the machine it will suggest you move the taps UP or DOWN if the coupling is not so good. They reported 'something' to the factory and the factory suggested they "move the tap down". Now the unit itself is suggesting they raise the tap. This is still not supposed to be a show stopper just an optimization.

If they put the machine into manual mode and jumper everything on the control terminals it appears to run OK. Yet if the control is handed back to the PLC and the terminal block on the rear, it has issues.

That's pretty much it. I'll be going back tomorrow when -theoretically- they won't be using the machine and I can.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Make sure that the 9kw of MW RF isn't getting into the controller.

But it'll probably be something completely unexpected. Good luck.

 

[ScottyUK]

I built one of these things a long time ago at university, although mine was at lower frequency. Some topologies rely upon a tuning capacitor to bring the work coil close to resonance. You might want check the condition of the tuning cap if this unit has one.

The second from last paragraph - do you mean that in manual mode the unit is performing normally? If so then that presumably indicates that power stage is working ok and the fault is in the control or possibly in the resonance detection if the unit has a self-trimming capability.

 

[IRstuff]

"Yet if the control is handed back to the PLC and the terminal block on the rear, it has issues. "

So, it sounds like there's something amiss in that block...

TTFN
faq731-376

 

[itsmoked]

What a day.

We sat there working as assembly workers all day waiting for the machine to have one of it's conniptions. This, under the theory of, "it will crap-out after we do a few". 350 later... Nothing. I put all the covers back on and fully installed control unit back into the rack as we continued to run another hour of product.

After an hour or so it started to act up. Instead of 400 Amps it dropped to about 385A, occasionally and the display put up an asterisk which signifies the output being more than 2% off target. So the controller knows about the problem.

The controller automatically alters the drive frequency to optimize the power transfer. After running at 266kHz all day it was now wandering around to lower values like 185kHz during these anomalies.

This whole thing hinges on a tank circuit that couples the RF power thru a transformer to the user's heating coil. There's a key capacitor involved. Can anyone think of a way for a cap to shift 'quality' or capacitance without failing outright?

Since it did this after an hour of having the covers on, it cast the possibility of temperature as a participant. So I got a temp gun and a heat gun and sequentially heated every major parts group or circuit board in the controller. I got them them up to about 130F. This did not cause the anomaly to ever appear. Note: Nothing in the controller seems to get over 82F in normal operation Everything major appears to be water cooled anyway.

Still puzzled.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

At 9kw it must be something almost the size of a softball or maybe a shoe-box. (?)

What type of capacitor? Fixed or variable? What type of capacitor-making technology?

 

[Skogsgurra]

Sorry to be late, Smoked.
Sorry that I cannot help much, I'm afraid.
I think that you may have a load problem. If you could check coil current an voltage on a scope, that would probably help. You can see if the compensation is OK, current and voltage are well aligned (phase angle near zero) and it is easy to see if there are any deviations. But, be careful with the current clamp. If they are not made for those frequencies, they may start a fire or even melt down.

Tip: make your own clamp for higher frequencies. I have a few of those that I needed for a special investigation. Shall see if I can find a description. BBL.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[VEBill]

...check coil current an voltage on a scope....current and voltage are well aligned (phase angle near zero)...

The E and I should be in phase (resistive matched load) at the input (transmitter side) to the matching circuit (assuming that the matching circuit is working correctly); but the coil itself (its current and applied voltage) should be on the inductive side of the Smith chart and ELI the ICE man would apply. Unless I'm confused (?) .

 

[Skogsgurra]

Here it is! Fresh from the oven.

http://www.gke.org/presentationer/files/Green ferrite current clamp PM.pdf

The pdf describes a current clamp that I needed lots of for a bearing current investigation. I made ten of them. This particular one has 50 turns secondary and a 100 milliohms burden resistor. It can easily measure 10 A, probably more. If you need to measure more current, adjust number of secondary turns accordingly.

The ferrite quality is a high permealibity type with around 1 MHz band-width. For phase comparison, I think you can use just about any decent quality.

Broken? Yes! I broke it. If you hit the toroid with a mallet, it will break nicely and the irregular cuts aids in keeping the halves in place. Use nylon strip to keep them together.

Of course, if you can break the current path, you do not need to break the toroid.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[btrueblood]

Smoked,

Is it possible that the power supply and induction coil are just fine and doing what they are supposed to do, and the fault instead lies in the part? You say the parts are tubular - are they welded into tubes or drawn? If welded, I wonder if linear flaws or variations like that in the weld seam interrupt the eddy currents that the induction coil is trying to generate. Might be worth doing that day-long experiment again, but this time grab samples of parts that heat ok to compare to parts that give the controller fits. Maybe it's as simple as wall thickness variations in the parts, or more complicated like local areas that are more heavily strain hardened, or have more oxide inclusions...dunno but it might be worth looking at. Finally, if the 385 amp parts still got hot enough, is there really an issue?

 

[itsmoked]

At 9kw it must be something almost the size of a softball or maybe a shoe-box. (?)

What type of capacitor? Fixed or variable? What type of capacitor-making technology?

Sorry VE1BLL. I'm not actually sure. It is a white labeled cylinder about the diameter of a coke can and about 3" long. It has a 1/4" stud sticking out the center of each epoxy potted end. Does that give us any clues?

Sorry to be late, Smoked.
Sorry that I cannot help much, I'm afraid.
I think that you may have a load problem. If you could check coil current an voltage on a scope, that would probably help. You can see if the compensation is OK, current and voltage are well aligned (phase angle near zero) and it is easy to see if there are any deviations. But, be careful with the current clamp. If they are not made for those frequencies, they may start a fire or even melt down.

Hi! Better late than never!
I actually don't know what voltage to expect or what a DMM will think of 400A at 250kHz. At that current I'm expecting a really low voltage, but I'm not sure. Will the DMM even give me a correct voltage value? Since I don't yet know the voltage, I've hesitated hooking a scope probe across the coil.

The E and I should be in phase (resistive matched load) at the input (transmitter side) to the matching circuit (assuming that the matching circuit is working correctly); but the coil itself (its current and applied voltage) should be on the inductive side of the Smith chart and ELI the ICE man would apply. Unless I'm confused (?)

"ELI the ICE man" haven't heard that one in a while! Current on the heating side should lead the voltage. Makes sense. Nice to know there shouldn't be much difference on the supply side. That's something I can check. Is that why the controller messes with the frequency? To try to avoid any phase difference so maximum power is transferred?

Here it is! Fresh from the oven.

http://www.gke.org/presentationer/files/Green fe...

The pdf describes a current clamp that I needed lots of for a bearing current investigation. I made ten of them. This particular one has 50 turns secondary and a 100 milliohms burden resistor. It can easily measure 10 A, probably more. If you need to measure more current, adjust number of secondary turns accordingly.

The ferrite quality is a high permealibity type with around 1 MHz band-width. For phase comparison, I think you can use just about any decent quality.

Broken? Yes! I broke it. If you hit the toroid with a mallet, it will break nicely and the irregular cuts aids in keeping the halves in place. Use nylon strip to keep them together.

Of course, if you can break the current path, you do not need to break the toroid.

Very cleaver Gunnar! Thanks MUCH for the Data Sheet. Where do I order these?

What are the nylon tubes for?, to tie-wrap the halves together?
10A huh? Did you notice I'd have to measure 400A? Suggestions?

Smoked,

Is it possible that the power supply and induction coil are just fine and doing what they are supposed to do, and the fault instead lies in the part? You say the parts are tubular - are they welded into tubes or drawn? If welded, I wonder if linear flaws or variations like that in the weld seam interrupt the eddy currents that the induction coil is trying to generate. Might be worth doing that day-long experiment again, but this time grab samples of parts that heat ok to compare to parts that give the controller fits. Maybe it's as simple as wall thickness variations in the parts, or more complicated like local areas that are more heavily strain hardened, or have more oxide inclusions...dunno but it might be worth looking at. Finally, if the 385 amp parts still got hot enough, is there really an issue?

Hail btrue; No chance. The work pieces are forged precision-machined aluminum block-things, not tubes as I thought. Tubes are rammed into the blocks once they're heated (expanded). Once the 'issue' occurs it hangs around,(so I'm told), which precludes the actual parts being the cause.

Added information discovery:

This thing runs on 208V 3ph. The factory tells us the only thing that would cause what we're describing would be if the line supply voltage is sagging. It does sag but from about 210 to 207.5V and we couldn't see any direct correlation between the sag and the output dribbling off the few minutes we saw the phenom.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

Smoked
I think that you can order them at the Flaming Systems Company somewhere near S.F, CA. Look them up in the Yellow Pages

"to tie-wrap the halves together?
10A huh? Did you notice I'd have to measure 400A? Suggestions?"

Yes, to tie-wrap the halves together.

400 A. I was wrong when I said 10 A. The actual measurement is 53 A and nothing says it can't take 100 A. Use 200 turns secondary and .1 ohm burden to get 400 A capability. Use AWG 20 or thereabouts.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

Have you looked at the temperature and flow of the cooling water?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[VEBill]

If they put the machine into manual mode and jumper everything on the control terminals it appears to run OK. Yet if the control is handed back to the PLC and the terminal block on the rear, it has issues.

The above quote seems to be the most important clue. Maybe it should be very carefully rechecked. If it's true, then it excludes many possible faults.

 

[itsmoked]

Skogs; Thanks! Will do.

Bill; Yes. This thing requires a chiller. The chiller is about 3 feet away with a large active temp display on the front. You can see the coolant temp at all times. You can watch as it climbs every cycle and the returns almost to pre-cycle temp. If production is continuous you can see it's steady state value rise a few degrees. It keeps the coolant around 76F. I'm guessing this prevents condensation issues.

VE1BLL; This was considered a priority. I essentially spent hours focused on the connector. I went so far as to replace it. They'd made a statement to me that, "When there's a problem the current drops to nothing", which seemed black and white. Later interrogation had them recant this, then I saw the 3~8% decline myself, confirming the reality that it doesn't, "drop to nothing", like an intermittent control signal might.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[btrueblood]

Keith,

Must admit, I have no clue how that auto-tuning circuit works. Units I have seen/used in the past were strictly of the turn it on and run it til the part gets hot variety*. What is the feedback that the PLC/controller is using to vary the frequency - the phase angle b/n current and voltage at the coil? Is that somehow correlated to the part's bulk temperature? Is there any direct measurement of part temperature? Sorry if this is of no help, but I'm just curious how you can get from coil RF measurements to the power dissipated in the part (I sorta see how that could work, but there are assumptions to that equation that I'd not take lightly) - but from power dissipated by the coil to part temperature rise would seem to require some more information, like part mass and specific heat, and cooling rate vs. temperature, and...

* along those lines - what prevents this device from being operated with manual/fixed settings for frequency and time, and just set it and forget it, once a few tests are run to find a "best" setting? You seemed to state that if the controller is left out of the loop, the unit functions just fine?

 

[itsmoked]

Btrue; A typical open loop method is using straight time as the control. That's what you see on the front display. Time, frequency, current set, current delivered. When GO is enabled the unit starts counting down to zero.

In this application a PLC is reading the temperature of the workpiece. There are two non-contact temperature sensors that are looking at the workpiece. Once they both reach temp the PLC commands OFF to the EASYHEAT.

This is all a mute point because the "anomaly" is something the PLC cannot cause. That is, the delivered current dribbling off. The PLC cannot, in any way, command or adjust the power setting. Only ON/OFF.

I think the reason the unit worked with the PLC removed from the equation was because the unit was uncycled for the time it took them to change the connectors out. Same as my experience. It just started to happen then we ran out of workpieces. By the time we got more it refused to crapout again. Hence that reported 'condition' was only seen once and not repeatable.

The frequency is automatic. The unit just adjusts it each cycle continuously. Normal deviation is about +/- 1kHz during each cycle. The unit adjusts it to maximize power output. When the anomaly appeared the freq started deviating widely. By ~100kHz. This makes me suspect something in the "power head" or other output stuff. It's just physically hard to look at that stuff, plus I am not good a fixing things that are working correctly. I've asked them for some special workpieces that we can rapidly cycle thru the machine to stress it and hopefully make the anomaly show up on demand in minutes instead of hours and depending on workpiece availability.

Keith Cress
kcress -

http://www.flaminsystems.com