EEVblog Electronics Community Forum
EEVblog => EEVblog Specific => Topic started by: EEVblog on March 19, 2021, 10:15:24 pm
-
What's inside a 10kW linear power supply for a 10W Argon Ion Laser? Teardown time!
Daryl Tewksbury: https://theamphour.com/521-outdoor-laser-projection-object-mapping-with-daryl-tewksbury/ (https://theamphour.com/521-outdoor-laser-projection-object-mapping-with-daryl-tewksbury/)
https://www.youtube.com/watch?v=N8hz3MGS01E (https://www.youtube.com/watch?v=N8hz3MGS01E)
-
The "diode bridges" near the main rectifier diodes may very well be current shunts.
Depending on the Laser model there the control board may have to do more control to the laser. There can be some movable optics to selct a wavelength. Those Ar-ion laser can use a few different lines and one may select one or all.
The large part on top of the filter box looks like a contactor - a kind of high power relay - kind of need such a thing to turn things off.
The filter inductor is surprisingly small - I had expected a larger chunk of iron.
If the supply is made for 208 V, how do they run it with a 240 V mains grid ?
-
I'd like to get my hands on that pass transistor PCB/module, looks easy enough to convert into an 200V rated DC Electronic Load.
-
Going to bet those transistor base and emitter leads each are into a PCB mounted socket, probably a solder pin from a socket, used so changing a single transistor is only undo 2 screws and pull it out, then solder in a new Littlefuse to get the new one operational again.
The dual SCR is used with the LC filter to regulate the main DC bus voltage, using the conduction angle to control the voltage. Rough pre regulation before the pass module to reduce power dissipation, more because the dissipation without would need a bigger case, and another pass module with 2 entire dies of Motorola silicon, combined with a bigger external water tank and radiator, than any power saving. Going to bet the coolant circuit used a 50% propylene glycol and distilled water blend, along with a little blend of extra modifiers like sodium silicate and borax to buffer the pH to close to 7.00, and then there was another water to water heat exchanger that used a mini Sulzer tower to handle the actual cooling.
The white crud on all the metal is aluminium and zinc oxide, as this was used in a humid environment, with a lot of dissimilar metal contact, so yes the aluminium zinc alloy would corrode, wherever it was in contact with pure aluminium from the cover, as it would have a permanent electrochemical cell there, and the tiny gap would hold any condensed moisture in contact. not helped by the massive brass block either, which likely is driving all the reactions, seeing how shiny it still is except at the base where it contacts bare aluminium.
I doubt they expected it to last 30 years....
-
Transistor sockes are indeed likely, when I saw it I was wonderin where they got transistors with so long pins :-DD :palm:.
The output for the cooling water is labeled "drain" so chances are this could be run from plain tap water in a once through mode. In areas where water is expensive they may also use a closed cycle, possibly with some addition. Likely not glycol, as this is more making corrosion worse. It mainly avoids freezing.
-
I didn't see the HV igniter circuit anywhere, you need many kV with a good wallop (because of the water's capacitance) on the tube.
-
I didn't see the HV igniter circuit anywhere, you need many kV with a good wallop (because of the water's capacitance) on the tube.
The igniter will be in the head right at the tube. They use a transformer to couple the ignition pulse while allowing the DC current to pass through.
-
The large part on top of the filter box looks like a contactor...
If the supply is made for 208 V, how do they run it with a 240 V mains grid ?
The part shown at the 29 minute mark, what Dave refers to as a circuit breaker, is definitely a contactor.
The large magnetic component just near the incoming mains is very likely a step down transformer or perhaps an auto-transformer to get the 415 V three phase down to the 208 three phase required. Northlake Engineering certainly made transformers.
-
It would be nice if Dave to upload pictures on www.flickr.com/photos/eevblog/ (http://www.flickr.com/photos/eevblog/)
At 18:57 two transformers can be seen and a couple of optocouplers grouped in two separate channels.
On the processor board, there are several Analog Devices chips. Interesting to see is if they are ADCs only or ADC +DACs.
Think that LM324 are only to sum current over all transistors not opam per single transistor to do local current regulation.
-
Juts to sum up the currents one would not need 1 OP per transistor. So it makes absolute sense to hve the OPs as current regulators. The transistors also need quite some base current - so the OPs also act as drivers to provode that current. Still with a current gain of usually less than 100 this could be relatively low current. So I wonder if there are some extra small transistors hidden somewhere.
With the tranformer near the input, I am with Dave: this is likely the transformer for the hot kathode.
A step down transformer from 400 to 208 V would need to be way larger.
-
With the tranformer near the input, I am with Dave: this is likely the transformer for the hot kathode.
A step down transformer from 400 to 208 V would need to be way larger.
You are of course correct. I am an idiot :) A 10 kW power supply would need a far, far larger transformer.
-
Imagine just how much smaller this could be with modern 1200V SiC FETs. I would venture a guess that 4 Wolfspeed gen 3 FETs would do the trick. The output filter would still need to be similarly sized but the power electronics could be so, so much smaller. Absolutely incredible to see how far we've come.
-
Imagine just how much smaller this could be with modern 1200V SiC FETs. I would venture a guess that 4 Wolfspeed gen 3 FETs would do the trick. The output filter would still need to be similarly sized but the power electronics could be so, so much smaller. Absolutely incredible to see how far we've come.
Would it? I thought BJTs were still king when it comes to linear applications. Of course a modern ion laser PSU would probably be a SMPS but back in the day linear was how these were done. I always liked ion lasers, big brutal things, it's still weird to me that it's possible to get multiple WATTS CW out of a single tiny laser diode these days but even so the modern diodes can't touch the beam quality or the multi-line ability of an ion laser.
-
Would it? I thought BJTs were still king when it comes to linear applications. Of course a modern ion laser PSU would probably be a SMPS but back in the day linear was how these were done. I always liked ion lasers, big brutal things, it's still weird to me that it's possible to get multiple WATTS CW out of a single tiny laser diode these days but even so the modern diodes can't touch the beam quality or the multi-line ability of an ion laser.
Oh yes, I think you're probably correct there. It could probably be done smaller with IGBTs if it needed to be linear, but at the time this laser was built, I would venture a guess that IGBT tech was not particularly great. Would this need to be a linear supply though? Considering that it's CW I would think SMPS should be fine, and then the SiC FETs would be IMO ideal.
-
IGBTs are usually rather poor for linear operation. At least nearly all have a really poor SOA.
Also don't get fooled with the power rating of some modern FETs - 200 W for a TO220 case is not really working in real life.
BJTs are still a relatively good choice if the voltage to drop is below some 60 V. At high voltage a electron tube may be an alternative - I have seen such a thing in a 10 kV - 1 A linear supply.
It also helps to split the power over multiple devices, this makes the cooling easier.
-
IGBTs are usually rather poor for linear operation. At least nearly all have a really poor SOA.
Also don't get fooled with the power rating of some modern FETs - 200 W for a TO220 case is not really working in real life.
BJTs are still a relatively good choice if the voltage to drop is below some 60 V. At high voltage a electron tube may be an alternative - I have seen such a thing in a 10 kV - 1 A linear supply.
It also helps to split the power over multiple devices, this makes the cooling easier.
You can push a good TO220 MOSFET to quite a lot of power if you keep the temperature below the thermal runaway region. BJTs are not that great to be honest. In theory they could be OKfor this sort of application, but their thermal resistance is much higher than modern FETs. I don;t thing there is any practical reason for this, it's just there isn't a demand for it so they dont bring the same production processes to BJTs.
There are some FETs with better than 0.5 K/W junction to case. And that's what you need for such an application. Lots of die area and the smallest thermal resistance. TO247 might be even better, and those SOT227 packages are the beasts.
-
I didn't see the HV igniter circuit anywhere, you need many kV with a good wallop (because of the water's capacitance) on the tube.
The igniter will be in the head right at the tube. They use a transformer to couple the ignition pulse while allowing the DC current to pass through.
Yes - the igniter is always in the head, as the fast risetime (usually from a sparkgap) would never make it down a cable.
This is an example of an igniter for this scale of tube - large inductor on top passes the DC while allowing the ignition pulse to get to the tube, you can just see the sparkgap on the left
(http://electricstuff.co.uk/fbiglas4.jpg)