Spellman DXM70N600X3547 power supplies on ebay -70kV 8.56MA

[Rich Feldman]

Received my PS, but haven't yet read enough of the manual to try powering up the unit. Found a connector in junk box that fits J2; might try the analog controls before the digital approach.

You know those four receptacle contacts in the HV connector? I verified that 2mm "bullet connectors" from the hobby store, which are stubby miniature banana plugs, plug in very nicely.

http://www.e-fliterc.com/Products/Defau ... =EFLAEC208 http://stores.aeromicro.com/ec2-device-connector-2/

Andrew, please tell Bern Bareis thanks from me for the PS tip. I hope we haven't seen the last of his posts at this forum, and that he didn't feel unwelcome here.

[edit]
Young William wrote: "The extent of my knowledge is that: I need this power source to transfer at the very minimum around 35kv to the vacuum MDC feedthrough. So what I need chords for are: firstly, from the plug on my wall to the power supply next from the power supply to a computer as that is how you control the output I believe and then finally from the power supply to the MDC feedthrough on the vacuum chamber."

Sorry, you have to show more progress on learning about electricity before getting help from me. It's not a thing to skip on the road to fusion, though a couple of neutron clubbers have gotten away with it.

[Jim Kovalchick]

Put in new line in fuses, and my unit now powers up and connects to the router. I can't test the HV yet because I'm staying in a corporate leased apartment and don't know the xfinity modem's user name. That keeps me from figuring out the supply address. I was thinking of guessing up to 255 times lol.

[Jim Kovalchick]

I've come to the conclusion that my unit is not obtaining an address from my router. It does not show up in network scans. The yellow light is lit on both my router and unit but green never flashes on the router. Only the unit. Any advice would be appreciated.

I suppose I can keep the unit on the chance that I can still set up control manually. Maybe more trouble for me than its worth.

[Jim Stead]

The manual suggests the IP of the supply is statically assigned. Assuming it hasn't been changed by the last owner, it should be 192.168.1.4

[Andrew Seltzman]

The yellow light staying on is normal behavior when it's not actively communicating with anything. The Ip should be 192.168.1.4 on all supplies by default. Try connecting to it with the matlab program or typing the Ip address into the web browser.

From the manual where it says: Ethernet Supported Operating Systems: Windows 98 2ED, Windows 2000 (SP2), Windows NT (SP6), Windows XP Professional, this refers to operation of the java applet over the web server only, ex the webserver won't launch the java control applet in win7 or win10, you should be able to connect to the supply in a web browser in all operating systems though and it should bring up the webserver page, just not launch the applet for all OS.

Ethernet communication works fine across all operating systems though, and matlab or any other TCP system should be able to control the supply.

[William Kelly]

Has there been any developments on the CA11 cable issue yet?
- William

[Liam David]

I have found the failure point in my power supply. A IRFBG30 mosfet blew out on the top board and came right off the board in pieces. I'll be ordering a replacement and will see what happens. I hope it didn't cause a chain of failures, or vise versa.

Location of the blowout




Removed mosfet

[Peter Schmelcher]

I bought 2 supplies, one works and one is dead (it even came with a toe tag). I’m considering fixing it and that is not something I undertake lightly. The circuit board is visibly burnt and some high current components are cooked. It’s a crapshoot without a schematic or phrased another way “Do you feel lucky… punk”. My impression is that it is a good modern DSP (Digital Signal Processor) design worth repairing -stay tuned.

Looking at the DXM brochure and manual I see the 300W model input voltage is 90 to 265VAC while the 600W and 1200W models input voltage is 180 to 264VAC. More input voltage equals more output power.

Our supply has a custom model suffix X3547 with nameplate input voltage of 115VAC and with output specifications implying 600W, so they are oddballs. Perhaps the 600W ratting at 115VAC is not continuous!

Given that some of these supplies have failed, caution might suggest that at 120VAC operation be restricted to 300W until more technical details are known.

-Peter

[Andrew Seltzman]

I was in contact with Cliff Scapellati at Spellman when I first got a supply of this model a year ago. The custom model X3547 is designed to operate at 115vac at full power continuously. Please post pictures of the damage to the board.

I believe it is possible to test the functionality of the digital control circuitry by applying 24Vdc to the power supply fan connector without plugging in the power supply to the AC mains. The fan connector is in parallel with the main 24v bus that gets regulated down for all the digital and control logic. This should in theory power up the control board, bypassing the AC power input circuitry that may be damaged. If the control board powers up, try to run the matlab program and post the output.

If the digital circuitry is unharmed it might be possible to debug the input power electronics and repair the supply, if the digital circuitry is bad, the supply is likely beyond repair.

[Rich Feldman]

Thanks for the 24 VDC jump-start tip, Andrew.

The club of DOA power supply owners includes me. Some observations and questions:

1. Top comes off w/ 9 identical screws. The small sheet metal box covering fuse and filter module is deluxe -- note welded corners.

2. Let's compare notes on fuses, in case the one I found blown is different from the type installed by Spellman. It says 10A 250 V 314. A Littelfuse 314-series fast blow ceramic body (powder filled) cartridge, like those in MWO's. I need to buy replacements, or use a plain glass 10 amp fuse before moving forward.

3. There are no obviously damaged components. Sniffing closely reveals no burned-electric-stuff smell. (The human nose is a valuable diagnostic tool.)

4. My unit's s/n is 101825126-A00370. Sticker also says REV:AW 10/13 (maybe that's the sticker rev). The remains of a hand-written annotation on the metal, near s/n sticker, says "installed 4/17/13".

5. This not-so-old model apparently isn't rare in junkyards. Here is a website with new units (rev L) for $81. Out of stock. https://www.nriparts.com/electrical/con ... wer-supply

[Jim Kovalchick]

I have been trying 192.168.1.4 since I first plugged the unit in. There was no response, that's why I was checking into whether my router was assigning an IP address dynamically. I see now that the manual implies that the unit must use a static IP. So, I think the failure mode possibilities are:

1. The static IP is not currently valid for my router and/or LAN. It's an Xfinity cable modem/router.
2. A previous user changed the IP. I believe this is possible since it would be required if more than one power supply was connected to the same LAN. The ominous part is that the manual suggests writing down changed IP's because there is no procedure for restoring the default.
3. There is an unspecified malfunction of the communications hardware in the unit.
4. Operator error.
5. Combination of more than one of items 1 through 4.

I believe item #1 is important to run to the ground. Based on the my limited understanding of the Xfinity documentation on the internet, it appears that you can use a static IP device directly connected to their modem but there is work I need to do in their 10.0.0.1 gateway to make the unit accessible to me on the LAN. Because they won't give me the username and password for the modem since it is paid for by my corporate account, the gateway is off limits. My options are to buy and install a router for a new LAN separate from the modem or to wait until my next trip back home where I already have a setup like that. The latter option is better for my wallet and therefore the best.
I'm not sure what to do in the case of failure mode leg #2.
For leg #3, I'm hopeful that the light status I'm seeing is a positive sign. The solid amber light indicates the connection is good and occasional green flashes show information is passing. I wish that I was also seeing green flashes on the router, but that could be an item #1 issue. When I scan the network with an app on my wireless connected phone, the unit immediately flashes for a few moments which I take as another good sign.
Leg #4 is always a possibility, but I can address that by spending more time with the manual.

Finally, I purchased a CAT5 crossover cable yesterday. I plan on attempting a direct connection to my laptop soon.

[Kuba Anglin]

I have also not been able to verify that my router identifies the power supply via direct ethernet connection. The IP address does not work. I logged into my Xfinity account at 10.0.0.1 and looked at all connected devices. The power supply was not listed anywhere. My supply does not indicate any signs of failure except this.

I also tried connecting to my computer through USB. This resulted in my computer stating that the USB device was malfunctioning. The manual of these machines was copyrighted in 2000, which makes me wonder how old these power supplies really are. Their software seems to operate on Visual Basic, the same language my grandfather used when learning computer science several decades ago. Something I want to try is installing a virtual machine on my computer that can run Windows XP, as the manual lists Windows XP as a compatible operating system. It is important to note that this virtual machine only works on Windows 7.

-Kuba

[Andrew Seltzman]

When I am connecting to the power supply my computer and the supply are both plugged into an ethernet switch that is separate from my router, the switch is also plugged into the router for internet/DHCP. Both supplies that I received were on 192.168.1.4, I might suspect the router if you are having problems connecting to the supply.

I would try the serial port, it's probably more straight forward then trying to get the USB to work.
default serial settings are 115.K 8N1
Serial commands start on page 94 of the manual PDF

You don't need to use visual basic, they are just using it as an example for code to communicate with the supply, matlab or any other software should also work

[Dan Knapp]

For those having problems with the digital interfaces, it might be easier to try analog control via J2. I'm still waiting for my unit to arrive. In the meantime, I sketched a circuit diagram for an analog controller (attached). I use a similar analog control box for another Spellman supply I have. This is a tentative circuit. I'll first try it without the LED indicators since I'm not sure how much extra current is available via pin 12 from the 15 volt supply when the enable input is jumpered. The manual implies that there is 15 ma available on pin 12, but it doesn't say how much the voltage enable connect (pin 11) draws. I'm guessing very little, but need to check it. I plan to pull the 10VDC reference voltage from one of the front panel pots (local filament limit control) by turning it full CW and using the pot wiper connection (pin 5) as the 10 volt supply. I don't know how much excess capacity the internal 10VDC supply has; I may have to disconnect the ground end of the front panel pot. for a really bare bones analog control to test the supply, one could just use a pot across a 9 volt battery to supply the control voltage to pin 3.
I haven't actually built the box yet in case my supply arrives DOA, in which case I plan to return it.

[Richard Hull]

I do not have one of these items and at the risk of butting in and looking the "outsider fool", I think Dan's post and PDF diagram, above, is the way to go. This is far too easy to hook up and check out via hard wired connection to determine whether any of you have a "pig in a poke" or the greatest thing since sliced bread. This hookup is very spellmanesque and has saved me hours in the few spellman supplies I have encountered. They do often offer a simple hardwired control! (very wise).

All of this computer hookup and control effort over the 6 pages of posts here has left me much bemused. Any HV supply that has a 132 pin micro-controller can become a real nightmare and effectively unrepairable at any level save for the spellman repair shop.

Hardwire the thing ala Dan's diagram!!...This should tell the tale. Even then, if the micro-controller is involved and needs a critcal signal that isn't there or is FUBAR, you may still be out of luck.

As a last resort, a very crafty work around might be possible if a genius in electronics can find a "bypass" to the micro-contoller. (all the best on that quest)

We are slowly creating a morass of new electronics that involves incredibly complex and multi-tenticled micro-controlled devices that are readily deemed absolutely unrepairable at any level! This is fine if the item is made super inexpensive by micro-controlled electronics......Just toss the worthless pig in the trash and buy a new one that may be cheaper, still. (toasters, coffee makers, cell phones, Flat panel TV's, etc.) However, a new $4,000 item that had been a classic 100 pound, click on a power switch and turn the dial power supply in 1958 and is now an 8 pound USB controlled, ethernet connectable, supply with 15 new safety inputs and interlocking custom signals to a micro-controller can easily become an albatross to the amateur.

To the professional user who spent the $4,000 for the new bells and whistles supply, a $1,248.15 repair will work fine and if hyper critical to his process, a "hot swap $4,000" spare pulled from his store room will be on hand while the original is off for the kilobuck repair. The amateur scientist is not so well handled here with his $200 bargain.

I would like to hear from the first guy in this "recent" mass purchase group who actually gets measured voltage pouring out of one of these gems. I hope every single one will work fine.

I am not talking through my hat here.....I am a retired electronics engineer and currently designing, making and selling micro-controlled electronics.

Richard Hull

[Dan Knapp]

I received my DXM supply today, but I fear I may also have a dead one. The fuses were good, the fan starts when plugged in, and the "power on" LED comes on. I can see it at 192.168.1.4 via the ethernet port and get the Spellman welcome page and software agreement acceptance, but the Java app will not run because "security settings blocked it from running." I need to figure out how to make a security exception to get the ethernet app to run.
I tried to control it via a DC control voltage into pin 3 of J2 (voltage program input) and pin 11 connected to pin 12 to enable, but I see no voltage on the HV output connector. I also don't get the "x-ray on" LED lighting up when I connect 11 and 12 to enable. I need to study the manual some more to make sure it's not operator error, but I fear this unit is DOA.
The seller has a 14 day DOA return policy, but the buyer pays for shipping both ways when executing this mode of return. These units were advertised as "fully functional" and therefore should come under the eBay guarantee where the buyer gets a full refund and the seller pays the return shipping. If I conclude that the unit is indeed dead, I will return it under the eBay warranty process. Since it appears several other peoples' units were received DOA, I'll let you know how things go with this.

[Jim Kovalchick]

I was able to make a direct connection to a laptop running Windows 10 using a crossover cable.

The applet starts to run but stops without loading the data. I am guessing this is a compatibility issue to newer versions of java and/or the browser.
I am happy to get this far and will probably wait to run this on an old pc at my home lab.

[Andrew Seltzman]

This custom version of the supply requires an additional HV enable signal on the 15 pin connector when running in local control mode(not over ethernet).

The high voltage interlock must first be enabled by connecting pins 11 and 12. In addition in local control mode(not required for Ethernet/RS232/USB) an HV enable signal must be given by connecting pins 15 and 12. If the interlock is opened in local mode then the enable signal must be re-asserted again(cycled open then closed) after the interlock is closed for operation.

The "x-ray on" light will not come on until some threshold voltage(not sure about actual number, but it's a few kV) is reached on the output.

The applet problem is due to the new versions of java, it won't work on any newer systems, you have to use the ethernet commands in the manual.

[Peter Schmelcher]

Andrew the 24V powering is an interesting diagnostic approach. I think I will fix the board first and then commission it using the 24V idea. Having a working supply my approach will be to compare the unpowered differences diode drops and resistances between various nodes in the effected area between the 2 supplies. Then powering up the good one I will scope similar nodes to better understand how the circuit failed.

The board was overheated at this via which is an easy fix with some copper wire and solder. Components visibly involved are K3, RV2, C149 and R218. Probably overstressed and also worth replacing include CR35 and CR34. Then the issue of finding the root cause of the failure immerges. About $70 of FETs are mounted on the internal heat sinks and if any were over stressed they should probably be replaced as a set. The supply is conservatively laid out and the RTV is a good indication of vibration testing for harsh industrial environments. Honestly it would be a pleasure to work on if I had a schematic. So it’s just finding the hours to do a methodical investigation/repair with a budget of $100 for parts.

My good unit works fine and the software hobbs meter reads 194 hours of on time (I don’t have matlab). I used a Fluke 8060A meter and Fluke 80k40 40kV probe. The probe easily fits into the CA11 socket and I tested the supply via Ethernet to 30kV in 100V steps while monitoring the actual output voltage with the probe. Wattage testing is for another day.

By chance did you inquire about a schematic?

-Peter

[Andrew Seltzman]

I have a schematic of the control board but have been asked not to publish or re-distribute it since it is a proprietary design of one of their flagship supplies. You could try contacting Cliff Scapellati at Spellman and see if he will send you a copy. In the meantime, if you send me a list of problems your supply is having I'll have a look and advise you what components to check.

I recommend the 24V test first since if the digital control system is fried, the power supply is likely damaged beyond repair, even with the schematic, and I might recommend returning it without trying the replace the power electronics. The 24v test should be very easy to accomplish and can be done without repairing the damage to the power input components.

R218 is a 0.05ohm, 1%, 5W resistor
K3 is a t90s1d42-24 POTTER&BRUMFIELD relay
RV2 is a CL70 NTC Thermistor
C149 is a 33uf(marked as 68uf in the default schematic but 33uf in my supply) 450V capacitor (Digi-Key Part Number 565-1423-ND )
CR35 is a GBPC3508W rectifier
CR34 is a MUR460 diode

RV2 is an inrush protection varistor, once the filter caps charge through the varistor, the relay K3, which is in parallel with RV2 should close, bypassing RV2. If this failed to happen, the full current of the supply was probably being drawn through the varistor causing it to burn. Try replacing the varistor and relay, in absence of any evidence of a short circuit or other damage this may get you up and running.

Could you send a picture of your full board showing damage to the other components?

[Rich Feldman]

Picked up fuses today, but didn't yet repeat the AC "power up" exercise.

Instead I tried the 24-VDC jump start trick, with gratifying success.
After separating fan from circuit board at the connector, ran fan and then circuit board from an adjustable bench power supply.
In the picture, PS common is connected to TP6 with an alligator clip. PS + is connected to fan header using a single receptacle contact.

When voltage was ramped up to 24V over a few seconds, a relay clicked and a bunch of LEDs came on, including several red front-panel FAULT indicators. I bet the ramp was too slow to trigger "power-on reset" behavior. Switching the PS off and back on did the trick: no LEDs on except front panel power indicator and one internal green LED. Current = 0.28 amps. All test points labeled with voltages (+3.3, +5, +10, +15, etc.) had the expected voltage values. One pin in bottom row of J2 (the DA-15 connector) had +15 V on it. So far so good.

Guess the next step is to reconnect the 120-VAC input module, with a fresh fuse, and plug it in to see what happens. Got no experience with troubleshooting any SMPS, much less a ZVS with PFC input. Could a current-limited startup using a series ballast (bypassable with a manual switch) do any harm?

[Andrew Seltzman]

I'd say try it without the ballast but with a lower amperage fuse, the power supply shouldn't draw much current to just power up the control circuits and would only need the rated fuse for running at full power. Perhaps start with a 2A or 4A quick blow.

If you plug in the Ethernet when 24V is applied, can you communicate with the supply?

[Stefan Gustafsson]

I was able to acquire one the supplies as well. To bad after shipping it over to Europe it was DOA. One of the fuses are bad and the VR2 thermistor is cooked.
It also looks like one the capacitors is broken. If I measure the resistance over the capacitors it's zero ohms.

I have been able to start the digital part up using a 24v bench supply and Ethernet commands works so I'm hopeful. Thank you very much for that information.
I rewrote the mathlab code into a C program if anyone is interested.

I will start removing broken components until I can find the short-circuit and try to rebuild the power parts.

The unit is marked with 115V but the product manual says 180-265V input.
Andrew, can you find any other changes to the selected components that would make it impossible for me to run my unit on 230V as here in Sweden?

[Jim Kovalchick]

I am not a coder and don't use MATLAB. Given that the GUI applet doesn't seem to be happy with modern versions of JAVA, I need to find a way to send TCP commands that works for a dummy like me. Surfing this morning I found a free utility called Packet Sender that it looks like it would work to send and receive data from a network based device like the DXM. I hope to try it soon, and I'll let the forum know how it turns out. I'm sure the trick will be in syntax.

[Richard Hull]

Could someone supply an image of the HV transformer block and the multiplier diode/capacitor area? Also where does the back of the HV connector go to. Is this thing a driver for a stack or an x-ray head?
Some of this stuff must be potted if in this rather smallish box!
Just curious.

Richard Hull