Fixing high voltage DC source for mass spectrometer

[sverl]

Hi all!

I’m chemist working with mass spectrometers (MS), instruments that weight ions really accurately. We have a ~30 year old MS where the electronics are not working quite right... I've done some troubleshooting, but I need help with the last details.

Mass spectrometers consist of an ion source and a detector. The ion source generates an ion beam at low pressure. The beam is accelerated and steered under high vacuum by static high voltage potentials into the detector. Unfortunately, I cannot see any ions in the detector and I'm sure the detector works because it has an internal ion source where I can detect ions. There are many different high voltages that control the path of the ions and among those are two pairs of voltages called XDFL and YDFL that deflect the ions in x- and y-direction to keep the beam in the right path. The software readout of the voltages are ~0 V no matter what I set the voltages to (-200 V to +200V). I have confirmed that the software readout matches measurements with multimeter. The XDFL voltage is also supposed to be pulsed about every second with an additional 18 V so the ion beam is deflected away from the detector when it is weighing the ions.

All of the high voltages are coming from one big PCB with 11 smaller identical PCBs mounted on it, connected with ribbon cables, see the first photo. My impression is that each of the smaller boards generate 3-5 V AC for the electronics underneath that control each of their own DC high voltages. I think it is done this way because some of the voltages float on the 3 kV that accelerate the ions. Coincidentally, the XDFL and YDFL voltages both come from underneath the same smaller PCB so it makes sense neither of them work correctly if something with the shared electronics is wrong. On the other hand, the mounted PCB appears to work correctly and, from what I can tell, the electronics underneath are separated so maybe there is the same fault on both parts of the circuit…


My understanding of the circuitry below, see second photo, is that a microprocessor controlled by software sends a digital signal of the desired voltage. This goes through an optocoupler to a DAC and the analog output is buffered by the OPAMP. There are also SSRs, voltage regulators and diodes, but it is not clear to me how they are connected and generate the +-200 V DC, but I assume some of it is generating 5 V DC for the logic from the 5 V AC from the transformers above.


When it comes to comparing to the parts that work (e.g. DPL4 voltages), when the voltage is set in software too 100 V, the OPAMP gets V+=8.8 V, V-=0 V, IN+=1.4 V, IN-=1.4 V and OUT=4.0 V, but for the faulty circuits V+=8.8 V, V-=0 V, IN+=1.4 V, IN-=8.2 V and OUT=0 V. IN- is connected to the AQW210EH SSR and K2700 voltage regulator. Does this indicate that the SSR or voltage regulator is toast? How can I test them?

I need to fix the circuitry and would like to know what can have gone wrong and how I can test what is wrong. I have access to oscilloscope and other equipment and can supply more photos or measurements if needed. Any more knowledgeable people that can help out a chemist?

The ICs are from left to right:
PS2501-4: https://www.renesas.com/us/en/document/dst/ps2501-1-ps2501-4-ps2501l-1-ps2501l-4-data-sheet
LTC1456: https://www.analog.com/media/en/technical-documentation/data-sheets/1456fs.pdf
OPA251: https://www.farnell.com/datasheets/81252.pdf
AQW210EH: https://www3.panasonic.biz/ac/e_download/control/relay/photomos/catalog/semi_eng_ge2a_aqw21_e.pdf
K2700: https://datasheetspdf.com/pdf-file/549187/ToshibaSemiconductor/K2700/1
78L05: https://www.ti.com/lit/gpn/UA78L
L78L09: https://www.st.com/resource/en/datasheet/l78l.pdf

[ace1903]

Equipment of that age in general has problem with capacitors. When SSR or high voltage transistor is broken that usually is accompanied with magic smoke and hole in the part body.
My advice is to replace capacitors in broken channel first. Also check for oxidation in connectors or on flat cables.
Has device some form of self check or some status report in software?

[sverl]

Thanks, I've already measured the capacitance of the caps on the smaller boards which are all good. Unfortunately, I'm unable to check the caps of the main board as I do not have access to the underside of the PCB, but they look fine and same goes for the SSR, transistors, voltage regulators and other components. There is a self check and status report which is supposed to report shorts etc. When I set the voltage much more or less than 0 V, it complains that the measured voltage differs from the set voltage. The manual of the unit just says that these kinds of errors are typical of a faulty PCB...

[ace1903]

IN+=1.4 V, IN-=8.2 V This looks like a problem. Usually inputs are very close to each other and amplifier works in linear regime.
One input is connected to output of the DAC through some resistor and other input is connected to output voltage through some resistive voltage divider.
I would guess that this IN- is connected through divider to output voltage.
When device is off can you check resistance from IN- to ground and resistance to output of that channel? It should be some ratio of 20:1.

[sverl]

Yeah, I also though that was weird, but I don't have much experience reading data sheets and I don't know how it is wired up either so I was not sure if it was "in range of amplification". I will update tomorrow with resistances.

[fzabkar]

I would trace the circuit around the op amp and SSR. We need to understand what is going on here. I suspect that both channels are not being switched on, in which case I'd be looking for a common enable signal from somewhere.

[sverl]

So as ace1903 suggested, I have measured the resistances between IN- and ground (15k) and OUT (3.3M) for the good channel (DPL4), on the bad channels (XDFL, YDFL), they're 6k and 3.3M. The two bad channels have some different resistors. However, as far as I can tell the circuit is wired differently than suggested.

Following fzabkar's suggestion, I tried tracing the circuit. This is what I have so far. It's quite tedious as I don't have access to the underside of the board... I can not find any feedback loop for the OPAMP. If you have suggestions for how things might be connected, I will try to check it and complete the diagram!

[ace1903]

Any chance for pictures with better quality of the channel at the corner of the board?
Also better picture of the resistors on the right side of the connector?
With a multimeter on continuity, you can check the gate of the transistor to which resistor is connected.
SSRs connection as connected on the schematic is wrong I think.
When you say the channel is bad is that means that the voltage is off by some percent or totaly missing or always 200V?
Can you post pictures of the boar with transformers?

[sverl]

Hope these photos show what you need!

I have revised the schematic, hopefully it is more correct now. I didn't have time to check the SSR connections, but will do tomorrow.

The bad channels show no voltage.

I can also add that the voltages on the ribbon cable are (roughly symmetric):
1 5.2 V AC
2 NC
3 2.8 V AC
4 3.0 V AC
5 NC
6 5.2 V AC
7 NC
------
8 NC
9 4.4 V AC
10 NC
11 3.3 V AC
12 3.4 V AC
13 NC
14 4.4 V AC

[fzabkar]

The op amp U31 is configured as a high current voltage follower, with Q8 being the pass transistor. If the source pin of Q8 is at 8V and its gate is at 0V, then either Q8 has a D-S short, or something is happening at the XDFL/YDFL outputs.

I guess the op amp may be faulty, but its input and output voltages are not inconsistent.

[james_s]

Measuring capacitance only tells you part of the story, the usual failure mode for electrolytics is rising ESR.

[ace1903]

@fzabkar Could you please elaborate more on which high current voltage follower circuit you think.
What bothers me is that 90M resistor that makes no sense on output of the amplifier.
I have idea that SSRs are only for output polarity switching. Do you agree?
Short between D-S is plausible but in my experience it is always accompanied with hole in such high voltage transistor.
Short test with multimeter on resistance can make thing clear. I would also check resistance between amplifier output and ground.

Capacitors ESR can not make output zero volts, must be something else.
Transformer look as regular 50hz one with gretz rectifier and smoothing capacitor. Large ESR can make more noise but no dead output.

[fzabkar]

The usual configuration for a voltage follower would directly connect pin #6 of U31 to pin #2. However, in this circuit the output current capability of U31 is insufficient to drive the load, so a pass transistor (Q8) is inserted in the loop.

Could R90 provide a minimum load to prevent instability in the op amp?

The rating of S15 is 120mA max, so I would think that Q8 would never dissipate more than 1W (8V x 120mA).

What bothers me is that there are two identical faults on two different channels with no apparent common factor.

I have no idea what DPL4-1 and DPL4-2 are actually controlling.

Edit:

If the op amp has a very high input impedance, and if S15 is switched off, would the D-S leakage current of the FET be sufficient to make it look like there was a D-S short? Perhaps it would be a good idea to check the voltages across the LEDs of the SSRs to see which, if any, of them are being switched on?

[sverl]

@fzabkar For the MOSFET, I measure 17M between drain and source for both good and bad channels. Nothing is (or should be) happening at the XDFL/YDFL or DPL4-1/DPL4-2 output, at the moment the contacts are disconnected and when they are connected, they should only be connected to plates in the MS. DPL4-1/DPL4-2 should be more or less identical to XDFL-1/XDFL-2 and YDFL-1/YDFL-2 just being plates for steering ions in the MS. The only difference I know is that the XDFL and YDFL voltages will be superimposed on +-300 kV DC for accelerating the ions. However I measure 0 V on XDFL and YDFL no matter what the high voltage is set to. The high voltage is also reported correctly in software.

@ace1903 Resistance for R90 was wrong, it's 15.6k, not 12.6M. (I read it backwards...) I had also traced it wrong between the MOSFET and OP AMP, hopefully it is correct now. It would physically/chemically make sense if the SSR only are for polarity switching since more or less all voltages going to the mass spectrometer needs to change polarity when you change the polarity of the ions you are measuring. I tried tracing the SSR circuit again and got the same result. I measured the resistance between OP AMP OUT and GND is 3M.

Would it be a good idea to try swapping the OP AMPS on the good and bad channels? It is quite a lot of work though...

[ace1903]

In general using broken components on channel that are working can cause new problems.
I would try to use set of SSRs and amplifier from working channel on non working one and see what will happen.

[sverl]

So I replaced the OP AMPs on the faulty channels with OP AMPs from good channels and it works now! Hurray, thanks for the help everyone!

However, it appears the OPA251PA (Burr-Brown) is no longer produced or in stock at major suppliers... Are there any equivalent ICs in 8-pin DIP that I can use that are in stock?

EDIT: Looks like the SOIC-8 package (OPA251UA) is available so I guess I can use something like SparkFun's SOIC to DIP Adapter (https://www.sparkfun.com/products/13655), right?

[ace1903]

Glad to hear that you have found the culprit. Think that soic version plus adapter will work.
There are pins called "Swiss Machine Pin male" that you can use with that adapter board to be able to insert adapter board into existing socket.