I thought I would write an update to this story. As Jonathan mentions above, we have been having trouble with these triplers. In fact we have fitted 13 of them and so far 10 have failed. Most have failed within a very short time, as you can see in the graph below.So far, we have identified two failure mechanisms:We had one tripler that started arcing immediately, here’s a video:
https://youtube.com/shorts/LpFF60f1RZk?feature=share
The arcing you can see is from the tripler to the adjacent earth wire. On other triplers, we had noticed arcing from the tripler to a ‘via’ on the PCB that was just underneath it. To avoid the PCB problem, on this particular tripler, we added a 7mm nylon pad to insulate it from the PCB. Clearly the earth wire was too close.
We then tried routing the earth wire away from the tripler. We first tried it with the original tripler with the arc damage. This arced to the PCB, through the old earth wire hole in the nylon pad. This is a jump of about 10mm – with a bang:
https://youtube.com/shorts/lCHHqtY2lyU?feature=shareThe tripler is made from only a few components, but they are buried in potting compound. I had a suspicion that the high voltage wire that carries 24,000 volts was too close to the casing. We managed to confirm this by carefully cutting the bottom off. You can see part of the wire – and with a meter we proved it was definitely the high voltage output:This means that the 24,000 volt wire only has 1mm of insulation, made from the tripler’s outer casing. We have a suspicion that more reliable triplers have this wire closer to the middle of the tripler, allowing the potting material to provide more insulation. With many plastics having a dielectric strength of about 10kV per mm, this makes this failure mode plausible.
We’ve now fitted another new tripler to this monitor, placed it on an insulating pad and carefully routed the earth wire away from the critical area. It has been working just less than a week. Let’s cross our fingers that it will last.
But, as the graph shows, it is likely that not all faults are due to arcing. Testing of individual components in the tripler is difficult as there is no way to probe each item in the potting compound. The diodes, being very high voltage types, do not have a normal 0.7V drop across them either. I set up a test circuit using a 40V power supply and a 110K resistor to give a slight load (note these are not high current devices). I was able to determine that 35 +/- 0.2 volts was dropped by the chain of five diodes. This was consistent across a selection of new triplers.
A failed tripler, with no apparent arcing damage, gave a voltage drop of 32.2V. This suggests to me that a diode had partly gone short-circuit or a capacitor had arced internally and made itself into a resistor. Of course, the full tripler chain, when tested at a reverse voltage of 1000V, still blocked OK.
I had hoped that the insulating pad would allow us to make these substandard triplers work. But the data suggests this is not the only fault. Unfortunately we have a lot of new triplers in stock, so we want to explore all options before deciding they aren’t worth it. We are going to do some more trials with insulated triplers and see if these are more reliable.
It is interesting to note that the triplers that have lasted the longest time were the ones we fitted first in early 2023. We have a suspicion that these are from a different batch, possibly from a different production line, so have been better built. The bad ones have all been coloured green. Other (better) batches may be a cream colour.Always think about safety when working on monitors!
Tom
https://youtube.com/shorts/LpFF60f1RZk?feature=share
The arcing you can see is from the tripler to the adjacent earth wire. On other triplers, we had noticed arcing from the tripler to a ‘via’ on the PCB that was just underneath it. To avoid the PCB problem, on this particular tripler, we added a 7mm nylon pad to insulate it from the PCB. Clearly the earth wire was too close.
We then tried routing the earth wire away from the tripler. We first tried it with the original tripler with the arc damage. This arced to the PCB, through the old earth wire hole in the nylon pad. This is a jump of about 10mm – with a bang:
https://youtube.com/shorts/lCHHqtY2lyU?feature=share
We’ve now fitted another new tripler to this monitor, placed it on an insulating pad and carefully routed the earth wire away from the critical area. It has been working just less than a week. Let’s cross our fingers that it will last.
But, as the graph shows, it is likely that not all faults are due to arcing. Testing of individual components in the tripler is difficult as there is no way to probe each item in the potting compound. The diodes, being very high voltage types, do not have a normal 0.7V drop across them either. I set up a test circuit using a 40V power supply and a 110K resistor to give a slight load (note these are not high current devices). I was able to determine that 35 +/- 0.2 volts was dropped by the chain of five diodes. This was consistent across a selection of new triplers.
A failed tripler, with no apparent arcing damage, gave a voltage drop of 32.2V. This suggests to me that a diode had partly gone short-circuit or a capacitor had arced internally and made itself into a resistor. Of course, the full tripler chain, when tested at a reverse voltage of 1000V, still blocked OK.
I had hoped that the insulating pad would allow us to make these substandard triplers work. But the data suggests this is not the only fault. Unfortunately we have a lot of new triplers in stock, so we want to explore all options before deciding they aren’t worth it. We are going to do some more trials with insulated triplers and see if these are more reliable.
It is interesting to note that the triplers that have lasted the longest time were the ones we fitted first in early 2023. We have a suspicion that these are from a different batch, possibly from a different production line, so have been better built. The bad ones have all been coloured green. Other (better) batches may be a cream colour.
Tom
Statistics: Posted by Issue7 — Fri Feb 16, 2024 11:36 pm