Many thanks all. I have the caps available and will source the TVS and try that.
@Richard - yes 12v 1amp cheapo power supplies from Amazon. Ironically I went with 12v as my simple logic assumed the 8200 would have to work less hard. On a bench supply the clock draws about 250mA. Can I ask which drivers you're using? I'm curious if the K155ID1 is a factor. My first ever clock using these had a different PSU and ran without issue. This is the first time I have combined the driver and 8200. I saw reference to a bleeder resistor (150K 1W) and 33uF 250v cap across the HV output on a DC-DC boost circuit. Could this afford additional protection to the PSU? Thanks On Monday, 16 February 2026 at 06:35:29 UTC gregebert wrote: > I admit I'm stubborn, and all of my nixie clocks use linear power > supplies, EXCEPT one, which was a nixie watch. Getting that booger to work > in a small volume was a lot of work and it never will run as reliably as an > old-school linear supply. The HV supply for my NIMO clock is also a > switcher, and has a ridiculous amount of control on it (yep, software can > specify how many kilovolts it wants to throw at the NIMO tubes, and can > check the temperature of the MOSFETs driving the converter). > > I've never had any sort of failure with the linear supplies, and I have > seen one DCDC converter fail (one of those direct-replacements for the > TO-220 LDO regulators). I really like how efficient they are (no heat > sinks!!), but they have not yet proven themselves to be as durable. The > latest rev of my main board for clock projects has A/D converters so I can > have software monitor the low-voltage power supplies (most of which are > derived from some sort of DCDC converter), and I'm planning to add > something so that the nixie supply can also be monitored. > > On Sunday, February 15, 2026 at 10:07:13 PM UTC-8 Jeff Walton wrote: > >> Protect your device by adding this to the input. Reduces noise, spikes >> and protects the NCH8200HV. The HV module is sensitive to spikes. >> Wall-wart power supplies are not all created equally. No failures since >> adding - had a couple nixie clock failures with the NCH8200HV before doing >> this. >> >> >> Jeff >> >> -------- Original message -------- >> From: 'Richard Scales' via neonixie-l <[email protected]> >> Date: 2/15/26 11:15 PM (GMT-06:00) >> To: neonixie-l <[email protected]> >> Subject: Re: [neonixie-l] HV Power Supply failues >> >> Hello, >> >> >> I had a number of NCH8200HV failures though specifically, this was when I >> was using a 12V power adapter, I have many more boards using a 5V supply >> and none of these have failed. >> >> I went straight to the designer who took a look at my board and came up >> with an answer. >> >> l *Root Cause:* Catastrophic MOSFET (NCH8200HV) failure observed across >> all modules. >> >> l *Power Delivery Concerns:* >> >> The 12V input bypasses critical bulk capacitance prior to distribution. >> >> Current PCB layout lacks immediate electrolytic buffer post power input >> >> l *Critical Design Recommendations* >> >> 1. *Capacitor Placement Protocol:* >> >> 12V rail must implement bulk electrolytic capacitor (≥470μF) within 10mm >> of input connector. >> Since then I have added a 470uf electrolytic as suggested and have had >> zero issues. >> >> What input voltage are you using: 12V, 5V, something else? >> >> - Richard >> >> >> >> On Sunday, 15 February 2026 at 11:58:25 UTC Max DN wrote: >> >>> Yes, I figured out electronics and windings of the transformer as I >>> wanted to embed it in my nixie clock design. Yes, the neon is just for fun, >>> as an indicator that HV is on. >>> >>> In my experience 8200 can take 30ma but not for long as the transformer >>> is very small and gets hot (I may be wrong). I think that even if you are >>> only using 13ma, the peak current (and in-rush current especially at start >>> up matters so much). If you have a way to check that, you’ll see. If you >>> don’t have an oscilloscope, try to use a power bench supply and limit the >>> current to, say 50mA. It’s a cheeky way to test peak current at start-up, >>> not accurate but gives you an idea. The power supply will enter into an >>> overcurrent mode, at least for a few seconds until the current stabilises >>> within the transformer. There are ways (hardware and software to minimise >>> in-rush current, but it’s not very simple and requires plenty of testing). >>> >>> >>> >>> On 15 Feb 2026, at 11:45, 'JBro63' via neonixie-l < >>> [email protected]> wrote: >>> >>> Thanks. They look similar in output to the 6300 but that would need a >>> new PCB :( >>> Curious what the neon is for? Power indicator? >>> >>> The 8200 is rated for 30ma - I measured 9ma on the prototype when it's >>> running 4 tubes, 13ma for 6 tubes so it should be within limits. >>> >>> On Sunday, 15 February 2026 at 11:19:34 UTC Max DN wrote: >>> >>>> You must have acceded peak current and the mosfet is gone. If they are >>>> both gone, no point repairing or replacing with same HVPS, it’ll happen >>>> again. >>>> >>>> If interested, I have a much more powerful power supply for sale here: >>>> High Voltage DC Power Supply Nixie Low Noise Audio Tubes 3.7-15V to >>>> 100-230V <https://www.ebay.co.uk/itm/276921910331> >>>> ebay.co.uk <https://www.ebay.co.uk/itm/276921910331> >>>> <https://www.ebay.co.uk/itm/276921910331> >>>> >>>> >>>> >>>> On 15 Feb 2026, at 10:54, 'JBro63' via neonixie-l < >>>> [email protected]> wrote: >>>> >>>> Hi all, looking for insight in to PSU problems. Have had two NCH8200s >>>> fail very close together after only a few weeks of use. >>>> >>>> In December, I finally got round to building four IN-12 4 tube clocks >>>> using a new PCB and some leftover tubes and drivers. They are driven using >>>> a K155ID1 for each tube (no multiplexing) and MJE340 for the INS-1 colons. >>>> Current limiting resistor is 18K. Each tube anode has a PC851 optocoupler >>>> to turn the tube off if no presence is detected. >>>> >>>> My previous IN-12, IN-14 & IN-18 clocks use the 8200 or 6300 and are >>>> still going strong. They are driven using HV5530s. >>>> >>>> Both 8200s have gone full short across the input. Anything about the >>>> K155ID1 / PC851 that is harsher on a HV supply? Anything I can check for >>>> with a scope? >>>> >>>> Thanks >>>> >>>> -- >>>> You received this message because you are subscribed to the Google >>>> Groups "neonixie-l" group. >>>> To unsubscribe from this group and stop receiving emails from it, send >>>> an email to [email protected]. >>>> To view this discussion, visit >>>> https://groups.google.com/d/msgid/neonixie-l/a93fbd9d-36aa-4791-bffa-54f7d2514cdfn%40googlegroups.com >>>> >>>> <https://groups.google.com/d/msgid/neonixie-l/a93fbd9d-36aa-4791-bffa-54f7d2514cdfn%40googlegroups.com?utm_medium=email&utm_source=footer> >>>> . >>>> >>>> >>>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "neonixie-l" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> >>> To view this discussion, visit >>> https://groups.google.com/d/msgid/neonixie-l/513d8a2a-cb3e-473e-a91f-45e6d50bb56cn%40googlegroups.com >>> >>> <https://groups.google.com/d/msgid/neonixie-l/513d8a2a-cb3e-473e-a91f-45e6d50bb56cn%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >>> >>> -- >> You received this message because you are subscribed to the Google Groups >> "neonixie-l" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To view this discussion, visit >> https://groups.google.com/d/msgid/neonixie-l/2fe0e94c-244f-494b-938c-db201d6fdf50n%40googlegroups.com >> >> <https://groups.google.com/d/msgid/neonixie-l/2fe0e94c-244f-494b-938c-db201d6fdf50n%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- You received this message because you are subscribed to the Google Groups "neonixie-l" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/d83a7de3-da83-4514-a170-2034c22e1c4cn%40googlegroups.com.
